diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.42345966546686425,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8464024662971497,
+      "learning_rate": 4e-05,
+      "loss": 0.3488,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5379390716552734,
+      "learning_rate": 8e-05,
+      "loss": 0.202,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6969972252845764,
+      "learning_rate": 0.00012,
+      "loss": 0.312,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.45794710516929626,
+      "learning_rate": 0.00016,
+      "loss": 0.2298,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.44517195224761963,
+      "learning_rate": 0.0002,
+      "loss": 0.1948,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.361826092004776,
+      "learning_rate": 0.00019979899497487438,
+      "loss": 0.2009,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.631228506565094,
+      "learning_rate": 0.00019959798994974876,
+      "loss": 0.2959,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6042431592941284,
+      "learning_rate": 0.00019939698492462313,
+      "loss": 0.2594,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6731159090995789,
+      "learning_rate": 0.0001991959798994975,
+      "loss": 0.3103,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.4393431544303894,
+      "learning_rate": 0.00019899497487437187,
+      "loss": 0.1662,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.47421371936798096,
+      "learning_rate": 0.00019879396984924622,
+      "loss": 0.2106,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5907872319221497,
+      "learning_rate": 0.00019859296482412062,
+      "loss": 0.351,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5357276201248169,
+      "learning_rate": 0.000198391959798995,
+      "loss": 0.2511,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8222237825393677,
+      "learning_rate": 0.00019819095477386937,
+      "loss": 0.3362,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4601736068725586,
+      "learning_rate": 0.0001979899497487437,
+      "loss": 0.2097,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5627934336662292,
+      "learning_rate": 0.0001977889447236181,
+      "loss": 0.2663,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6441662907600403,
+      "learning_rate": 0.00019758793969849249,
+      "loss": 0.2665,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5793730020523071,
+      "learning_rate": 0.00019738693467336683,
+      "loss": 0.2452,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5800638794898987,
+      "learning_rate": 0.0001971859296482412,
+      "loss": 0.2573,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6255142092704773,
+      "learning_rate": 0.0001969849246231156,
+      "loss": 0.3046,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5359498858451843,
+      "learning_rate": 0.00019678391959798995,
+      "loss": 0.2445,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5056596398353577,
+      "learning_rate": 0.00019658291457286432,
+      "loss": 0.2487,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7259389162063599,
+      "learning_rate": 0.0001963819095477387,
+      "loss": 0.3341,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5617966055870056,
+      "learning_rate": 0.0001961809045226131,
+      "loss": 0.2495,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5566623210906982,
+      "learning_rate": 0.00019597989949748744,
+      "loss": 0.2732,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4777694344520569,
+      "learning_rate": 0.00019577889447236181,
+      "loss": 0.2614,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4215550422668457,
+      "learning_rate": 0.0001955778894472362,
+      "loss": 0.2205,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5983753204345703,
+      "learning_rate": 0.00019537688442211056,
+      "loss": 0.3514,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5450871586799622,
+      "learning_rate": 0.00019517587939698493,
+      "loss": 0.2667,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.488443523645401,
+      "learning_rate": 0.0001949748743718593,
+      "loss": 0.2691,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5746788382530212,
+      "learning_rate": 0.00019477386934673368,
+      "loss": 0.3224,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.517463207244873,
+      "learning_rate": 0.00019457286432160805,
+      "loss": 0.2269,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.3751838803291321,
+      "learning_rate": 0.00019437185929648243,
+      "loss": 0.1812,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6401920318603516,
+      "learning_rate": 0.0001941708542713568,
+      "loss": 0.224,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8179358243942261,
+      "learning_rate": 0.00019396984924623117,
+      "loss": 0.2426,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5015102028846741,
+      "learning_rate": 0.00019376884422110552,
+      "loss": 0.2673,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.558474063873291,
+      "learning_rate": 0.00019356783919597992,
+      "loss": 0.2309,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.48645198345184326,
+      "learning_rate": 0.0001933668341708543,
+      "loss": 0.2097,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6228073835372925,
+      "learning_rate": 0.00019316582914572864,
+      "loss": 0.223,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7079476714134216,
+      "learning_rate": 0.000192964824120603,
+      "loss": 0.358,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.47177523374557495,
+      "learning_rate": 0.0001927638190954774,
+      "loss": 0.1826,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5007954835891724,
+      "learning_rate": 0.00019256281407035178,
+      "loss": 0.235,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6503958702087402,
+      "learning_rate": 0.00019236180904522613,
+      "loss": 0.2588,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5159966349601746,
+      "learning_rate": 0.0001921608040201005,
+      "loss": 0.2266,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6967421770095825,
+      "learning_rate": 0.0001919597989949749,
+      "loss": 0.2165,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8722981810569763,
+      "learning_rate": 0.00019175879396984925,
+      "loss": 0.3716,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.45119017362594604,
+      "learning_rate": 0.00019155778894472362,
+      "loss": 0.2343,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.49875113368034363,
+      "learning_rate": 0.000191356783919598,
+      "loss": 0.1764,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6528317332267761,
+      "learning_rate": 0.0001911557788944724,
+      "loss": 0.2664,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5773115754127502,
+      "learning_rate": 0.00019095477386934674,
+      "loss": 0.2688,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6044211387634277,
+      "learning_rate": 0.0001907537688442211,
+      "loss": 0.2723,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5186501741409302,
+      "learning_rate": 0.00019055276381909548,
+      "loss": 0.2437,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5406463742256165,
+      "learning_rate": 0.00019035175879396986,
+      "loss": 0.2152,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6606569290161133,
+      "learning_rate": 0.00019015075376884423,
+      "loss": 0.25,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.45699411630630493,
+      "learning_rate": 0.0001899497487437186,
+      "loss": 0.1749,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5122879147529602,
+      "learning_rate": 0.00018974874371859298,
+      "loss": 0.2857,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5184636116027832,
+      "learning_rate": 0.00018954773869346732,
+      "loss": 0.2414,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.40956321358680725,
+      "learning_rate": 0.00018934673366834172,
+      "loss": 0.177,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.38545408844947815,
+      "learning_rate": 0.0001891457286432161,
+      "loss": 0.1856,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.575831949710846,
+      "learning_rate": 0.00018894472361809047,
+      "loss": 0.2726,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6056025624275208,
+      "learning_rate": 0.00018874371859296481,
+      "loss": 0.161,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.43295279145240784,
+      "learning_rate": 0.00018854271356783921,
+      "loss": 0.1989,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6881011724472046,
+      "learning_rate": 0.0001883417085427136,
+      "loss": 0.3154,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.548842191696167,
+      "learning_rate": 0.00018814070351758793,
+      "loss": 0.2496,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7280535697937012,
+      "learning_rate": 0.0001879396984924623,
+      "loss": 0.2954,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4259982407093048,
+      "learning_rate": 0.0001877386934673367,
+      "loss": 0.2039,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5318437218666077,
+      "learning_rate": 0.00018753768844221108,
+      "loss": 0.2795,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4900606572628021,
+      "learning_rate": 0.00018733668341708543,
+      "loss": 0.2276,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.72746342420578,
+      "learning_rate": 0.0001871356783919598,
+      "loss": 0.3113,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48556947708129883,
+      "learning_rate": 0.0001869346733668342,
+      "loss": 0.2374,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5738449692726135,
+      "learning_rate": 0.00018673366834170854,
+      "loss": 0.313,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44591739773750305,
+      "learning_rate": 0.00018653266331658292,
+      "loss": 0.2157,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.578846275806427,
+      "learning_rate": 0.0001863316582914573,
+      "loss": 0.3421,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.37167009711265564,
+      "learning_rate": 0.0001861306532663317,
+      "loss": 0.2151,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3555445671081543,
+      "learning_rate": 0.00018592964824120604,
+      "loss": 0.1793,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7459139227867126,
+      "learning_rate": 0.0001857286432160804,
+      "loss": 0.4165,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5642420649528503,
+      "learning_rate": 0.00018552763819095478,
+      "loss": 0.2745,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7679473161697388,
+      "learning_rate": 0.00018532663316582915,
+      "loss": 0.323,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.2970917820930481,
+      "learning_rate": 0.00018512562814070353,
+      "loss": 0.1476,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.2861027121543884,
+      "learning_rate": 0.0001849246231155779,
+      "loss": 0.1457,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7647842168807983,
+      "learning_rate": 0.00018472361809045227,
+      "loss": 0.3267,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.598876416683197,
+      "learning_rate": 0.00018452261306532662,
+      "loss": 0.2377,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46886759996414185,
+      "learning_rate": 0.00018432160804020102,
+      "loss": 0.2444,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6426663398742676,
+      "learning_rate": 0.0001841206030150754,
+      "loss": 0.2771,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.593606173992157,
+      "learning_rate": 0.00018391959798994977,
+      "loss": 0.2985,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5843706130981445,
+      "learning_rate": 0.0001837185929648241,
+      "loss": 0.2548,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5532352328300476,
+      "learning_rate": 0.0001835175879396985,
+      "loss": 0.2587,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6341427564620972,
+      "learning_rate": 0.00018331658291457288,
+      "loss": 0.3246,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6387240886688232,
+      "learning_rate": 0.00018311557788944723,
+      "loss": 0.3389,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4927945137023926,
+      "learning_rate": 0.0001829145728643216,
+      "loss": 0.2141,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.552358090877533,
+      "learning_rate": 0.000182713567839196,
+      "loss": 0.3409,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5346209406852722,
+      "learning_rate": 0.00018251256281407038,
+      "loss": 0.236,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5484887957572937,
+      "learning_rate": 0.00018231155778894472,
+      "loss": 0.2121,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45301929116249084,
+      "learning_rate": 0.0001821105527638191,
+      "loss": 0.196,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5568594336509705,
+      "learning_rate": 0.0001819095477386935,
+      "loss": 0.2479,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5509335994720459,
+      "learning_rate": 0.00018170854271356784,
+      "loss": 0.2424,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4819609820842743,
+      "learning_rate": 0.00018150753768844221,
+      "loss": 0.2526,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4724488854408264,
+      "learning_rate": 0.0001813065326633166,
+      "loss": 0.2192,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.554046094417572,
+      "learning_rate": 0.00018110552763819096,
+      "loss": 0.3003,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42843714356422424,
+      "learning_rate": 0.00018090452261306533,
+      "loss": 0.1914,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4597126543521881,
+      "learning_rate": 0.0001807035175879397,
+      "loss": 0.2409,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6792829036712646,
+      "learning_rate": 0.00018050251256281408,
+      "loss": 0.2929,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6163105368614197,
+      "learning_rate": 0.00018030150753768845,
+      "loss": 0.2995,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5015051960945129,
+      "learning_rate": 0.00018010050251256282,
+      "loss": 0.2628,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5530080795288086,
+      "learning_rate": 0.0001798994974874372,
+      "loss": 0.2616,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5682195425033569,
+      "learning_rate": 0.00017969849246231157,
+      "loss": 0.2844,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5251821875572205,
+      "learning_rate": 0.00017949748743718592,
+      "loss": 0.2502,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6445398330688477,
+      "learning_rate": 0.00017929648241206032,
+      "loss": 0.2709,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3485439419746399,
+      "learning_rate": 0.0001790954773869347,
+      "loss": 0.15,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4932727813720703,
+      "learning_rate": 0.00017889447236180906,
+      "loss": 0.2608,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6365306377410889,
+      "learning_rate": 0.0001786934673366834,
+      "loss": 0.212,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6077302694320679,
+      "learning_rate": 0.0001784924623115578,
+      "loss": 0.2978,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5916979908943176,
+      "learning_rate": 0.00017829145728643218,
+      "loss": 0.2934,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.634833037853241,
+      "learning_rate": 0.00017809045226130653,
+      "loss": 0.2495,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49833521246910095,
+      "learning_rate": 0.0001778894472361809,
+      "loss": 0.2329,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5283430218696594,
+      "learning_rate": 0.0001776884422110553,
+      "loss": 0.2317,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45649638772010803,
+      "learning_rate": 0.00017748743718592967,
+      "loss": 0.2078,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4763767123222351,
+      "learning_rate": 0.00017728643216080402,
+      "loss": 0.2394,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42383307218551636,
+      "learning_rate": 0.0001770854271356784,
+      "loss": 0.2294,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46265044808387756,
+      "learning_rate": 0.0001768844221105528,
+      "loss": 0.2321,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6166386604309082,
+      "learning_rate": 0.00017668341708542714,
+      "loss": 0.2763,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4315739870071411,
+      "learning_rate": 0.0001764824120603015,
+      "loss": 0.235,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5360255241394043,
+      "learning_rate": 0.00017628140703517588,
+      "loss": 0.2765,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.24499636888504028,
+      "learning_rate": 0.00017608040201005026,
+      "loss": 0.1312,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6798754930496216,
+      "learning_rate": 0.00017587939698492463,
+      "loss": 0.2664,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3721320331096649,
+      "learning_rate": 0.000175678391959799,
+      "loss": 0.1916,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46311238408088684,
+      "learning_rate": 0.00017547738693467338,
+      "loss": 0.2669,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6015424728393555,
+      "learning_rate": 0.00017527638190954775,
+      "loss": 0.2961,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4419313669204712,
+      "learning_rate": 0.00017507537688442212,
+      "loss": 0.2165,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4701198935508728,
+      "learning_rate": 0.0001748743718592965,
+      "loss": 0.2125,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.655662477016449,
+      "learning_rate": 0.00017467336683417087,
+      "loss": 0.2876,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.516601026058197,
+      "learning_rate": 0.00017447236180904521,
+      "loss": 0.2212,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41010063886642456,
+      "learning_rate": 0.00017427135678391961,
+      "loss": 0.193,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5393264889717102,
+      "learning_rate": 0.000174070351758794,
+      "loss": 0.2937,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7648224830627441,
+      "learning_rate": 0.00017386934673366836,
+      "loss": 0.2454,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.33873265981674194,
+      "learning_rate": 0.0001736683417085427,
+      "loss": 0.1509,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4198644161224365,
+      "learning_rate": 0.0001734673366834171,
+      "loss": 0.2785,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48141562938690186,
+      "learning_rate": 0.00017326633165829148,
+      "loss": 0.2382,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.503093421459198,
+      "learning_rate": 0.00017306532663316582,
+      "loss": 0.1948,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5640888810157776,
+      "learning_rate": 0.0001728643216080402,
+      "loss": 0.3005,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7310622334480286,
+      "learning_rate": 0.0001726633165829146,
+      "loss": 0.2385,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4998970925807953,
+      "learning_rate": 0.00017246231155778897,
+      "loss": 0.1729,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7067146897315979,
+      "learning_rate": 0.00017226130653266332,
+      "loss": 0.2471,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.50519198179245,
+      "learning_rate": 0.0001720603015075377,
+      "loss": 0.2125,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.740380048751831,
+      "learning_rate": 0.00017185929648241206,
+      "loss": 0.3141,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5456702709197998,
+      "learning_rate": 0.00017165829145728644,
+      "loss": 0.1892,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.42799606919288635,
+      "learning_rate": 0.0001714572864321608,
+      "loss": 0.1578,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4013369381427765,
+      "learning_rate": 0.00017125628140703518,
+      "loss": 0.1967,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4266522228717804,
+      "learning_rate": 0.00017105527638190955,
+      "loss": 0.2127,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6991720795631409,
+      "learning_rate": 0.00017085427135678393,
+      "loss": 0.1862,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5733714699745178,
+      "learning_rate": 0.0001706532663316583,
+      "loss": 0.2042,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5416551232337952,
+      "learning_rate": 0.00017045226130653267,
+      "loss": 0.2362,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5882843732833862,
+      "learning_rate": 0.00017025125628140705,
+      "loss": 0.2472,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5776798129081726,
+      "learning_rate": 0.00017005025125628142,
+      "loss": 0.2371,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4624580442905426,
+      "learning_rate": 0.0001698492462311558,
+      "loss": 0.1732,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6425135731697083,
+      "learning_rate": 0.00016964824120603016,
+      "loss": 0.2333,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5413373708724976,
+      "learning_rate": 0.0001694472361809045,
+      "loss": 0.309,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.485989511013031,
+      "learning_rate": 0.0001692462311557789,
+      "loss": 0.2429,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.397554874420166,
+      "learning_rate": 0.00016904522613065328,
+      "loss": 0.1851,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3770751357078552,
+      "learning_rate": 0.00016884422110552766,
+      "loss": 0.1868,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4285247325897217,
+      "learning_rate": 0.000168643216080402,
+      "loss": 0.2162,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5541285276412964,
+      "learning_rate": 0.0001684422110552764,
+      "loss": 0.222,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.46622520685195923,
+      "learning_rate": 0.00016824120603015078,
+      "loss": 0.2261,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6006364822387695,
+      "learning_rate": 0.00016804020100502512,
+      "loss": 0.1759,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5816208720207214,
+      "learning_rate": 0.0001678391959798995,
+      "loss": 0.2334,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5955444574356079,
+      "learning_rate": 0.0001676381909547739,
+      "loss": 0.266,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6741182804107666,
+      "learning_rate": 0.00016743718592964827,
+      "loss": 0.2282,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5965931415557861,
+      "learning_rate": 0.0001672361809045226,
+      "loss": 0.2768,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5617109537124634,
+      "learning_rate": 0.00016703517587939699,
+      "loss": 0.2306,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4135819673538208,
+      "learning_rate": 0.00016683417085427136,
+      "loss": 0.1766,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6315815448760986,
+      "learning_rate": 0.00016663316582914573,
+      "loss": 0.2797,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.47556841373443604,
+      "learning_rate": 0.0001664321608040201,
+      "loss": 0.2466,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4448830783367157,
+      "learning_rate": 0.00016623115577889448,
+      "loss": 0.1973,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4126035273075104,
+      "learning_rate": 0.00016603015075376885,
+      "loss": 0.1983,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3132772445678711,
+      "learning_rate": 0.00016582914572864322,
+      "loss": 0.1092,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.35979005694389343,
+      "learning_rate": 0.0001656281407035176,
+      "loss": 0.1578,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.180555820465088,
+      "learning_rate": 0.00016542713567839197,
+      "loss": 0.2077,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.419072687625885,
+      "learning_rate": 0.00016522613065326634,
+      "loss": 0.2021,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3829838037490845,
+      "learning_rate": 0.00016502512562814072,
+      "loss": 0.2232,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.46572762727737427,
+      "learning_rate": 0.0001648241206030151,
+      "loss": 0.24,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5526518821716309,
+      "learning_rate": 0.00016462311557788946,
+      "loss": 0.1901,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5201070308685303,
+      "learning_rate": 0.0001644221105527638,
+      "loss": 0.239,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8323246836662292,
+      "learning_rate": 0.0001642211055276382,
+      "loss": 0.2358,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4965194761753082,
+      "learning_rate": 0.00016402010050251258,
+      "loss": 0.2338,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6949551701545715,
+      "learning_rate": 0.00016381909547738695,
+      "loss": 0.2852,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5701273679733276,
+      "learning_rate": 0.0001636180904522613,
+      "loss": 0.2168,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5376931428909302,
+      "learning_rate": 0.0001634170854271357,
+      "loss": 0.1949,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.71084064245224,
+      "learning_rate": 0.00016321608040201007,
+      "loss": 0.2778,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4446854591369629,
+      "learning_rate": 0.00016301507537688442,
+      "loss": 0.1724,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4885426163673401,
+      "learning_rate": 0.0001628140703517588,
+      "loss": 0.2089,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6394463181495667,
+      "learning_rate": 0.00016261306532663316,
+      "loss": 0.3154,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3582012355327606,
+      "learning_rate": 0.00016241206030150756,
+      "loss": 0.1843,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4599544107913971,
+      "learning_rate": 0.0001622110552763819,
+      "loss": 0.2417,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4549001157283783,
+      "learning_rate": 0.00016201005025125628,
+      "loss": 0.1791,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6934012770652771,
+      "learning_rate": 0.00016180904522613066,
+      "loss": 0.3607,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6693351864814758,
+      "learning_rate": 0.00016160804020100503,
+      "loss": 0.2773,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5981723666191101,
+      "learning_rate": 0.0001614070351758794,
+      "loss": 0.2875,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4262889623641968,
+      "learning_rate": 0.00016120603015075378,
+      "loss": 0.2016,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.35076144337654114,
+      "learning_rate": 0.00016100502512562815,
+      "loss": 0.1483,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5418830513954163,
+      "learning_rate": 0.00016080402010050252,
+      "loss": 0.2251,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4608229696750641,
+      "learning_rate": 0.0001606030150753769,
+      "loss": 0.2225,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6093661785125732,
+      "learning_rate": 0.00016040201005025127,
+      "loss": 0.2892,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5232911705970764,
+      "learning_rate": 0.00016020100502512564,
+      "loss": 0.2725,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4431813359260559,
+      "learning_rate": 0.00016,
+      "loss": 0.1911,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3894941806793213,
+      "learning_rate": 0.00015979899497487439,
+      "loss": 0.1865,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5542987585067749,
+      "learning_rate": 0.00015959798994974876,
+      "loss": 0.2471,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4694427251815796,
+      "learning_rate": 0.0001593969849246231,
+      "loss": 0.1946,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6252871155738831,
+      "learning_rate": 0.0001591959798994975,
+      "loss": 0.274,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4899269640445709,
+      "learning_rate": 0.00015899497487437188,
+      "loss": 0.2558,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.470769464969635,
+      "learning_rate": 0.00015879396984924625,
+      "loss": 0.1648,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5405287742614746,
+      "learning_rate": 0.0001585929648241206,
+      "loss": 0.258,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7451691627502441,
+      "learning_rate": 0.000158391959798995,
+      "loss": 0.3103,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5479302406311035,
+      "learning_rate": 0.00015819095477386937,
+      "loss": 0.2905,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.442613810300827,
+      "learning_rate": 0.00015798994974874372,
+      "loss": 0.1913,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5008172392845154,
+      "learning_rate": 0.0001577889447236181,
+      "loss": 0.1954,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7391428351402283,
+      "learning_rate": 0.00015758793969849246,
+      "loss": 0.2254,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5006258487701416,
+      "learning_rate": 0.00015738693467336686,
+      "loss": 0.172,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5547099113464355,
+      "learning_rate": 0.0001571859296482412,
+      "loss": 0.1796,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6475242376327515,
+      "learning_rate": 0.00015698492462311558,
+      "loss": 0.2977,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6442254185676575,
+      "learning_rate": 0.00015678391959798995,
+      "loss": 0.2228,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6465861797332764,
+      "learning_rate": 0.00015658291457286433,
+      "loss": 0.341,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45650309324264526,
+      "learning_rate": 0.0001563819095477387,
+      "loss": 0.251,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4780837893486023,
+      "learning_rate": 0.00015618090452261307,
+      "loss": 0.1468,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.464327871799469,
+      "learning_rate": 0.00015597989949748745,
+      "loss": 0.2534,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3201116621494293,
+      "learning_rate": 0.00015577889447236182,
+      "loss": 0.1893,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4082617461681366,
+      "learning_rate": 0.0001555778894472362,
+      "loss": 0.2026,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.35708555579185486,
+      "learning_rate": 0.00015537688442211056,
+      "loss": 0.1746,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4080864489078522,
+      "learning_rate": 0.00015517587939698494,
+      "loss": 0.1585,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.471068799495697,
+      "learning_rate": 0.0001549748743718593,
+      "loss": 0.2246,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.450642466545105,
+      "learning_rate": 0.00015477386934673368,
+      "loss": 0.1979,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5421797633171082,
+      "learning_rate": 0.00015457286432160806,
+      "loss": 0.2056,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.41133153438568115,
+      "learning_rate": 0.0001543718592964824,
+      "loss": 0.2068,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45967182517051697,
+      "learning_rate": 0.0001541708542713568,
+      "loss": 0.2182,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.49316030740737915,
+      "learning_rate": 0.00015396984924623117,
+      "loss": 0.1856,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8641966581344604,
+      "learning_rate": 0.00015376884422110555,
+      "loss": 0.3491,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5350483655929565,
+      "learning_rate": 0.0001535678391959799,
+      "loss": 0.2346,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7115418314933777,
+      "learning_rate": 0.00015336683417085427,
+      "loss": 0.2561,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6251563429832458,
+      "learning_rate": 0.00015316582914572867,
+      "loss": 0.2683,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5587172508239746,
+      "learning_rate": 0.000152964824120603,
+      "loss": 0.227,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5323506593704224,
+      "learning_rate": 0.00015276381909547739,
+      "loss": 0.1803,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4084484577178955,
+      "learning_rate": 0.00015256281407035176,
+      "loss": 0.1665,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6423041224479675,
+      "learning_rate": 0.00015236180904522613,
+      "loss": 0.3084,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3014443516731262,
+      "learning_rate": 0.0001521608040201005,
+      "loss": 0.1687,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.437581866979599,
+      "learning_rate": 0.00015195979899497488,
+      "loss": 0.2215,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.49017801880836487,
+      "learning_rate": 0.00015175879396984925,
+      "loss": 0.2294,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.31686311960220337,
+      "learning_rate": 0.00015155778894472362,
+      "loss": 0.1547,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.47575148940086365,
+      "learning_rate": 0.000151356783919598,
+      "loss": 0.1956,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5036256909370422,
+      "learning_rate": 0.00015115577889447237,
+      "loss": 0.2365,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6078888177871704,
+      "learning_rate": 0.00015095477386934674,
+      "loss": 0.3543,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5907254219055176,
+      "learning_rate": 0.00015075376884422112,
+      "loss": 0.3306,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5760742425918579,
+      "learning_rate": 0.0001505527638190955,
+      "loss": 0.3161,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5956338047981262,
+      "learning_rate": 0.00015035175879396986,
+      "loss": 0.3077,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.30162495374679565,
+      "learning_rate": 0.00015015075376884423,
+      "loss": 0.1726,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5765033960342407,
+      "learning_rate": 0.0001499497487437186,
+      "loss": 0.2605,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.607266902923584,
+      "learning_rate": 0.00014974874371859298,
+      "loss": 0.25,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5608094930648804,
+      "learning_rate": 0.00014954773869346735,
+      "loss": 0.2919,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.48317083716392517,
+      "learning_rate": 0.0001493467336683417,
+      "loss": 0.2096,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4825833737850189,
+      "learning_rate": 0.0001491457286432161,
+      "loss": 0.2043,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4634295105934143,
+      "learning_rate": 0.00014894472361809047,
+      "loss": 0.2384,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4482268691062927,
+      "learning_rate": 0.00014874371859296482,
+      "loss": 0.2357,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46578094363212585,
+      "learning_rate": 0.0001485427135678392,
+      "loss": 0.2681,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5069098472595215,
+      "learning_rate": 0.00014834170854271356,
+      "loss": 0.2637,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3945165276527405,
+      "learning_rate": 0.00014814070351758796,
+      "loss": 0.1639,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.30467841029167175,
+      "learning_rate": 0.0001479396984924623,
+      "loss": 0.1697,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4881652891635895,
+      "learning_rate": 0.00014773869346733668,
+      "loss": 0.2332,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4942628741264343,
+      "learning_rate": 0.00014753768844221106,
+      "loss": 0.2342,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.42789214849472046,
+      "learning_rate": 0.00014733668341708543,
+      "loss": 0.1512,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6232418417930603,
+      "learning_rate": 0.0001471356783919598,
+      "loss": 0.3112,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.528900146484375,
+      "learning_rate": 0.00014693467336683417,
+      "loss": 0.2127,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6290087103843689,
+      "learning_rate": 0.00014673366834170855,
+      "loss": 0.2726,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.42025241255760193,
+      "learning_rate": 0.00014653266331658292,
+      "loss": 0.1842,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.41428080201148987,
+      "learning_rate": 0.0001463316582914573,
+      "loss": 0.2146,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3822866976261139,
+      "learning_rate": 0.00014613065326633167,
+      "loss": 0.157,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5606590509414673,
+      "learning_rate": 0.00014592964824120604,
+      "loss": 0.2617,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5164960026741028,
+      "learning_rate": 0.0001457286432160804,
+      "loss": 0.2353,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.47546109557151794,
+      "learning_rate": 0.00014552763819095479,
+      "loss": 0.1443,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6333800554275513,
+      "learning_rate": 0.00014532663316582916,
+      "loss": 0.2557,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5148735642433167,
+      "learning_rate": 0.00014512562814070353,
+      "loss": 0.2115,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.46984830498695374,
+      "learning_rate": 0.0001449246231155779,
+      "loss": 0.2134,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5740910768508911,
+      "learning_rate": 0.00014472361809045228,
+      "loss": 0.2558,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3297008275985718,
+      "learning_rate": 0.00014452261306532665,
+      "loss": 0.1292,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5956743359565735,
+      "learning_rate": 0.000144321608040201,
+      "loss": 0.1468,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7425946593284607,
+      "learning_rate": 0.00014412060301507537,
+      "loss": 0.3394,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6454773545265198,
+      "learning_rate": 0.00014391959798994977,
+      "loss": 0.223,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6458725333213806,
+      "learning_rate": 0.00014371859296482411,
+      "loss": 0.3424,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.440280944108963,
+      "learning_rate": 0.0001435175879396985,
+      "loss": 0.1558,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4852254390716553,
+      "learning_rate": 0.00014331658291457286,
+      "loss": 0.2069,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8012005090713501,
+      "learning_rate": 0.00014311557788944726,
+      "loss": 0.247,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.46413126587867737,
+      "learning_rate": 0.0001429145728643216,
+      "loss": 0.2691,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3412766456604004,
+      "learning_rate": 0.00014271356783919598,
+      "loss": 0.1581,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4218701422214508,
+      "learning_rate": 0.00014251256281407035,
+      "loss": 0.2102,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.569365382194519,
+      "learning_rate": 0.00014231155778894473,
+      "loss": 0.2343,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4240405261516571,
+      "learning_rate": 0.0001421105527638191,
+      "loss": 0.2201,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5527992844581604,
+      "learning_rate": 0.00014190954773869347,
+      "loss": 0.3343,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5937294960021973,
+      "learning_rate": 0.00014170854271356784,
+      "loss": 0.2895,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43181246519088745,
+      "learning_rate": 0.00014150753768844222,
+      "loss": 0.2039,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5217970013618469,
+      "learning_rate": 0.0001413065326633166,
+      "loss": 0.2642,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.496409147977829,
+      "learning_rate": 0.00014110552763819096,
+      "loss": 0.2071,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.38781318068504333,
+      "learning_rate": 0.00014090452261306534,
+      "loss": 0.1885,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.48885929584503174,
+      "learning_rate": 0.0001407035175879397,
+      "loss": 0.2035,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3917163014411926,
+      "learning_rate": 0.00014050251256281408,
+      "loss": 0.1853,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5597235560417175,
+      "learning_rate": 0.00014030150753768846,
+      "loss": 0.2303,
+      "step": 302
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.656257688999176,
+      "learning_rate": 0.0001401005025125628,
+      "loss": 0.3848,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4178178608417511,
+      "learning_rate": 0.0001398994974874372,
+      "loss": 0.1598,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5133432745933533,
+      "learning_rate": 0.00013969849246231157,
+      "loss": 0.2427,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.39841488003730774,
+      "learning_rate": 0.00013949748743718595,
+      "loss": 0.2144,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.30867376923561096,
+      "learning_rate": 0.0001392964824120603,
+      "loss": 0.12,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5577691197395325,
+      "learning_rate": 0.00013909547738693467,
+      "loss": 0.2167,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4248412847518921,
+      "learning_rate": 0.00013889447236180907,
+      "loss": 0.2066,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.41768893599510193,
+      "learning_rate": 0.0001386934673366834,
+      "loss": 0.1845,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.40006211400032043,
+      "learning_rate": 0.00013849246231155778,
+      "loss": 0.2085,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.479081928730011,
+      "learning_rate": 0.00013829145728643216,
+      "loss": 0.2253,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6218622922897339,
+      "learning_rate": 0.00013809045226130656,
+      "loss": 0.2304,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3523159921169281,
+      "learning_rate": 0.0001378894472361809,
+      "loss": 0.1487,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5560610294342041,
+      "learning_rate": 0.00013768844221105528,
+      "loss": 0.2568,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.40717363357543945,
+      "learning_rate": 0.00013748743718592965,
+      "loss": 0.1973,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5085451006889343,
+      "learning_rate": 0.00013728643216080402,
+      "loss": 0.2091,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4253947138786316,
+      "learning_rate": 0.0001370854271356784,
+      "loss": 0.1766,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.569270133972168,
+      "learning_rate": 0.00013688442211055277,
+      "loss": 0.224,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.637840986251831,
+      "learning_rate": 0.00013668341708542714,
+      "loss": 0.2496,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.38835230469703674,
+      "learning_rate": 0.00013648241206030151,
+      "loss": 0.1423,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4860616624355316,
+      "learning_rate": 0.0001362814070351759,
+      "loss": 0.1989,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.35445964336395264,
+      "learning_rate": 0.00013608040201005026,
+      "loss": 0.1497,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3935982584953308,
+      "learning_rate": 0.00013587939698492463,
+      "loss": 0.1486,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.49870574474334717,
+      "learning_rate": 0.000135678391959799,
+      "loss": 0.2537,
+      "step": 325
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5200526118278503,
+      "learning_rate": 0.00013547738693467338,
+      "loss": 0.229,
+      "step": 326
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.32679593563079834,
+      "learning_rate": 0.00013527638190954775,
+      "loss": 0.1545,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5571056604385376,
+      "learning_rate": 0.0001350753768844221,
+      "loss": 0.2033,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6576252579689026,
+      "learning_rate": 0.00013487437185929647,
+      "loss": 0.2517,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5548920631408691,
+      "learning_rate": 0.00013467336683417087,
+      "loss": 0.288,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6064463257789612,
+      "learning_rate": 0.00013447236180904524,
+      "loss": 0.3079,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.31661224365234375,
+      "learning_rate": 0.0001342713567839196,
+      "loss": 0.1413,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5535265803337097,
+      "learning_rate": 0.00013407035175879396,
+      "loss": 0.2165,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6876853108406067,
+      "learning_rate": 0.00013386934673366836,
+      "loss": 0.2922,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4079119861125946,
+      "learning_rate": 0.0001336683417085427,
+      "loss": 0.2241,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6994270086288452,
+      "learning_rate": 0.00013346733668341708,
+      "loss": 0.2341,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44559770822525024,
+      "learning_rate": 0.00013326633165829146,
+      "loss": 0.223,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.45040494203567505,
+      "learning_rate": 0.00013306532663316586,
+      "loss": 0.2135,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3731357753276825,
+      "learning_rate": 0.0001328643216080402,
+      "loss": 0.1656,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3673096299171448,
+      "learning_rate": 0.00013266331658291457,
+      "loss": 0.1443,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3495692014694214,
+      "learning_rate": 0.00013246231155778895,
+      "loss": 0.1709,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5194134712219238,
+      "learning_rate": 0.00013226130653266332,
+      "loss": 0.2019,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4705762565135956,
+      "learning_rate": 0.0001320603015075377,
+      "loss": 0.2321,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5076374411582947,
+      "learning_rate": 0.00013185929648241207,
+      "loss": 0.1972,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.48760005831718445,
+      "learning_rate": 0.00013165829145728644,
+      "loss": 0.1725,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5571461319923401,
+      "learning_rate": 0.0001314572864321608,
+      "loss": 0.1909,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.45073091983795166,
+      "learning_rate": 0.00013125628140703518,
+      "loss": 0.215,
+      "step": 347
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3622073531150818,
+      "learning_rate": 0.00013105527638190956,
+      "loss": 0.1711,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.62618488073349,
+      "learning_rate": 0.00013085427135678393,
+      "loss": 0.2632,
+      "step": 349
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4555056095123291,
+      "learning_rate": 0.0001306532663316583,
+      "loss": 0.2097,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6761806607246399,
+      "learning_rate": 0.00013045226130653268,
+      "loss": 0.2299,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5507418513298035,
+      "learning_rate": 0.00013025125628140705,
+      "loss": 0.2501,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5329481363296509,
+      "learning_rate": 0.0001300502512562814,
+      "loss": 0.2035,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5065528750419617,
+      "learning_rate": 0.00012984924623115577,
+      "loss": 0.1878,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.42520710825920105,
+      "learning_rate": 0.00012964824120603017,
+      "loss": 0.1861,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5527524948120117,
+      "learning_rate": 0.00012944723618090454,
+      "loss": 0.2651,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.38461214303970337,
+      "learning_rate": 0.0001292462311557789,
+      "loss": 0.1386,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.31263014674186707,
+      "learning_rate": 0.00012904522613065326,
+      "loss": 0.1746,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.38606059551239014,
+      "learning_rate": 0.00012884422110552766,
+      "loss": 0.1502,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.37648263573646545,
+      "learning_rate": 0.000128643216080402,
+      "loss": 0.1662,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.550672173500061,
+      "learning_rate": 0.00012844221105527638,
+      "loss": 0.2232,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.42321744561195374,
+      "learning_rate": 0.00012824120603015075,
+      "loss": 0.1622,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7612898945808411,
+      "learning_rate": 0.00012804020100502515,
+      "loss": 0.317,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3842167556285858,
+      "learning_rate": 0.0001278391959798995,
+      "loss": 0.1435,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40250709652900696,
+      "learning_rate": 0.00012763819095477387,
+      "loss": 0.2155,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5009702444076538,
+      "learning_rate": 0.00012743718592964824,
+      "loss": 0.2199,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4045158922672272,
+      "learning_rate": 0.00012723618090452262,
+      "loss": 0.1815,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6103566884994507,
+      "learning_rate": 0.000127035175879397,
+      "loss": 0.2449,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.42164695262908936,
+      "learning_rate": 0.00012683417085427136,
+      "loss": 0.2092,
+      "step": 369
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.30808696150779724,
+      "learning_rate": 0.00012663316582914574,
+      "loss": 0.1614,
+      "step": 370
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4597577452659607,
+      "learning_rate": 0.0001264321608040201,
+      "loss": 0.1898,
+      "step": 371
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6087939143180847,
+      "learning_rate": 0.00012623115577889448,
+      "loss": 0.2863,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.42504316568374634,
+      "learning_rate": 0.00012603015075376885,
+      "loss": 0.1619,
+      "step": 373
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.45179635286331177,
+      "learning_rate": 0.00012582914572864323,
+      "loss": 0.2398,
+      "step": 374
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3470228612422943,
+      "learning_rate": 0.0001256281407035176,
+      "loss": 0.1546,
+      "step": 375
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4229773283004761,
+      "learning_rate": 0.00012542713567839197,
+      "loss": 0.178,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3456679880619049,
+      "learning_rate": 0.00012522613065326635,
+      "loss": 0.1597,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4441853165626526,
+      "learning_rate": 0.0001250251256281407,
+      "loss": 0.2628,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.640588104724884,
+      "learning_rate": 0.00012482412060301507,
+      "loss": 0.3056,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4708755612373352,
+      "learning_rate": 0.00012462311557788947,
+      "loss": 0.1812,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5513359308242798,
+      "learning_rate": 0.00012442211055276384,
+      "loss": 0.268,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4472403824329376,
+      "learning_rate": 0.00012422110552763818,
+      "loss": 0.2437,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3861692249774933,
+      "learning_rate": 0.00012402010050251256,
+      "loss": 0.2121,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39574527740478516,
+      "learning_rate": 0.00012381909547738696,
+      "loss": 0.1584,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3973689377307892,
+      "learning_rate": 0.0001236180904522613,
+      "loss": 0.1864,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3852390944957733,
+      "learning_rate": 0.00012341708542713568,
+      "loss": 0.1936,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7175599336624146,
+      "learning_rate": 0.00012321608040201005,
+      "loss": 0.3122,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.38100698590278625,
+      "learning_rate": 0.00012301507537688445,
+      "loss": 0.1986,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5062547326087952,
+      "learning_rate": 0.0001228140703517588,
+      "loss": 0.2042,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.596307098865509,
+      "learning_rate": 0.00012261306532663317,
+      "loss": 0.2033,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5087392926216125,
+      "learning_rate": 0.00012241206030150754,
+      "loss": 0.2427,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39303168654441833,
+      "learning_rate": 0.00012221105527638191,
+      "loss": 0.2413,
+      "step": 392
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5655479431152344,
+      "learning_rate": 0.00012201005025125629,
+      "loss": 0.2434,
+      "step": 393
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.37353000044822693,
+      "learning_rate": 0.00012180904522613066,
+      "loss": 0.1971,
+      "step": 394
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4436759948730469,
+      "learning_rate": 0.00012160804020100502,
+      "loss": 0.2238,
+      "step": 395
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3882415294647217,
+      "learning_rate": 0.00012140703517587942,
+      "loss": 0.1489,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5103885531425476,
+      "learning_rate": 0.00012120603015075378,
+      "loss": 0.2176,
+      "step": 397
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.30824366211891174,
+      "learning_rate": 0.00012100502512562815,
+      "loss": 0.1488,
+      "step": 398
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6127710342407227,
+      "learning_rate": 0.00012080402010050251,
+      "loss": 0.3003,
+      "step": 399
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7671993374824524,
+      "learning_rate": 0.00012060301507537688,
+      "loss": 0.3096,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5708138942718506,
+      "learning_rate": 0.00012040201005025127,
+      "loss": 0.2392,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42668575048446655,
+      "learning_rate": 0.00012020100502512563,
+      "loss": 0.1882,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5934561491012573,
+      "learning_rate": 0.00012,
+      "loss": 0.2632,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.43244239687919617,
+      "learning_rate": 0.00011979899497487436,
+      "loss": 0.2458,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.47703075408935547,
+      "learning_rate": 0.00011959798994974876,
+      "loss": 0.2011,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.722775399684906,
+      "learning_rate": 0.00011939698492462312,
+      "loss": 0.2793,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7606828212738037,
+      "learning_rate": 0.0001191959798994975,
+      "loss": 0.2416,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4368147552013397,
+      "learning_rate": 0.00011899497487437185,
+      "loss": 0.2078,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5169294476509094,
+      "learning_rate": 0.00011879396984924624,
+      "loss": 0.2636,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6035071611404419,
+      "learning_rate": 0.00011859296482412061,
+      "loss": 0.2767,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5761665105819702,
+      "learning_rate": 0.00011839195979899497,
+      "loss": 0.2485,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.37852367758750916,
+      "learning_rate": 0.00011819095477386935,
+      "loss": 0.182,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5448431968688965,
+      "learning_rate": 0.00011798994974874373,
+      "loss": 0.2761,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.46888917684555054,
+      "learning_rate": 0.0001177889447236181,
+      "loss": 0.193,
+      "step": 414
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.42235323786735535,
+      "learning_rate": 0.00011758793969849247,
+      "loss": 0.1648,
+      "step": 415
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6363163590431213,
+      "learning_rate": 0.00011738693467336684,
+      "loss": 0.2847,
+      "step": 416
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.49301624298095703,
+      "learning_rate": 0.00011718592964824122,
+      "loss": 0.2491,
+      "step": 417
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38518691062927246,
+      "learning_rate": 0.00011698492462311558,
+      "loss": 0.1776,
+      "step": 418
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7396824955940247,
+      "learning_rate": 0.00011678391959798996,
+      "loss": 0.1623,
+      "step": 419
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.43891581892967224,
+      "learning_rate": 0.00011658291457286432,
+      "loss": 0.1766,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.28952324390411377,
+      "learning_rate": 0.00011638190954773872,
+      "loss": 0.1274,
+      "step": 421
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3468928039073944,
+      "learning_rate": 0.00011618090452261308,
+      "loss": 0.1249,
+      "step": 422
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.49591997265815735,
+      "learning_rate": 0.00011597989949748745,
+      "loss": 0.1978,
+      "step": 423
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4968617260456085,
+      "learning_rate": 0.00011577889447236181,
+      "loss": 0.2065,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.47323212027549744,
+      "learning_rate": 0.00011557788944723618,
+      "loss": 0.2298,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6332685947418213,
+      "learning_rate": 0.00011537688442211057,
+      "loss": 0.2474,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5754814743995667,
+      "learning_rate": 0.00011517587939698493,
+      "loss": 0.2771,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5415279269218445,
+      "learning_rate": 0.0001149748743718593,
+      "loss": 0.2068,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5374659895896912,
+      "learning_rate": 0.00011477386934673366,
+      "loss": 0.2339,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5732151865959167,
+      "learning_rate": 0.00011457286432160806,
+      "loss": 0.1922,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5028908252716064,
+      "learning_rate": 0.00011437185929648242,
+      "loss": 0.1981,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6984832286834717,
+      "learning_rate": 0.00011417085427135679,
+      "loss": 0.2972,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4545430839061737,
+      "learning_rate": 0.00011396984924623115,
+      "loss": 0.172,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5130873918533325,
+      "learning_rate": 0.00011376884422110554,
+      "loss": 0.1968,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6122608184814453,
+      "learning_rate": 0.00011356783919597991,
+      "loss": 0.2781,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8052571415901184,
+      "learning_rate": 0.00011336683417085427,
+      "loss": 0.2175,
+      "step": 436
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5444523096084595,
+      "learning_rate": 0.00011316582914572864,
+      "loss": 0.2295,
+      "step": 437
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4526660442352295,
+      "learning_rate": 0.00011296482412060303,
+      "loss": 0.1815,
+      "step": 438
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3232409656047821,
+      "learning_rate": 0.0001127638190954774,
+      "loss": 0.1794,
+      "step": 439
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6111499071121216,
+      "learning_rate": 0.00011256281407035176,
+      "loss": 0.2706,
+      "step": 440
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.45419833064079285,
+      "learning_rate": 0.00011236180904522614,
+      "loss": 0.2081,
+      "step": 441
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4061923623085022,
+      "learning_rate": 0.00011216080402010052,
+      "loss": 0.1677,
+      "step": 442
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3937043845653534,
+      "learning_rate": 0.00011195979899497488,
+      "loss": 0.1704,
+      "step": 443
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38572314381599426,
+      "learning_rate": 0.00011175879396984925,
+      "loss": 0.18,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6058553457260132,
+      "learning_rate": 0.00011155778894472361,
+      "loss": 0.2878,
+      "step": 445
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.43011635541915894,
+      "learning_rate": 0.00011135678391959799,
+      "loss": 0.1797,
+      "step": 446
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.44222158193588257,
+      "learning_rate": 0.00011115577889447237,
+      "loss": 0.1713,
+      "step": 447
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38344600796699524,
+      "learning_rate": 0.00011095477386934675,
+      "loss": 0.1663,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6274626851081848,
+      "learning_rate": 0.0001107537688442211,
+      "loss": 0.2461,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5568159222602844,
+      "learning_rate": 0.00011055276381909548,
+      "loss": 0.2774,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3897724747657776,
+      "learning_rate": 0.00011035175879396986,
+      "loss": 0.1874,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.42583590745925903,
+      "learning_rate": 0.00011015075376884422,
+      "loss": 0.1591,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3379085659980774,
+      "learning_rate": 0.0001099497487437186,
+      "loss": 0.1679,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38965246081352234,
+      "learning_rate": 0.00010974874371859296,
+      "loss": 0.1503,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5767619609832764,
+      "learning_rate": 0.00010954773869346736,
+      "loss": 0.2193,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.44832924008369446,
+      "learning_rate": 0.00010934673366834172,
+      "loss": 0.1931,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.43048256635665894,
+      "learning_rate": 0.00010914572864321609,
+      "loss": 0.1937,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.35769662261009216,
+      "learning_rate": 0.00010894472361809045,
+      "loss": 0.164,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3910098373889923,
+      "learning_rate": 0.00010874371859296483,
+      "loss": 0.1505,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5478328466415405,
+      "learning_rate": 0.00010854271356783921,
+      "loss": 0.2857,
+      "step": 460
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4424973130226135,
+      "learning_rate": 0.00010834170854271357,
+      "loss": 0.2499,
+      "step": 461
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6312743425369263,
+      "learning_rate": 0.00010814070351758794,
+      "loss": 0.2633,
+      "step": 462
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4388771653175354,
+      "learning_rate": 0.00010793969849246233,
+      "loss": 0.1829,
+      "step": 463
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.391568660736084,
+      "learning_rate": 0.0001077386934673367,
+      "loss": 0.1809,
+      "step": 464
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.696977972984314,
+      "learning_rate": 0.00010753768844221106,
+      "loss": 0.2329,
+      "step": 465
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5525539517402649,
+      "learning_rate": 0.00010733668341708543,
+      "loss": 0.2176,
+      "step": 466
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.42988142371177673,
+      "learning_rate": 0.00010713567839195982,
+      "loss": 0.2135,
+      "step": 467
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.43678900599479675,
+      "learning_rate": 0.00010693467336683418,
+      "loss": 0.1607,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.40529149770736694,
+      "learning_rate": 0.00010673366834170855,
+      "loss": 0.1831,
+      "step": 469
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.47344741225242615,
+      "learning_rate": 0.00010653266331658291,
+      "loss": 0.1809,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4426578879356384,
+      "learning_rate": 0.00010633165829145728,
+      "loss": 0.213,
+      "step": 471
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3296988904476166,
+      "learning_rate": 0.00010613065326633167,
+      "loss": 0.151,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6979312300682068,
+      "learning_rate": 0.00010592964824120604,
+      "loss": 0.2797,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3550786077976227,
+      "learning_rate": 0.0001057286432160804,
+      "loss": 0.2019,
+      "step": 474
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5628048181533813,
+      "learning_rate": 0.00010552763819095478,
+      "loss": 0.252,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3417305052280426,
+      "learning_rate": 0.00010532663316582916,
+      "loss": 0.1568,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5894454121589661,
+      "learning_rate": 0.00010512562814070352,
+      "loss": 0.2911,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7387587428092957,
+      "learning_rate": 0.0001049246231155779,
+      "loss": 0.316,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.585311770439148,
+      "learning_rate": 0.00010472361809045225,
+      "loss": 0.2556,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.43730881810188293,
+      "learning_rate": 0.00010452261306532664,
+      "loss": 0.2182,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6188883185386658,
+      "learning_rate": 0.00010432160804020101,
+      "loss": 0.3168,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4580572545528412,
+      "learning_rate": 0.00010412060301507539,
+      "loss": 0.1873,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.46275004744529724,
+      "learning_rate": 0.00010391959798994975,
+      "loss": 0.2342,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6386345624923706,
+      "learning_rate": 0.00010371859296482413,
+      "loss": 0.1601,
+      "step": 484
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4647381901741028,
+      "learning_rate": 0.0001035175879396985,
+      "loss": 0.2553,
+      "step": 485
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.392676442861557,
+      "learning_rate": 0.00010331658291457286,
+      "loss": 0.1727,
+      "step": 486
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6019845008850098,
+      "learning_rate": 0.00010311557788944724,
+      "loss": 0.3154,
+      "step": 487
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4485069215297699,
+      "learning_rate": 0.00010291457286432162,
+      "loss": 0.197,
+      "step": 488
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5247390866279602,
+      "learning_rate": 0.00010271356783919598,
+      "loss": 0.2212,
+      "step": 489
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5502858757972717,
+      "learning_rate": 0.00010251256281407036,
+      "loss": 0.2542,
+      "step": 490
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5899858474731445,
+      "learning_rate": 0.00010231155778894473,
+      "loss": 0.2944,
+      "step": 491
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3429397642612457,
+      "learning_rate": 0.00010211055276381909,
+      "loss": 0.1621,
+      "step": 492
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5685318112373352,
+      "learning_rate": 0.00010190954773869348,
+      "loss": 0.274,
+      "step": 493
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4867880046367645,
+      "learning_rate": 0.00010170854271356785,
+      "loss": 0.2044,
+      "step": 494
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5068891644477844,
+      "learning_rate": 0.00010150753768844221,
+      "loss": 0.2473,
+      "step": 495
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5069141983985901,
+      "learning_rate": 0.00010130653266331658,
+      "loss": 0.2217,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5071575045585632,
+      "learning_rate": 0.00010110552763819097,
+      "loss": 0.255,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.37730205059051514,
+      "learning_rate": 0.00010090452261306533,
+      "loss": 0.1576,
+      "step": 498
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3155452609062195,
+      "learning_rate": 0.0001007035175879397,
+      "loss": 0.1367,
+      "step": 499
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.43023696541786194,
+      "learning_rate": 0.00010050251256281407,
+      "loss": 0.2032,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3815259635448456,
+      "learning_rate": 0.00010030150753768846,
+      "loss": 0.1504,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4405147135257721,
+      "learning_rate": 0.00010010050251256282,
+      "loss": 0.1929,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5072082877159119,
+      "learning_rate": 9.989949748743719e-05,
+      "loss": 0.2101,
+      "step": 503
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.42251914739608765,
+      "learning_rate": 9.969849246231156e-05,
+      "loss": 0.2453,
+      "step": 504
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5401110053062439,
+      "learning_rate": 9.949748743718594e-05,
+      "loss": 0.2435,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6041588187217712,
+      "learning_rate": 9.929648241206031e-05,
+      "loss": 0.262,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.518314778804779,
+      "learning_rate": 9.909547738693468e-05,
+      "loss": 0.2278,
+      "step": 507
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4269994795322418,
+      "learning_rate": 9.889447236180906e-05,
+      "loss": 0.1982,
+      "step": 508
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.41217511892318726,
+      "learning_rate": 9.869346733668342e-05,
+      "loss": 0.2239,
+      "step": 509
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3829472064971924,
+      "learning_rate": 9.84924623115578e-05,
+      "loss": 0.1467,
+      "step": 510
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.44718316197395325,
+      "learning_rate": 9.829145728643216e-05,
+      "loss": 0.2209,
+      "step": 511
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3699786961078644,
+      "learning_rate": 9.809045226130655e-05,
+      "loss": 0.1533,
+      "step": 512
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.42279669642448425,
+      "learning_rate": 9.788944723618091e-05,
+      "loss": 0.1862,
+      "step": 513
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2697815895080566,
+      "learning_rate": 9.768844221105528e-05,
+      "loss": 0.2211,
+      "step": 514
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4647808074951172,
+      "learning_rate": 9.748743718592965e-05,
+      "loss": 0.2384,
+      "step": 515
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.45755669474601746,
+      "learning_rate": 9.728643216080403e-05,
+      "loss": 0.2094,
+      "step": 516
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6136307716369629,
+      "learning_rate": 9.70854271356784e-05,
+      "loss": 0.2892,
+      "step": 517
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4117870628833771,
+      "learning_rate": 9.688442211055276e-05,
+      "loss": 0.1885,
+      "step": 518
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.43240073323249817,
+      "learning_rate": 9.668341708542715e-05,
+      "loss": 0.2072,
+      "step": 519
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4392816722393036,
+      "learning_rate": 9.64824120603015e-05,
+      "loss": 0.188,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.31601712107658386,
+      "learning_rate": 9.628140703517589e-05,
+      "loss": 0.1547,
+      "step": 521
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.38773342967033386,
+      "learning_rate": 9.608040201005025e-05,
+      "loss": 0.1645,
+      "step": 522
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4962211847305298,
+      "learning_rate": 9.587939698492462e-05,
+      "loss": 0.212,
+      "step": 523
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.49664270877838135,
+      "learning_rate": 9.5678391959799e-05,
+      "loss": 0.2379,
+      "step": 524
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5445250272750854,
+      "learning_rate": 9.547738693467337e-05,
+      "loss": 0.2364,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5319061279296875,
+      "learning_rate": 9.527638190954774e-05,
+      "loss": 0.2105,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4577290713787079,
+      "learning_rate": 9.507537688442212e-05,
+      "loss": 0.1805,
+      "step": 527
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5664304494857788,
+      "learning_rate": 9.487437185929649e-05,
+      "loss": 0.1455,
+      "step": 528
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4801952838897705,
+      "learning_rate": 9.467336683417086e-05,
+      "loss": 0.259,
+      "step": 529
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.735759973526001,
+      "learning_rate": 9.447236180904523e-05,
+      "loss": 0.3454,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4495560824871063,
+      "learning_rate": 9.427135678391961e-05,
+      "loss": 0.2254,
+      "step": 531
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41473764181137085,
+      "learning_rate": 9.407035175879397e-05,
+      "loss": 0.1851,
+      "step": 532
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3230622112751007,
+      "learning_rate": 9.386934673366835e-05,
+      "loss": 0.1518,
+      "step": 533
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6365942358970642,
+      "learning_rate": 9.366834170854271e-05,
+      "loss": 0.2762,
+      "step": 534
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4550526738166809,
+      "learning_rate": 9.34673366834171e-05,
+      "loss": 0.1956,
+      "step": 535
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5498403310775757,
+      "learning_rate": 9.326633165829146e-05,
+      "loss": 0.2066,
+      "step": 536
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5257768034934998,
+      "learning_rate": 9.306532663316585e-05,
+      "loss": 0.2014,
+      "step": 537
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.49358436465263367,
+      "learning_rate": 9.28643216080402e-05,
+      "loss": 0.221,
+      "step": 538
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.33671697974205017,
+      "learning_rate": 9.266331658291458e-05,
+      "loss": 0.1663,
+      "step": 539
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5135949850082397,
+      "learning_rate": 9.246231155778895e-05,
+      "loss": 0.2232,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.38979530334472656,
+      "learning_rate": 9.226130653266331e-05,
+      "loss": 0.1623,
+      "step": 541
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.29901430010795593,
+      "learning_rate": 9.20603015075377e-05,
+      "loss": 0.1219,
+      "step": 542
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5453833341598511,
+      "learning_rate": 9.185929648241206e-05,
+      "loss": 0.2261,
+      "step": 543
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5688655972480774,
+      "learning_rate": 9.165829145728644e-05,
+      "loss": 0.2187,
+      "step": 544
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5330553650856018,
+      "learning_rate": 9.14572864321608e-05,
+      "loss": 0.2126,
+      "step": 545
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4559541940689087,
+      "learning_rate": 9.125628140703519e-05,
+      "loss": 0.2291,
+      "step": 546
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3465883135795593,
+      "learning_rate": 9.105527638190955e-05,
+      "loss": 0.198,
+      "step": 547
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.33882659673690796,
+      "learning_rate": 9.085427135678392e-05,
+      "loss": 0.168,
+      "step": 548
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.42802536487579346,
+      "learning_rate": 9.06532663316583e-05,
+      "loss": 0.2351,
+      "step": 549
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5641934275627136,
+      "learning_rate": 9.045226130653267e-05,
+      "loss": 0.2587,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5979325175285339,
+      "learning_rate": 9.025125628140704e-05,
+      "loss": 0.2807,
+      "step": 551
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.43575671315193176,
+      "learning_rate": 9.005025125628141e-05,
+      "loss": 0.1673,
+      "step": 552
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.49646639823913574,
+      "learning_rate": 8.984924623115579e-05,
+      "loss": 0.2079,
+      "step": 553
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6471161246299744,
+      "learning_rate": 8.964824120603016e-05,
+      "loss": 0.2519,
+      "step": 554
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6349927186965942,
+      "learning_rate": 8.944723618090453e-05,
+      "loss": 0.2993,
+      "step": 555
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.46648383140563965,
+      "learning_rate": 8.92462311557789e-05,
+      "loss": 0.2477,
+      "step": 556
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.38790756464004517,
+      "learning_rate": 8.904522613065326e-05,
+      "loss": 0.1964,
+      "step": 557
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.38935327529907227,
+      "learning_rate": 8.884422110552765e-05,
+      "loss": 0.1942,
+      "step": 558
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.45376303791999817,
+      "learning_rate": 8.864321608040201e-05,
+      "loss": 0.2201,
+      "step": 559
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.301658570766449,
+      "learning_rate": 8.84422110552764e-05,
+      "loss": 0.1673,
+      "step": 560
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.45029112696647644,
+      "learning_rate": 8.824120603015076e-05,
+      "loss": 0.247,
+      "step": 561
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.31792956590652466,
+      "learning_rate": 8.804020100502513e-05,
+      "loss": 0.1804,
+      "step": 562
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5120775103569031,
+      "learning_rate": 8.78391959798995e-05,
+      "loss": 0.2089,
+      "step": 563
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.36469754576683044,
+      "learning_rate": 8.763819095477387e-05,
+      "loss": 0.1186,
+      "step": 564
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5263894200325012,
+      "learning_rate": 8.743718592964825e-05,
+      "loss": 0.2857,
+      "step": 565
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5383757948875427,
+      "learning_rate": 8.723618090452261e-05,
+      "loss": 0.2347,
+      "step": 566
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.39517709612846375,
+      "learning_rate": 8.7035175879397e-05,
+      "loss": 0.1794,
+      "step": 567
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.44381988048553467,
+      "learning_rate": 8.683417085427135e-05,
+      "loss": 0.2065,
+      "step": 568
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3555888235569,
+      "learning_rate": 8.663316582914574e-05,
+      "loss": 0.1825,
+      "step": 569
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.565974235534668,
+      "learning_rate": 8.64321608040201e-05,
+      "loss": 0.2543,
+      "step": 570
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.37384283542633057,
+      "learning_rate": 8.623115577889449e-05,
+      "loss": 0.1703,
+      "step": 571
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4946988821029663,
+      "learning_rate": 8.603015075376884e-05,
+      "loss": 0.241,
+      "step": 572
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3822475075721741,
+      "learning_rate": 8.582914572864322e-05,
+      "loss": 0.1982,
+      "step": 573
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5194388628005981,
+      "learning_rate": 8.562814070351759e-05,
+      "loss": 0.2266,
+      "step": 574
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3615367114543915,
+      "learning_rate": 8.542713567839196e-05,
+      "loss": 0.1643,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.47169193625450134,
+      "learning_rate": 8.522613065326634e-05,
+      "loss": 0.2381,
+      "step": 576
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.38423123955726624,
+      "learning_rate": 8.502512562814071e-05,
+      "loss": 0.2342,
+      "step": 577
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4775454103946686,
+      "learning_rate": 8.482412060301508e-05,
+      "loss": 0.2582,
+      "step": 578
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.33534660935401917,
+      "learning_rate": 8.462311557788946e-05,
+      "loss": 0.1475,
+      "step": 579
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5885127186775208,
+      "learning_rate": 8.442211055276383e-05,
+      "loss": 0.2584,
+      "step": 580
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5246654748916626,
+      "learning_rate": 8.42211055276382e-05,
+      "loss": 0.2736,
+      "step": 581
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5678746104240417,
+      "learning_rate": 8.402010050251256e-05,
+      "loss": 0.1765,
+      "step": 582
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.485567182302475,
+      "learning_rate": 8.381909547738695e-05,
+      "loss": 0.1889,
+      "step": 583
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.49088627099990845,
+      "learning_rate": 8.36180904522613e-05,
+      "loss": 0.2282,
+      "step": 584
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.43999719619750977,
+      "learning_rate": 8.341708542713568e-05,
+      "loss": 0.1682,
+      "step": 585
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4806053340435028,
+      "learning_rate": 8.321608040201005e-05,
+      "loss": 0.2292,
+      "step": 586
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3324624300003052,
+      "learning_rate": 8.301507537688443e-05,
+      "loss": 0.1654,
+      "step": 587
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4315497577190399,
+      "learning_rate": 8.28140703517588e-05,
+      "loss": 0.2407,
+      "step": 588
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.30554506182670593,
+      "learning_rate": 8.261306532663317e-05,
+      "loss": 0.1315,
+      "step": 589
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.45997923612594604,
+      "learning_rate": 8.241206030150754e-05,
+      "loss": 0.1902,
+      "step": 590
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4228266179561615,
+      "learning_rate": 8.22110552763819e-05,
+      "loss": 0.1868,
+      "step": 591
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.31143641471862793,
+      "learning_rate": 8.201005025125629e-05,
+      "loss": 0.158,
+      "step": 592
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4374869167804718,
+      "learning_rate": 8.180904522613065e-05,
+      "loss": 0.1965,
+      "step": 593
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.35473427176475525,
+      "learning_rate": 8.160804020100504e-05,
+      "loss": 0.14,
+      "step": 594
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3932483196258545,
+      "learning_rate": 8.14070351758794e-05,
+      "loss": 0.1904,
+      "step": 595
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4485935568809509,
+      "learning_rate": 8.120603015075378e-05,
+      "loss": 0.2045,
+      "step": 596
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.47243842482566833,
+      "learning_rate": 8.100502512562814e-05,
+      "loss": 0.2102,
+      "step": 597
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5314076542854309,
+      "learning_rate": 8.080402010050251e-05,
+      "loss": 0.2139,
+      "step": 598
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4143196642398834,
+      "learning_rate": 8.060301507537689e-05,
+      "loss": 0.192,
+      "step": 599
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8205462098121643,
+      "learning_rate": 8.040201005025126e-05,
+      "loss": 0.2679,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5004960298538208,
+      "learning_rate": 8.020100502512563e-05,
+      "loss": 0.2012,
+      "step": 601
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3939405083656311,
+      "learning_rate": 8e-05,
+      "loss": 0.1939,
+      "step": 602
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4227248430252075,
+      "learning_rate": 7.979899497487438e-05,
+      "loss": 0.2312,
+      "step": 603
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.328396201133728,
+      "learning_rate": 7.959798994974875e-05,
+      "loss": 0.1883,
+      "step": 604
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3216836154460907,
+      "learning_rate": 7.939698492462313e-05,
+      "loss": 0.1427,
+      "step": 605
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5136520862579346,
+      "learning_rate": 7.91959798994975e-05,
+      "loss": 0.1962,
+      "step": 606
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3919290006160736,
+      "learning_rate": 7.899497487437186e-05,
+      "loss": 0.1825,
+      "step": 607
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.42786556482315063,
+      "learning_rate": 7.879396984924623e-05,
+      "loss": 0.201,
+      "step": 608
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.40081706643104553,
+      "learning_rate": 7.85929648241206e-05,
+      "loss": 0.1494,
+      "step": 609
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5619733929634094,
+      "learning_rate": 7.839195979899498e-05,
+      "loss": 0.2593,
+      "step": 610
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.35712411999702454,
+      "learning_rate": 7.819095477386935e-05,
+      "loss": 0.1535,
+      "step": 611
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.42240583896636963,
+      "learning_rate": 7.798994974874372e-05,
+      "loss": 0.2288,
+      "step": 612
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.360383003950119,
+      "learning_rate": 7.77889447236181e-05,
+      "loss": 0.1702,
+      "step": 613
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.38333067297935486,
+      "learning_rate": 7.758793969849247e-05,
+      "loss": 0.18,
+      "step": 614
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3969596028327942,
+      "learning_rate": 7.738693467336684e-05,
+      "loss": 0.1836,
+      "step": 615
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5108949542045593,
+      "learning_rate": 7.71859296482412e-05,
+      "loss": 0.2344,
+      "step": 616
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.700456976890564,
+      "learning_rate": 7.698492462311559e-05,
+      "loss": 0.3223,
+      "step": 617
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3161924481391907,
+      "learning_rate": 7.678391959798995e-05,
+      "loss": 0.1758,
+      "step": 618
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3729405105113983,
+      "learning_rate": 7.658291457286433e-05,
+      "loss": 0.1738,
+      "step": 619
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4030870795249939,
+      "learning_rate": 7.638190954773869e-05,
+      "loss": 0.1905,
+      "step": 620
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.44550663232803345,
+      "learning_rate": 7.618090452261307e-05,
+      "loss": 0.2326,
+      "step": 621
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5030593872070312,
+      "learning_rate": 7.597989949748744e-05,
+      "loss": 0.2007,
+      "step": 622
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7457550764083862,
+      "learning_rate": 7.577889447236181e-05,
+      "loss": 0.2438,
+      "step": 623
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3807595670223236,
+      "learning_rate": 7.557788944723618e-05,
+      "loss": 0.1561,
+      "step": 624
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5035004019737244,
+      "learning_rate": 7.537688442211056e-05,
+      "loss": 0.2496,
+      "step": 625
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.32844075560569763,
+      "learning_rate": 7.517587939698493e-05,
+      "loss": 0.2129,
+      "step": 626
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4441528916358948,
+      "learning_rate": 7.49748743718593e-05,
+      "loss": 0.2066,
+      "step": 627
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4826768934726715,
+      "learning_rate": 7.477386934673368e-05,
+      "loss": 0.2123,
+      "step": 628
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.632175624370575,
+      "learning_rate": 7.457286432160805e-05,
+      "loss": 0.2659,
+      "step": 629
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.42496258020401,
+      "learning_rate": 7.437185929648241e-05,
+      "loss": 0.2294,
+      "step": 630
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5259225368499756,
+      "learning_rate": 7.417085427135678e-05,
+      "loss": 0.2113,
+      "step": 631
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.425467312335968,
+      "learning_rate": 7.396984924623115e-05,
+      "loss": 0.194,
+      "step": 632
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.43209174275398254,
+      "learning_rate": 7.376884422110553e-05,
+      "loss": 0.1925,
+      "step": 633
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3306873142719269,
+      "learning_rate": 7.35678391959799e-05,
+      "loss": 0.1812,
+      "step": 634
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.40517109632492065,
+      "learning_rate": 7.336683417085427e-05,
+      "loss": 0.1464,
+      "step": 635
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4161038100719452,
+      "learning_rate": 7.316582914572865e-05,
+      "loss": 0.1761,
+      "step": 636
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.42401084303855896,
+      "learning_rate": 7.296482412060302e-05,
+      "loss": 0.18,
+      "step": 637
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5300508141517639,
+      "learning_rate": 7.276381909547739e-05,
+      "loss": 0.1941,
+      "step": 638
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.502947986125946,
+      "learning_rate": 7.256281407035177e-05,
+      "loss": 0.2022,
+      "step": 639
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4637830853462219,
+      "learning_rate": 7.236180904522614e-05,
+      "loss": 0.2126,
+      "step": 640
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5627738833427429,
+      "learning_rate": 7.21608040201005e-05,
+      "loss": 0.2623,
+      "step": 641
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.35125699639320374,
+      "learning_rate": 7.195979899497488e-05,
+      "loss": 0.1693,
+      "step": 642
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2746927738189697,
+      "learning_rate": 7.175879396984924e-05,
+      "loss": 0.1302,
+      "step": 643
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3855142891407013,
+      "learning_rate": 7.155778894472363e-05,
+      "loss": 0.1871,
+      "step": 644
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4590088129043579,
+      "learning_rate": 7.135678391959799e-05,
+      "loss": 0.1887,
+      "step": 645
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.36200350522994995,
+      "learning_rate": 7.115577889447236e-05,
+      "loss": 0.1729,
+      "step": 646
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.49908801913261414,
+      "learning_rate": 7.095477386934674e-05,
+      "loss": 0.2256,
+      "step": 647
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5674376487731934,
+      "learning_rate": 7.075376884422111e-05,
+      "loss": 0.2588,
+      "step": 648
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.512169599533081,
+      "learning_rate": 7.055276381909548e-05,
+      "loss": 0.1977,
+      "step": 649
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.49708080291748047,
+      "learning_rate": 7.035175879396985e-05,
+      "loss": 0.2387,
+      "step": 650
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4566737413406372,
+      "learning_rate": 7.015075376884423e-05,
+      "loss": 0.207,
+      "step": 651
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5629884004592896,
+      "learning_rate": 6.99497487437186e-05,
+      "loss": 0.2256,
+      "step": 652
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4460812211036682,
+      "learning_rate": 6.974874371859297e-05,
+      "loss": 0.2072,
+      "step": 653
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5162737369537354,
+      "learning_rate": 6.954773869346733e-05,
+      "loss": 0.2438,
+      "step": 654
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.37910178303718567,
+      "learning_rate": 6.93467336683417e-05,
+      "loss": 0.143,
+      "step": 655
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.49443650245666504,
+      "learning_rate": 6.914572864321608e-05,
+      "loss": 0.1988,
+      "step": 656
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4551987648010254,
+      "learning_rate": 6.894472361809045e-05,
+      "loss": 0.1939,
+      "step": 657
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3725804090499878,
+      "learning_rate": 6.874371859296482e-05,
+      "loss": 0.1934,
+      "step": 658
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5743696689605713,
+      "learning_rate": 6.85427135678392e-05,
+      "loss": 0.2682,
+      "step": 659
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4046136140823364,
+      "learning_rate": 6.834170854271357e-05,
+      "loss": 0.1855,
+      "step": 660
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5430765748023987,
+      "learning_rate": 6.814070351758794e-05,
+      "loss": 0.2382,
+      "step": 661
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.27529728412628174,
+      "learning_rate": 6.793969849246232e-05,
+      "loss": 0.1014,
+      "step": 662
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3833965063095093,
+      "learning_rate": 6.773869346733669e-05,
+      "loss": 0.1588,
+      "step": 663
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3148491680622101,
+      "learning_rate": 6.753768844221105e-05,
+      "loss": 0.1547,
+      "step": 664
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.46502408385276794,
+      "learning_rate": 6.733668341708544e-05,
+      "loss": 0.2111,
+      "step": 665
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3772050142288208,
+      "learning_rate": 6.71356783919598e-05,
+      "loss": 0.1872,
+      "step": 666
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3045889735221863,
+      "learning_rate": 6.693467336683418e-05,
+      "loss": 0.1099,
+      "step": 667
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6520645022392273,
+      "learning_rate": 6.673366834170854e-05,
+      "loss": 0.2573,
+      "step": 668
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3540431261062622,
+      "learning_rate": 6.653266331658293e-05,
+      "loss": 0.1699,
+      "step": 669
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.41358596086502075,
+      "learning_rate": 6.633165829145729e-05,
+      "loss": 0.2099,
+      "step": 670
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5491157174110413,
+      "learning_rate": 6.613065326633166e-05,
+      "loss": 0.2703,
+      "step": 671
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4711921513080597,
+      "learning_rate": 6.592964824120603e-05,
+      "loss": 0.1761,
+      "step": 672
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.33731257915496826,
+      "learning_rate": 6.57286432160804e-05,
+      "loss": 0.1713,
+      "step": 673
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.423803448677063,
+      "learning_rate": 6.552763819095478e-05,
+      "loss": 0.1764,
+      "step": 674
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.33093374967575073,
+      "learning_rate": 6.532663316582915e-05,
+      "loss": 0.1334,
+      "step": 675
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4147561490535736,
+      "learning_rate": 6.512562814070352e-05,
+      "loss": 0.1557,
+      "step": 676
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3329070210456848,
+      "learning_rate": 6.492462311557788e-05,
+      "loss": 0.1688,
+      "step": 677
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.48782962560653687,
+      "learning_rate": 6.472361809045227e-05,
+      "loss": 0.2571,
+      "step": 678
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.38781771063804626,
+      "learning_rate": 6.452261306532663e-05,
+      "loss": 0.2022,
+      "step": 679
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3884035646915436,
+      "learning_rate": 6.4321608040201e-05,
+      "loss": 0.1418,
+      "step": 680
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.48443275690078735,
+      "learning_rate": 6.412060301507538e-05,
+      "loss": 0.2499,
+      "step": 681
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.38683968782424927,
+      "learning_rate": 6.391959798994975e-05,
+      "loss": 0.2103,
+      "step": 682
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.31988051533699036,
+      "learning_rate": 6.371859296482412e-05,
+      "loss": 0.172,
+      "step": 683
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.351112425327301,
+      "learning_rate": 6.35175879396985e-05,
+      "loss": 0.1635,
+      "step": 684
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.40820181369781494,
+      "learning_rate": 6.331658291457287e-05,
+      "loss": 0.1636,
+      "step": 685
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5173302888870239,
+      "learning_rate": 6.311557788944724e-05,
+      "loss": 0.2135,
+      "step": 686
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5712177753448486,
+      "learning_rate": 6.291457286432161e-05,
+      "loss": 0.1266,
+      "step": 687
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.44783997535705566,
+      "learning_rate": 6.271356783919599e-05,
+      "loss": 0.218,
+      "step": 688
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.31131142377853394,
+      "learning_rate": 6.251256281407035e-05,
+      "loss": 0.1404,
+      "step": 689
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.39051759243011475,
+      "learning_rate": 6.231155778894473e-05,
+      "loss": 0.1428,
+      "step": 690
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4637140929698944,
+      "learning_rate": 6.211055276381909e-05,
+      "loss": 0.1846,
+      "step": 691
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4765917956829071,
+      "learning_rate": 6.190954773869348e-05,
+      "loss": 0.1501,
+      "step": 692
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.45205971598625183,
+      "learning_rate": 6.170854271356784e-05,
+      "loss": 0.1898,
+      "step": 693
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5680738687515259,
+      "learning_rate": 6.150753768844222e-05,
+      "loss": 0.2216,
+      "step": 694
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3384318947792053,
+      "learning_rate": 6.130653266331658e-05,
+      "loss": 0.1613,
+      "step": 695
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5899595618247986,
+      "learning_rate": 6.110552763819096e-05,
+      "loss": 0.2388,
+      "step": 696
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.44541215896606445,
+      "learning_rate": 6.090452261306533e-05,
+      "loss": 0.1734,
+      "step": 697
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4676991105079651,
+      "learning_rate": 6.070351758793971e-05,
+      "loss": 0.221,
+      "step": 698
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6239830851554871,
+      "learning_rate": 6.0502512562814076e-05,
+      "loss": 0.2688,
+      "step": 699
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.43823522329330444,
+      "learning_rate": 6.030150753768844e-05,
+      "loss": 0.1981,
+      "step": 700
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3032325804233551,
+      "learning_rate": 6.0100502512562815e-05,
+      "loss": 0.1329,
+      "step": 701
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.33419281244277954,
+      "learning_rate": 5.989949748743718e-05,
+      "loss": 0.1361,
+      "step": 702
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5834112167358398,
+      "learning_rate": 5.969849246231156e-05,
+      "loss": 0.2134,
+      "step": 703
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.47133928537368774,
+      "learning_rate": 5.949748743718593e-05,
+      "loss": 0.2316,
+      "step": 704
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4087996184825897,
+      "learning_rate": 5.929648241206031e-05,
+      "loss": 0.1973,
+      "step": 705
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.38998982310295105,
+      "learning_rate": 5.909547738693467e-05,
+      "loss": 0.1719,
+      "step": 706
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40847426652908325,
+      "learning_rate": 5.889447236180905e-05,
+      "loss": 0.1913,
+      "step": 707
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.37600865960121155,
+      "learning_rate": 5.869346733668342e-05,
+      "loss": 0.1841,
+      "step": 708
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3851812183856964,
+      "learning_rate": 5.849246231155779e-05,
+      "loss": 0.1692,
+      "step": 709
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5539536476135254,
+      "learning_rate": 5.829145728643216e-05,
+      "loss": 0.2312,
+      "step": 710
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.33982518315315247,
+      "learning_rate": 5.809045226130654e-05,
+      "loss": 0.1547,
+      "step": 711
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.44841164350509644,
+      "learning_rate": 5.7889447236180904e-05,
+      "loss": 0.2319,
+      "step": 712
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4287841022014618,
+      "learning_rate": 5.7688442211055284e-05,
+      "loss": 0.2062,
+      "step": 713
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4406593441963196,
+      "learning_rate": 5.748743718592965e-05,
+      "loss": 0.2187,
+      "step": 714
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3746339976787567,
+      "learning_rate": 5.728643216080403e-05,
+      "loss": 0.2016,
+      "step": 715
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7049340009689331,
+      "learning_rate": 5.7085427135678396e-05,
+      "loss": 0.2532,
+      "step": 716
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5307794213294983,
+      "learning_rate": 5.688442211055277e-05,
+      "loss": 0.2093,
+      "step": 717
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.31563398241996765,
+      "learning_rate": 5.6683417085427135e-05,
+      "loss": 0.1565,
+      "step": 718
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5347794890403748,
+      "learning_rate": 5.6482412060301515e-05,
+      "loss": 0.2473,
+      "step": 719
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3637223541736603,
+      "learning_rate": 5.628140703517588e-05,
+      "loss": 0.1663,
+      "step": 720
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6633073091506958,
+      "learning_rate": 5.608040201005026e-05,
+      "loss": 0.2649,
+      "step": 721
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.36611008644104004,
+      "learning_rate": 5.587939698492463e-05,
+      "loss": 0.1544,
+      "step": 722
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.43361905217170715,
+      "learning_rate": 5.567839195979899e-05,
+      "loss": 0.1687,
+      "step": 723
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.48814257979393005,
+      "learning_rate": 5.547738693467337e-05,
+      "loss": 0.2227,
+      "step": 724
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3796922564506531,
+      "learning_rate": 5.527638190954774e-05,
+      "loss": 0.1915,
+      "step": 725
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.36174800992012024,
+      "learning_rate": 5.507537688442211e-05,
+      "loss": 0.1723,
+      "step": 726
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.36948803067207336,
+      "learning_rate": 5.487437185929648e-05,
+      "loss": 0.1525,
+      "step": 727
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4576868414878845,
+      "learning_rate": 5.467336683417086e-05,
+      "loss": 0.1663,
+      "step": 728
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.38392430543899536,
+      "learning_rate": 5.4472361809045224e-05,
+      "loss": 0.1431,
+      "step": 729
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.44476935267448425,
+      "learning_rate": 5.4271356783919604e-05,
+      "loss": 0.2179,
+      "step": 730
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.40011876821517944,
+      "learning_rate": 5.407035175879397e-05,
+      "loss": 0.1557,
+      "step": 731
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.42617473006248474,
+      "learning_rate": 5.386934673366835e-05,
+      "loss": 0.2283,
+      "step": 732
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.45721670985221863,
+      "learning_rate": 5.3668341708542716e-05,
+      "loss": 0.2026,
+      "step": 733
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5912308096885681,
+      "learning_rate": 5.346733668341709e-05,
+      "loss": 0.2628,
+      "step": 734
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.44696366786956787,
+      "learning_rate": 5.3266331658291455e-05,
+      "loss": 0.1878,
+      "step": 735
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.36148297786712646,
+      "learning_rate": 5.3065326633165835e-05,
+      "loss": 0.164,
+      "step": 736
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3467967212200165,
+      "learning_rate": 5.28643216080402e-05,
+      "loss": 0.1657,
+      "step": 737
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6262069940567017,
+      "learning_rate": 5.266331658291458e-05,
+      "loss": 0.2245,
+      "step": 738
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.42778122425079346,
+      "learning_rate": 5.246231155778895e-05,
+      "loss": 0.1717,
+      "step": 739
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.51275634765625,
+      "learning_rate": 5.226130653266332e-05,
+      "loss": 0.2183,
+      "step": 740
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.49025994539260864,
+      "learning_rate": 5.206030150753769e-05,
+      "loss": 0.2648,
+      "step": 741
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5096611380577087,
+      "learning_rate": 5.1859296482412066e-05,
+      "loss": 0.2544,
+      "step": 742
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4826028645038605,
+      "learning_rate": 5.165829145728643e-05,
+      "loss": 0.2256,
+      "step": 743
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.54673832654953,
+      "learning_rate": 5.145728643216081e-05,
+      "loss": 0.2344,
+      "step": 744
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.46049684286117554,
+      "learning_rate": 5.125628140703518e-05,
+      "loss": 0.186,
+      "step": 745
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5595524311065674,
+      "learning_rate": 5.1055276381909544e-05,
+      "loss": 0.2271,
+      "step": 746
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.42573636770248413,
+      "learning_rate": 5.0854271356783924e-05,
+      "loss": 0.2002,
+      "step": 747
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.37139996886253357,
+      "learning_rate": 5.065326633165829e-05,
+      "loss": 0.199,
+      "step": 748
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.46777206659317017,
+      "learning_rate": 5.045226130653266e-05,
+      "loss": 0.2271,
+      "step": 749
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4091203212738037,
+      "learning_rate": 5.0251256281407036e-05,
+      "loss": 0.2002,
+      "step": 750
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4840552806854248,
+      "learning_rate": 5.005025125628141e-05,
+      "loss": 0.2238,
+      "step": 751
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3718883693218231,
+      "learning_rate": 4.984924623115578e-05,
+      "loss": 0.2041,
+      "step": 752
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4906879663467407,
+      "learning_rate": 4.9648241206030155e-05,
+      "loss": 0.2466,
+      "step": 753
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.582410991191864,
+      "learning_rate": 4.944723618090453e-05,
+      "loss": 0.1961,
+      "step": 754
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40116363763809204,
+      "learning_rate": 4.92462311557789e-05,
+      "loss": 0.2109,
+      "step": 755
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.39108482003211975,
+      "learning_rate": 4.9045226130653274e-05,
+      "loss": 0.1804,
+      "step": 756
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5282275676727295,
+      "learning_rate": 4.884422110552764e-05,
+      "loss": 0.175,
+      "step": 757
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4697287976741791,
+      "learning_rate": 4.864321608040201e-05,
+      "loss": 0.168,
+      "step": 758
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6108348369598389,
+      "learning_rate": 4.844221105527638e-05,
+      "loss": 0.1997,
+      "step": 759
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.31648653745651245,
+      "learning_rate": 4.824120603015075e-05,
+      "loss": 0.148,
+      "step": 760
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7432184219360352,
+      "learning_rate": 4.8040201005025125e-05,
+      "loss": 0.3999,
+      "step": 761
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5415976643562317,
+      "learning_rate": 4.78391959798995e-05,
+      "loss": 0.2219,
+      "step": 762
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.47961345314979553,
+      "learning_rate": 4.763819095477387e-05,
+      "loss": 0.2282,
+      "step": 763
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4368441104888916,
+      "learning_rate": 4.7437185929648244e-05,
+      "loss": 0.1894,
+      "step": 764
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.36008429527282715,
+      "learning_rate": 4.723618090452262e-05,
+      "loss": 0.1797,
+      "step": 765
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3960406184196472,
+      "learning_rate": 4.703517587939698e-05,
+      "loss": 0.1641,
+      "step": 766
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5364041328430176,
+      "learning_rate": 4.6834170854271356e-05,
+      "loss": 0.2232,
+      "step": 767
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5432655811309814,
+      "learning_rate": 4.663316582914573e-05,
+      "loss": 0.2936,
+      "step": 768
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5242969393730164,
+      "learning_rate": 4.64321608040201e-05,
+      "loss": 0.2779,
+      "step": 769
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.597951352596283,
+      "learning_rate": 4.6231155778894475e-05,
+      "loss": 0.2731,
+      "step": 770
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.459395170211792,
+      "learning_rate": 4.603015075376885e-05,
+      "loss": 0.2536,
+      "step": 771
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3068626821041107,
+      "learning_rate": 4.582914572864322e-05,
+      "loss": 0.1761,
+      "step": 772
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.456222265958786,
+      "learning_rate": 4.5628140703517594e-05,
+      "loss": 0.2256,
+      "step": 773
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4786625802516937,
+      "learning_rate": 4.542713567839196e-05,
+      "loss": 0.2026,
+      "step": 774
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6046634912490845,
+      "learning_rate": 4.522613065326633e-05,
+      "loss": 0.2876,
+      "step": 775
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3308449387550354,
+      "learning_rate": 4.5025125628140706e-05,
+      "loss": 0.1685,
+      "step": 776
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5068316459655762,
+      "learning_rate": 4.482412060301508e-05,
+      "loss": 0.2652,
+      "step": 777
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6490353941917419,
+      "learning_rate": 4.462311557788945e-05,
+      "loss": 0.2178,
+      "step": 778
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4765259921550751,
+      "learning_rate": 4.4422110552763825e-05,
+      "loss": 0.2673,
+      "step": 779
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4649331271648407,
+      "learning_rate": 4.42211055276382e-05,
+      "loss": 0.2806,
+      "step": 780
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5351561903953552,
+      "learning_rate": 4.4020100502512564e-05,
+      "loss": 0.263,
+      "step": 781
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.49080169200897217,
+      "learning_rate": 4.381909547738694e-05,
+      "loss": 0.2438,
+      "step": 782
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3744328022003174,
+      "learning_rate": 4.3618090452261303e-05,
+      "loss": 0.1542,
+      "step": 783
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5244365930557251,
+      "learning_rate": 4.3417085427135676e-05,
+      "loss": 0.1754,
+      "step": 784
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.24274443089962006,
+      "learning_rate": 4.321608040201005e-05,
+      "loss": 0.1502,
+      "step": 785
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.41415631771087646,
+      "learning_rate": 4.301507537688442e-05,
+      "loss": 0.2229,
+      "step": 786
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3658473491668701,
+      "learning_rate": 4.2814070351758795e-05,
+      "loss": 0.2292,
+      "step": 787
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.22280116379261017,
+      "learning_rate": 4.261306532663317e-05,
+      "loss": 0.1124,
+      "step": 788
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.44349828362464905,
+      "learning_rate": 4.241206030150754e-05,
+      "loss": 0.2261,
+      "step": 789
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7576150298118591,
+      "learning_rate": 4.2211055276381914e-05,
+      "loss": 0.2237,
+      "step": 790
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.41980302333831787,
+      "learning_rate": 4.201005025125628e-05,
+      "loss": 0.2102,
+      "step": 791
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7608839273452759,
+      "learning_rate": 4.180904522613065e-05,
+      "loss": 0.3884,
+      "step": 792
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6635060906410217,
+      "learning_rate": 4.1608040201005026e-05,
+      "loss": 0.3462,
+      "step": 793
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5781263113021851,
+      "learning_rate": 4.14070351758794e-05,
+      "loss": 0.2675,
+      "step": 794
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.354825496673584,
+      "learning_rate": 4.120603015075377e-05,
+      "loss": 0.19,
+      "step": 795
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.516942024230957,
+      "learning_rate": 4.1005025125628145e-05,
+      "loss": 0.2157,
+      "step": 796
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3210093677043915,
+      "learning_rate": 4.080402010050252e-05,
+      "loss": 0.167,
+      "step": 797
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.36717918515205383,
+      "learning_rate": 4.060301507537689e-05,
+      "loss": 0.176,
+      "step": 798
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.33211013674736023,
+      "learning_rate": 4.040201005025126e-05,
+      "loss": 0.1659,
+      "step": 799
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.522240161895752,
+      "learning_rate": 4.020100502512563e-05,
+      "loss": 0.257,
+      "step": 800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5097737312316895,
+      "learning_rate": 4e-05,
+      "loss": 0.2492,
+      "step": 801
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5827239751815796,
+      "learning_rate": 3.9798994974874376e-05,
+      "loss": 0.287,
+      "step": 802
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4194738268852234,
+      "learning_rate": 3.959798994974875e-05,
+      "loss": 0.1696,
+      "step": 803
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.49148353934288025,
+      "learning_rate": 3.9396984924623115e-05,
+      "loss": 0.2323,
+      "step": 804
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5025193095207214,
+      "learning_rate": 3.919597989949749e-05,
+      "loss": 0.2973,
+      "step": 805
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3152432143688202,
+      "learning_rate": 3.899497487437186e-05,
+      "loss": 0.181,
+      "step": 806
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5241256356239319,
+      "learning_rate": 3.8793969849246234e-05,
+      "loss": 0.2553,
+      "step": 807
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.29373928904533386,
+      "learning_rate": 3.85929648241206e-05,
+      "loss": 0.156,
+      "step": 808
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5213522911071777,
+      "learning_rate": 3.8391959798994973e-05,
+      "loss": 0.3123,
+      "step": 809
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2721177339553833,
+      "learning_rate": 3.8190954773869346e-05,
+      "loss": 0.1192,
+      "step": 810
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4929058253765106,
+      "learning_rate": 3.798994974874372e-05,
+      "loss": 0.2325,
+      "step": 811
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.553525984287262,
+      "learning_rate": 3.778894472361809e-05,
+      "loss": 0.2711,
+      "step": 812
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.32851460576057434,
+      "learning_rate": 3.7587939698492465e-05,
+      "loss": 0.189,
+      "step": 813
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.40988585352897644,
+      "learning_rate": 3.738693467336684e-05,
+      "loss": 0.2204,
+      "step": 814
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5577826499938965,
+      "learning_rate": 3.7185929648241204e-05,
+      "loss": 0.2279,
+      "step": 815
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3908914029598236,
+      "learning_rate": 3.698492462311558e-05,
+      "loss": 0.1832,
+      "step": 816
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.44297468662261963,
+      "learning_rate": 3.678391959798995e-05,
+      "loss": 0.2031,
+      "step": 817
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.46193307638168335,
+      "learning_rate": 3.658291457286432e-05,
+      "loss": 0.2362,
+      "step": 818
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.449309378862381,
+      "learning_rate": 3.6381909547738696e-05,
+      "loss": 0.2741,
+      "step": 819
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.47377705574035645,
+      "learning_rate": 3.618090452261307e-05,
+      "loss": 0.2441,
+      "step": 820
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5499680638313293,
+      "learning_rate": 3.597989949748744e-05,
+      "loss": 0.2853,
+      "step": 821
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4443001449108124,
+      "learning_rate": 3.5778894472361815e-05,
+      "loss": 0.2124,
+      "step": 822
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.31199464201927185,
+      "learning_rate": 3.557788944723618e-05,
+      "loss": 0.1669,
+      "step": 823
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4009048342704773,
+      "learning_rate": 3.5376884422110554e-05,
+      "loss": 0.1822,
+      "step": 824
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.35128268599510193,
+      "learning_rate": 3.517587939698493e-05,
+      "loss": 0.2077,
+      "step": 825
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6096314787864685,
+      "learning_rate": 3.49748743718593e-05,
+      "loss": 0.319,
+      "step": 826
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.49398887157440186,
+      "learning_rate": 3.4773869346733667e-05,
+      "loss": 0.3382,
+      "step": 827
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.46479663252830505,
+      "learning_rate": 3.457286432160804e-05,
+      "loss": 0.2339,
+      "step": 828
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3224971294403076,
+      "learning_rate": 3.437185929648241e-05,
+      "loss": 0.1626,
+      "step": 829
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5267730951309204,
+      "learning_rate": 3.4170854271356785e-05,
+      "loss": 0.2813,
+      "step": 830
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5544459223747253,
+      "learning_rate": 3.396984924623116e-05,
+      "loss": 0.3294,
+      "step": 831
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5161508917808533,
+      "learning_rate": 3.3768844221105525e-05,
+      "loss": 0.2428,
+      "step": 832
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5045228004455566,
+      "learning_rate": 3.35678391959799e-05,
+      "loss": 0.2672,
+      "step": 833
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.33565956354141235,
+      "learning_rate": 3.336683417085427e-05,
+      "loss": 0.1272,
+      "step": 834
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.27528029680252075,
+      "learning_rate": 3.3165829145728643e-05,
+      "loss": 0.1325,
+      "step": 835
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3082004189491272,
+      "learning_rate": 3.2964824120603016e-05,
+      "loss": 0.173,
+      "step": 836
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.38152727484703064,
+      "learning_rate": 3.276381909547739e-05,
+      "loss": 0.1784,
+      "step": 837
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5074553489685059,
+      "learning_rate": 3.256281407035176e-05,
+      "loss": 0.2878,
+      "step": 838
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.37047079205513,
+      "learning_rate": 3.2361809045226135e-05,
+      "loss": 0.167,
+      "step": 839
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4814290702342987,
+      "learning_rate": 3.21608040201005e-05,
+      "loss": 0.2293,
+      "step": 840
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.37862345576286316,
+      "learning_rate": 3.1959798994974875e-05,
+      "loss": 0.2462,
+      "step": 841
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.302799791097641,
+      "learning_rate": 3.175879396984925e-05,
+      "loss": 0.1738,
+      "step": 842
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.35300469398498535,
+      "learning_rate": 3.155778894472362e-05,
+      "loss": 0.1652,
+      "step": 843
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.45807746052742004,
+      "learning_rate": 3.1356783919597993e-05,
+      "loss": 0.2621,
+      "step": 844
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6076413989067078,
+      "learning_rate": 3.1155778894472366e-05,
+      "loss": 0.3297,
+      "step": 845
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.35612595081329346,
+      "learning_rate": 3.095477386934674e-05,
+      "loss": 0.1883,
+      "step": 846
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.28016719222068787,
+      "learning_rate": 3.075376884422111e-05,
+      "loss": 0.1264,
+      "step": 847
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4207148253917694,
+      "learning_rate": 3.055276381909548e-05,
+      "loss": 0.2308,
+      "step": 848
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3742695748806,
+      "learning_rate": 3.0351758793969855e-05,
+      "loss": 0.1903,
+      "step": 849
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5703909993171692,
+      "learning_rate": 3.015075376884422e-05,
+      "loss": 0.1784,
+      "step": 850
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3696439266204834,
+      "learning_rate": 2.994974874371859e-05,
+      "loss": 0.1743,
+      "step": 851
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.557442843914032,
+      "learning_rate": 2.9748743718592964e-05,
+      "loss": 0.2909,
+      "step": 852
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6531166434288025,
+      "learning_rate": 2.9547738693467337e-05,
+      "loss": 0.2273,
+      "step": 853
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6506884098052979,
+      "learning_rate": 2.934673366834171e-05,
+      "loss": 0.2878,
+      "step": 854
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4518173933029175,
+      "learning_rate": 2.914572864321608e-05,
+      "loss": 0.2037,
+      "step": 855
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4443354606628418,
+      "learning_rate": 2.8944723618090452e-05,
+      "loss": 0.2299,
+      "step": 856
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.23116129636764526,
+      "learning_rate": 2.8743718592964825e-05,
+      "loss": 0.1179,
+      "step": 857
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.40902626514434814,
+      "learning_rate": 2.8542713567839198e-05,
+      "loss": 0.2197,
+      "step": 858
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5275218486785889,
+      "learning_rate": 2.8341708542713568e-05,
+      "loss": 0.2427,
+      "step": 859
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3060431480407715,
+      "learning_rate": 2.814070351758794e-05,
+      "loss": 0.162,
+      "step": 860
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6324546933174133,
+      "learning_rate": 2.7939698492462314e-05,
+      "loss": 0.2545,
+      "step": 861
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4083345830440521,
+      "learning_rate": 2.7738693467336686e-05,
+      "loss": 0.1776,
+      "step": 862
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7091811299324036,
+      "learning_rate": 2.7537688442211056e-05,
+      "loss": 0.335,
+      "step": 863
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3959479033946991,
+      "learning_rate": 2.733668341708543e-05,
+      "loss": 0.2122,
+      "step": 864
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.37795379757881165,
+      "learning_rate": 2.7135678391959802e-05,
+      "loss": 0.1956,
+      "step": 865
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.34550997614860535,
+      "learning_rate": 2.6934673366834175e-05,
+      "loss": 0.1586,
+      "step": 866
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3692917823791504,
+      "learning_rate": 2.6733668341708545e-05,
+      "loss": 0.2125,
+      "step": 867
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.48720014095306396,
+      "learning_rate": 2.6532663316582917e-05,
+      "loss": 0.2308,
+      "step": 868
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6313906908035278,
+      "learning_rate": 2.633165829145729e-05,
+      "loss": 0.2026,
+      "step": 869
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3886338770389557,
+      "learning_rate": 2.613065326633166e-05,
+      "loss": 0.2053,
+      "step": 870
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6648421883583069,
+      "learning_rate": 2.5929648241206033e-05,
+      "loss": 0.385,
+      "step": 871
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5134801864624023,
+      "learning_rate": 2.5728643216080406e-05,
+      "loss": 0.1886,
+      "step": 872
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40226277709007263,
+      "learning_rate": 2.5527638190954772e-05,
+      "loss": 0.1778,
+      "step": 873
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3925773799419403,
+      "learning_rate": 2.5326633165829145e-05,
+      "loss": 0.2397,
+      "step": 874
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4807821214199066,
+      "learning_rate": 2.5125628140703518e-05,
+      "loss": 0.1851,
+      "step": 875
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.47010698914527893,
+      "learning_rate": 2.492462311557789e-05,
+      "loss": 0.2635,
+      "step": 876
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.43276745080947876,
+      "learning_rate": 2.4723618090452264e-05,
+      "loss": 0.1813,
+      "step": 877
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4495302736759186,
+      "learning_rate": 2.4522613065326637e-05,
+      "loss": 0.2207,
+      "step": 878
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.30723971128463745,
+      "learning_rate": 2.4321608040201007e-05,
+      "loss": 0.2037,
+      "step": 879
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3359546661376953,
+      "learning_rate": 2.4120603015075376e-05,
+      "loss": 0.1761,
+      "step": 880
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9301133155822754,
+      "learning_rate": 2.391959798994975e-05,
+      "loss": 0.182,
+      "step": 881
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3476134240627289,
+      "learning_rate": 2.3718592964824122e-05,
+      "loss": 0.1796,
+      "step": 882
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.28115808963775635,
+      "learning_rate": 2.351758793969849e-05,
+      "loss": 0.1378,
+      "step": 883
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.46458545327186584,
+      "learning_rate": 2.3316582914572865e-05,
+      "loss": 0.2178,
+      "step": 884
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3719908595085144,
+      "learning_rate": 2.3115577889447238e-05,
+      "loss": 0.2225,
+      "step": 885
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4343881905078888,
+      "learning_rate": 2.291457286432161e-05,
+      "loss": 0.1423,
+      "step": 886
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3831157982349396,
+      "learning_rate": 2.271356783919598e-05,
+      "loss": 0.2008,
+      "step": 887
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3542693853378296,
+      "learning_rate": 2.2512562814070353e-05,
+      "loss": 0.239,
+      "step": 888
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3413107693195343,
+      "learning_rate": 2.2311557788944726e-05,
+      "loss": 0.1751,
+      "step": 889
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6149652004241943,
+      "learning_rate": 2.21105527638191e-05,
+      "loss": 0.239,
+      "step": 890
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.39706915616989136,
+      "learning_rate": 2.190954773869347e-05,
+      "loss": 0.189,
+      "step": 891
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4143030345439911,
+      "learning_rate": 2.1708542713567838e-05,
+      "loss": 0.2113,
+      "step": 892
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.46678248047828674,
+      "learning_rate": 2.150753768844221e-05,
+      "loss": 0.2433,
+      "step": 893
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5161128640174866,
+      "learning_rate": 2.1306532663316584e-05,
+      "loss": 0.2854,
+      "step": 894
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4151817560195923,
+      "learning_rate": 2.1105527638190957e-05,
+      "loss": 0.175,
+      "step": 895
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3842785954475403,
+      "learning_rate": 2.0904522613065327e-05,
+      "loss": 0.208,
+      "step": 896
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2840658724308014,
+      "learning_rate": 2.07035175879397e-05,
+      "loss": 0.1405,
+      "step": 897
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3122474253177643,
+      "learning_rate": 2.0502512562814073e-05,
+      "loss": 0.1946,
+      "step": 898
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.48670870065689087,
+      "learning_rate": 2.0301507537688446e-05,
+      "loss": 0.2515,
+      "step": 899
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3537992238998413,
+      "learning_rate": 2.0100502512562815e-05,
+      "loss": 0.1731,
+      "step": 900
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.894879937171936,
+      "learning_rate": 1.9899497487437188e-05,
+      "loss": 0.2873,
+      "step": 901
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.820382297039032,
+      "learning_rate": 1.9698492462311558e-05,
+      "loss": 0.2662,
+      "step": 902
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4541124999523163,
+      "learning_rate": 1.949748743718593e-05,
+      "loss": 0.2322,
+      "step": 903
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2983895540237427,
+      "learning_rate": 1.92964824120603e-05,
+      "loss": 0.126,
+      "step": 904
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2186942845582962,
+      "learning_rate": 1.9095477386934673e-05,
+      "loss": 0.1206,
+      "step": 905
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6614108085632324,
+      "learning_rate": 1.8894472361809046e-05,
+      "loss": 0.2525,
+      "step": 906
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6433589458465576,
+      "learning_rate": 1.869346733668342e-05,
+      "loss": 0.2856,
+      "step": 907
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3145541548728943,
+      "learning_rate": 1.849246231155779e-05,
+      "loss": 0.1602,
+      "step": 908
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4058758318424225,
+      "learning_rate": 1.829145728643216e-05,
+      "loss": 0.2108,
+      "step": 909
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.31698545813560486,
+      "learning_rate": 1.8090452261306535e-05,
+      "loss": 0.1923,
+      "step": 910
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2648496627807617,
+      "learning_rate": 1.7889447236180908e-05,
+      "loss": 0.1372,
+      "step": 911
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.38380950689315796,
+      "learning_rate": 1.7688442211055277e-05,
+      "loss": 0.1771,
+      "step": 912
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6659491658210754,
+      "learning_rate": 1.748743718592965e-05,
+      "loss": 0.2835,
+      "step": 913
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5000067353248596,
+      "learning_rate": 1.728643216080402e-05,
+      "loss": 0.205,
+      "step": 914
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2938293516635895,
+      "learning_rate": 1.7085427135678393e-05,
+      "loss": 0.1422,
+      "step": 915
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2897784113883972,
+      "learning_rate": 1.6884422110552762e-05,
+      "loss": 0.1297,
+      "step": 916
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3621378242969513,
+      "learning_rate": 1.6683417085427135e-05,
+      "loss": 0.1713,
+      "step": 917
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.25102704763412476,
+      "learning_rate": 1.6482412060301508e-05,
+      "loss": 0.112,
+      "step": 918
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.33787477016448975,
+      "learning_rate": 1.628140703517588e-05,
+      "loss": 0.1573,
+      "step": 919
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5255904197692871,
+      "learning_rate": 1.608040201005025e-05,
+      "loss": 0.2663,
+      "step": 920
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.43443411588668823,
+      "learning_rate": 1.5879396984924624e-05,
+      "loss": 0.1731,
+      "step": 921
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.371311753988266,
+      "learning_rate": 1.5678391959798997e-05,
+      "loss": 0.194,
+      "step": 922
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.46048715710639954,
+      "learning_rate": 1.547738693467337e-05,
+      "loss": 0.2677,
+      "step": 923
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4153316020965576,
+      "learning_rate": 1.527638190954774e-05,
+      "loss": 0.1633,
+      "step": 924
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4759661555290222,
+      "learning_rate": 1.507537688442211e-05,
+      "loss": 0.2009,
+      "step": 925
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.44403836131095886,
+      "learning_rate": 1.4874371859296482e-05,
+      "loss": 0.213,
+      "step": 926
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5492647886276245,
+      "learning_rate": 1.4673366834170855e-05,
+      "loss": 0.2041,
+      "step": 927
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.43692106008529663,
+      "learning_rate": 1.4472361809045226e-05,
+      "loss": 0.2253,
+      "step": 928
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.46681973338127136,
+      "learning_rate": 1.4271356783919599e-05,
+      "loss": 0.2476,
+      "step": 929
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2761835753917694,
+      "learning_rate": 1.407035175879397e-05,
+      "loss": 0.1159,
+      "step": 930
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.32756364345550537,
+      "learning_rate": 1.3869346733668343e-05,
+      "loss": 0.1649,
+      "step": 931
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26436007022857666,
+      "learning_rate": 1.3668341708542715e-05,
+      "loss": 0.1635,
+      "step": 932
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.40596601366996765,
+      "learning_rate": 1.3467336683417087e-05,
+      "loss": 0.2097,
+      "step": 933
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2804524898529053,
+      "learning_rate": 1.3266331658291459e-05,
+      "loss": 0.1472,
+      "step": 934
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3760283589363098,
+      "learning_rate": 1.306532663316583e-05,
+      "loss": 0.1813,
+      "step": 935
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.29557615518569946,
+      "learning_rate": 1.2864321608040203e-05,
+      "loss": 0.1399,
+      "step": 936
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6540164351463318,
+      "learning_rate": 1.2663316582914573e-05,
+      "loss": 0.2999,
+      "step": 937
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.44547465443611145,
+      "learning_rate": 1.2462311557788946e-05,
+      "loss": 0.1994,
+      "step": 938
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6938953995704651,
+      "learning_rate": 1.2261306532663318e-05,
+      "loss": 0.3192,
+      "step": 939
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7034733891487122,
+      "learning_rate": 1.2060301507537688e-05,
+      "loss": 0.352,
+      "step": 940
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6876631379127502,
+      "learning_rate": 1.1859296482412061e-05,
+      "loss": 0.313,
+      "step": 941
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4050236642360687,
+      "learning_rate": 1.1658291457286432e-05,
+      "loss": 0.2126,
+      "step": 942
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.24842841923236847,
+      "learning_rate": 1.1457286432160805e-05,
+      "loss": 0.1333,
+      "step": 943
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.38990554213523865,
+      "learning_rate": 1.1256281407035177e-05,
+      "loss": 0.2205,
+      "step": 944
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.34424301981925964,
+      "learning_rate": 1.105527638190955e-05,
+      "loss": 0.1987,
+      "step": 945
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3523971736431122,
+      "learning_rate": 1.0854271356783919e-05,
+      "loss": 0.1813,
+      "step": 946
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4130074083805084,
+      "learning_rate": 1.0653266331658292e-05,
+      "loss": 0.2409,
+      "step": 947
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.614470899105072,
+      "learning_rate": 1.0452261306532663e-05,
+      "loss": 0.275,
+      "step": 948
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3006427586078644,
+      "learning_rate": 1.0251256281407036e-05,
+      "loss": 0.1483,
+      "step": 949
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4929478168487549,
+      "learning_rate": 1.0050251256281408e-05,
+      "loss": 0.2328,
+      "step": 950
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4946010112762451,
+      "learning_rate": 9.849246231155779e-06,
+      "loss": 0.2307,
+      "step": 951
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.47001707553863525,
+      "learning_rate": 9.64824120603015e-06,
+      "loss": 0.2133,
+      "step": 952
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5728352069854736,
+      "learning_rate": 9.447236180904523e-06,
+      "loss": 0.2261,
+      "step": 953
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.45893171429634094,
+      "learning_rate": 9.246231155778894e-06,
+      "loss": 0.257,
+      "step": 954
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4369294345378876,
+      "learning_rate": 9.045226130653267e-06,
+      "loss": 0.2485,
+      "step": 955
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3858097493648529,
+      "learning_rate": 8.844221105527639e-06,
+      "loss": 0.1506,
+      "step": 956
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4247424304485321,
+      "learning_rate": 8.64321608040201e-06,
+      "loss": 0.1968,
+      "step": 957
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6413017511367798,
+      "learning_rate": 8.442211055276381e-06,
+      "loss": 0.3137,
+      "step": 958
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.47462546825408936,
+      "learning_rate": 8.241206030150754e-06,
+      "loss": 0.2323,
+      "step": 959
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4901590347290039,
+      "learning_rate": 8.040201005025125e-06,
+      "loss": 0.2167,
+      "step": 960
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.46188369393348694,
+      "learning_rate": 7.839195979899498e-06,
+      "loss": 0.2188,
+      "step": 961
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5853116512298584,
+      "learning_rate": 7.63819095477387e-06,
+      "loss": 0.2758,
+      "step": 962
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5120633840560913,
+      "learning_rate": 7.437185929648241e-06,
+      "loss": 0.2467,
+      "step": 963
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.525434672832489,
+      "learning_rate": 7.236180904522613e-06,
+      "loss": 0.2374,
+      "step": 964
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5119476914405823,
+      "learning_rate": 7.035175879396985e-06,
+      "loss": 0.2483,
+      "step": 965
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2736360728740692,
+      "learning_rate": 6.834170854271357e-06,
+      "loss": 0.1504,
+      "step": 966
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4709615111351013,
+      "learning_rate": 6.633165829145729e-06,
+      "loss": 0.2252,
+      "step": 967
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6076705455780029,
+      "learning_rate": 6.4321608040201015e-06,
+      "loss": 0.2818,
+      "step": 968
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.29698169231414795,
+      "learning_rate": 6.231155778894473e-06,
+      "loss": 0.1461,
+      "step": 969
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.26702263951301575,
+      "learning_rate": 6.030150753768844e-06,
+      "loss": 0.125,
+      "step": 970
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.43181735277175903,
+      "learning_rate": 5.829145728643216e-06,
+      "loss": 0.1723,
+      "step": 971
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3504771888256073,
+      "learning_rate": 5.628140703517588e-06,
+      "loss": 0.1807,
+      "step": 972
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4474789798259735,
+      "learning_rate": 5.4271356783919595e-06,
+      "loss": 0.2144,
+      "step": 973
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.29683947563171387,
+      "learning_rate": 5.226130653266332e-06,
+      "loss": 0.1741,
+      "step": 974
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.461755633354187,
+      "learning_rate": 5.025125628140704e-06,
+      "loss": 0.1881,
+      "step": 975
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.37622129917144775,
+      "learning_rate": 4.824120603015075e-06,
+      "loss": 0.2117,
+      "step": 976
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4308166801929474,
+      "learning_rate": 4.623115577889447e-06,
+      "loss": 0.2124,
+      "step": 977
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4249589741230011,
+      "learning_rate": 4.422110552763819e-06,
+      "loss": 0.1813,
+      "step": 978
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3398144245147705,
+      "learning_rate": 4.2211055276381906e-06,
+      "loss": 0.1889,
+      "step": 979
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3491975963115692,
+      "learning_rate": 4.020100502512563e-06,
+      "loss": 0.1784,
+      "step": 980
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3642324209213257,
+      "learning_rate": 3.819095477386935e-06,
+      "loss": 0.1502,
+      "step": 981
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.31112533807754517,
+      "learning_rate": 3.6180904522613065e-06,
+      "loss": 0.1326,
+      "step": 982
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4723191261291504,
+      "learning_rate": 3.4170854271356786e-06,
+      "loss": 0.2128,
+      "step": 983
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3740854561328888,
+      "learning_rate": 3.2160804020100507e-06,
+      "loss": 0.1296,
+      "step": 984
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4681090712547302,
+      "learning_rate": 3.015075376884422e-06,
+      "loss": 0.2435,
+      "step": 985
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4755851924419403,
+      "learning_rate": 2.814070351758794e-06,
+      "loss": 0.192,
+      "step": 986
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7064125537872314,
+      "learning_rate": 2.613065326633166e-06,
+      "loss": 0.3695,
+      "step": 987
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3578234314918518,
+      "learning_rate": 2.4120603015075375e-06,
+      "loss": 0.1774,
+      "step": 988
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4586274027824402,
+      "learning_rate": 2.2110552763819096e-06,
+      "loss": 0.1662,
+      "step": 989
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.9307977557182312,
+      "learning_rate": 2.0100502512562813e-06,
+      "loss": 0.3233,
+      "step": 990
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4612172842025757,
+      "learning_rate": 1.8090452261306533e-06,
+      "loss": 0.1946,
+      "step": 991
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.35452717542648315,
+      "learning_rate": 1.6080402010050254e-06,
+      "loss": 0.2148,
+      "step": 992
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.41886764764785767,
+      "learning_rate": 1.407035175879397e-06,
+      "loss": 0.2024,
+      "step": 993
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.36347076296806335,
+      "learning_rate": 1.2060301507537688e-06,
+      "loss": 0.2238,
+      "step": 994
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.30956777930259705,
+      "learning_rate": 1.0050251256281407e-06,
+      "loss": 0.1816,
+      "step": 995
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.29292166233062744,
+      "learning_rate": 8.040201005025127e-07,
+      "loss": 0.1565,
+      "step": 996
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5314776301383972,
+      "learning_rate": 6.030150753768844e-07,
+      "loss": 0.2801,
+      "step": 997
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5279101133346558,
+      "learning_rate": 4.0201005025125634e-07,
+      "loss": 0.2125,
+      "step": 998
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5519235730171204,
+      "learning_rate": 2.0100502512562817e-07,
+      "loss": 0.2095,
+      "step": 999
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4091325104236603,
+      "learning_rate": 0.0,
+      "loss": 0.2152,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 6.759802199059661e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}