diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,23665 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.9991099243435695,
+  "eval_steps": 500,
+  "global_step": 3370,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.998483180999756,
+      "learning_rate": 9.99999921600369e-06,
+      "loss": 3.4553,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.117033004760742,
+      "learning_rate": 9.999996864014995e-06,
+      "loss": 3.408,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.321349143981934,
+      "learning_rate": 9.99999294403466e-06,
+      "loss": 3.3582,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.385439872741699,
+      "learning_rate": 9.999987456063911e-06,
+      "loss": 3.3623,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.095900535583496,
+      "learning_rate": 9.999980400104472e-06,
+      "loss": 3.2493,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.9515767097473145,
+      "learning_rate": 9.999971776158553e-06,
+      "loss": 3.2612,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.227610111236572,
+      "learning_rate": 9.999961584228859e-06,
+      "loss": 3.2378,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.344577789306641,
+      "learning_rate": 9.999949824318586e-06,
+      "loss": 3.3015,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.722344875335693,
+      "learning_rate": 9.999936496431424e-06,
+      "loss": 3.0527,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.678772449493408,
+      "learning_rate": 9.99992160057155e-06,
+      "loss": 3.0421,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.044250011444092,
+      "learning_rate": 9.999905136743635e-06,
+      "loss": 3.0885,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.970749378204346,
+      "learning_rate": 9.999887104952844e-06,
+      "loss": 3.0309,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.793628215789795,
+      "learning_rate": 9.999867505204833e-06,
+      "loss": 3.0985,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.272780418395996,
+      "learning_rate": 9.999846337505742e-06,
+      "loss": 3.1106,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.186373710632324,
+      "learning_rate": 9.999823601862217e-06,
+      "loss": 3.0215,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.143390655517578,
+      "learning_rate": 9.999799298281385e-06,
+      "loss": 3.0563,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.6611127853393555,
+      "learning_rate": 9.999773426770864e-06,
+      "loss": 3.0693,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 6.581606864929199,
+      "learning_rate": 9.999745987338774e-06,
+      "loss": 3.0197,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.401041507720947,
+      "learning_rate": 9.999716979993714e-06,
+      "loss": 2.9102,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.319998741149902,
+      "learning_rate": 9.999686404744782e-06,
+      "loss": 2.7516,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.082314968109131,
+      "learning_rate": 9.99965426160157e-06,
+      "loss": 2.9609,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.889050006866455,
+      "learning_rate": 9.999620550574155e-06,
+      "loss": 2.9817,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.612086296081543,
+      "learning_rate": 9.999585271673106e-06,
+      "loss": 2.9325,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.3004326820373535,
+      "learning_rate": 9.999548424909492e-06,
+      "loss": 3.1249,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.294023513793945,
+      "learning_rate": 9.999510010294864e-06,
+      "loss": 2.7454,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 6.1827311515808105,
+      "learning_rate": 9.999470027841273e-06,
+      "loss": 2.8555,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.101302623748779,
+      "learning_rate": 9.999428477561253e-06,
+      "loss": 2.7815,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.572377681732178,
+      "learning_rate": 9.999385359467835e-06,
+      "loss": 2.7861,
+      "step": 28
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.560662746429443,
+      "learning_rate": 9.999340673574544e-06,
+      "loss": 2.5217,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.479100704193115,
+      "learning_rate": 9.999294419895389e-06,
+      "loss": 2.6327,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.259706974029541,
+      "learning_rate": 9.999246598444878e-06,
+      "loss": 2.7057,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.7917375564575195,
+      "learning_rate": 9.999197209238008e-06,
+      "loss": 2.809,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.3115434646606445,
+      "learning_rate": 9.999146252290264e-06,
+      "loss": 2.8267,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.73801326751709,
+      "learning_rate": 9.99909372761763e-06,
+      "loss": 2.9467,
+      "step": 34
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.860792636871338,
+      "learning_rate": 9.999039635236576e-06,
+      "loss": 2.7271,
+      "step": 35
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.7296319007873535,
+      "learning_rate": 9.998983975164064e-06,
+      "loss": 2.647,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.758646011352539,
+      "learning_rate": 9.99892674741755e-06,
+      "loss": 2.4287,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 9.516364097595215,
+      "learning_rate": 9.998867952014983e-06,
+      "loss": 2.4565,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 7.628922939300537,
+      "learning_rate": 9.998807588974798e-06,
+      "loss": 2.4196,
+      "step": 39
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.621899604797363,
+      "learning_rate": 9.998745658315924e-06,
+      "loss": 2.576,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.799045085906982,
+      "learning_rate": 9.998682160057784e-06,
+      "loss": 2.6646,
+      "step": 41
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.699931621551514,
+      "learning_rate": 9.998617094220294e-06,
+      "loss": 2.5334,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.873729228973389,
+      "learning_rate": 9.998550460823853e-06,
+      "loss": 2.4829,
+      "step": 43
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.7138261795043945,
+      "learning_rate": 9.99848225988936e-06,
+      "loss": 2.3673,
+      "step": 44
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.80642032623291,
+      "learning_rate": 9.998412491438201e-06,
+      "loss": 2.6323,
+      "step": 45
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.193628311157227,
+      "learning_rate": 9.998341155492258e-06,
+      "loss": 2.5356,
+      "step": 46
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 7.456587314605713,
+      "learning_rate": 9.998268252073901e-06,
+      "loss": 2.4061,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.375697135925293,
+      "learning_rate": 9.998193781205991e-06,
+      "loss": 2.294,
+      "step": 48
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.281881332397461,
+      "learning_rate": 9.998117742911883e-06,
+      "loss": 2.422,
+      "step": 49
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.819869041442871,
+      "learning_rate": 9.998040137215423e-06,
+      "loss": 2.37,
+      "step": 50
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.839913368225098,
+      "learning_rate": 9.997960964140946e-06,
+      "loss": 2.1717,
+      "step": 51
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.189290046691895,
+      "learning_rate": 9.997880223713283e-06,
+      "loss": 2.1404,
+      "step": 52
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 9.0689058303833,
+      "learning_rate": 9.997797915957754e-06,
+      "loss": 2.0891,
+      "step": 53
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.522842407226562,
+      "learning_rate": 9.997714040900167e-06,
+      "loss": 2.3077,
+      "step": 54
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.40389633178711,
+      "learning_rate": 9.99762859856683e-06,
+      "loss": 2.2429,
+      "step": 55
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.828639030456543,
+      "learning_rate": 9.997541588984534e-06,
+      "loss": 2.0804,
+      "step": 56
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.22755241394043,
+      "learning_rate": 9.997453012180566e-06,
+      "loss": 1.9986,
+      "step": 57
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 7.72681999206543,
+      "learning_rate": 9.997362868182706e-06,
+      "loss": 2.1589,
+      "step": 58
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.323179244995117,
+      "learning_rate": 9.99727115701922e-06,
+      "loss": 1.7307,
+      "step": 59
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 26.13371467590332,
+      "learning_rate": 9.99717787871887e-06,
+      "loss": 1.9167,
+      "step": 60
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.804581642150879,
+      "learning_rate": 9.997083033310905e-06,
+      "loss": 2.0497,
+      "step": 61
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 8.681260108947754,
+      "learning_rate": 9.996986620825071e-06,
+      "loss": 1.7433,
+      "step": 62
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.992922306060791,
+      "learning_rate": 9.996888641291605e-06,
+      "loss": 2.0109,
+      "step": 63
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 9.218756675720215,
+      "learning_rate": 9.996789094741229e-06,
+      "loss": 2.031,
+      "step": 64
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.618896961212158,
+      "learning_rate": 9.996687981205163e-06,
+      "loss": 1.7019,
+      "step": 65
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.921351909637451,
+      "learning_rate": 9.996585300715117e-06,
+      "loss": 1.672,
+      "step": 66
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.27910041809082,
+      "learning_rate": 9.996481053303287e-06,
+      "loss": 1.7383,
+      "step": 67
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.485714435577393,
+      "learning_rate": 9.99637523900237e-06,
+      "loss": 1.6599,
+      "step": 68
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.049826145172119,
+      "learning_rate": 9.996267857845546e-06,
+      "loss": 1.7378,
+      "step": 69
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.234002113342285,
+      "learning_rate": 9.99615890986649e-06,
+      "loss": 1.6104,
+      "step": 70
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.838518142700195,
+      "learning_rate": 9.996048395099369e-06,
+      "loss": 1.717,
+      "step": 71
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.925804138183594,
+      "learning_rate": 9.99593631357884e-06,
+      "loss": 1.5898,
+      "step": 72
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.905149459838867,
+      "learning_rate": 9.99582266534005e-06,
+      "loss": 1.7803,
+      "step": 73
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.42403793334961,
+      "learning_rate": 9.99570745041864e-06,
+      "loss": 1.6539,
+      "step": 74
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 10.385242462158203,
+      "learning_rate": 9.995590668850745e-06,
+      "loss": 1.5529,
+      "step": 75
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.211100578308105,
+      "learning_rate": 9.995472320672982e-06,
+      "loss": 1.6531,
+      "step": 76
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.415603637695312,
+      "learning_rate": 9.995352405922467e-06,
+      "loss": 1.6093,
+      "step": 77
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.058953285217285,
+      "learning_rate": 9.995230924636806e-06,
+      "loss": 1.4958,
+      "step": 78
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.037732124328613,
+      "learning_rate": 9.995107876854092e-06,
+      "loss": 1.4137,
+      "step": 79
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 9.352761268615723,
+      "learning_rate": 9.994983262612916e-06,
+      "loss": 1.517,
+      "step": 80
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 10.336080551147461,
+      "learning_rate": 9.994857081952354e-06,
+      "loss": 1.4686,
+      "step": 81
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 14.166356086730957,
+      "learning_rate": 9.994729334911981e-06,
+      "loss": 1.6135,
+      "step": 82
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 13.83187484741211,
+      "learning_rate": 9.994600021531853e-06,
+      "loss": 1.5985,
+      "step": 83
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 11.849742889404297,
+      "learning_rate": 9.994469141852525e-06,
+      "loss": 1.5944,
+      "step": 84
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 12.22191047668457,
+      "learning_rate": 9.994336695915041e-06,
+      "loss": 1.4375,
+      "step": 85
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 12.994050979614258,
+      "learning_rate": 9.994202683760933e-06,
+      "loss": 1.3686,
+      "step": 86
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 12.180079460144043,
+      "learning_rate": 9.994067105432232e-06,
+      "loss": 1.1665,
+      "step": 87
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 15.327380180358887,
+      "learning_rate": 9.99392996097145e-06,
+      "loss": 1.6539,
+      "step": 88
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 9.95577335357666,
+      "learning_rate": 9.993791250421599e-06,
+      "loss": 1.486,
+      "step": 89
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 11.751002311706543,
+      "learning_rate": 9.993650973826177e-06,
+      "loss": 1.2462,
+      "step": 90
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 20.796630859375,
+      "learning_rate": 9.993509131229173e-06,
+      "loss": 1.3681,
+      "step": 91
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 9.534221649169922,
+      "learning_rate": 9.99336572267507e-06,
+      "loss": 1.2109,
+      "step": 92
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 12.138165473937988,
+      "learning_rate": 9.993220748208843e-06,
+      "loss": 1.4657,
+      "step": 93
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 12.162806510925293,
+      "learning_rate": 9.993074207875953e-06,
+      "loss": 1.3509,
+      "step": 94
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 10.419145584106445,
+      "learning_rate": 9.992926101722355e-06,
+      "loss": 1.1257,
+      "step": 95
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 11.593511581420898,
+      "learning_rate": 9.992776429794495e-06,
+      "loss": 1.1489,
+      "step": 96
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 11.043091773986816,
+      "learning_rate": 9.992625192139312e-06,
+      "loss": 1.6359,
+      "step": 97
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 11.715983390808105,
+      "learning_rate": 9.99247238880423e-06,
+      "loss": 1.1387,
+      "step": 98
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 11.284956932067871,
+      "learning_rate": 9.992318019837171e-06,
+      "loss": 1.0889,
+      "step": 99
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 9.603680610656738,
+      "learning_rate": 9.992162085286543e-06,
+      "loss": 1.3152,
+      "step": 100
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 16.600419998168945,
+      "learning_rate": 9.99200458520125e-06,
+      "loss": 1.1983,
+      "step": 101
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 10.15400505065918,
+      "learning_rate": 9.991845519630679e-06,
+      "loss": 1.0162,
+      "step": 102
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 9.521133422851562,
+      "learning_rate": 9.991684888624717e-06,
+      "loss": 1.13,
+      "step": 103
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 11.79595947265625,
+      "learning_rate": 9.991522692233735e-06,
+      "loss": 1.3088,
+      "step": 104
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 8.877530097961426,
+      "learning_rate": 9.991358930508599e-06,
+      "loss": 1.2175,
+      "step": 105
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 10.289795875549316,
+      "learning_rate": 9.991193603500664e-06,
+      "loss": 1.2656,
+      "step": 106
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 11.021401405334473,
+      "learning_rate": 9.991026711261776e-06,
+      "loss": 1.1163,
+      "step": 107
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 10.933035850524902,
+      "learning_rate": 9.990858253844273e-06,
+      "loss": 1.088,
+      "step": 108
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 9.310973167419434,
+      "learning_rate": 9.99068823130098e-06,
+      "loss": 1.1521,
+      "step": 109
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 12.622461318969727,
+      "learning_rate": 9.990516643685222e-06,
+      "loss": 1.2251,
+      "step": 110
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 8.76944351196289,
+      "learning_rate": 9.990343491050803e-06,
+      "loss": 1.1529,
+      "step": 111
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 10.389765739440918,
+      "learning_rate": 9.990168773452024e-06,
+      "loss": 1.2113,
+      "step": 112
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 10.373367309570312,
+      "learning_rate": 9.989992490943678e-06,
+      "loss": 1.1699,
+      "step": 113
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 14.425374984741211,
+      "learning_rate": 9.989814643581048e-06,
+      "loss": 1.2659,
+      "step": 114
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 10.276859283447266,
+      "learning_rate": 9.989635231419903e-06,
+      "loss": 1.2199,
+      "step": 115
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 12.483972549438477,
+      "learning_rate": 9.989454254516511e-06,
+      "loss": 1.0677,
+      "step": 116
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 8.567665100097656,
+      "learning_rate": 9.989271712927621e-06,
+      "loss": 0.9301,
+      "step": 117
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 16.923992156982422,
+      "learning_rate": 9.989087606710481e-06,
+      "loss": 1.113,
+      "step": 118
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 11.599892616271973,
+      "learning_rate": 9.988901935922826e-06,
+      "loss": 1.0904,
+      "step": 119
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 10.569705963134766,
+      "learning_rate": 9.988714700622882e-06,
+      "loss": 0.9523,
+      "step": 120
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 9.551050186157227,
+      "learning_rate": 9.988525900869366e-06,
+      "loss": 1.1525,
+      "step": 121
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 11.35373306274414,
+      "learning_rate": 9.988335536721483e-06,
+      "loss": 1.0901,
+      "step": 122
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 17.783721923828125,
+      "learning_rate": 9.988143608238934e-06,
+      "loss": 1.1597,
+      "step": 123
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 9.60888385772705,
+      "learning_rate": 9.987950115481905e-06,
+      "loss": 1.3822,
+      "step": 124
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 12.48217487335205,
+      "learning_rate": 9.987755058511079e-06,
+      "loss": 0.9674,
+      "step": 125
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 10.974038124084473,
+      "learning_rate": 9.987558437387622e-06,
+      "loss": 1.0471,
+      "step": 126
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 9.559907913208008,
+      "learning_rate": 9.987360252173193e-06,
+      "loss": 1.0244,
+      "step": 127
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 9.782307624816895,
+      "learning_rate": 9.987160502929946e-06,
+      "loss": 1.038,
+      "step": 128
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.9509172439575195,
+      "learning_rate": 9.98695918972052e-06,
+      "loss": 0.7776,
+      "step": 129
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 10.83304500579834,
+      "learning_rate": 9.986756312608048e-06,
+      "loss": 0.8654,
+      "step": 130
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.926746368408203,
+      "learning_rate": 9.986551871656152e-06,
+      "loss": 0.9711,
+      "step": 131
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 8.738396644592285,
+      "learning_rate": 9.98634586692894e-06,
+      "loss": 0.9712,
+      "step": 132
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 8.782489776611328,
+      "learning_rate": 9.986138298491022e-06,
+      "loss": 0.8712,
+      "step": 133
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 12.451414108276367,
+      "learning_rate": 9.985929166407486e-06,
+      "loss": 1.1765,
+      "step": 134
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 11.359150886535645,
+      "learning_rate": 9.985718470743916e-06,
+      "loss": 0.9356,
+      "step": 135
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 10.417428016662598,
+      "learning_rate": 9.985506211566388e-06,
+      "loss": 0.9794,
+      "step": 136
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 9.114100456237793,
+      "learning_rate": 9.985292388941465e-06,
+      "loss": 0.9145,
+      "step": 137
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 14.217204093933105,
+      "learning_rate": 9.985077002936202e-06,
+      "loss": 0.8567,
+      "step": 138
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 8.123570442199707,
+      "learning_rate": 9.984860053618141e-06,
+      "loss": 0.8589,
+      "step": 139
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 9.934085845947266,
+      "learning_rate": 9.98464154105532e-06,
+      "loss": 0.9975,
+      "step": 140
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 10.983957290649414,
+      "learning_rate": 9.984421465316263e-06,
+      "loss": 0.9911,
+      "step": 141
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 10.371049880981445,
+      "learning_rate": 9.984199826469986e-06,
+      "loss": 1.0466,
+      "step": 142
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 9.365143775939941,
+      "learning_rate": 9.983976624585996e-06,
+      "loss": 0.6556,
+      "step": 143
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 10.723000526428223,
+      "learning_rate": 9.983751859734286e-06,
+      "loss": 1.1202,
+      "step": 144
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 9.906790733337402,
+      "learning_rate": 9.983525531985343e-06,
+      "loss": 0.8182,
+      "step": 145
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 13.778162002563477,
+      "learning_rate": 9.983297641410143e-06,
+      "loss": 0.8171,
+      "step": 146
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 12.216089248657227,
+      "learning_rate": 9.983068188080154e-06,
+      "loss": 0.7732,
+      "step": 147
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 11.112737655639648,
+      "learning_rate": 9.982837172067328e-06,
+      "loss": 0.9184,
+      "step": 148
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 9.897744178771973,
+      "learning_rate": 9.982604593444116e-06,
+      "loss": 0.714,
+      "step": 149
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 9.966035842895508,
+      "learning_rate": 9.982370452283451e-06,
+      "loss": 0.9946,
+      "step": 150
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 15.138193130493164,
+      "learning_rate": 9.982134748658761e-06,
+      "loss": 1.2197,
+      "step": 151
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 12.098490715026855,
+      "learning_rate": 9.981897482643961e-06,
+      "loss": 0.8251,
+      "step": 152
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 12.232954978942871,
+      "learning_rate": 9.981658654313458e-06,
+      "loss": 0.6857,
+      "step": 153
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 12.435187339782715,
+      "learning_rate": 9.981418263742148e-06,
+      "loss": 0.9909,
+      "step": 154
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 16.95964813232422,
+      "learning_rate": 9.981176311005419e-06,
+      "loss": 0.941,
+      "step": 155
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 10.133909225463867,
+      "learning_rate": 9.980932796179143e-06,
+      "loss": 0.7963,
+      "step": 156
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 10.93771743774414,
+      "learning_rate": 9.98068771933969e-06,
+      "loss": 1.1203,
+      "step": 157
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 11.151179313659668,
+      "learning_rate": 9.980441080563915e-06,
+      "loss": 1.028,
+      "step": 158
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 10.619705200195312,
+      "learning_rate": 9.98019287992916e-06,
+      "loss": 0.7607,
+      "step": 159
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 10.821515083312988,
+      "learning_rate": 9.979943117513265e-06,
+      "loss": 0.8384,
+      "step": 160
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 10.00261116027832,
+      "learning_rate": 9.979691793394551e-06,
+      "loss": 0.8587,
+      "step": 161
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 17.43947982788086,
+      "learning_rate": 9.979438907651836e-06,
+      "loss": 1.2322,
+      "step": 162
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 11.888916015625,
+      "learning_rate": 9.979184460364423e-06,
+      "loss": 0.7515,
+      "step": 163
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 13.577738761901855,
+      "learning_rate": 9.97892845161211e-06,
+      "loss": 0.9162,
+      "step": 164
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 13.197647094726562,
+      "learning_rate": 9.978670881475173e-06,
+      "loss": 0.7018,
+      "step": 165
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 12.1685209274292,
+      "learning_rate": 9.978411750034394e-06,
+      "loss": 1.0033,
+      "step": 166
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 10.492837905883789,
+      "learning_rate": 9.97815105737103e-06,
+      "loss": 0.7894,
+      "step": 167
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 12.857146263122559,
+      "learning_rate": 9.977888803566837e-06,
+      "loss": 0.8061,
+      "step": 168
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 11.049395561218262,
+      "learning_rate": 9.977624988704058e-06,
+      "loss": 0.5457,
+      "step": 169
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 12.154472351074219,
+      "learning_rate": 9.977359612865424e-06,
+      "loss": 0.7394,
+      "step": 170
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 12.737568855285645,
+      "learning_rate": 9.977092676134156e-06,
+      "loss": 0.7695,
+      "step": 171
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 11.109391212463379,
+      "learning_rate": 9.976824178593964e-06,
+      "loss": 0.8161,
+      "step": 172
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 15.20705509185791,
+      "learning_rate": 9.976554120329051e-06,
+      "loss": 0.8499,
+      "step": 173
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 9.756624221801758,
+      "learning_rate": 9.976282501424106e-06,
+      "loss": 0.9027,
+      "step": 174
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 11.121670722961426,
+      "learning_rate": 9.976009321964306e-06,
+      "loss": 0.8642,
+      "step": 175
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 9.139627456665039,
+      "learning_rate": 9.975734582035323e-06,
+      "loss": 0.7916,
+      "step": 176
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 13.86678409576416,
+      "learning_rate": 9.975458281723314e-06,
+      "loss": 0.9667,
+      "step": 177
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 10.30676555633545,
+      "learning_rate": 9.975180421114924e-06,
+      "loss": 0.8639,
+      "step": 178
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 11.133285522460938,
+      "learning_rate": 9.974901000297293e-06,
+      "loss": 0.6169,
+      "step": 179
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 9.716639518737793,
+      "learning_rate": 9.974620019358046e-06,
+      "loss": 0.6931,
+      "step": 180
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 11.775928497314453,
+      "learning_rate": 9.974337478385296e-06,
+      "loss": 0.8292,
+      "step": 181
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 15.081896781921387,
+      "learning_rate": 9.97405337746765e-06,
+      "loss": 0.8854,
+      "step": 182
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 10.911590576171875,
+      "learning_rate": 9.973767716694201e-06,
+      "loss": 1.041,
+      "step": 183
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 12.45612907409668,
+      "learning_rate": 9.973480496154533e-06,
+      "loss": 0.7548,
+      "step": 184
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 11.014619827270508,
+      "learning_rate": 9.973191715938715e-06,
+      "loss": 0.7244,
+      "step": 185
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 11.48668384552002,
+      "learning_rate": 9.97290137613731e-06,
+      "loss": 0.8406,
+      "step": 186
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 11.985112190246582,
+      "learning_rate": 9.972609476841368e-06,
+      "loss": 0.9085,
+      "step": 187
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 12.708012580871582,
+      "learning_rate": 9.972316018142427e-06,
+      "loss": 0.855,
+      "step": 188
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 12.260651588439941,
+      "learning_rate": 9.972021000132516e-06,
+      "loss": 0.7806,
+      "step": 189
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 14.270858764648438,
+      "learning_rate": 9.971724422904154e-06,
+      "loss": 0.9596,
+      "step": 190
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 10.415160179138184,
+      "learning_rate": 9.971426286550345e-06,
+      "loss": 0.8398,
+      "step": 191
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 14.872221946716309,
+      "learning_rate": 9.971126591164583e-06,
+      "loss": 0.8252,
+      "step": 192
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 14.003769874572754,
+      "learning_rate": 9.970825336840857e-06,
+      "loss": 0.7967,
+      "step": 193
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 11.029498100280762,
+      "learning_rate": 9.970522523673634e-06,
+      "loss": 0.63,
+      "step": 194
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 12.037678718566895,
+      "learning_rate": 9.97021815175788e-06,
+      "loss": 0.687,
+      "step": 195
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 9.419631004333496,
+      "learning_rate": 9.969912221189043e-06,
+      "loss": 0.5312,
+      "step": 196
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 12.24401569366455,
+      "learning_rate": 9.969604732063062e-06,
+      "loss": 0.8819,
+      "step": 197
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 12.276981353759766,
+      "learning_rate": 9.96929568447637e-06,
+      "loss": 0.8092,
+      "step": 198
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 13.441302299499512,
+      "learning_rate": 9.968985078525876e-06,
+      "loss": 0.772,
+      "step": 199
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 17.710548400878906,
+      "learning_rate": 9.968672914308995e-06,
+      "loss": 0.6268,
+      "step": 200
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 10.37500286102295,
+      "learning_rate": 9.968359191923612e-06,
+      "loss": 0.8091,
+      "step": 201
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 13.829612731933594,
+      "learning_rate": 9.968043911468115e-06,
+      "loss": 0.7759,
+      "step": 202
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 10.291707038879395,
+      "learning_rate": 9.967727073041376e-06,
+      "loss": 0.8665,
+      "step": 203
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 12.719216346740723,
+      "learning_rate": 9.96740867674275e-06,
+      "loss": 0.7747,
+      "step": 204
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 10.045609474182129,
+      "learning_rate": 9.967088722672094e-06,
+      "loss": 0.6309,
+      "step": 205
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 9.315113067626953,
+      "learning_rate": 9.966767210929738e-06,
+      "loss": 0.837,
+      "step": 206
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 11.884369850158691,
+      "learning_rate": 9.96644414161651e-06,
+      "loss": 0.7996,
+      "step": 207
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 8.947154998779297,
+      "learning_rate": 9.966119514833723e-06,
+      "loss": 0.7034,
+      "step": 208
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 10.427030563354492,
+      "learning_rate": 9.965793330683182e-06,
+      "loss": 0.8521,
+      "step": 209
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 11.855607032775879,
+      "learning_rate": 9.965465589267176e-06,
+      "loss": 0.9155,
+      "step": 210
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 13.313431739807129,
+      "learning_rate": 9.965136290688484e-06,
+      "loss": 0.851,
+      "step": 211
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 9.773415565490723,
+      "learning_rate": 9.964805435050374e-06,
+      "loss": 0.9887,
+      "step": 212
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 12.979085922241211,
+      "learning_rate": 9.964473022456602e-06,
+      "loss": 0.8102,
+      "step": 213
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 10.189248085021973,
+      "learning_rate": 9.964139053011413e-06,
+      "loss": 0.6867,
+      "step": 214
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 28.07099151611328,
+      "learning_rate": 9.963803526819537e-06,
+      "loss": 0.9846,
+      "step": 215
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 13.008727073669434,
+      "learning_rate": 9.963466443986198e-06,
+      "loss": 0.7857,
+      "step": 216
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 10.71626091003418,
+      "learning_rate": 9.963127804617102e-06,
+      "loss": 0.7243,
+      "step": 217
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 11.85562515258789,
+      "learning_rate": 9.962787608818444e-06,
+      "loss": 0.8522,
+      "step": 218
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 10.415912628173828,
+      "learning_rate": 9.962445856696912e-06,
+      "loss": 0.7506,
+      "step": 219
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 13.334424018859863,
+      "learning_rate": 9.96210254835968e-06,
+      "loss": 0.7621,
+      "step": 220
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 11.701071739196777,
+      "learning_rate": 9.961757683914406e-06,
+      "loss": 0.9153,
+      "step": 221
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 13.26212215423584,
+      "learning_rate": 9.96141126346924e-06,
+      "loss": 0.7741,
+      "step": 222
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 11.683025360107422,
+      "learning_rate": 9.96106328713282e-06,
+      "loss": 0.6063,
+      "step": 223
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 8.18270206451416,
+      "learning_rate": 9.96071375501427e-06,
+      "loss": 0.5223,
+      "step": 224
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 12.783717155456543,
+      "learning_rate": 9.960362667223202e-06,
+      "loss": 1.1241,
+      "step": 225
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 9.316211700439453,
+      "learning_rate": 9.960010023869717e-06,
+      "loss": 0.7491,
+      "step": 226
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 12.756223678588867,
+      "learning_rate": 9.959655825064405e-06,
+      "loss": 0.928,
+      "step": 227
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 10.876447677612305,
+      "learning_rate": 9.959300070918341e-06,
+      "loss": 0.6095,
+      "step": 228
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.959899425506592,
+      "learning_rate": 9.95894276154309e-06,
+      "loss": 0.5664,
+      "step": 229
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 16.20490837097168,
+      "learning_rate": 9.9585838970507e-06,
+      "loss": 1.1435,
+      "step": 230
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 12.675557136535645,
+      "learning_rate": 9.958223477553715e-06,
+      "loss": 1.0454,
+      "step": 231
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 9.572710990905762,
+      "learning_rate": 9.957861503165158e-06,
+      "loss": 0.5651,
+      "step": 232
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 10.701699256896973,
+      "learning_rate": 9.957497973998549e-06,
+      "loss": 0.5692,
+      "step": 233
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 9.034031867980957,
+      "learning_rate": 9.957132890167884e-06,
+      "loss": 0.6909,
+      "step": 234
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 13.133191108703613,
+      "learning_rate": 9.956766251787657e-06,
+      "loss": 0.7923,
+      "step": 235
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 12.245318412780762,
+      "learning_rate": 9.956398058972844e-06,
+      "loss": 0.7072,
+      "step": 236
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 12.623373985290527,
+      "learning_rate": 9.95602831183891e-06,
+      "loss": 0.6772,
+      "step": 237
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 13.936652183532715,
+      "learning_rate": 9.955657010501807e-06,
+      "loss": 0.8354,
+      "step": 238
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 10.412864685058594,
+      "learning_rate": 9.955284155077974e-06,
+      "loss": 0.7798,
+      "step": 239
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 14.492460250854492,
+      "learning_rate": 9.954909745684339e-06,
+      "loss": 0.6906,
+      "step": 240
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 12.190313339233398,
+      "learning_rate": 9.954533782438314e-06,
+      "loss": 0.7468,
+      "step": 241
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 11.922014236450195,
+      "learning_rate": 9.954156265457801e-06,
+      "loss": 0.8672,
+      "step": 242
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 12.213216781616211,
+      "learning_rate": 9.953777194861193e-06,
+      "loss": 0.7113,
+      "step": 243
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 14.39379596710205,
+      "learning_rate": 9.953396570767362e-06,
+      "loss": 0.6564,
+      "step": 244
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 11.251302719116211,
+      "learning_rate": 9.95301439329567e-06,
+      "loss": 0.6515,
+      "step": 245
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 12.369184494018555,
+      "learning_rate": 9.95263066256597e-06,
+      "loss": 0.6687,
+      "step": 246
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 12.323153495788574,
+      "learning_rate": 9.9522453786986e-06,
+      "loss": 0.7793,
+      "step": 247
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 8.590582847595215,
+      "learning_rate": 9.951858541814383e-06,
+      "loss": 0.5628,
+      "step": 248
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 11.145419120788574,
+      "learning_rate": 9.951470152034629e-06,
+      "loss": 0.5389,
+      "step": 249
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 11.325778007507324,
+      "learning_rate": 9.951080209481138e-06,
+      "loss": 0.8552,
+      "step": 250
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 8.35358715057373,
+      "learning_rate": 9.950688714276198e-06,
+      "loss": 0.6837,
+      "step": 251
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 14.441850662231445,
+      "learning_rate": 9.950295666542576e-06,
+      "loss": 0.6671,
+      "step": 252
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 11.948869705200195,
+      "learning_rate": 9.949901066403536e-06,
+      "loss": 0.7998,
+      "step": 253
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 13.227315902709961,
+      "learning_rate": 9.949504913982822e-06,
+      "loss": 0.6869,
+      "step": 254
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 18.817630767822266,
+      "learning_rate": 9.949107209404664e-06,
+      "loss": 0.6608,
+      "step": 255
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 14.170276641845703,
+      "learning_rate": 9.948707952793787e-06,
+      "loss": 0.8439,
+      "step": 256
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 9.641496658325195,
+      "learning_rate": 9.948307144275393e-06,
+      "loss": 0.7697,
+      "step": 257
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 12.959053039550781,
+      "learning_rate": 9.947904783975177e-06,
+      "loss": 0.6831,
+      "step": 258
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 10.54731559753418,
+      "learning_rate": 9.94750087201932e-06,
+      "loss": 0.57,
+      "step": 259
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 11.34838581085205,
+      "learning_rate": 9.947095408534483e-06,
+      "loss": 0.6339,
+      "step": 260
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 9.583577156066895,
+      "learning_rate": 9.946688393647825e-06,
+      "loss": 0.7819,
+      "step": 261
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 10.245312690734863,
+      "learning_rate": 9.946279827486982e-06,
+      "loss": 0.7041,
+      "step": 262
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 11.53131103515625,
+      "learning_rate": 9.94586971018008e-06,
+      "loss": 0.881,
+      "step": 263
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 8.526832580566406,
+      "learning_rate": 9.945458041855732e-06,
+      "loss": 0.4542,
+      "step": 264
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 9.491833686828613,
+      "learning_rate": 9.945044822643033e-06,
+      "loss": 0.682,
+      "step": 265
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 7.513880729675293,
+      "learning_rate": 9.944630052671573e-06,
+      "loss": 0.4408,
+      "step": 266
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 9.057185173034668,
+      "learning_rate": 9.94421373207142e-06,
+      "loss": 0.5609,
+      "step": 267
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 12.214913368225098,
+      "learning_rate": 9.943795860973135e-06,
+      "loss": 0.8278,
+      "step": 268
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 12.763900756835938,
+      "learning_rate": 9.943376439507757e-06,
+      "loss": 0.7878,
+      "step": 269
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 10.350016593933105,
+      "learning_rate": 9.94295546780682e-06,
+      "loss": 0.7557,
+      "step": 270
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 10.582222938537598,
+      "learning_rate": 9.942532946002338e-06,
+      "loss": 0.6368,
+      "step": 271
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 14.16856575012207,
+      "learning_rate": 9.942108874226812e-06,
+      "loss": 0.5797,
+      "step": 272
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 10.295936584472656,
+      "learning_rate": 9.941683252613235e-06,
+      "loss": 0.7154,
+      "step": 273
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 10.523819923400879,
+      "learning_rate": 9.941256081295077e-06,
+      "loss": 0.6309,
+      "step": 274
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 12.695225715637207,
+      "learning_rate": 9.940827360406297e-06,
+      "loss": 0.7226,
+      "step": 275
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 19.4698429107666,
+      "learning_rate": 9.940397090081346e-06,
+      "loss": 0.8634,
+      "step": 276
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 21.5108642578125,
+      "learning_rate": 9.939965270455154e-06,
+      "loss": 0.7936,
+      "step": 277
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 11.360212326049805,
+      "learning_rate": 9.93953190166314e-06,
+      "loss": 0.6152,
+      "step": 278
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 12.101655006408691,
+      "learning_rate": 9.939096983841206e-06,
+      "loss": 0.7098,
+      "step": 279
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 12.0094633102417,
+      "learning_rate": 9.93866051712574e-06,
+      "loss": 0.454,
+      "step": 280
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 9.73945426940918,
+      "learning_rate": 9.938222501653622e-06,
+      "loss": 0.5169,
+      "step": 281
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 12.925312995910645,
+      "learning_rate": 9.93778293756221e-06,
+      "loss": 0.681,
+      "step": 282
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 11.55838680267334,
+      "learning_rate": 9.937341824989351e-06,
+      "loss": 0.6995,
+      "step": 283
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 15.32224178314209,
+      "learning_rate": 9.936899164073379e-06,
+      "loss": 0.894,
+      "step": 284
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 15.470325469970703,
+      "learning_rate": 9.936454954953108e-06,
+      "loss": 0.6718,
+      "step": 285
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 10.36946964263916,
+      "learning_rate": 9.936009197767847e-06,
+      "loss": 0.5171,
+      "step": 286
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 9.612388610839844,
+      "learning_rate": 9.935561892657378e-06,
+      "loss": 0.7533,
+      "step": 287
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 13.342158317565918,
+      "learning_rate": 9.935113039761981e-06,
+      "loss": 0.6849,
+      "step": 288
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 10.606959342956543,
+      "learning_rate": 9.934662639222412e-06,
+      "loss": 0.5407,
+      "step": 289
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 11.100262641906738,
+      "learning_rate": 9.934210691179918e-06,
+      "loss": 0.6904,
+      "step": 290
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 12.644224166870117,
+      "learning_rate": 9.933757195776228e-06,
+      "loss": 0.7679,
+      "step": 291
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 12.337888717651367,
+      "learning_rate": 9.93330215315356e-06,
+      "loss": 0.7587,
+      "step": 292
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 12.88183879852295,
+      "learning_rate": 9.93284556345461e-06,
+      "loss": 0.6961,
+      "step": 293
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 15.309942245483398,
+      "learning_rate": 9.932387426822568e-06,
+      "loss": 0.8338,
+      "step": 294
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 11.499650955200195,
+      "learning_rate": 9.931927743401102e-06,
+      "loss": 0.8564,
+      "step": 295
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 12.307567596435547,
+      "learning_rate": 9.931466513334372e-06,
+      "loss": 0.7564,
+      "step": 296
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 10.340993881225586,
+      "learning_rate": 9.931003736767013e-06,
+      "loss": 0.5756,
+      "step": 297
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 10.656303405761719,
+      "learning_rate": 9.930539413844158e-06,
+      "loss": 0.6316,
+      "step": 298
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 10.219998359680176,
+      "learning_rate": 9.930073544711414e-06,
+      "loss": 0.6506,
+      "step": 299
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 11.718817710876465,
+      "learning_rate": 9.929606129514875e-06,
+      "loss": 0.74,
+      "step": 300
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 20.799468994140625,
+      "learning_rate": 9.929137168401127e-06,
+      "loss": 0.6129,
+      "step": 301
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 10.624800682067871,
+      "learning_rate": 9.928666661517232e-06,
+      "loss": 0.6487,
+      "step": 302
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 11.067184448242188,
+      "learning_rate": 9.928194609010742e-06,
+      "loss": 0.7472,
+      "step": 303
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 11.581793785095215,
+      "learning_rate": 9.92772101102969e-06,
+      "loss": 0.7927,
+      "step": 304
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 10.539834976196289,
+      "learning_rate": 9.927245867722596e-06,
+      "loss": 0.5098,
+      "step": 305
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 10.228087425231934,
+      "learning_rate": 9.926769179238467e-06,
+      "loss": 0.7291,
+      "step": 306
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 9.015642166137695,
+      "learning_rate": 9.926290945726786e-06,
+      "loss": 0.5307,
+      "step": 307
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 11.048788070678711,
+      "learning_rate": 9.925811167337533e-06,
+      "loss": 0.7233,
+      "step": 308
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 11.979606628417969,
+      "learning_rate": 9.925329844221163e-06,
+      "loss": 0.6876,
+      "step": 309
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 9.841987609863281,
+      "learning_rate": 9.924846976528618e-06,
+      "loss": 0.5381,
+      "step": 310
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 8.488785743713379,
+      "learning_rate": 9.924362564411321e-06,
+      "loss": 0.6779,
+      "step": 311
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 10.88672161102295,
+      "learning_rate": 9.923876608021191e-06,
+      "loss": 0.5152,
+      "step": 312
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 11.39985179901123,
+      "learning_rate": 9.923389107510616e-06,
+      "loss": 0.5701,
+      "step": 313
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 9.675251007080078,
+      "learning_rate": 9.92290006303248e-06,
+      "loss": 0.5302,
+      "step": 314
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 9.422196388244629,
+      "learning_rate": 9.922409474740142e-06,
+      "loss": 0.6425,
+      "step": 315
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 9.374537467956543,
+      "learning_rate": 9.921917342787455e-06,
+      "loss": 0.6549,
+      "step": 316
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 17.770401000976562,
+      "learning_rate": 9.921423667328749e-06,
+      "loss": 0.8845,
+      "step": 317
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 10.37729263305664,
+      "learning_rate": 9.920928448518837e-06,
+      "loss": 0.6182,
+      "step": 318
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 13.116642951965332,
+      "learning_rate": 9.920431686513023e-06,
+      "loss": 0.7849,
+      "step": 319
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 9.81547737121582,
+      "learning_rate": 9.919933381467088e-06,
+      "loss": 0.5847,
+      "step": 320
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 10.248363494873047,
+      "learning_rate": 9.9194335335373e-06,
+      "loss": 0.5394,
+      "step": 321
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 11.28476619720459,
+      "learning_rate": 9.918932142880414e-06,
+      "loss": 0.7492,
+      "step": 322
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 15.136175155639648,
+      "learning_rate": 9.918429209653662e-06,
+      "loss": 0.8457,
+      "step": 323
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 10.415621757507324,
+      "learning_rate": 9.917924734014762e-06,
+      "loss": 0.799,
+      "step": 324
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 9.711078643798828,
+      "learning_rate": 9.91741871612192e-06,
+      "loss": 0.546,
+      "step": 325
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 8.69312572479248,
+      "learning_rate": 9.916911156133822e-06,
+      "loss": 0.6325,
+      "step": 326
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 22.05565071105957,
+      "learning_rate": 9.916402054209635e-06,
+      "loss": 0.5789,
+      "step": 327
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 11.203454971313477,
+      "learning_rate": 9.915891410509015e-06,
+      "loss": 0.6598,
+      "step": 328
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 12.977240562438965,
+      "learning_rate": 9.9153792251921e-06,
+      "loss": 0.5412,
+      "step": 329
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 9.482515335083008,
+      "learning_rate": 9.91486549841951e-06,
+      "loss": 0.5408,
+      "step": 330
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 8.682055473327637,
+      "learning_rate": 9.914350230352347e-06,
+      "loss": 0.5951,
+      "step": 331
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 8.145601272583008,
+      "learning_rate": 9.913833421152202e-06,
+      "loss": 0.4326,
+      "step": 332
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 8.323835372924805,
+      "learning_rate": 9.91331507098114e-06,
+      "loss": 0.3722,
+      "step": 333
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 11.824346542358398,
+      "learning_rate": 9.91279518000172e-06,
+      "loss": 0.7692,
+      "step": 334
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 8.8349609375,
+      "learning_rate": 9.912273748376976e-06,
+      "loss": 0.5369,
+      "step": 335
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 9.979643821716309,
+      "learning_rate": 9.911750776270433e-06,
+      "loss": 0.7026,
+      "step": 336
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 15.253710746765137,
+      "learning_rate": 9.911226263846086e-06,
+      "loss": 0.6657,
+      "step": 337
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 13.147467613220215,
+      "learning_rate": 9.91070021126843e-06,
+      "loss": 0.7037,
+      "step": 338
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 11.478538513183594,
+      "learning_rate": 9.910172618702427e-06,
+      "loss": 0.6179,
+      "step": 339
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 8.337403297424316,
+      "learning_rate": 9.909643486313533e-06,
+      "loss": 0.5105,
+      "step": 340
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 11.949843406677246,
+      "learning_rate": 9.909112814267686e-06,
+      "loss": 0.6893,
+      "step": 341
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 8.313594818115234,
+      "learning_rate": 9.908580602731297e-06,
+      "loss": 0.4339,
+      "step": 342
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 10.916149139404297,
+      "learning_rate": 9.908046851871272e-06,
+      "loss": 0.6769,
+      "step": 343
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 8.36259937286377,
+      "learning_rate": 9.907511561854994e-06,
+      "loss": 0.4293,
+      "step": 344
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 12.71483325958252,
+      "learning_rate": 9.906974732850327e-06,
+      "loss": 0.7968,
+      "step": 345
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 12.12741756439209,
+      "learning_rate": 9.90643636502562e-06,
+      "loss": 0.6241,
+      "step": 346
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 9.419227600097656,
+      "learning_rate": 9.905896458549706e-06,
+      "loss": 0.6176,
+      "step": 347
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 10.552946090698242,
+      "learning_rate": 9.905355013591898e-06,
+      "loss": 0.5052,
+      "step": 348
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 11.505782127380371,
+      "learning_rate": 9.904812030321992e-06,
+      "loss": 0.6718,
+      "step": 349
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 12.304089546203613,
+      "learning_rate": 9.904267508910269e-06,
+      "loss": 0.7981,
+      "step": 350
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 11.282035827636719,
+      "learning_rate": 9.903721449527485e-06,
+      "loss": 0.5435,
+      "step": 351
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 10.82486629486084,
+      "learning_rate": 9.903173852344889e-06,
+      "loss": 0.5752,
+      "step": 352
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 10.979615211486816,
+      "learning_rate": 9.902624717534205e-06,
+      "loss": 0.5758,
+      "step": 353
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 9.440935134887695,
+      "learning_rate": 9.90207404526764e-06,
+      "loss": 0.6779,
+      "step": 354
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 11.809123992919922,
+      "learning_rate": 9.901521835717882e-06,
+      "loss": 0.6372,
+      "step": 355
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 13.38632583618164,
+      "learning_rate": 9.900968089058108e-06,
+      "loss": 0.8324,
+      "step": 356
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 7.28113317489624,
+      "learning_rate": 9.900412805461968e-06,
+      "loss": 0.3272,
+      "step": 357
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 11.599934577941895,
+      "learning_rate": 9.899855985103599e-06,
+      "loss": 0.6348,
+      "step": 358
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 11.989489555358887,
+      "learning_rate": 9.899297628157622e-06,
+      "loss": 0.828,
+      "step": 359
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 10.861467361450195,
+      "learning_rate": 9.898737734799134e-06,
+      "loss": 0.7714,
+      "step": 360
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 13.150429725646973,
+      "learning_rate": 9.898176305203715e-06,
+      "loss": 0.6172,
+      "step": 361
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 10.835484504699707,
+      "learning_rate": 9.897613339547433e-06,
+      "loss": 0.8008,
+      "step": 362
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 12.96138858795166,
+      "learning_rate": 9.89704883800683e-06,
+      "loss": 0.7168,
+      "step": 363
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 11.503767013549805,
+      "learning_rate": 9.896482800758934e-06,
+      "loss": 0.5061,
+      "step": 364
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 12.663115501403809,
+      "learning_rate": 9.895915227981254e-06,
+      "loss": 0.7793,
+      "step": 365
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 10.445459365844727,
+      "learning_rate": 9.895346119851778e-06,
+      "loss": 0.5299,
+      "step": 366
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 10.617250442504883,
+      "learning_rate": 9.89477547654898e-06,
+      "loss": 0.7179,
+      "step": 367
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 8.971922874450684,
+      "learning_rate": 9.894203298251813e-06,
+      "loss": 0.7426,
+      "step": 368
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 10.164106369018555,
+      "learning_rate": 9.893629585139705e-06,
+      "loss": 0.7137,
+      "step": 369
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 11.568824768066406,
+      "learning_rate": 9.89305433739258e-06,
+      "loss": 0.6297,
+      "step": 370
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 9.161955833435059,
+      "learning_rate": 9.892477555190831e-06,
+      "loss": 0.568,
+      "step": 371
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 10.648917198181152,
+      "learning_rate": 9.891899238715336e-06,
+      "loss": 0.4608,
+      "step": 372
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 13.20228385925293,
+      "learning_rate": 9.891319388147456e-06,
+      "loss": 0.7565,
+      "step": 373
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 10.443532943725586,
+      "learning_rate": 9.890738003669029e-06,
+      "loss": 0.5893,
+      "step": 374
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 9.268345832824707,
+      "learning_rate": 9.890155085462376e-06,
+      "loss": 0.5996,
+      "step": 375
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 9.987162590026855,
+      "learning_rate": 9.889570633710302e-06,
+      "loss": 0.6525,
+      "step": 376
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 9.322579383850098,
+      "learning_rate": 9.888984648596088e-06,
+      "loss": 0.644,
+      "step": 377
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 11.039369583129883,
+      "learning_rate": 9.8883971303035e-06,
+      "loss": 0.5142,
+      "step": 378
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 7.706017017364502,
+      "learning_rate": 9.88780807901678e-06,
+      "loss": 0.4618,
+      "step": 379
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 9.697425842285156,
+      "learning_rate": 9.887217494920655e-06,
+      "loss": 0.5649,
+      "step": 380
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 13.707801818847656,
+      "learning_rate": 9.886625378200332e-06,
+      "loss": 0.5986,
+      "step": 381
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 8.144953727722168,
+      "learning_rate": 9.886031729041498e-06,
+      "loss": 0.4626,
+      "step": 382
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 14.33889389038086,
+      "learning_rate": 9.885436547630321e-06,
+      "loss": 0.7316,
+      "step": 383
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 7.756721019744873,
+      "learning_rate": 9.884839834153446e-06,
+      "loss": 0.4317,
+      "step": 384
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 14.532642364501953,
+      "learning_rate": 9.884241588798004e-06,
+      "loss": 0.5249,
+      "step": 385
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 11.781455039978027,
+      "learning_rate": 9.883641811751603e-06,
+      "loss": 0.7816,
+      "step": 386
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 8.37305736541748,
+      "learning_rate": 9.883040503202334e-06,
+      "loss": 0.6454,
+      "step": 387
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 15.635860443115234,
+      "learning_rate": 9.882437663338764e-06,
+      "loss": 0.7687,
+      "step": 388
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 13.760857582092285,
+      "learning_rate": 9.881833292349943e-06,
+      "loss": 0.8003,
+      "step": 389
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 11.014084815979004,
+      "learning_rate": 9.881227390425404e-06,
+      "loss": 0.4841,
+      "step": 390
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 11.339284896850586,
+      "learning_rate": 9.880619957755151e-06,
+      "loss": 0.5932,
+      "step": 391
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 19.062419891357422,
+      "learning_rate": 9.880010994529679e-06,
+      "loss": 0.6057,
+      "step": 392
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 15.597712516784668,
+      "learning_rate": 9.879400500939958e-06,
+      "loss": 0.6503,
+      "step": 393
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 13.804299354553223,
+      "learning_rate": 9.878788477177434e-06,
+      "loss": 0.7872,
+      "step": 394
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 12.089165687561035,
+      "learning_rate": 9.87817492343404e-06,
+      "loss": 0.6189,
+      "step": 395
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 10.769774436950684,
+      "learning_rate": 9.877559839902185e-06,
+      "loss": 0.6641,
+      "step": 396
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 14.993908882141113,
+      "learning_rate": 9.876943226774757e-06,
+      "loss": 0.7799,
+      "step": 397
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 9.348165512084961,
+      "learning_rate": 9.876325084245127e-06,
+      "loss": 0.6365,
+      "step": 398
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 8.602823257446289,
+      "learning_rate": 9.87570541250714e-06,
+      "loss": 0.5097,
+      "step": 399
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 10.00500774383545,
+      "learning_rate": 9.875084211755127e-06,
+      "loss": 0.58,
+      "step": 400
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 10.007948875427246,
+      "learning_rate": 9.874461482183898e-06,
+      "loss": 0.6256,
+      "step": 401
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 14.605157852172852,
+      "learning_rate": 9.873837223988737e-06,
+      "loss": 0.681,
+      "step": 402
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 9.300259590148926,
+      "learning_rate": 9.873211437365409e-06,
+      "loss": 0.7034,
+      "step": 403
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 12.702817916870117,
+      "learning_rate": 9.872584122510162e-06,
+      "loss": 0.6353,
+      "step": 404
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 13.831355094909668,
+      "learning_rate": 9.871955279619721e-06,
+      "loss": 0.6951,
+      "step": 405
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 12.047754287719727,
+      "learning_rate": 9.87132490889129e-06,
+      "loss": 0.7845,
+      "step": 406
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 12.68343448638916,
+      "learning_rate": 9.870693010522552e-06,
+      "loss": 0.5288,
+      "step": 407
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 9.743354797363281,
+      "learning_rate": 9.870059584711668e-06,
+      "loss": 0.4557,
+      "step": 408
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 10.424694061279297,
+      "learning_rate": 9.869424631657283e-06,
+      "loss": 0.6526,
+      "step": 409
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 11.782444953918457,
+      "learning_rate": 9.868788151558513e-06,
+      "loss": 0.603,
+      "step": 410
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 9.839704513549805,
+      "learning_rate": 9.868150144614961e-06,
+      "loss": 0.5565,
+      "step": 411
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 11.640275955200195,
+      "learning_rate": 9.867510611026702e-06,
+      "loss": 0.7496,
+      "step": 412
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 11.834068298339844,
+      "learning_rate": 9.866869550994296e-06,
+      "loss": 0.6316,
+      "step": 413
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 10.161084175109863,
+      "learning_rate": 9.866226964718777e-06,
+      "loss": 0.5098,
+      "step": 414
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 10.799976348876953,
+      "learning_rate": 9.865582852401659e-06,
+      "loss": 0.7824,
+      "step": 415
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 13.717724800109863,
+      "learning_rate": 9.864937214244931e-06,
+      "loss": 0.6852,
+      "step": 416
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 13.076543807983398,
+      "learning_rate": 9.86429005045107e-06,
+      "loss": 0.739,
+      "step": 417
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 11.7551851272583,
+      "learning_rate": 9.863641361223025e-06,
+      "loss": 0.7613,
+      "step": 418
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 12.113115310668945,
+      "learning_rate": 9.86299114676422e-06,
+      "loss": 0.6223,
+      "step": 419
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 13.434591293334961,
+      "learning_rate": 9.862339407278564e-06,
+      "loss": 0.596,
+      "step": 420
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 11.071924209594727,
+      "learning_rate": 9.861686142970443e-06,
+      "loss": 0.5875,
+      "step": 421
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 11.253523826599121,
+      "learning_rate": 9.861031354044716e-06,
+      "loss": 0.6798,
+      "step": 422
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 9.049861907958984,
+      "learning_rate": 9.860375040706726e-06,
+      "loss": 0.5753,
+      "step": 423
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 9.950443267822266,
+      "learning_rate": 9.859717203162292e-06,
+      "loss": 0.4762,
+      "step": 424
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 10.762567520141602,
+      "learning_rate": 9.859057841617709e-06,
+      "loss": 0.549,
+      "step": 425
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 8.997856140136719,
+      "learning_rate": 9.858396956279755e-06,
+      "loss": 0.3977,
+      "step": 426
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 10.30014705657959,
+      "learning_rate": 9.857734547355679e-06,
+      "loss": 0.7045,
+      "step": 427
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 9.953113555908203,
+      "learning_rate": 9.857070615053216e-06,
+      "loss": 0.4615,
+      "step": 428
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 8.847273826599121,
+      "learning_rate": 9.85640515958057e-06,
+      "loss": 0.4162,
+      "step": 429
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 8.176438331604004,
+      "learning_rate": 9.855738181146427e-06,
+      "loss": 0.5631,
+      "step": 430
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 11.96515941619873,
+      "learning_rate": 9.855069679959956e-06,
+      "loss": 0.6706,
+      "step": 431
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 10.118613243103027,
+      "learning_rate": 9.85439965623079e-06,
+      "loss": 0.6603,
+      "step": 432
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 10.819546699523926,
+      "learning_rate": 9.853728110169054e-06,
+      "loss": 0.4968,
+      "step": 433
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 12.731653213500977,
+      "learning_rate": 9.85305504198534e-06,
+      "loss": 0.7621,
+      "step": 434
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 9.242257118225098,
+      "learning_rate": 9.852380451890723e-06,
+      "loss": 0.5765,
+      "step": 435
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.451205253601074,
+      "learning_rate": 9.851704340096752e-06,
+      "loss": 0.4249,
+      "step": 436
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.890918254852295,
+      "learning_rate": 9.851026706815456e-06,
+      "loss": 0.4218,
+      "step": 437
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 9.179704666137695,
+      "learning_rate": 9.850347552259339e-06,
+      "loss": 0.4587,
+      "step": 438
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.036215305328369,
+      "learning_rate": 9.849666876641384e-06,
+      "loss": 0.4922,
+      "step": 439
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 9.643997192382812,
+      "learning_rate": 9.848984680175049e-06,
+      "loss": 0.4673,
+      "step": 440
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 11.018205642700195,
+      "learning_rate": 9.848300963074269e-06,
+      "loss": 0.7134,
+      "step": 441
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 9.255934715270996,
+      "learning_rate": 9.847615725553457e-06,
+      "loss": 0.6708,
+      "step": 442
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 9.3004150390625,
+      "learning_rate": 9.846928967827505e-06,
+      "loss": 0.6113,
+      "step": 443
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 12.182660102844238,
+      "learning_rate": 9.846240690111774e-06,
+      "loss": 0.7192,
+      "step": 444
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 13.014025688171387,
+      "learning_rate": 9.845550892622113e-06,
+      "loss": 0.5448,
+      "step": 445
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 11.434033393859863,
+      "learning_rate": 9.844859575574838e-06,
+      "loss": 0.5677,
+      "step": 446
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 8.435720443725586,
+      "learning_rate": 9.844166739186744e-06,
+      "loss": 0.5463,
+      "step": 447
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 6.199049472808838,
+      "learning_rate": 9.843472383675108e-06,
+      "loss": 0.3331,
+      "step": 448
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 8.172830581665039,
+      "learning_rate": 9.842776509257674e-06,
+      "loss": 0.4321,
+      "step": 449
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 8.728845596313477,
+      "learning_rate": 9.84207911615267e-06,
+      "loss": 0.5158,
+      "step": 450
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 12.642927169799805,
+      "learning_rate": 9.841380204578795e-06,
+      "loss": 0.7075,
+      "step": 451
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 7.548518180847168,
+      "learning_rate": 9.84067977475523e-06,
+      "loss": 0.5879,
+      "step": 452
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 8.492765426635742,
+      "learning_rate": 9.839977826901625e-06,
+      "loss": 0.5269,
+      "step": 453
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 7.718329429626465,
+      "learning_rate": 9.839274361238113e-06,
+      "loss": 0.3395,
+      "step": 454
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 9.875319480895996,
+      "learning_rate": 9.838569377985298e-06,
+      "loss": 0.5866,
+      "step": 455
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 8.861202239990234,
+      "learning_rate": 9.837862877364264e-06,
+      "loss": 0.5658,
+      "step": 456
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 9.063336372375488,
+      "learning_rate": 9.837154859596566e-06,
+      "loss": 0.4624,
+      "step": 457
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 10.135499000549316,
+      "learning_rate": 9.836445324904237e-06,
+      "loss": 0.5767,
+      "step": 458
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 9.92232608795166,
+      "learning_rate": 9.835734273509787e-06,
+      "loss": 0.5572,
+      "step": 459
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 10.1605224609375,
+      "learning_rate": 9.835021705636201e-06,
+      "loss": 0.7387,
+      "step": 460
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 9.25823974609375,
+      "learning_rate": 9.83430762150694e-06,
+      "loss": 0.5537,
+      "step": 461
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 12.154031753540039,
+      "learning_rate": 9.833592021345938e-06,
+      "loss": 0.6136,
+      "step": 462
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 10.395689964294434,
+      "learning_rate": 9.832874905377608e-06,
+      "loss": 0.399,
+      "step": 463
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 13.269356727600098,
+      "learning_rate": 9.832156273826834e-06,
+      "loss": 0.676,
+      "step": 464
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.385758876800537,
+      "learning_rate": 9.83143612691898e-06,
+      "loss": 0.4284,
+      "step": 465
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 10.52908706665039,
+      "learning_rate": 9.830714464879883e-06,
+      "loss": 0.6201,
+      "step": 466
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 12.051213264465332,
+      "learning_rate": 9.829991287935855e-06,
+      "loss": 0.5787,
+      "step": 467
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 11.557876586914062,
+      "learning_rate": 9.82926659631368e-06,
+      "loss": 0.6621,
+      "step": 468
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 21.905012130737305,
+      "learning_rate": 9.828540390240624e-06,
+      "loss": 0.496,
+      "step": 469
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 8.631387710571289,
+      "learning_rate": 9.827812669944423e-06,
+      "loss": 0.5854,
+      "step": 470
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 8.464282035827637,
+      "learning_rate": 9.82708343565329e-06,
+      "loss": 0.3075,
+      "step": 471
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 12.647194862365723,
+      "learning_rate": 9.82635268759591e-06,
+      "loss": 0.5056,
+      "step": 472
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 9.318597793579102,
+      "learning_rate": 9.825620426001446e-06,
+      "loss": 0.3516,
+      "step": 473
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 12.450582504272461,
+      "learning_rate": 9.824886651099534e-06,
+      "loss": 0.6391,
+      "step": 474
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 15.280452728271484,
+      "learning_rate": 9.824151363120283e-06,
+      "loss": 0.7899,
+      "step": 475
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 13.64406681060791,
+      "learning_rate": 9.82341456229428e-06,
+      "loss": 0.9056,
+      "step": 476
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 9.222602844238281,
+      "learning_rate": 9.822676248852582e-06,
+      "loss": 0.5428,
+      "step": 477
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 10.81840705871582,
+      "learning_rate": 9.821936423026728e-06,
+      "loss": 0.8339,
+      "step": 478
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 8.712065696716309,
+      "learning_rate": 9.821195085048723e-06,
+      "loss": 0.3326,
+      "step": 479
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 16.628429412841797,
+      "learning_rate": 9.82045223515105e-06,
+      "loss": 0.8353,
+      "step": 480
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 11.709946632385254,
+      "learning_rate": 9.819707873566664e-06,
+      "loss": 0.5778,
+      "step": 481
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 10.946529388427734,
+      "learning_rate": 9.818962000528997e-06,
+      "loss": 0.5669,
+      "step": 482
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 10.898555755615234,
+      "learning_rate": 9.818214616271957e-06,
+      "loss": 0.4419,
+      "step": 483
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 8.391704559326172,
+      "learning_rate": 9.817465721029916e-06,
+      "loss": 0.4047,
+      "step": 484
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 9.908906936645508,
+      "learning_rate": 9.816715315037733e-06,
+      "loss": 0.4892,
+      "step": 485
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 12.4695463180542,
+      "learning_rate": 9.815963398530727e-06,
+      "loss": 0.6535,
+      "step": 486
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.847550868988037,
+      "learning_rate": 9.815209971744707e-06,
+      "loss": 0.2293,
+      "step": 487
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 16.817188262939453,
+      "learning_rate": 9.814455034915939e-06,
+      "loss": 0.8089,
+      "step": 488
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 10.427989959716797,
+      "learning_rate": 9.813698588281174e-06,
+      "loss": 0.656,
+      "step": 489
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 10.519749641418457,
+      "learning_rate": 9.812940632077629e-06,
+      "loss": 0.5331,
+      "step": 490
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 10.185182571411133,
+      "learning_rate": 9.812181166543002e-06,
+      "loss": 0.6294,
+      "step": 491
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 11.053422927856445,
+      "learning_rate": 9.811420191915457e-06,
+      "loss": 0.6077,
+      "step": 492
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 13.634664535522461,
+      "learning_rate": 9.810657708433637e-06,
+      "loss": 0.8089,
+      "step": 493
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 8.612808227539062,
+      "learning_rate": 9.809893716336654e-06,
+      "loss": 0.4303,
+      "step": 494
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 13.521270751953125,
+      "learning_rate": 9.809128215864096e-06,
+      "loss": 0.847,
+      "step": 495
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 11.675257682800293,
+      "learning_rate": 9.808361207256024e-06,
+      "loss": 0.6721,
+      "step": 496
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 11.107856750488281,
+      "learning_rate": 9.807592690752965e-06,
+      "loss": 0.6187,
+      "step": 497
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 10.13960075378418,
+      "learning_rate": 9.806822666595931e-06,
+      "loss": 0.5581,
+      "step": 498
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 12.868432998657227,
+      "learning_rate": 9.806051135026398e-06,
+      "loss": 0.6246,
+      "step": 499
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 8.669515609741211,
+      "learning_rate": 9.805278096286318e-06,
+      "loss": 0.4433,
+      "step": 500
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 6.546693325042725,
+      "learning_rate": 9.804503550618113e-06,
+      "loss": 0.4212,
+      "step": 501
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 12.599363327026367,
+      "learning_rate": 9.803727498264681e-06,
+      "loss": 0.458,
+      "step": 502
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 12.589111328125,
+      "learning_rate": 9.80294993946939e-06,
+      "loss": 0.5701,
+      "step": 503
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 10.536702156066895,
+      "learning_rate": 9.802170874476082e-06,
+      "loss": 0.5958,
+      "step": 504
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 10.091132164001465,
+      "learning_rate": 9.80139030352907e-06,
+      "loss": 0.6376,
+      "step": 505
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 10.614989280700684,
+      "learning_rate": 9.800608226873143e-06,
+      "loss": 0.79,
+      "step": 506
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 14.539179801940918,
+      "learning_rate": 9.799824644753552e-06,
+      "loss": 0.6776,
+      "step": 507
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 11.357399940490723,
+      "learning_rate": 9.799039557416034e-06,
+      "loss": 0.5703,
+      "step": 508
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 7.345721244812012,
+      "learning_rate": 9.798252965106787e-06,
+      "loss": 0.3331,
+      "step": 509
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 13.442143440246582,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 0.6761,
+      "step": 510
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 10.102835655212402,
+      "learning_rate": 9.796675266560281e-06,
+      "loss": 0.4306,
+      "step": 511
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 9.243897438049316,
+      "learning_rate": 9.795884160817786e-06,
+      "loss": 0.326,
+      "step": 512
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 9.08454418182373,
+      "learning_rate": 9.79509155109309e-06,
+      "loss": 0.5289,
+      "step": 513
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 12.854437828063965,
+      "learning_rate": 9.794297437634757e-06,
+      "loss": 0.5036,
+      "step": 514
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 11.14797592163086,
+      "learning_rate": 9.793501820691818e-06,
+      "loss": 0.7425,
+      "step": 515
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 10.359697341918945,
+      "learning_rate": 9.792704700513777e-06,
+      "loss": 0.3959,
+      "step": 516
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 9.362765312194824,
+      "learning_rate": 9.791906077350613e-06,
+      "loss": 0.5938,
+      "step": 517
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 9.751148223876953,
+      "learning_rate": 9.791105951452769e-06,
+      "loss": 0.5045,
+      "step": 518
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 11.859084129333496,
+      "learning_rate": 9.790304323071164e-06,
+      "loss": 0.8976,
+      "step": 519
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 11.962151527404785,
+      "learning_rate": 9.789501192457188e-06,
+      "loss": 0.4148,
+      "step": 520
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 9.350335121154785,
+      "learning_rate": 9.788696559862702e-06,
+      "loss": 0.3899,
+      "step": 521
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 10.396106719970703,
+      "learning_rate": 9.787890425540038e-06,
+      "loss": 0.5377,
+      "step": 522
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 8.526636123657227,
+      "learning_rate": 9.787082789742e-06,
+      "loss": 0.4437,
+      "step": 523
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 15.801200866699219,
+      "learning_rate": 9.786273652721857e-06,
+      "loss": 0.7673,
+      "step": 524
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 9.015732765197754,
+      "learning_rate": 9.785463014733356e-06,
+      "loss": 0.5309,
+      "step": 525
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 6.979864120483398,
+      "learning_rate": 9.784650876030713e-06,
+      "loss": 0.3826,
+      "step": 526
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 9.423440933227539,
+      "learning_rate": 9.78383723686861e-06,
+      "loss": 0.4922,
+      "step": 527
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 11.650758743286133,
+      "learning_rate": 9.783022097502204e-06,
+      "loss": 0.6451,
+      "step": 528
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 12.18952465057373,
+      "learning_rate": 9.782205458187126e-06,
+      "loss": 0.7661,
+      "step": 529
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 11.730113983154297,
+      "learning_rate": 9.781387319179465e-06,
+      "loss": 0.6267,
+      "step": 530
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 11.342617988586426,
+      "learning_rate": 9.780567680735796e-06,
+      "loss": 0.861,
+      "step": 531
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 9.465302467346191,
+      "learning_rate": 9.779746543113151e-06,
+      "loss": 0.4578,
+      "step": 532
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 11.096479415893555,
+      "learning_rate": 9.778923906569042e-06,
+      "loss": 0.6042,
+      "step": 533
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.44420862197876,
+      "learning_rate": 9.778099771361442e-06,
+      "loss": 0.4282,
+      "step": 534
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 16.6076602935791,
+      "learning_rate": 9.777274137748802e-06,
+      "loss": 0.6464,
+      "step": 535
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 11.399652481079102,
+      "learning_rate": 9.776447005990038e-06,
+      "loss": 0.3849,
+      "step": 536
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 9.213934898376465,
+      "learning_rate": 9.775618376344536e-06,
+      "loss": 0.5164,
+      "step": 537
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 12.120831489562988,
+      "learning_rate": 9.774788249072156e-06,
+      "loss": 0.6199,
+      "step": 538
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 9.66297721862793,
+      "learning_rate": 9.773956624433224e-06,
+      "loss": 0.4337,
+      "step": 539
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 8.792263984680176,
+      "learning_rate": 9.773123502688532e-06,
+      "loss": 0.4816,
+      "step": 540
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 112.53068542480469,
+      "learning_rate": 9.772288884099353e-06,
+      "loss": 0.4564,
+      "step": 541
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 12.131253242492676,
+      "learning_rate": 9.771452768927416e-06,
+      "loss": 0.5091,
+      "step": 542
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 12.005109786987305,
+      "learning_rate": 9.770615157434928e-06,
+      "loss": 0.5646,
+      "step": 543
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 8.583512306213379,
+      "learning_rate": 9.769776049884564e-06,
+      "loss": 0.3536,
+      "step": 544
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 10.094390869140625,
+      "learning_rate": 9.768935446539464e-06,
+      "loss": 0.4301,
+      "step": 545
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 9.595136642456055,
+      "learning_rate": 9.768093347663243e-06,
+      "loss": 0.3571,
+      "step": 546
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 8.435619354248047,
+      "learning_rate": 9.76724975351998e-06,
+      "loss": 0.5013,
+      "step": 547
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 7.663536071777344,
+      "learning_rate": 9.766404664374225e-06,
+      "loss": 0.426,
+      "step": 548
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 11.494606018066406,
+      "learning_rate": 9.765558080490998e-06,
+      "loss": 0.6269,
+      "step": 549
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 11.298256874084473,
+      "learning_rate": 9.764710002135784e-06,
+      "loss": 0.5164,
+      "step": 550
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 12.774913787841797,
+      "learning_rate": 9.763860429574542e-06,
+      "loss": 0.7405,
+      "step": 551
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 9.756369590759277,
+      "learning_rate": 9.763009363073694e-06,
+      "loss": 0.4333,
+      "step": 552
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 11.240097045898438,
+      "learning_rate": 9.762156802900133e-06,
+      "loss": 0.3907,
+      "step": 553
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 11.007864952087402,
+      "learning_rate": 9.761302749321223e-06,
+      "loss": 0.4882,
+      "step": 554
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 7.719780445098877,
+      "learning_rate": 9.760447202604796e-06,
+      "loss": 0.3746,
+      "step": 555
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 10.753983497619629,
+      "learning_rate": 9.759590163019146e-06,
+      "loss": 0.3713,
+      "step": 556
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 9.289199829101562,
+      "learning_rate": 9.758731630833039e-06,
+      "loss": 0.5148,
+      "step": 557
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 10.315267562866211,
+      "learning_rate": 9.757871606315712e-06,
+      "loss": 0.6647,
+      "step": 558
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 11.545065879821777,
+      "learning_rate": 9.757010089736867e-06,
+      "loss": 0.5028,
+      "step": 559
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 12.856269836425781,
+      "learning_rate": 9.756147081366673e-06,
+      "loss": 0.4087,
+      "step": 560
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 9.536337852478027,
+      "learning_rate": 9.755282581475769e-06,
+      "loss": 0.5003,
+      "step": 561
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8520866349709456,
+      "eval_loss": 0.4738004505634308,
+      "eval_runtime": 11.4331,
+      "eval_samples_per_second": 165.572,
+      "eval_steps_per_second": 20.729,
+      "step": 561
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 8.779393196105957,
+      "learning_rate": 9.75441659033526e-06,
+      "loss": 0.4464,
+      "step": 562
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.718747138977051,
+      "learning_rate": 9.753549108216722e-06,
+      "loss": 0.3785,
+      "step": 563
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 9.057177543640137,
+      "learning_rate": 9.752680135392193e-06,
+      "loss": 0.4617,
+      "step": 564
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 8.231096267700195,
+      "learning_rate": 9.751809672134184e-06,
+      "loss": 0.3711,
+      "step": 565
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 10.373970985412598,
+      "learning_rate": 9.75093771871567e-06,
+      "loss": 0.4725,
+      "step": 566
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 6.778740406036377,
+      "learning_rate": 9.750064275410095e-06,
+      "loss": 0.3607,
+      "step": 567
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 12.065496444702148,
+      "learning_rate": 9.749189342491368e-06,
+      "loss": 0.4806,
+      "step": 568
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 6.326473236083984,
+      "learning_rate": 9.748312920233867e-06,
+      "loss": 0.1989,
+      "step": 569
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 9.346095085144043,
+      "learning_rate": 9.747435008912438e-06,
+      "loss": 0.3478,
+      "step": 570
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 7.086182594299316,
+      "learning_rate": 9.74655560880239e-06,
+      "loss": 0.3261,
+      "step": 571
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 7.893068313598633,
+      "learning_rate": 9.745674720179507e-06,
+      "loss": 0.404,
+      "step": 572
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 6.949780464172363,
+      "learning_rate": 9.744792343320028e-06,
+      "loss": 0.3275,
+      "step": 573
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 12.822416305541992,
+      "learning_rate": 9.74390847850067e-06,
+      "loss": 0.4621,
+      "step": 574
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 9.68514633178711,
+      "learning_rate": 9.74302312599861e-06,
+      "loss": 0.5081,
+      "step": 575
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.721839189529419,
+      "learning_rate": 9.742136286091493e-06,
+      "loss": 0.2119,
+      "step": 576
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 8.837911605834961,
+      "learning_rate": 9.74124795905743e-06,
+      "loss": 0.3899,
+      "step": 577
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 10.308550834655762,
+      "learning_rate": 9.740358145174999e-06,
+      "loss": 0.3589,
+      "step": 578
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 9.589912414550781,
+      "learning_rate": 9.739466844723245e-06,
+      "loss": 0.3763,
+      "step": 579
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 10.515608787536621,
+      "learning_rate": 9.73857405798168e-06,
+      "loss": 0.4701,
+      "step": 580
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 10.156450271606445,
+      "learning_rate": 9.737679785230277e-06,
+      "loss": 0.3007,
+      "step": 581
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 9.349661827087402,
+      "learning_rate": 9.736784026749481e-06,
+      "loss": 0.4323,
+      "step": 582
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 13.339727401733398,
+      "learning_rate": 9.735886782820202e-06,
+      "loss": 0.4102,
+      "step": 583
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 11.36646842956543,
+      "learning_rate": 9.734988053723811e-06,
+      "loss": 0.5166,
+      "step": 584
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 14.438636779785156,
+      "learning_rate": 9.734087839742152e-06,
+      "loss": 0.5596,
+      "step": 585
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 10.601497650146484,
+      "learning_rate": 9.733186141157526e-06,
+      "loss": 0.3824,
+      "step": 586
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 10.539875984191895,
+      "learning_rate": 9.732282958252709e-06,
+      "loss": 0.437,
+      "step": 587
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 11.75239086151123,
+      "learning_rate": 9.731378291310934e-06,
+      "loss": 0.5644,
+      "step": 588
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 6.822336196899414,
+      "learning_rate": 9.730472140615907e-06,
+      "loss": 0.2866,
+      "step": 589
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 12.240249633789062,
+      "learning_rate": 9.729564506451791e-06,
+      "loss": 0.6873,
+      "step": 590
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 12.426766395568848,
+      "learning_rate": 9.728655389103223e-06,
+      "loss": 0.4458,
+      "step": 591
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 7.946753025054932,
+      "learning_rate": 9.7277447888553e-06,
+      "loss": 0.3531,
+      "step": 592
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 10.869359016418457,
+      "learning_rate": 9.72683270599358e-06,
+      "loss": 0.5187,
+      "step": 593
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 8.237872123718262,
+      "learning_rate": 9.7259191408041e-06,
+      "loss": 0.4352,
+      "step": 594
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 11.390605926513672,
+      "learning_rate": 9.725004093573343e-06,
+      "loss": 0.3982,
+      "step": 595
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 11.030872344970703,
+      "learning_rate": 9.724087564588275e-06,
+      "loss": 0.4749,
+      "step": 596
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 10.00020980834961,
+      "learning_rate": 9.723169554136309e-06,
+      "loss": 0.4152,
+      "step": 597
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 7.669042587280273,
+      "learning_rate": 9.722250062505341e-06,
+      "loss": 0.3994,
+      "step": 598
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 12.858966827392578,
+      "learning_rate": 9.721329089983716e-06,
+      "loss": 0.432,
+      "step": 599
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 7.33831787109375,
+      "learning_rate": 9.720406636860252e-06,
+      "loss": 0.4376,
+      "step": 600
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 32.82384490966797,
+      "learning_rate": 9.71948270342423e-06,
+      "loss": 0.4774,
+      "step": 601
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 10.197738647460938,
+      "learning_rate": 9.718557289965391e-06,
+      "loss": 0.6837,
+      "step": 602
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 9.729262351989746,
+      "learning_rate": 9.717630396773946e-06,
+      "loss": 0.3321,
+      "step": 603
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 7.790341377258301,
+      "learning_rate": 9.716702024140566e-06,
+      "loss": 0.3411,
+      "step": 604
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 9.120753288269043,
+      "learning_rate": 9.715772172356388e-06,
+      "loss": 0.3244,
+      "step": 605
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 8.661083221435547,
+      "learning_rate": 9.714840841713011e-06,
+      "loss": 0.2214,
+      "step": 606
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 11.357797622680664,
+      "learning_rate": 9.713908032502501e-06,
+      "loss": 0.581,
+      "step": 607
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 16.81509780883789,
+      "learning_rate": 9.712973745017384e-06,
+      "loss": 0.3764,
+      "step": 608
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 9.057856559753418,
+      "learning_rate": 9.712037979550651e-06,
+      "loss": 0.4304,
+      "step": 609
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 7.538285255432129,
+      "learning_rate": 9.711100736395758e-06,
+      "loss": 0.2704,
+      "step": 610
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 7.614073753356934,
+      "learning_rate": 9.710162015846623e-06,
+      "loss": 0.2224,
+      "step": 611
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 8.92867660522461,
+      "learning_rate": 9.709221818197626e-06,
+      "loss": 0.4772,
+      "step": 612
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 9.771719932556152,
+      "learning_rate": 9.708280143743609e-06,
+      "loss": 0.4526,
+      "step": 613
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 6.499321937561035,
+      "learning_rate": 9.707336992779887e-06,
+      "loss": 0.3727,
+      "step": 614
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 10.07966423034668,
+      "learning_rate": 9.706392365602224e-06,
+      "loss": 0.3928,
+      "step": 615
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 9.511795043945312,
+      "learning_rate": 9.705446262506858e-06,
+      "loss": 0.385,
+      "step": 616
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 12.598077774047852,
+      "learning_rate": 9.704498683790484e-06,
+      "loss": 0.4833,
+      "step": 617
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 8.100473403930664,
+      "learning_rate": 9.703549629750262e-06,
+      "loss": 0.4341,
+      "step": 618
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 9.722856521606445,
+      "learning_rate": 9.702599100683811e-06,
+      "loss": 0.4654,
+      "step": 619
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 10.565521240234375,
+      "learning_rate": 9.70164709688922e-06,
+      "loss": 0.3699,
+      "step": 620
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 6.691813945770264,
+      "learning_rate": 9.70069361866503e-06,
+      "loss": 0.2695,
+      "step": 621
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 15.809154510498047,
+      "learning_rate": 9.699738666310259e-06,
+      "loss": 0.4821,
+      "step": 622
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 8.641901016235352,
+      "learning_rate": 9.698782240124371e-06,
+      "loss": 0.5473,
+      "step": 623
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 10.854053497314453,
+      "learning_rate": 9.697824340407303e-06,
+      "loss": 0.4452,
+      "step": 624
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 6.259403705596924,
+      "learning_rate": 9.69686496745945e-06,
+      "loss": 0.2891,
+      "step": 625
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 11.982359886169434,
+      "learning_rate": 9.69590412158167e-06,
+      "loss": 0.5453,
+      "step": 626
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 7.000669956207275,
+      "learning_rate": 9.694941803075285e-06,
+      "loss": 0.4641,
+      "step": 627
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 8.62365436553955,
+      "learning_rate": 9.693978012242073e-06,
+      "loss": 0.3298,
+      "step": 628
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 14.987152099609375,
+      "learning_rate": 9.69301274938428e-06,
+      "loss": 0.4965,
+      "step": 629
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 9.6792573928833,
+      "learning_rate": 9.69204601480461e-06,
+      "loss": 0.4923,
+      "step": 630
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 7.610067367553711,
+      "learning_rate": 9.69107780880623e-06,
+      "loss": 0.2267,
+      "step": 631
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 10.01244831085205,
+      "learning_rate": 9.690108131692768e-06,
+      "loss": 0.4459,
+      "step": 632
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 8.989778518676758,
+      "learning_rate": 9.689136983768312e-06,
+      "loss": 0.3434,
+      "step": 633
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 10.87840461730957,
+      "learning_rate": 9.688164365337416e-06,
+      "loss": 0.3879,
+      "step": 634
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 9.524523735046387,
+      "learning_rate": 9.687190276705088e-06,
+      "loss": 0.448,
+      "step": 635
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 12.467297554016113,
+      "learning_rate": 9.686214718176804e-06,
+      "loss": 0.5103,
+      "step": 636
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 6.580789089202881,
+      "learning_rate": 9.685237690058494e-06,
+      "loss": 0.2874,
+      "step": 637
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 11.867737770080566,
+      "learning_rate": 9.684259192656554e-06,
+      "loss": 0.4601,
+      "step": 638
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 10.610692977905273,
+      "learning_rate": 9.68327922627784e-06,
+      "loss": 0.5744,
+      "step": 639
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 13.323002815246582,
+      "learning_rate": 9.682297791229668e-06,
+      "loss": 0.5905,
+      "step": 640
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 9.821512222290039,
+      "learning_rate": 9.681314887819815e-06,
+      "loss": 0.353,
+      "step": 641
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 16.294443130493164,
+      "learning_rate": 9.680330516356517e-06,
+      "loss": 0.717,
+      "step": 642
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 9.918705940246582,
+      "learning_rate": 9.679344677148473e-06,
+      "loss": 0.4926,
+      "step": 643
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 12.107922554016113,
+      "learning_rate": 9.678357370504838e-06,
+      "loss": 0.557,
+      "step": 644
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 13.941534042358398,
+      "learning_rate": 9.677368596735232e-06,
+      "loss": 0.5098,
+      "step": 645
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 14.436763763427734,
+      "learning_rate": 9.676378356149733e-06,
+      "loss": 0.6408,
+      "step": 646
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 9.680541038513184,
+      "learning_rate": 9.675386649058878e-06,
+      "loss": 0.3551,
+      "step": 647
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 16.916828155517578,
+      "learning_rate": 9.674393475773667e-06,
+      "loss": 0.4862,
+      "step": 648
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 15.478622436523438,
+      "learning_rate": 9.673398836605554e-06,
+      "loss": 0.6139,
+      "step": 649
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 11.439444541931152,
+      "learning_rate": 9.67240273186646e-06,
+      "loss": 0.4052,
+      "step": 650
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 10.493536949157715,
+      "learning_rate": 9.67140516186876e-06,
+      "loss": 0.3327,
+      "step": 651
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 7.983470439910889,
+      "learning_rate": 9.67040612692529e-06,
+      "loss": 0.2905,
+      "step": 652
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 10.680438995361328,
+      "learning_rate": 9.669405627349346e-06,
+      "loss": 0.4487,
+      "step": 653
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 7.990166187286377,
+      "learning_rate": 9.668403663454685e-06,
+      "loss": 0.272,
+      "step": 654
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 8.564766883850098,
+      "learning_rate": 9.66740023555552e-06,
+      "loss": 0.2863,
+      "step": 655
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 9.973396301269531,
+      "learning_rate": 9.666395343966526e-06,
+      "loss": 0.4507,
+      "step": 656
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 9.524887084960938,
+      "learning_rate": 9.665388989002835e-06,
+      "loss": 0.4279,
+      "step": 657
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 9.859561920166016,
+      "learning_rate": 9.664381170980034e-06,
+      "loss": 0.3875,
+      "step": 658
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 12.0888671875,
+      "learning_rate": 9.66337189021418e-06,
+      "loss": 0.5257,
+      "step": 659
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 8.582508087158203,
+      "learning_rate": 9.66236114702178e-06,
+      "loss": 0.3055,
+      "step": 660
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 9.119597434997559,
+      "learning_rate": 9.661348941719799e-06,
+      "loss": 0.3254,
+      "step": 661
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 8.26047420501709,
+      "learning_rate": 9.660335274625665e-06,
+      "loss": 0.3512,
+      "step": 662
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 9.378521919250488,
+      "learning_rate": 9.659320146057263e-06,
+      "loss": 0.4168,
+      "step": 663
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 8.383490562438965,
+      "learning_rate": 9.658303556332932e-06,
+      "loss": 0.2867,
+      "step": 664
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 7.669978618621826,
+      "learning_rate": 9.65728550577148e-06,
+      "loss": 0.4441,
+      "step": 665
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 9.815823554992676,
+      "learning_rate": 9.65626599469216e-06,
+      "loss": 0.5105,
+      "step": 666
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 10.764351844787598,
+      "learning_rate": 9.655245023414693e-06,
+      "loss": 0.3309,
+      "step": 667
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 10.087464332580566,
+      "learning_rate": 9.65422259225925e-06,
+      "loss": 0.2864,
+      "step": 668
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 9.725699424743652,
+      "learning_rate": 9.653198701546469e-06,
+      "loss": 0.4579,
+      "step": 669
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 10.332109451293945,
+      "learning_rate": 9.652173351597435e-06,
+      "loss": 0.3474,
+      "step": 670
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 9.60068130493164,
+      "learning_rate": 9.651146542733702e-06,
+      "loss": 0.2882,
+      "step": 671
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 9.192439079284668,
+      "learning_rate": 9.650118275277272e-06,
+      "loss": 0.308,
+      "step": 672
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 11.568072319030762,
+      "learning_rate": 9.64908854955061e-06,
+      "loss": 0.3589,
+      "step": 673
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 13.34099006652832,
+      "learning_rate": 9.648057365876634e-06,
+      "loss": 0.4515,
+      "step": 674
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 10.70632553100586,
+      "learning_rate": 9.647024724578724e-06,
+      "loss": 0.2433,
+      "step": 675
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 9.742579460144043,
+      "learning_rate": 9.645990625980714e-06,
+      "loss": 0.3414,
+      "step": 676
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 14.165474891662598,
+      "learning_rate": 9.644955070406897e-06,
+      "loss": 0.3345,
+      "step": 677
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 16.842557907104492,
+      "learning_rate": 9.643918058182017e-06,
+      "loss": 0.5414,
+      "step": 678
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 11.285589218139648,
+      "learning_rate": 9.642879589631286e-06,
+      "loss": 0.4586,
+      "step": 679
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 5.416263103485107,
+      "learning_rate": 9.641839665080363e-06,
+      "loss": 0.1571,
+      "step": 680
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 9.182520866394043,
+      "learning_rate": 9.640798284855368e-06,
+      "loss": 0.3089,
+      "step": 681
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 8.085102081298828,
+      "learning_rate": 9.639755449282874e-06,
+      "loss": 0.2895,
+      "step": 682
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 9.769516944885254,
+      "learning_rate": 9.638711158689915e-06,
+      "loss": 0.3176,
+      "step": 683
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 6.7729949951171875,
+      "learning_rate": 9.637665413403978e-06,
+      "loss": 0.2365,
+      "step": 684
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 8.425326347351074,
+      "learning_rate": 9.636618213753006e-06,
+      "loss": 0.4389,
+      "step": 685
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 8.229381561279297,
+      "learning_rate": 9.635569560065401e-06,
+      "loss": 0.3916,
+      "step": 686
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 5.542147159576416,
+      "learning_rate": 9.634519452670019e-06,
+      "loss": 0.2344,
+      "step": 687
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 13.222643852233887,
+      "learning_rate": 9.633467891896171e-06,
+      "loss": 0.4774,
+      "step": 688
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 8.360149383544922,
+      "learning_rate": 9.632414878073628e-06,
+      "loss": 0.4886,
+      "step": 689
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 10.396231651306152,
+      "learning_rate": 9.631360411532609e-06,
+      "loss": 0.4441,
+      "step": 690
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 21.4537353515625,
+      "learning_rate": 9.630304492603797e-06,
+      "loss": 0.4215,
+      "step": 691
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 8.892533302307129,
+      "learning_rate": 9.629247121618323e-06,
+      "loss": 0.347,
+      "step": 692
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 8.009398460388184,
+      "learning_rate": 9.628188298907782e-06,
+      "loss": 0.4411,
+      "step": 693
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 10.975576400756836,
+      "learning_rate": 9.627128024804216e-06,
+      "loss": 0.3507,
+      "step": 694
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 7.532343864440918,
+      "learning_rate": 9.626066299640124e-06,
+      "loss": 0.1973,
+      "step": 695
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 5.477530002593994,
+      "learning_rate": 9.625003123748464e-06,
+      "loss": 0.2126,
+      "step": 696
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 12.35208797454834,
+      "learning_rate": 9.623938497462647e-06,
+      "loss": 0.5135,
+      "step": 697
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 13.38844108581543,
+      "learning_rate": 9.622872421116535e-06,
+      "loss": 0.5107,
+      "step": 698
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 8.146617889404297,
+      "learning_rate": 9.621804895044451e-06,
+      "loss": 0.3064,
+      "step": 699
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 12.739921569824219,
+      "learning_rate": 9.620735919581168e-06,
+      "loss": 0.7149,
+      "step": 700
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 9.771955490112305,
+      "learning_rate": 9.619665495061916e-06,
+      "loss": 0.3301,
+      "step": 701
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 7.802317142486572,
+      "learning_rate": 9.618593621822376e-06,
+      "loss": 0.1931,
+      "step": 702
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 10.133605003356934,
+      "learning_rate": 9.61752030019869e-06,
+      "loss": 0.4213,
+      "step": 703
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 9.60625171661377,
+      "learning_rate": 9.616445530527448e-06,
+      "loss": 0.4962,
+      "step": 704
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 12.223292350769043,
+      "learning_rate": 9.615369313145695e-06,
+      "loss": 0.431,
+      "step": 705
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 14.382707595825195,
+      "learning_rate": 9.614291648390933e-06,
+      "loss": 0.5547,
+      "step": 706
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 9.862970352172852,
+      "learning_rate": 9.613212536601116e-06,
+      "loss": 0.3576,
+      "step": 707
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 11.678335189819336,
+      "learning_rate": 9.612131978114652e-06,
+      "loss": 0.4317,
+      "step": 708
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 9.259819984436035,
+      "learning_rate": 9.6110499732704e-06,
+      "loss": 0.389,
+      "step": 709
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 13.46289348602295,
+      "learning_rate": 9.609966522407678e-06,
+      "loss": 0.4013,
+      "step": 710
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 13.409856796264648,
+      "learning_rate": 9.608881625866253e-06,
+      "loss": 0.4901,
+      "step": 711
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 18.925800323486328,
+      "learning_rate": 9.607795283986348e-06,
+      "loss": 0.3749,
+      "step": 712
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 18.489225387573242,
+      "learning_rate": 9.606707497108639e-06,
+      "loss": 0.6695,
+      "step": 713
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 12.039801597595215,
+      "learning_rate": 9.60561826557425e-06,
+      "loss": 0.4421,
+      "step": 714
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 11.478935241699219,
+      "learning_rate": 9.60452758972477e-06,
+      "loss": 0.4467,
+      "step": 715
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 9.511488914489746,
+      "learning_rate": 9.603435469902225e-06,
+      "loss": 0.4553,
+      "step": 716
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 9.71642780303955,
+      "learning_rate": 9.602341906449108e-06,
+      "loss": 0.36,
+      "step": 717
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 11.606756210327148,
+      "learning_rate": 9.601246899708355e-06,
+      "loss": 0.3519,
+      "step": 718
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 13.155590057373047,
+      "learning_rate": 9.600150450023363e-06,
+      "loss": 0.3156,
+      "step": 719
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 10.377179145812988,
+      "learning_rate": 9.599052557737973e-06,
+      "loss": 0.335,
+      "step": 720
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 15.308124542236328,
+      "learning_rate": 9.597953223196485e-06,
+      "loss": 0.4007,
+      "step": 721
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 7.4133620262146,
+      "learning_rate": 9.596852446743647e-06,
+      "loss": 0.2745,
+      "step": 722
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 8.160500526428223,
+      "learning_rate": 9.595750228724661e-06,
+      "loss": 0.2438,
+      "step": 723
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 9.987648010253906,
+      "learning_rate": 9.594646569485179e-06,
+      "loss": 0.348,
+      "step": 724
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 13.39833927154541,
+      "learning_rate": 9.593541469371313e-06,
+      "loss": 0.6109,
+      "step": 725
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 12.48079776763916,
+      "learning_rate": 9.592434928729617e-06,
+      "loss": 0.2987,
+      "step": 726
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 12.24653148651123,
+      "learning_rate": 9.591326947907097e-06,
+      "loss": 0.3818,
+      "step": 727
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 6.557040214538574,
+      "learning_rate": 9.590217527251221e-06,
+      "loss": 0.3063,
+      "step": 728
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 16.49262046813965,
+      "learning_rate": 9.589106667109897e-06,
+      "loss": 0.7196,
+      "step": 729
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 11.748454093933105,
+      "learning_rate": 9.58799436783149e-06,
+      "loss": 0.6179,
+      "step": 730
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 7.200084686279297,
+      "learning_rate": 9.586880629764817e-06,
+      "loss": 0.2513,
+      "step": 731
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 21.336111068725586,
+      "learning_rate": 9.585765453259144e-06,
+      "loss": 0.4404,
+      "step": 732
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 9.39238166809082,
+      "learning_rate": 9.584648838664187e-06,
+      "loss": 0.2966,
+      "step": 733
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 14.571737289428711,
+      "learning_rate": 9.583530786330115e-06,
+      "loss": 0.5261,
+      "step": 734
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 8.34515380859375,
+      "learning_rate": 9.58241129660755e-06,
+      "loss": 0.2378,
+      "step": 735
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 11.729763984680176,
+      "learning_rate": 9.581290369847562e-06,
+      "loss": 0.294,
+      "step": 736
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 12.128561973571777,
+      "learning_rate": 9.58016800640167e-06,
+      "loss": 0.4317,
+      "step": 737
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 11.640527725219727,
+      "learning_rate": 9.579044206621844e-06,
+      "loss": 0.5247,
+      "step": 738
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 12.281421661376953,
+      "learning_rate": 9.577918970860509e-06,
+      "loss": 0.4627,
+      "step": 739
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 12.351391792297363,
+      "learning_rate": 9.576792299470537e-06,
+      "loss": 0.4098,
+      "step": 740
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 13.537297248840332,
+      "learning_rate": 9.575664192805252e-06,
+      "loss": 0.586,
+      "step": 741
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 14.388579368591309,
+      "learning_rate": 9.574534651218421e-06,
+      "loss": 0.5819,
+      "step": 742
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 12.514777183532715,
+      "learning_rate": 9.573403675064272e-06,
+      "loss": 0.5127,
+      "step": 743
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 14.211150169372559,
+      "learning_rate": 9.572271264697476e-06,
+      "loss": 0.4805,
+      "step": 744
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 11.094466209411621,
+      "learning_rate": 9.571137420473154e-06,
+      "loss": 0.5127,
+      "step": 745
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 15.266779899597168,
+      "learning_rate": 9.57000214274688e-06,
+      "loss": 0.4929,
+      "step": 746
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 13.651497840881348,
+      "learning_rate": 9.568865431874673e-06,
+      "loss": 0.5759,
+      "step": 747
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 10.449331283569336,
+      "learning_rate": 9.567727288213005e-06,
+      "loss": 0.4218,
+      "step": 748
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 9.863807678222656,
+      "learning_rate": 9.566587712118797e-06,
+      "loss": 0.4041,
+      "step": 749
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 9.361985206604004,
+      "learning_rate": 9.565446703949417e-06,
+      "loss": 0.3421,
+      "step": 750
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 21.406614303588867,
+      "learning_rate": 9.564304264062684e-06,
+      "loss": 0.4843,
+      "step": 751
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 15.96340560913086,
+      "learning_rate": 9.563160392816866e-06,
+      "loss": 0.4748,
+      "step": 752
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 8.502838134765625,
+      "learning_rate": 9.562015090570677e-06,
+      "loss": 0.4517,
+      "step": 753
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 10.185551643371582,
+      "learning_rate": 9.560868357683287e-06,
+      "loss": 0.337,
+      "step": 754
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 9.848067283630371,
+      "learning_rate": 9.559720194514303e-06,
+      "loss": 0.4812,
+      "step": 755
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 10.440569877624512,
+      "learning_rate": 9.558570601423794e-06,
+      "loss": 0.4051,
+      "step": 756
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 9.96517562866211,
+      "learning_rate": 9.557419578772266e-06,
+      "loss": 0.3031,
+      "step": 757
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 11.18921184539795,
+      "learning_rate": 9.556267126920679e-06,
+      "loss": 0.531,
+      "step": 758
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 12.531009674072266,
+      "learning_rate": 9.555113246230443e-06,
+      "loss": 0.6628,
+      "step": 759
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 6.8645339012146,
+      "learning_rate": 9.55395793706341e-06,
+      "loss": 0.5201,
+      "step": 760
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 10.674735069274902,
+      "learning_rate": 9.552801199781884e-06,
+      "loss": 0.5631,
+      "step": 761
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 9.336393356323242,
+      "learning_rate": 9.551643034748616e-06,
+      "loss": 0.3099,
+      "step": 762
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 12.368720054626465,
+      "learning_rate": 9.550483442326807e-06,
+      "loss": 0.3885,
+      "step": 763
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 8.859185218811035,
+      "learning_rate": 9.5493224228801e-06,
+      "loss": 0.2443,
+      "step": 764
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 19.69757080078125,
+      "learning_rate": 9.548159976772593e-06,
+      "loss": 0.6071,
+      "step": 765
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 8.464173316955566,
+      "learning_rate": 9.546996104368824e-06,
+      "loss": 0.2546,
+      "step": 766
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 10.251245498657227,
+      "learning_rate": 9.545830806033783e-06,
+      "loss": 0.4179,
+      "step": 767
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 15.080374717712402,
+      "learning_rate": 9.544664082132904e-06,
+      "loss": 0.5226,
+      "step": 768
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 9.251607894897461,
+      "learning_rate": 9.543495933032074e-06,
+      "loss": 0.4988,
+      "step": 769
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 7.7165422439575195,
+      "learning_rate": 9.542326359097619e-06,
+      "loss": 0.2638,
+      "step": 770
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 10.366019248962402,
+      "learning_rate": 9.541155360696319e-06,
+      "loss": 0.4128,
+      "step": 771
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 6.815579414367676,
+      "learning_rate": 9.539982938195394e-06,
+      "loss": 0.2159,
+      "step": 772
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 7.597775459289551,
+      "learning_rate": 9.538809091962515e-06,
+      "loss": 0.4332,
+      "step": 773
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 8.600618362426758,
+      "learning_rate": 9.537633822365798e-06,
+      "loss": 0.4133,
+      "step": 774
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 11.554625511169434,
+      "learning_rate": 9.536457129773808e-06,
+      "loss": 0.372,
+      "step": 775
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 13.994037628173828,
+      "learning_rate": 9.535279014555552e-06,
+      "loss": 0.4815,
+      "step": 776
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 11.78786563873291,
+      "learning_rate": 9.534099477080487e-06,
+      "loss": 0.5358,
+      "step": 777
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 15.103484153747559,
+      "learning_rate": 9.532918517718512e-06,
+      "loss": 0.4052,
+      "step": 778
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 11.817140579223633,
+      "learning_rate": 9.531736136839978e-06,
+      "loss": 0.4216,
+      "step": 779
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 11.051838874816895,
+      "learning_rate": 9.530552334815672e-06,
+      "loss": 0.5904,
+      "step": 780
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 9.852317810058594,
+      "learning_rate": 9.529367112016836e-06,
+      "loss": 0.2019,
+      "step": 781
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 12.343427658081055,
+      "learning_rate": 9.528180468815155e-06,
+      "loss": 0.4859,
+      "step": 782
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 8.242925643920898,
+      "learning_rate": 9.526992405582757e-06,
+      "loss": 0.4809,
+      "step": 783
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 9.652149200439453,
+      "learning_rate": 9.525802922692217e-06,
+      "loss": 0.4297,
+      "step": 784
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 8.826685905456543,
+      "learning_rate": 9.524612020516556e-06,
+      "loss": 0.5167,
+      "step": 785
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 9.229496002197266,
+      "learning_rate": 9.523419699429238e-06,
+      "loss": 0.4212,
+      "step": 786
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 7.060328960418701,
+      "learning_rate": 9.522225959804175e-06,
+      "loss": 0.2109,
+      "step": 787
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 11.581299781799316,
+      "learning_rate": 9.521030802015719e-06,
+      "loss": 0.4952,
+      "step": 788
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 12.459565162658691,
+      "learning_rate": 9.519834226438671e-06,
+      "loss": 0.5033,
+      "step": 789
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 9.49426555633545,
+      "learning_rate": 9.518636233448276e-06,
+      "loss": 0.5072,
+      "step": 790
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 8.173099517822266,
+      "learning_rate": 9.517436823420223e-06,
+      "loss": 0.3413,
+      "step": 791
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 9.29448127746582,
+      "learning_rate": 9.516235996730645e-06,
+      "loss": 0.3345,
+      "step": 792
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 13.833184242248535,
+      "learning_rate": 9.51503375375612e-06,
+      "loss": 0.4414,
+      "step": 793
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 12.314870834350586,
+      "learning_rate": 9.513830094873667e-06,
+      "loss": 0.3805,
+      "step": 794
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 10.87198543548584,
+      "learning_rate": 9.512625020460754e-06,
+      "loss": 0.4382,
+      "step": 795
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 12.025700569152832,
+      "learning_rate": 9.51141853089529e-06,
+      "loss": 0.5597,
+      "step": 796
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 10.935479164123535,
+      "learning_rate": 9.51021062655563e-06,
+      "loss": 0.3417,
+      "step": 797
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 9.5264892578125,
+      "learning_rate": 9.509001307820568e-06,
+      "loss": 0.3336,
+      "step": 798
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 7.905416488647461,
+      "learning_rate": 9.507790575069347e-06,
+      "loss": 0.3419,
+      "step": 799
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 9.485803604125977,
+      "learning_rate": 9.506578428681648e-06,
+      "loss": 0.3415,
+      "step": 800
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 12.703716278076172,
+      "learning_rate": 9.505364869037602e-06,
+      "loss": 0.4308,
+      "step": 801
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 7.652472496032715,
+      "learning_rate": 9.504149896517778e-06,
+      "loss": 0.3044,
+      "step": 802
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 8.386664390563965,
+      "learning_rate": 9.502933511503187e-06,
+      "loss": 0.2049,
+      "step": 803
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 10.396738052368164,
+      "learning_rate": 9.50171571437529e-06,
+      "loss": 0.3725,
+      "step": 804
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 9.804624557495117,
+      "learning_rate": 9.500496505515986e-06,
+      "loss": 0.4009,
+      "step": 805
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 16.55901336669922,
+      "learning_rate": 9.499275885307612e-06,
+      "loss": 0.5164,
+      "step": 806
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 9.77879524230957,
+      "learning_rate": 9.498053854132958e-06,
+      "loss": 0.3551,
+      "step": 807
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 9.307594299316406,
+      "learning_rate": 9.496830412375248e-06,
+      "loss": 0.3976,
+      "step": 808
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 11.895771980285645,
+      "learning_rate": 9.495605560418155e-06,
+      "loss": 0.4554,
+      "step": 809
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 10.190606117248535,
+      "learning_rate": 9.494379298645788e-06,
+      "loss": 0.5117,
+      "step": 810
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 7.086047649383545,
+      "learning_rate": 9.4931516274427e-06,
+      "loss": 0.3215,
+      "step": 811
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 8.68492317199707,
+      "learning_rate": 9.49192254719389e-06,
+      "loss": 0.4246,
+      "step": 812
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 8.77740478515625,
+      "learning_rate": 9.490692058284795e-06,
+      "loss": 0.2877,
+      "step": 813
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 11.375448226928711,
+      "learning_rate": 9.489460161101291e-06,
+      "loss": 0.5247,
+      "step": 814
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 10.664010047912598,
+      "learning_rate": 9.488226856029704e-06,
+      "loss": 0.3172,
+      "step": 815
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 10.560041427612305,
+      "learning_rate": 9.486992143456792e-06,
+      "loss": 0.4305,
+      "step": 816
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 8.990984916687012,
+      "learning_rate": 9.485756023769763e-06,
+      "loss": 0.4148,
+      "step": 817
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 12.359453201293945,
+      "learning_rate": 9.48451849735626e-06,
+      "loss": 0.4097,
+      "step": 818
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 9.949883460998535,
+      "learning_rate": 9.483279564604374e-06,
+      "loss": 0.4632,
+      "step": 819
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 14.918018341064453,
+      "learning_rate": 9.482039225902623e-06,
+      "loss": 0.6416,
+      "step": 820
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 9.546536445617676,
+      "learning_rate": 9.480797481639985e-06,
+      "loss": 0.3248,
+      "step": 821
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 6.98854398727417,
+      "learning_rate": 9.479554332205865e-06,
+      "loss": 0.2718,
+      "step": 822
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 12.81551742553711,
+      "learning_rate": 9.478309777990111e-06,
+      "loss": 0.4946,
+      "step": 823
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 9.111442565917969,
+      "learning_rate": 9.477063819383017e-06,
+      "loss": 0.4223,
+      "step": 824
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 11.319433212280273,
+      "learning_rate": 9.475816456775313e-06,
+      "loss": 0.4324,
+      "step": 825
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 11.37824821472168,
+      "learning_rate": 9.474567690558168e-06,
+      "loss": 0.407,
+      "step": 826
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 12.735369682312012,
+      "learning_rate": 9.473317521123195e-06,
+      "loss": 0.3475,
+      "step": 827
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 11.709153175354004,
+      "learning_rate": 9.472065948862445e-06,
+      "loss": 0.426,
+      "step": 828
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 12.897237777709961,
+      "learning_rate": 9.470812974168409e-06,
+      "loss": 0.3785,
+      "step": 829
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 10.927692413330078,
+      "learning_rate": 9.469558597434018e-06,
+      "loss": 0.3216,
+      "step": 830
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 10.907355308532715,
+      "learning_rate": 9.468302819052643e-06,
+      "loss": 0.3684,
+      "step": 831
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 11.495699882507324,
+      "learning_rate": 9.467045639418093e-06,
+      "loss": 0.3774,
+      "step": 832
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 8.682900428771973,
+      "learning_rate": 9.46578705892462e-06,
+      "loss": 0.2996,
+      "step": 833
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 19.422754287719727,
+      "learning_rate": 9.464527077966912e-06,
+      "loss": 0.542,
+      "step": 834
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 10.318727493286133,
+      "learning_rate": 9.463265696940095e-06,
+      "loss": 0.4165,
+      "step": 835
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 11.116280555725098,
+      "learning_rate": 9.46200291623974e-06,
+      "loss": 0.283,
+      "step": 836
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 13.530844688415527,
+      "learning_rate": 9.46073873626185e-06,
+      "loss": 0.3528,
+      "step": 837
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 13.350300788879395,
+      "learning_rate": 9.459473157402871e-06,
+      "loss": 0.5507,
+      "step": 838
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 11.107094764709473,
+      "learning_rate": 9.458206180059689e-06,
+      "loss": 0.404,
+      "step": 839
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 13.291266441345215,
+      "learning_rate": 9.456937804629623e-06,
+      "loss": 0.3754,
+      "step": 840
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 9.819769859313965,
+      "learning_rate": 9.455668031510434e-06,
+      "loss": 0.4533,
+      "step": 841
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 12.221588134765625,
+      "learning_rate": 9.454396861100323e-06,
+      "loss": 0.3944,
+      "step": 842
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 9.889158248901367,
+      "learning_rate": 9.453124293797926e-06,
+      "loss": 0.4616,
+      "step": 843
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 11.957451820373535,
+      "learning_rate": 9.451850330002318e-06,
+      "loss": 0.4559,
+      "step": 844
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 11.827383041381836,
+      "learning_rate": 9.45057497011301e-06,
+      "loss": 0.3172,
+      "step": 845
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 13.511688232421875,
+      "learning_rate": 9.44929821452996e-06,
+      "loss": 0.3302,
+      "step": 846
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 12.380231857299805,
+      "learning_rate": 9.44802006365355e-06,
+      "loss": 0.5164,
+      "step": 847
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 10.965176582336426,
+      "learning_rate": 9.446740517884606e-06,
+      "loss": 0.3623,
+      "step": 848
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 9.935227394104004,
+      "learning_rate": 9.445459577624397e-06,
+      "loss": 0.4944,
+      "step": 849
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 14.081665992736816,
+      "learning_rate": 9.444177243274619e-06,
+      "loss": 0.401,
+      "step": 850
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 9.65530776977539,
+      "learning_rate": 9.44289351523741e-06,
+      "loss": 0.2176,
+      "step": 851
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 8.725106239318848,
+      "learning_rate": 9.441608393915352e-06,
+      "loss": 0.1872,
+      "step": 852
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 9.504620552062988,
+      "learning_rate": 9.440321879711449e-06,
+      "loss": 0.458,
+      "step": 853
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 13.653008460998535,
+      "learning_rate": 9.439033973029155e-06,
+      "loss": 0.3139,
+      "step": 854
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 8.611638069152832,
+      "learning_rate": 9.437744674272353e-06,
+      "loss": 0.3091,
+      "step": 855
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 8.568094253540039,
+      "learning_rate": 9.436453983845367e-06,
+      "loss": 0.2619,
+      "step": 856
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 25.40271759033203,
+      "learning_rate": 9.435161902152955e-06,
+      "loss": 0.2856,
+      "step": 857
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 10.678138732910156,
+      "learning_rate": 9.43386842960031e-06,
+      "loss": 0.4684,
+      "step": 858
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 9.498933792114258,
+      "learning_rate": 9.432573566593066e-06,
+      "loss": 0.2929,
+      "step": 859
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 12.769570350646973,
+      "learning_rate": 9.43127731353729e-06,
+      "loss": 0.5847,
+      "step": 860
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 9.36600112915039,
+      "learning_rate": 9.429979670839481e-06,
+      "loss": 0.3014,
+      "step": 861
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 8.011848449707031,
+      "learning_rate": 9.428680638906585e-06,
+      "loss": 0.3755,
+      "step": 862
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 8.896998405456543,
+      "learning_rate": 9.427380218145969e-06,
+      "loss": 0.3198,
+      "step": 863
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 7.280299663543701,
+      "learning_rate": 9.426078408965446e-06,
+      "loss": 0.3366,
+      "step": 864
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 10.357894897460938,
+      "learning_rate": 9.424775211773263e-06,
+      "loss": 0.3158,
+      "step": 865
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 18.96193504333496,
+      "learning_rate": 9.4234706269781e-06,
+      "loss": 0.9163,
+      "step": 866
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 13.815512657165527,
+      "learning_rate": 9.422164654989073e-06,
+      "loss": 0.4943,
+      "step": 867
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 10.81698989868164,
+      "learning_rate": 9.42085729621573e-06,
+      "loss": 0.3018,
+      "step": 868
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 13.414194107055664,
+      "learning_rate": 9.419548551068061e-06,
+      "loss": 0.4866,
+      "step": 869
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 10.096065521240234,
+      "learning_rate": 9.418238419956484e-06,
+      "loss": 0.4585,
+      "step": 870
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 10.861050605773926,
+      "learning_rate": 9.416926903291855e-06,
+      "loss": 0.4053,
+      "step": 871
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 9.277708053588867,
+      "learning_rate": 9.415614001485462e-06,
+      "loss": 0.3299,
+      "step": 872
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 11.696009635925293,
+      "learning_rate": 9.414299714949031e-06,
+      "loss": 0.3861,
+      "step": 873
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 10.554948806762695,
+      "learning_rate": 9.412984044094722e-06,
+      "loss": 0.3066,
+      "step": 874
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 10.93411636352539,
+      "learning_rate": 9.411666989335123e-06,
+      "loss": 0.575,
+      "step": 875
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 10.087698936462402,
+      "learning_rate": 9.410348551083264e-06,
+      "loss": 0.4311,
+      "step": 876
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 14.454934120178223,
+      "learning_rate": 9.409028729752604e-06,
+      "loss": 0.5489,
+      "step": 877
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 9.015199661254883,
+      "learning_rate": 9.407707525757036e-06,
+      "loss": 0.3095,
+      "step": 878
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 11.484269142150879,
+      "learning_rate": 9.406384939510888e-06,
+      "loss": 0.5324,
+      "step": 879
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 14.328922271728516,
+      "learning_rate": 9.405060971428924e-06,
+      "loss": 0.4184,
+      "step": 880
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 14.995519638061523,
+      "learning_rate": 9.403735621926334e-06,
+      "loss": 0.3623,
+      "step": 881
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 11.692059516906738,
+      "learning_rate": 9.40240889141875e-06,
+      "loss": 0.4,
+      "step": 882
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 10.946962356567383,
+      "learning_rate": 9.401080780322229e-06,
+      "loss": 0.466,
+      "step": 883
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 11.723125457763672,
+      "learning_rate": 9.399751289053267e-06,
+      "loss": 0.3561,
+      "step": 884
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 12.1270112991333,
+      "learning_rate": 9.398420418028789e-06,
+      "loss": 0.4878,
+      "step": 885
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 12.791010856628418,
+      "learning_rate": 9.397088167666156e-06,
+      "loss": 0.4641,
+      "step": 886
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 9.945054054260254,
+      "learning_rate": 9.395754538383157e-06,
+      "loss": 0.5316,
+      "step": 887
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 10.215434074401855,
+      "learning_rate": 9.39441953059802e-06,
+      "loss": 0.3277,
+      "step": 888
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 12.030582427978516,
+      "learning_rate": 9.393083144729398e-06,
+      "loss": 0.5057,
+      "step": 889
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 9.404858589172363,
+      "learning_rate": 9.391745381196382e-06,
+      "loss": 0.3728,
+      "step": 890
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 11.535151481628418,
+      "learning_rate": 9.39040624041849e-06,
+      "loss": 0.4832,
+      "step": 891
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 12.403703689575195,
+      "learning_rate": 9.389065722815678e-06,
+      "loss": 0.3646,
+      "step": 892
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 10.468901634216309,
+      "learning_rate": 9.387723828808326e-06,
+      "loss": 0.3831,
+      "step": 893
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 14.21241283416748,
+      "learning_rate": 9.386380558817257e-06,
+      "loss": 0.4908,
+      "step": 894
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 8.63652229309082,
+      "learning_rate": 9.38503591326371e-06,
+      "loss": 0.2946,
+      "step": 895
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 15.784919738769531,
+      "learning_rate": 9.38368989256937e-06,
+      "loss": 0.5847,
+      "step": 896
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 9.040005683898926,
+      "learning_rate": 9.382342497156344e-06,
+      "loss": 0.3845,
+      "step": 897
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 8.899520874023438,
+      "learning_rate": 9.380993727447174e-06,
+      "loss": 0.3095,
+      "step": 898
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 8.127609252929688,
+      "learning_rate": 9.379643583864832e-06,
+      "loss": 0.2849,
+      "step": 899
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 9.3443603515625,
+      "learning_rate": 9.378292066832723e-06,
+      "loss": 0.3933,
+      "step": 900
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 9.566666603088379,
+      "learning_rate": 9.376939176774678e-06,
+      "loss": 0.3722,
+      "step": 901
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 6.643682479858398,
+      "learning_rate": 9.375584914114963e-06,
+      "loss": 0.2718,
+      "step": 902
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 11.019343376159668,
+      "learning_rate": 9.374229279278272e-06,
+      "loss": 0.4383,
+      "step": 903
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 6.726876735687256,
+      "learning_rate": 9.372872272689733e-06,
+      "loss": 0.3436,
+      "step": 904
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 10.366143226623535,
+      "learning_rate": 9.371513894774894e-06,
+      "loss": 0.49,
+      "step": 905
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 17.453453063964844,
+      "learning_rate": 9.370154145959748e-06,
+      "loss": 0.5759,
+      "step": 906
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 9.634583473205566,
+      "learning_rate": 9.368793026670708e-06,
+      "loss": 0.3035,
+      "step": 907
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 7.39124870300293,
+      "learning_rate": 9.367430537334617e-06,
+      "loss": 0.2482,
+      "step": 908
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 14.085043907165527,
+      "learning_rate": 9.36606667837875e-06,
+      "loss": 0.3275,
+      "step": 909
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 16.50173568725586,
+      "learning_rate": 9.364701450230813e-06,
+      "loss": 0.3624,
+      "step": 910
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 11.782758712768555,
+      "learning_rate": 9.363334853318937e-06,
+      "loss": 0.53,
+      "step": 911
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 8.769566535949707,
+      "learning_rate": 9.361966888071687e-06,
+      "loss": 0.4453,
+      "step": 912
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 13.31277084350586,
+      "learning_rate": 9.360597554918055e-06,
+      "loss": 0.3724,
+      "step": 913
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 8.946446418762207,
+      "learning_rate": 9.35922685428746e-06,
+      "loss": 0.3581,
+      "step": 914
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 13.202397346496582,
+      "learning_rate": 9.357854786609754e-06,
+      "loss": 0.5437,
+      "step": 915
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 12.494349479675293,
+      "learning_rate": 9.356481352315215e-06,
+      "loss": 0.3541,
+      "step": 916
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 8.650835037231445,
+      "learning_rate": 9.355106551834544e-06,
+      "loss": 0.3011,
+      "step": 917
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 10.84846305847168,
+      "learning_rate": 9.353730385598887e-06,
+      "loss": 0.4084,
+      "step": 918
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 10.4662504196167,
+      "learning_rate": 9.3523528540398e-06,
+      "loss": 0.2422,
+      "step": 919
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 10.166276931762695,
+      "learning_rate": 9.350973957589278e-06,
+      "loss": 0.4647,
+      "step": 920
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 10.154031753540039,
+      "learning_rate": 9.349593696679739e-06,
+      "loss": 0.4037,
+      "step": 921
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 9.689313888549805,
+      "learning_rate": 9.348212071744032e-06,
+      "loss": 0.2694,
+      "step": 922
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 11.248096466064453,
+      "learning_rate": 9.346829083215432e-06,
+      "loss": 0.3446,
+      "step": 923
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 14.71185302734375,
+      "learning_rate": 9.345444731527642e-06,
+      "loss": 0.4066,
+      "step": 924
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 8.16855239868164,
+      "learning_rate": 9.344059017114796e-06,
+      "loss": 0.2871,
+      "step": 925
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 10.573026657104492,
+      "learning_rate": 9.342671940411445e-06,
+      "loss": 0.3928,
+      "step": 926
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 11.692510604858398,
+      "learning_rate": 9.341283501852581e-06,
+      "loss": 0.4834,
+      "step": 927
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 12.358516693115234,
+      "learning_rate": 9.339893701873612e-06,
+      "loss": 0.5677,
+      "step": 928
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 8.96460247039795,
+      "learning_rate": 9.338502540910379e-06,
+      "loss": 0.4149,
+      "step": 929
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 9.153914451599121,
+      "learning_rate": 9.33711001939915e-06,
+      "loss": 0.3559,
+      "step": 930
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 10.490222930908203,
+      "learning_rate": 9.335716137776614e-06,
+      "loss": 0.3643,
+      "step": 931
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 11.842387199401855,
+      "learning_rate": 9.334320896479891e-06,
+      "loss": 0.5491,
+      "step": 932
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 9.103155136108398,
+      "learning_rate": 9.332924295946528e-06,
+      "loss": 0.5902,
+      "step": 933
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 14.908157348632812,
+      "learning_rate": 9.331526336614496e-06,
+      "loss": 0.4039,
+      "step": 934
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 8.916375160217285,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 0.4306,
+      "step": 935
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 8.566136360168457,
+      "learning_rate": 9.328726343308446e-06,
+      "loss": 0.3571,
+      "step": 936
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 8.376839637756348,
+      "learning_rate": 9.327324310212498e-06,
+      "loss": 0.2454,
+      "step": 937
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 11.007676124572754,
+      "learning_rate": 9.325920920074031e-06,
+      "loss": 0.2805,
+      "step": 938
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 12.792417526245117,
+      "learning_rate": 9.324516173333142e-06,
+      "loss": 0.4966,
+      "step": 939
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 11.823273658752441,
+      "learning_rate": 9.32311007043036e-06,
+      "loss": 0.6791,
+      "step": 940
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 7.2026753425598145,
+      "learning_rate": 9.321702611806637e-06,
+      "loss": 0.2797,
+      "step": 941
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 10.273791313171387,
+      "learning_rate": 9.320293797903348e-06,
+      "loss": 0.305,
+      "step": 942
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 9.314358711242676,
+      "learning_rate": 9.318883629162294e-06,
+      "loss": 0.5165,
+      "step": 943
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 9.2044677734375,
+      "learning_rate": 9.317472106025706e-06,
+      "loss": 0.5579,
+      "step": 944
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 12.091362953186035,
+      "learning_rate": 9.316059228936231e-06,
+      "loss": 0.3511,
+      "step": 945
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 9.036722183227539,
+      "learning_rate": 9.31464499833695e-06,
+      "loss": 0.5667,
+      "step": 946
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 8.445394515991211,
+      "learning_rate": 9.31322941467136e-06,
+      "loss": 0.4823,
+      "step": 947
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 10.47631549835205,
+      "learning_rate": 9.311812478383386e-06,
+      "loss": 0.6102,
+      "step": 948
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 14.514805793762207,
+      "learning_rate": 9.310394189917377e-06,
+      "loss": 0.3757,
+      "step": 949
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 8.314196586608887,
+      "learning_rate": 9.30897454971811e-06,
+      "loss": 0.2103,
+      "step": 950
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 6.912398338317871,
+      "learning_rate": 9.307553558230777e-06,
+      "loss": 0.2487,
+      "step": 951
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 7.86810827255249,
+      "learning_rate": 9.306131215901004e-06,
+      "loss": 0.2035,
+      "step": 952
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 10.778363227844238,
+      "learning_rate": 9.304707523174831e-06,
+      "loss": 0.4206,
+      "step": 953
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 21.13538360595703,
+      "learning_rate": 9.303282480498728e-06,
+      "loss": 0.335,
+      "step": 954
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 11.219478607177734,
+      "learning_rate": 9.301856088319584e-06,
+      "loss": 0.3563,
+      "step": 955
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 10.69076156616211,
+      "learning_rate": 9.300428347084718e-06,
+      "loss": 0.4398,
+      "step": 956
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 12.649370193481445,
+      "learning_rate": 9.298999257241862e-06,
+      "loss": 0.3718,
+      "step": 957
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 11.996403694152832,
+      "learning_rate": 9.297568819239182e-06,
+      "loss": 0.6114,
+      "step": 958
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 9.951818466186523,
+      "learning_rate": 9.296137033525256e-06,
+      "loss": 0.3397,
+      "step": 959
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 12.51860237121582,
+      "learning_rate": 9.294703900549096e-06,
+      "loss": 0.4175,
+      "step": 960
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 14.227179527282715,
+      "learning_rate": 9.293269420760125e-06,
+      "loss": 0.5373,
+      "step": 961
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 9.520474433898926,
+      "learning_rate": 9.291833594608194e-06,
+      "loss": 0.4111,
+      "step": 962
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 9.914743423461914,
+      "learning_rate": 9.290396422543579e-06,
+      "loss": 0.464,
+      "step": 963
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 11.476784706115723,
+      "learning_rate": 9.288957905016975e-06,
+      "loss": 0.3418,
+      "step": 964
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 12.75112247467041,
+      "learning_rate": 9.287518042479495e-06,
+      "loss": 0.4667,
+      "step": 965
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 9.181680679321289,
+      "learning_rate": 9.286076835382682e-06,
+      "loss": 0.3569,
+      "step": 966
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 10.829106330871582,
+      "learning_rate": 9.284634284178496e-06,
+      "loss": 0.4677,
+      "step": 967
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 8.990328788757324,
+      "learning_rate": 9.283190389319315e-06,
+      "loss": 0.2908,
+      "step": 968
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 9.712418556213379,
+      "learning_rate": 9.281745151257946e-06,
+      "loss": 0.228,
+      "step": 969
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 6.227433681488037,
+      "learning_rate": 9.280298570447612e-06,
+      "loss": 0.1702,
+      "step": 970
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 8.63801383972168,
+      "learning_rate": 9.278850647341959e-06,
+      "loss": 0.3824,
+      "step": 971
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 10.084657669067383,
+      "learning_rate": 9.277401382395055e-06,
+      "loss": 0.2055,
+      "step": 972
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 11.15583610534668,
+      "learning_rate": 9.275950776061387e-06,
+      "loss": 0.3804,
+      "step": 973
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 14.605420112609863,
+      "learning_rate": 9.274498828795859e-06,
+      "loss": 0.5436,
+      "step": 974
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 7.999274253845215,
+      "learning_rate": 9.273045541053805e-06,
+      "loss": 0.4115,
+      "step": 975
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 9.766670227050781,
+      "learning_rate": 9.27159091329097e-06,
+      "loss": 0.3851,
+      "step": 976
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 13.886274337768555,
+      "learning_rate": 9.270134945963525e-06,
+      "loss": 0.4903,
+      "step": 977
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 12.445950508117676,
+      "learning_rate": 9.268677639528058e-06,
+      "loss": 0.533,
+      "step": 978
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 14.963641166687012,
+      "learning_rate": 9.26721899444158e-06,
+      "loss": 0.4364,
+      "step": 979
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 13.528694152832031,
+      "learning_rate": 9.265759011161519e-06,
+      "loss": 0.6435,
+      "step": 980
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 12.292191505432129,
+      "learning_rate": 9.264297690145722e-06,
+      "loss": 0.3739,
+      "step": 981
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 15.456099510192871,
+      "learning_rate": 9.262835031852462e-06,
+      "loss": 0.4686,
+      "step": 982
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 10.406805992126465,
+      "learning_rate": 9.261371036740421e-06,
+      "loss": 0.3508,
+      "step": 983
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 14.224225044250488,
+      "learning_rate": 9.259905705268705e-06,
+      "loss": 0.5707,
+      "step": 984
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 7.832348823547363,
+      "learning_rate": 9.258439037896846e-06,
+      "loss": 0.3039,
+      "step": 985
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 12.104130744934082,
+      "learning_rate": 9.256971035084786e-06,
+      "loss": 0.3366,
+      "step": 986
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 6.054928302764893,
+      "learning_rate": 9.255501697292886e-06,
+      "loss": 0.2184,
+      "step": 987
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 11.262185096740723,
+      "learning_rate": 9.25403102498193e-06,
+      "loss": 0.3812,
+      "step": 988
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 14.063738822937012,
+      "learning_rate": 9.252559018613118e-06,
+      "loss": 0.5521,
+      "step": 989
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 13.673068046569824,
+      "learning_rate": 9.251085678648072e-06,
+      "loss": 0.4072,
+      "step": 990
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 10.417755126953125,
+      "learning_rate": 9.249611005548825e-06,
+      "loss": 0.5259,
+      "step": 991
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 14.696074485778809,
+      "learning_rate": 9.248134999777836e-06,
+      "loss": 0.6522,
+      "step": 992
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 13.853986740112305,
+      "learning_rate": 9.246657661797975e-06,
+      "loss": 0.3731,
+      "step": 993
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 9.62138843536377,
+      "learning_rate": 9.245178992072534e-06,
+      "loss": 0.3015,
+      "step": 994
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 10.689718246459961,
+      "learning_rate": 9.243698991065222e-06,
+      "loss": 0.4203,
+      "step": 995
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 9.153942108154297,
+      "learning_rate": 9.242217659240167e-06,
+      "loss": 0.4231,
+      "step": 996
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 7.398947715759277,
+      "learning_rate": 9.24073499706191e-06,
+      "loss": 0.2462,
+      "step": 997
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 11.215632438659668,
+      "learning_rate": 9.239251004995411e-06,
+      "loss": 0.4332,
+      "step": 998
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 10.08993148803711,
+      "learning_rate": 9.237765683506051e-06,
+      "loss": 0.3275,
+      "step": 999
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 9.370858192443848,
+      "learning_rate": 9.236279033059622e-06,
+      "loss": 0.2515,
+      "step": 1000
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 10.58672046661377,
+      "learning_rate": 9.234791054122336e-06,
+      "loss": 0.3134,
+      "step": 1001
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 10.335655212402344,
+      "learning_rate": 9.233301747160822e-06,
+      "loss": 0.2831,
+      "step": 1002
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 6.278431415557861,
+      "learning_rate": 9.231811112642121e-06,
+      "loss": 0.258,
+      "step": 1003
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 7.914210796356201,
+      "learning_rate": 9.2303191510337e-06,
+      "loss": 0.1328,
+      "step": 1004
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 9.013747215270996,
+      "learning_rate": 9.22882586280343e-06,
+      "loss": 0.4805,
+      "step": 1005
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 12.271598815917969,
+      "learning_rate": 9.227331248419607e-06,
+      "loss": 0.5016,
+      "step": 1006
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 11.01476764678955,
+      "learning_rate": 9.22583530835094e-06,
+      "loss": 0.5436,
+      "step": 1007
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 9.752090454101562,
+      "learning_rate": 9.224338043066551e-06,
+      "loss": 0.3487,
+      "step": 1008
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 17.13884735107422,
+      "learning_rate": 9.222839453035983e-06,
+      "loss": 0.4311,
+      "step": 1009
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 6.206357002258301,
+      "learning_rate": 9.221339538729191e-06,
+      "loss": 0.2411,
+      "step": 1010
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 4.821128845214844,
+      "learning_rate": 9.219838300616547e-06,
+      "loss": 0.1562,
+      "step": 1011
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 10.753043174743652,
+      "learning_rate": 9.218335739168833e-06,
+      "loss": 0.3314,
+      "step": 1012
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 12.49563980102539,
+      "learning_rate": 9.216831854857253e-06,
+      "loss": 0.5401,
+      "step": 1013
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 13.130536079406738,
+      "learning_rate": 9.215326648153422e-06,
+      "loss": 0.3582,
+      "step": 1014
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 9.94231128692627,
+      "learning_rate": 9.213820119529372e-06,
+      "loss": 0.4336,
+      "step": 1015
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 7.196554660797119,
+      "learning_rate": 9.212312269457548e-06,
+      "loss": 0.2216,
+      "step": 1016
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 9.980297088623047,
+      "learning_rate": 9.210803098410808e-06,
+      "loss": 0.308,
+      "step": 1017
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 10.505521774291992,
+      "learning_rate": 9.209292606862428e-06,
+      "loss": 0.4731,
+      "step": 1018
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 9.656366348266602,
+      "learning_rate": 9.207780795286092e-06,
+      "loss": 0.4082,
+      "step": 1019
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 10.515889167785645,
+      "learning_rate": 9.206267664155906e-06,
+      "loss": 0.2972,
+      "step": 1020
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 11.882335662841797,
+      "learning_rate": 9.204753213946384e-06,
+      "loss": 0.3528,
+      "step": 1021
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 9.498517990112305,
+      "learning_rate": 9.203237445132456e-06,
+      "loss": 0.4146,
+      "step": 1022
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 7.5630903244018555,
+      "learning_rate": 9.201720358189464e-06,
+      "loss": 0.3477,
+      "step": 1023
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 8.425713539123535,
+      "learning_rate": 9.200201953593165e-06,
+      "loss": 0.2385,
+      "step": 1024
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 10.989039421081543,
+      "learning_rate": 9.198682231819727e-06,
+      "loss": 0.4025,
+      "step": 1025
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 10.628202438354492,
+      "learning_rate": 9.197161193345735e-06,
+      "loss": 0.4688,
+      "step": 1026
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 10.5223970413208,
+      "learning_rate": 9.195638838648183e-06,
+      "loss": 0.4465,
+      "step": 1027
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 8.814974784851074,
+      "learning_rate": 9.194115168204477e-06,
+      "loss": 0.2842,
+      "step": 1028
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 9.587896347045898,
+      "learning_rate": 9.192590182492442e-06,
+      "loss": 0.3584,
+      "step": 1029
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 12.46182918548584,
+      "learning_rate": 9.191063881990308e-06,
+      "loss": 0.2971,
+      "step": 1030
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 11.11462116241455,
+      "learning_rate": 9.189536267176722e-06,
+      "loss": 0.3127,
+      "step": 1031
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 11.511488914489746,
+      "learning_rate": 9.188007338530742e-06,
+      "loss": 0.4087,
+      "step": 1032
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 10.705217361450195,
+      "learning_rate": 9.186477096531838e-06,
+      "loss": 0.5039,
+      "step": 1033
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 7.113280296325684,
+      "learning_rate": 9.18494554165989e-06,
+      "loss": 0.318,
+      "step": 1034
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 12.60806655883789,
+      "learning_rate": 9.183412674395193e-06,
+      "loss": 0.4479,
+      "step": 1035
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 10.459493637084961,
+      "learning_rate": 9.18187849521845e-06,
+      "loss": 0.4922,
+      "step": 1036
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 12.748026847839355,
+      "learning_rate": 9.18034300461078e-06,
+      "loss": 0.3516,
+      "step": 1037
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 9.650742530822754,
+      "learning_rate": 9.178806203053709e-06,
+      "loss": 0.4054,
+      "step": 1038
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 9.995577812194824,
+      "learning_rate": 9.177268091029174e-06,
+      "loss": 0.3252,
+      "step": 1039
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 9.052395820617676,
+      "learning_rate": 9.17572866901953e-06,
+      "loss": 0.3386,
+      "step": 1040
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 9.171243667602539,
+      "learning_rate": 9.17418793750753e-06,
+      "loss": 0.3866,
+      "step": 1041
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 9.999776840209961,
+      "learning_rate": 9.172645896976352e-06,
+      "loss": 0.3253,
+      "step": 1042
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 11.148120880126953,
+      "learning_rate": 9.171102547909574e-06,
+      "loss": 0.3841,
+      "step": 1043
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 16.198596954345703,
+      "learning_rate": 9.169557890791188e-06,
+      "loss": 0.6822,
+      "step": 1044
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 12.12597370147705,
+      "learning_rate": 9.168011926105598e-06,
+      "loss": 0.5451,
+      "step": 1045
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 8.961607933044434,
+      "learning_rate": 9.166464654337615e-06,
+      "loss": 0.2072,
+      "step": 1046
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 11.609721183776855,
+      "learning_rate": 9.164916075972462e-06,
+      "loss": 0.3372,
+      "step": 1047
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 7.592455863952637,
+      "learning_rate": 9.163366191495768e-06,
+      "loss": 0.2855,
+      "step": 1048
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 13.81691837310791,
+      "learning_rate": 9.16181500139358e-06,
+      "loss": 0.3369,
+      "step": 1049
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 10.63996696472168,
+      "learning_rate": 9.160262506152343e-06,
+      "loss": 0.3554,
+      "step": 1050
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 9.461499214172363,
+      "learning_rate": 9.15870870625892e-06,
+      "loss": 0.432,
+      "step": 1051
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 15.497865676879883,
+      "learning_rate": 9.15715360220058e-06,
+      "loss": 0.4531,
+      "step": 1052
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 15.934529304504395,
+      "learning_rate": 9.155597194465003e-06,
+      "loss": 0.5064,
+      "step": 1053
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 9.329200744628906,
+      "learning_rate": 9.154039483540273e-06,
+      "loss": 0.4149,
+      "step": 1054
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 8.372329711914062,
+      "learning_rate": 9.15248046991489e-06,
+      "loss": 0.3966,
+      "step": 1055
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 9.679365158081055,
+      "learning_rate": 9.150920154077753e-06,
+      "loss": 0.368,
+      "step": 1056
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 12.956290245056152,
+      "learning_rate": 9.14935853651818e-06,
+      "loss": 0.5139,
+      "step": 1057
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 11.64295768737793,
+      "learning_rate": 9.147795617725889e-06,
+      "loss": 0.435,
+      "step": 1058
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 10.94347095489502,
+      "learning_rate": 9.14623139819101e-06,
+      "loss": 0.5361,
+      "step": 1059
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 12.231947898864746,
+      "learning_rate": 9.14466587840408e-06,
+      "loss": 0.3128,
+      "step": 1060
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 10.399981498718262,
+      "learning_rate": 9.143099058856042e-06,
+      "loss": 0.3558,
+      "step": 1061
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 9.469118118286133,
+      "learning_rate": 9.14153094003825e-06,
+      "loss": 0.31,
+      "step": 1062
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 10.397907257080078,
+      "learning_rate": 9.139961522442463e-06,
+      "loss": 0.3092,
+      "step": 1063
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 11.786216735839844,
+      "learning_rate": 9.138390806560849e-06,
+      "loss": 0.2645,
+      "step": 1064
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 9.598930358886719,
+      "learning_rate": 9.136818792885981e-06,
+      "loss": 0.2721,
+      "step": 1065
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 12.209094047546387,
+      "learning_rate": 9.135245481910842e-06,
+      "loss": 0.4819,
+      "step": 1066
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 21.54408073425293,
+      "learning_rate": 9.133670874128818e-06,
+      "loss": 0.3651,
+      "step": 1067
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 15.686665534973145,
+      "learning_rate": 9.132094970033706e-06,
+      "loss": 0.3611,
+      "step": 1068
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 12.140487670898438,
+      "learning_rate": 9.130517770119703e-06,
+      "loss": 0.3015,
+      "step": 1069
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 8.695904731750488,
+      "learning_rate": 9.12893927488142e-06,
+      "loss": 0.2479,
+      "step": 1070
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 12.503198623657227,
+      "learning_rate": 9.12735948481387e-06,
+      "loss": 0.6306,
+      "step": 1071
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 13.261467933654785,
+      "learning_rate": 9.125778400412474e-06,
+      "loss": 0.2994,
+      "step": 1072
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 24.588016510009766,
+      "learning_rate": 9.124196022173055e-06,
+      "loss": 0.48,
+      "step": 1073
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 12.782401084899902,
+      "learning_rate": 9.122612350591847e-06,
+      "loss": 0.2947,
+      "step": 1074
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 5.862837314605713,
+      "learning_rate": 9.121027386165487e-06,
+      "loss": 0.2799,
+      "step": 1075
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 8.968135833740234,
+      "learning_rate": 9.119441129391015e-06,
+      "loss": 0.4663,
+      "step": 1076
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 12.288074493408203,
+      "learning_rate": 9.117853580765881e-06,
+      "loss": 0.3198,
+      "step": 1077
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 14.748156547546387,
+      "learning_rate": 9.116264740787937e-06,
+      "loss": 0.333,
+      "step": 1078
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 10.767653465270996,
+      "learning_rate": 9.11467460995544e-06,
+      "loss": 0.2529,
+      "step": 1079
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 10.967245101928711,
+      "learning_rate": 9.113083188767057e-06,
+      "loss": 0.3507,
+      "step": 1080
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 8.183197021484375,
+      "learning_rate": 9.11149047772185e-06,
+      "loss": 0.2549,
+      "step": 1081
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 11.416047096252441,
+      "learning_rate": 9.109896477319293e-06,
+      "loss": 0.4891,
+      "step": 1082
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 16.294586181640625,
+      "learning_rate": 9.108301188059263e-06,
+      "loss": 0.3412,
+      "step": 1083
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 12.315059661865234,
+      "learning_rate": 9.10670461044204e-06,
+      "loss": 0.4574,
+      "step": 1084
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 10.211481094360352,
+      "learning_rate": 9.105106744968308e-06,
+      "loss": 0.5343,
+      "step": 1085
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 10.7732515335083,
+      "learning_rate": 9.103507592139154e-06,
+      "loss": 0.3864,
+      "step": 1086
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 7.140637397766113,
+      "learning_rate": 9.101907152456072e-06,
+      "loss": 0.281,
+      "step": 1087
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 13.590165138244629,
+      "learning_rate": 9.100305426420957e-06,
+      "loss": 0.3397,
+      "step": 1088
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 9.327340126037598,
+      "learning_rate": 9.098702414536107e-06,
+      "loss": 0.2212,
+      "step": 1089
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 15.51944351196289,
+      "learning_rate": 9.097098117304223e-06,
+      "loss": 0.5312,
+      "step": 1090
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 11.880060195922852,
+      "learning_rate": 9.095492535228414e-06,
+      "loss": 0.3152,
+      "step": 1091
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 9.794878005981445,
+      "learning_rate": 9.093885668812187e-06,
+      "loss": 0.6127,
+      "step": 1092
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 6.516966819763184,
+      "learning_rate": 9.092277518559452e-06,
+      "loss": 0.2333,
+      "step": 1093
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 9.361181259155273,
+      "learning_rate": 9.090668084974521e-06,
+      "loss": 0.3256,
+      "step": 1094
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 9.577570915222168,
+      "learning_rate": 9.089057368562113e-06,
+      "loss": 0.3706,
+      "step": 1095
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 10.970612525939941,
+      "learning_rate": 9.087445369827346e-06,
+      "loss": 0.5033,
+      "step": 1096
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 12.50491714477539,
+      "learning_rate": 9.085832089275737e-06,
+      "loss": 0.4938,
+      "step": 1097
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 9.271127700805664,
+      "learning_rate": 9.084217527413213e-06,
+      "loss": 0.4831,
+      "step": 1098
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 7.5547285079956055,
+      "learning_rate": 9.082601684746094e-06,
+      "loss": 0.5193,
+      "step": 1099
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 14.993159294128418,
+      "learning_rate": 9.08098456178111e-06,
+      "loss": 0.5224,
+      "step": 1100
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 10.327338218688965,
+      "learning_rate": 9.079366159025384e-06,
+      "loss": 0.4799,
+      "step": 1101
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 11.896102905273438,
+      "learning_rate": 9.077746476986449e-06,
+      "loss": 0.4142,
+      "step": 1102
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 12.115591049194336,
+      "learning_rate": 9.076125516172231e-06,
+      "loss": 0.3038,
+      "step": 1103
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 15.554551124572754,
+      "learning_rate": 9.074503277091065e-06,
+      "loss": 0.4671,
+      "step": 1104
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 8.743322372436523,
+      "learning_rate": 9.07287976025168e-06,
+      "loss": 0.3173,
+      "step": 1105
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 8.40587329864502,
+      "learning_rate": 9.071254966163208e-06,
+      "loss": 0.2994,
+      "step": 1106
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 12.941961288452148,
+      "learning_rate": 9.069628895335183e-06,
+      "loss": 0.5453,
+      "step": 1107
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 8.423136711120605,
+      "learning_rate": 9.06800154827754e-06,
+      "loss": 0.2844,
+      "step": 1108
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 16.55547523498535,
+      "learning_rate": 9.066372925500611e-06,
+      "loss": 0.5409,
+      "step": 1109
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 13.243544578552246,
+      "learning_rate": 9.064743027515127e-06,
+      "loss": 0.5216,
+      "step": 1110
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 12.495841979980469,
+      "learning_rate": 9.06311185483223e-06,
+      "loss": 0.2632,
+      "step": 1111
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 11.274785041809082,
+      "learning_rate": 9.061479407963443e-06,
+      "loss": 0.4644,
+      "step": 1112
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 13.965361595153809,
+      "learning_rate": 9.059845687420707e-06,
+      "loss": 0.4542,
+      "step": 1113
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 12.29924201965332,
+      "learning_rate": 9.058210693716348e-06,
+      "loss": 0.42,
+      "step": 1114
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 8.953022003173828,
+      "learning_rate": 9.056574427363102e-06,
+      "loss": 0.282,
+      "step": 1115
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 10.225995063781738,
+      "learning_rate": 9.0549368888741e-06,
+      "loss": 0.3179,
+      "step": 1116
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 28.219970703125,
+      "learning_rate": 9.05329807876287e-06,
+      "loss": 0.4842,
+      "step": 1117
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 12.042773246765137,
+      "learning_rate": 9.051657997543337e-06,
+      "loss": 0.3976,
+      "step": 1118
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 13.176989555358887,
+      "learning_rate": 9.050016645729835e-06,
+      "loss": 0.3755,
+      "step": 1119
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 12.683111190795898,
+      "learning_rate": 9.048374023837086e-06,
+      "loss": 0.4818,
+      "step": 1120
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 10.363407135009766,
+      "learning_rate": 9.046730132380213e-06,
+      "loss": 0.3862,
+      "step": 1121
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 7.114845275878906,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 0.2611,
+      "step": 1122
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 7.369638442993164,
+      "learning_rate": 9.043438542836583e-06,
+      "loss": 0.2868,
+      "step": 1123
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8605388272583201,
+      "eval_loss": 0.4313778579235077,
+      "eval_runtime": 11.3273,
+      "eval_samples_per_second": 167.119,
+      "eval_steps_per_second": 20.923,
+      "step": 1123
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 11.533417701721191,
+      "learning_rate": 9.041790845782064e-06,
+      "loss": 0.373,
+      "step": 1124
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 8.024828910827637,
+      "learning_rate": 9.040141881227897e-06,
+      "loss": 0.1695,
+      "step": 1125
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 13.246979713439941,
+      "learning_rate": 9.038491649691195e-06,
+      "loss": 0.2955,
+      "step": 1126
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 8.021769523620605,
+      "learning_rate": 9.036840151689466e-06,
+      "loss": 0.2456,
+      "step": 1127
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 7.556529998779297,
+      "learning_rate": 9.03518738774062e-06,
+      "loss": 0.2107,
+      "step": 1128
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 8.473773002624512,
+      "learning_rate": 9.033533358362963e-06,
+      "loss": 0.2732,
+      "step": 1129
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 7.329180717468262,
+      "learning_rate": 9.03187806407519e-06,
+      "loss": 0.2771,
+      "step": 1130
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 9.312518119812012,
+      "learning_rate": 9.030221505396405e-06,
+      "loss": 0.362,
+      "step": 1131
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 12.08809757232666,
+      "learning_rate": 9.028563682846099e-06,
+      "loss": 0.4072,
+      "step": 1132
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 10.633515357971191,
+      "learning_rate": 9.026904596944163e-06,
+      "loss": 0.3829,
+      "step": 1133
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 5.71665620803833,
+      "learning_rate": 9.025244248210887e-06,
+      "loss": 0.2718,
+      "step": 1134
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 4.534217357635498,
+      "learning_rate": 9.023582637166948e-06,
+      "loss": 0.1723,
+      "step": 1135
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 9.380720138549805,
+      "learning_rate": 9.02191976433343e-06,
+      "loss": 0.4177,
+      "step": 1136
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 11.654775619506836,
+      "learning_rate": 9.020255630231808e-06,
+      "loss": 0.3273,
+      "step": 1137
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 12.624345779418945,
+      "learning_rate": 9.018590235383945e-06,
+      "loss": 0.3607,
+      "step": 1138
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 7.771665096282959,
+      "learning_rate": 9.016923580312114e-06,
+      "loss": 0.3323,
+      "step": 1139
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 5.511932373046875,
+      "learning_rate": 9.015255665538972e-06,
+      "loss": 0.174,
+      "step": 1140
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 8.729756355285645,
+      "learning_rate": 9.013586491587576e-06,
+      "loss": 0.2926,
+      "step": 1141
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 8.394304275512695,
+      "learning_rate": 9.011916058981374e-06,
+      "loss": 0.2063,
+      "step": 1142
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 7.647059917449951,
+      "learning_rate": 9.010244368244214e-06,
+      "loss": 0.2179,
+      "step": 1143
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 6.844130039215088,
+      "learning_rate": 9.008571419900334e-06,
+      "loss": 0.1195,
+      "step": 1144
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 11.671330451965332,
+      "learning_rate": 9.00689721447437e-06,
+      "loss": 0.2289,
+      "step": 1145
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 9.100790023803711,
+      "learning_rate": 9.005221752491348e-06,
+      "loss": 0.2791,
+      "step": 1146
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 10.05386734008789,
+      "learning_rate": 9.003545034476691e-06,
+      "loss": 0.3041,
+      "step": 1147
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 7.347108840942383,
+      "learning_rate": 9.001867060956217e-06,
+      "loss": 0.3546,
+      "step": 1148
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 12.807306289672852,
+      "learning_rate": 9.000187832456135e-06,
+      "loss": 0.2563,
+      "step": 1149
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 5.8612141609191895,
+      "learning_rate": 8.998507349503048e-06,
+      "loss": 0.1691,
+      "step": 1150
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 6.4996819496154785,
+      "learning_rate": 8.996825612623953e-06,
+      "loss": 0.1923,
+      "step": 1151
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 6.166618824005127,
+      "learning_rate": 8.995142622346241e-06,
+      "loss": 0.2134,
+      "step": 1152
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 5.040030479431152,
+      "learning_rate": 8.993458379197695e-06,
+      "loss": 0.1518,
+      "step": 1153
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 7.301490783691406,
+      "learning_rate": 8.991772883706492e-06,
+      "loss": 0.2013,
+      "step": 1154
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 10.020792961120605,
+      "learning_rate": 8.990086136401199e-06,
+      "loss": 0.2619,
+      "step": 1155
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 7.523735046386719,
+      "learning_rate": 8.988398137810778e-06,
+      "loss": 0.2057,
+      "step": 1156
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 7.224781036376953,
+      "learning_rate": 8.986708888464583e-06,
+      "loss": 0.2176,
+      "step": 1157
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 8.363834381103516,
+      "learning_rate": 8.98501838889236e-06,
+      "loss": 0.1711,
+      "step": 1158
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 7.733609676361084,
+      "learning_rate": 8.98332663962425e-06,
+      "loss": 0.2901,
+      "step": 1159
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 11.951311111450195,
+      "learning_rate": 8.981633641190779e-06,
+      "loss": 0.4038,
+      "step": 1160
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 8.471813201904297,
+      "learning_rate": 8.979939394122872e-06,
+      "loss": 0.2364,
+      "step": 1161
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 7.482769966125488,
+      "learning_rate": 8.97824389895184e-06,
+      "loss": 0.2467,
+      "step": 1162
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 4.975108623504639,
+      "learning_rate": 8.976547156209387e-06,
+      "loss": 0.1105,
+      "step": 1163
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 13.38056468963623,
+      "learning_rate": 8.974849166427613e-06,
+      "loss": 0.3615,
+      "step": 1164
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 9.41634750366211,
+      "learning_rate": 8.973149930139e-06,
+      "loss": 0.1569,
+      "step": 1165
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 9.200654029846191,
+      "learning_rate": 8.97144944787643e-06,
+      "loss": 0.2859,
+      "step": 1166
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 7.796980381011963,
+      "learning_rate": 8.969747720173172e-06,
+      "loss": 0.3047,
+      "step": 1167
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 9.568486213684082,
+      "learning_rate": 8.968044747562883e-06,
+      "loss": 0.1802,
+      "step": 1168
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 6.297219276428223,
+      "learning_rate": 8.966340530579614e-06,
+      "loss": 0.2157,
+      "step": 1169
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 9.961592674255371,
+      "learning_rate": 8.964635069757803e-06,
+      "loss": 0.1791,
+      "step": 1170
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 6.563395023345947,
+      "learning_rate": 8.96292836563228e-06,
+      "loss": 0.1101,
+      "step": 1171
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 6.622359275817871,
+      "learning_rate": 8.961220418738269e-06,
+      "loss": 0.1926,
+      "step": 1172
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 6.84287166595459,
+      "learning_rate": 8.959511229611377e-06,
+      "loss": 0.1412,
+      "step": 1173
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 11.35946273803711,
+      "learning_rate": 8.957800798787601e-06,
+      "loss": 0.2157,
+      "step": 1174
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 10.371221542358398,
+      "learning_rate": 8.956089126803333e-06,
+      "loss": 0.1872,
+      "step": 1175
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 9.268054962158203,
+      "learning_rate": 8.95437621419535e-06,
+      "loss": 0.2903,
+      "step": 1176
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 7.964203834533691,
+      "learning_rate": 8.952662061500817e-06,
+      "loss": 0.1664,
+      "step": 1177
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 9.089916229248047,
+      "learning_rate": 8.950946669257293e-06,
+      "loss": 0.4381,
+      "step": 1178
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 13.699150085449219,
+      "learning_rate": 8.949230038002719e-06,
+      "loss": 0.5032,
+      "step": 1179
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 6.789539813995361,
+      "learning_rate": 8.94751216827543e-06,
+      "loss": 0.2011,
+      "step": 1180
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 11.783308029174805,
+      "learning_rate": 8.945793060614148e-06,
+      "loss": 0.2978,
+      "step": 1181
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 12.203374862670898,
+      "learning_rate": 8.94407271555798e-06,
+      "loss": 0.2984,
+      "step": 1182
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 13.663761138916016,
+      "learning_rate": 8.942351133646426e-06,
+      "loss": 0.4515,
+      "step": 1183
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 11.635826110839844,
+      "learning_rate": 8.940628315419371e-06,
+      "loss": 0.4037,
+      "step": 1184
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 11.321142196655273,
+      "learning_rate": 8.938904261417088e-06,
+      "loss": 0.3338,
+      "step": 1185
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 19.302474975585938,
+      "learning_rate": 8.937178972180239e-06,
+      "loss": 0.2801,
+      "step": 1186
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 10.215286254882812,
+      "learning_rate": 8.93545244824987e-06,
+      "loss": 0.1913,
+      "step": 1187
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 9.472397804260254,
+      "learning_rate": 8.933724690167417e-06,
+      "loss": 0.2274,
+      "step": 1188
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 8.840088844299316,
+      "learning_rate": 8.931995698474703e-06,
+      "loss": 0.1721,
+      "step": 1189
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 9.048388481140137,
+      "learning_rate": 8.930265473713939e-06,
+      "loss": 0.2436,
+      "step": 1190
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 12.446209907531738,
+      "learning_rate": 8.928534016427717e-06,
+      "loss": 0.4639,
+      "step": 1191
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 11.182536125183105,
+      "learning_rate": 8.926801327159022e-06,
+      "loss": 0.3981,
+      "step": 1192
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 11.084848403930664,
+      "learning_rate": 8.925067406451223e-06,
+      "loss": 0.2392,
+      "step": 1193
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 9.990920066833496,
+      "learning_rate": 8.923332254848072e-06,
+      "loss": 0.263,
+      "step": 1194
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 11.235258102416992,
+      "learning_rate": 8.921595872893714e-06,
+      "loss": 0.2809,
+      "step": 1195
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 4.07004451751709,
+      "learning_rate": 8.919858261132674e-06,
+      "loss": 0.0921,
+      "step": 1196
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 7.323151588439941,
+      "learning_rate": 8.918119420109865e-06,
+      "loss": 0.179,
+      "step": 1197
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 10.538361549377441,
+      "learning_rate": 8.916379350370583e-06,
+      "loss": 0.3192,
+      "step": 1198
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 11.130135536193848,
+      "learning_rate": 8.914638052460515e-06,
+      "loss": 0.3267,
+      "step": 1199
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 13.74011516571045,
+      "learning_rate": 8.912895526925726e-06,
+      "loss": 0.2458,
+      "step": 1200
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 3.807673692703247,
+      "learning_rate": 8.911151774312671e-06,
+      "loss": 0.0674,
+      "step": 1201
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 7.931633949279785,
+      "learning_rate": 8.909406795168188e-06,
+      "loss": 0.1704,
+      "step": 1202
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 6.864401340484619,
+      "learning_rate": 8.907660590039499e-06,
+      "loss": 0.1662,
+      "step": 1203
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 14.097731590270996,
+      "learning_rate": 8.905913159474213e-06,
+      "loss": 0.4204,
+      "step": 1204
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 13.610170364379883,
+      "learning_rate": 8.904164504020321e-06,
+      "loss": 0.3626,
+      "step": 1205
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 4.352688789367676,
+      "learning_rate": 8.902414624226197e-06,
+      "loss": 0.0875,
+      "step": 1206
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 5.519301891326904,
+      "learning_rate": 8.900663520640605e-06,
+      "loss": 0.2186,
+      "step": 1207
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 6.228072643280029,
+      "learning_rate": 8.898911193812684e-06,
+      "loss": 0.1575,
+      "step": 1208
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 9.225931167602539,
+      "learning_rate": 8.897157644291963e-06,
+      "loss": 0.145,
+      "step": 1209
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 10.350606918334961,
+      "learning_rate": 8.895402872628352e-06,
+      "loss": 0.3085,
+      "step": 1210
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 13.643526077270508,
+      "learning_rate": 8.893646879372148e-06,
+      "loss": 0.3608,
+      "step": 1211
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 7.906716346740723,
+      "learning_rate": 8.891889665074023e-06,
+      "loss": 0.1404,
+      "step": 1212
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 10.033092498779297,
+      "learning_rate": 8.890131230285039e-06,
+      "loss": 0.2569,
+      "step": 1213
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 13.337040901184082,
+      "learning_rate": 8.88837157555664e-06,
+      "loss": 0.3182,
+      "step": 1214
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 8.610502243041992,
+      "learning_rate": 8.886610701440648e-06,
+      "loss": 0.164,
+      "step": 1215
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 12.29676628112793,
+      "learning_rate": 8.884848608489272e-06,
+      "loss": 0.2904,
+      "step": 1216
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 10.676018714904785,
+      "learning_rate": 8.883085297255104e-06,
+      "loss": 0.278,
+      "step": 1217
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 10.659815788269043,
+      "learning_rate": 8.881320768291113e-06,
+      "loss": 0.3236,
+      "step": 1218
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 11.427408218383789,
+      "learning_rate": 8.879555022150653e-06,
+      "loss": 0.2651,
+      "step": 1219
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 6.821954250335693,
+      "learning_rate": 8.87778805938746e-06,
+      "loss": 0.2661,
+      "step": 1220
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 13.103662490844727,
+      "learning_rate": 8.87601988055565e-06,
+      "loss": 0.2872,
+      "step": 1221
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 13.563529014587402,
+      "learning_rate": 8.874250486209723e-06,
+      "loss": 0.1961,
+      "step": 1222
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 7.77569055557251,
+      "learning_rate": 8.872479876904556e-06,
+      "loss": 0.1783,
+      "step": 1223
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 7.041471481323242,
+      "learning_rate": 8.870708053195414e-06,
+      "loss": 0.1766,
+      "step": 1224
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 6.260490417480469,
+      "learning_rate": 8.868935015637932e-06,
+      "loss": 0.1174,
+      "step": 1225
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 6.749988555908203,
+      "learning_rate": 8.867160764788138e-06,
+      "loss": 0.137,
+      "step": 1226
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 10.987727165222168,
+      "learning_rate": 8.86538530120243e-06,
+      "loss": 0.2125,
+      "step": 1227
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 11.454978942871094,
+      "learning_rate": 8.863608625437591e-06,
+      "loss": 0.2835,
+      "step": 1228
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 9.580530166625977,
+      "learning_rate": 8.861830738050789e-06,
+      "loss": 0.227,
+      "step": 1229
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 7.352167129516602,
+      "learning_rate": 8.86005163959956e-06,
+      "loss": 0.1345,
+      "step": 1230
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 7.825957775115967,
+      "learning_rate": 8.858271330641832e-06,
+      "loss": 0.1812,
+      "step": 1231
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 7.947144985198975,
+      "learning_rate": 8.856489811735904e-06,
+      "loss": 0.1508,
+      "step": 1232
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 8.71175765991211,
+      "learning_rate": 8.85470708344046e-06,
+      "loss": 0.3362,
+      "step": 1233
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 10.28480052947998,
+      "learning_rate": 8.852923146314558e-06,
+      "loss": 0.4531,
+      "step": 1234
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 10.718217849731445,
+      "learning_rate": 8.851138000917641e-06,
+      "loss": 0.3444,
+      "step": 1235
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 13.521519660949707,
+      "learning_rate": 8.849351647809527e-06,
+      "loss": 0.3278,
+      "step": 1236
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 8.510748863220215,
+      "learning_rate": 8.847564087550413e-06,
+      "loss": 0.2135,
+      "step": 1237
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 11.96247673034668,
+      "learning_rate": 8.845775320700874e-06,
+      "loss": 0.274,
+      "step": 1238
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 9.094640731811523,
+      "learning_rate": 8.843985347821868e-06,
+      "loss": 0.2238,
+      "step": 1239
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 10.756121635437012,
+      "learning_rate": 8.842194169474727e-06,
+      "loss": 0.2226,
+      "step": 1240
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 11.273198127746582,
+      "learning_rate": 8.84040178622116e-06,
+      "loss": 0.2577,
+      "step": 1241
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 10.762075424194336,
+      "learning_rate": 8.838608198623257e-06,
+      "loss": 0.2152,
+      "step": 1242
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 10.329598426818848,
+      "learning_rate": 8.836813407243485e-06,
+      "loss": 0.3075,
+      "step": 1243
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 6.838067531585693,
+      "learning_rate": 8.835017412644688e-06,
+      "loss": 0.1776,
+      "step": 1244
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 9.11226749420166,
+      "learning_rate": 8.833220215390085e-06,
+      "loss": 0.3739,
+      "step": 1245
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 12.054835319519043,
+      "learning_rate": 8.831421816043277e-06,
+      "loss": 0.2206,
+      "step": 1246
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 12.13608169555664,
+      "learning_rate": 8.829622215168237e-06,
+      "loss": 0.2809,
+      "step": 1247
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 9.254409790039062,
+      "learning_rate": 8.827821413329318e-06,
+      "loss": 0.3116,
+      "step": 1248
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 8.833691596984863,
+      "learning_rate": 8.826019411091251e-06,
+      "loss": 0.3471,
+      "step": 1249
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 9.737935066223145,
+      "learning_rate": 8.824216209019139e-06,
+      "loss": 0.1619,
+      "step": 1250
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 10.543744087219238,
+      "learning_rate": 8.822411807678464e-06,
+      "loss": 0.1581,
+      "step": 1251
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 7.761831760406494,
+      "learning_rate": 8.820606207635083e-06,
+      "loss": 0.2553,
+      "step": 1252
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 11.810262680053711,
+      "learning_rate": 8.81879940945523e-06,
+      "loss": 0.2497,
+      "step": 1253
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 12.561576843261719,
+      "learning_rate": 8.816991413705515e-06,
+      "loss": 0.3423,
+      "step": 1254
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 8.79633617401123,
+      "learning_rate": 8.815182220952922e-06,
+      "loss": 0.1353,
+      "step": 1255
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 12.708853721618652,
+      "learning_rate": 8.813371831764812e-06,
+      "loss": 0.3147,
+      "step": 1256
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 14.389960289001465,
+      "learning_rate": 8.81156024670892e-06,
+      "loss": 0.3576,
+      "step": 1257
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 6.529843330383301,
+      "learning_rate": 8.809747466353356e-06,
+      "loss": 0.2904,
+      "step": 1258
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 10.682621955871582,
+      "learning_rate": 8.807933491266603e-06,
+      "loss": 0.3237,
+      "step": 1259
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 11.00978946685791,
+      "learning_rate": 8.806118322017525e-06,
+      "loss": 0.2535,
+      "step": 1260
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 9.682811737060547,
+      "learning_rate": 8.804301959175357e-06,
+      "loss": 0.3713,
+      "step": 1261
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 14.459606170654297,
+      "learning_rate": 8.802484403309702e-06,
+      "loss": 0.2874,
+      "step": 1262
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 8.354414939880371,
+      "learning_rate": 8.800665654990547e-06,
+      "loss": 0.1225,
+      "step": 1263
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 10.286574363708496,
+      "learning_rate": 8.798845714788249e-06,
+      "loss": 0.2194,
+      "step": 1264
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 9.369383811950684,
+      "learning_rate": 8.797024583273536e-06,
+      "loss": 0.3114,
+      "step": 1265
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 10.379693031311035,
+      "learning_rate": 8.795202261017515e-06,
+      "loss": 0.2817,
+      "step": 1266
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 5.303670883178711,
+      "learning_rate": 8.793378748591662e-06,
+      "loss": 0.0965,
+      "step": 1267
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 6.093077182769775,
+      "learning_rate": 8.79155404656783e-06,
+      "loss": 0.1405,
+      "step": 1268
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 11.653606414794922,
+      "learning_rate": 8.789728155518237e-06,
+      "loss": 0.2958,
+      "step": 1269
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 8.521370887756348,
+      "learning_rate": 8.787901076015487e-06,
+      "loss": 0.2037,
+      "step": 1270
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 11.210200309753418,
+      "learning_rate": 8.786072808632545e-06,
+      "loss": 0.1908,
+      "step": 1271
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 7.997415065765381,
+      "learning_rate": 8.784243353942753e-06,
+      "loss": 0.1279,
+      "step": 1272
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 8.447843551635742,
+      "learning_rate": 8.78241271251983e-06,
+      "loss": 0.1959,
+      "step": 1273
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 6.531437873840332,
+      "learning_rate": 8.780580884937855e-06,
+      "loss": 0.1537,
+      "step": 1274
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 11.632950782775879,
+      "learning_rate": 8.778747871771293e-06,
+      "loss": 0.3401,
+      "step": 1275
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 5.624758720397949,
+      "learning_rate": 8.776913673594968e-06,
+      "loss": 0.0792,
+      "step": 1276
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 27.29320526123047,
+      "learning_rate": 8.77507829098409e-06,
+      "loss": 0.3142,
+      "step": 1277
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 12.736369132995605,
+      "learning_rate": 8.773241724514226e-06,
+      "loss": 0.2523,
+      "step": 1278
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 8.978994369506836,
+      "learning_rate": 8.77140397476132e-06,
+      "loss": 0.2667,
+      "step": 1279
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 11.665748596191406,
+      "learning_rate": 8.769565042301692e-06,
+      "loss": 0.2785,
+      "step": 1280
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 7.332759380340576,
+      "learning_rate": 8.767724927712025e-06,
+      "loss": 0.2564,
+      "step": 1281
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 8.431818962097168,
+      "learning_rate": 8.765883631569379e-06,
+      "loss": 0.2046,
+      "step": 1282
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 10.260088920593262,
+      "learning_rate": 8.76404115445118e-06,
+      "loss": 0.1897,
+      "step": 1283
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 37.9990348815918,
+      "learning_rate": 8.762197496935225e-06,
+      "loss": 0.4192,
+      "step": 1284
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 8.519453048706055,
+      "learning_rate": 8.760352659599684e-06,
+      "loss": 0.2182,
+      "step": 1285
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 6.582061290740967,
+      "learning_rate": 8.758506643023096e-06,
+      "loss": 0.1497,
+      "step": 1286
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 8.28431224822998,
+      "learning_rate": 8.756659447784367e-06,
+      "loss": 0.2274,
+      "step": 1287
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 9.054848670959473,
+      "learning_rate": 8.754811074462776e-06,
+      "loss": 0.2324,
+      "step": 1288
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 5.161393165588379,
+      "learning_rate": 8.752961523637971e-06,
+      "loss": 0.2025,
+      "step": 1289
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 8.921879768371582,
+      "learning_rate": 8.751110795889966e-06,
+      "loss": 0.1965,
+      "step": 1290
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 9.026376724243164,
+      "learning_rate": 8.749258891799148e-06,
+      "loss": 0.2881,
+      "step": 1291
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 5.975406169891357,
+      "learning_rate": 8.747405811946272e-06,
+      "loss": 0.2218,
+      "step": 1292
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 10.702713012695312,
+      "learning_rate": 8.74555155691246e-06,
+      "loss": 0.2349,
+      "step": 1293
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 12.182295799255371,
+      "learning_rate": 8.743696127279202e-06,
+      "loss": 0.2317,
+      "step": 1294
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 7.7584547996521,
+      "learning_rate": 8.741839523628361e-06,
+      "loss": 0.2121,
+      "step": 1295
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 10.4217529296875,
+      "learning_rate": 8.739981746542164e-06,
+      "loss": 0.2282,
+      "step": 1296
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 10.402688026428223,
+      "learning_rate": 8.738122796603208e-06,
+      "loss": 0.33,
+      "step": 1297
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 7.520321369171143,
+      "learning_rate": 8.736262674394455e-06,
+      "loss": 0.2247,
+      "step": 1298
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 9.412546157836914,
+      "learning_rate": 8.73440138049924e-06,
+      "loss": 0.2048,
+      "step": 1299
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 8.540048599243164,
+      "learning_rate": 8.732538915501257e-06,
+      "loss": 0.3416,
+      "step": 1300
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 8.41686725616455,
+      "learning_rate": 8.730675279984577e-06,
+      "loss": 0.1772,
+      "step": 1301
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 11.474688529968262,
+      "learning_rate": 8.728810474533631e-06,
+      "loss": 0.3511,
+      "step": 1302
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 6.242444038391113,
+      "learning_rate": 8.726944499733218e-06,
+      "loss": 0.1008,
+      "step": 1303
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 10.841634750366211,
+      "learning_rate": 8.725077356168509e-06,
+      "loss": 0.3892,
+      "step": 1304
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 7.444168567657471,
+      "learning_rate": 8.723209044425034e-06,
+      "loss": 0.1607,
+      "step": 1305
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 10.83259391784668,
+      "learning_rate": 8.721339565088694e-06,
+      "loss": 0.2754,
+      "step": 1306
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 13.299782752990723,
+      "learning_rate": 8.719468918745754e-06,
+      "loss": 0.4905,
+      "step": 1307
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 6.928757190704346,
+      "learning_rate": 8.717597105982847e-06,
+      "loss": 0.1887,
+      "step": 1308
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 6.5458831787109375,
+      "learning_rate": 8.715724127386971e-06,
+      "loss": 0.1103,
+      "step": 1309
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 5.190580368041992,
+      "learning_rate": 8.71384998354549e-06,
+      "loss": 0.1307,
+      "step": 1310
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 10.123420715332031,
+      "learning_rate": 8.71197467504613e-06,
+      "loss": 0.1684,
+      "step": 1311
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 6.483027458190918,
+      "learning_rate": 8.710098202476986e-06,
+      "loss": 0.2884,
+      "step": 1312
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 8.743311882019043,
+      "learning_rate": 8.70822056642652e-06,
+      "loss": 0.2537,
+      "step": 1313
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 10.952034950256348,
+      "learning_rate": 8.706341767483552e-06,
+      "loss": 0.386,
+      "step": 1314
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 8.191468238830566,
+      "learning_rate": 8.704461806237272e-06,
+      "loss": 0.4183,
+      "step": 1315
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 9.176653861999512,
+      "learning_rate": 8.702580683277234e-06,
+      "loss": 0.2747,
+      "step": 1316
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 6.677629470825195,
+      "learning_rate": 8.700698399193352e-06,
+      "loss": 0.2136,
+      "step": 1317
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 11.435298919677734,
+      "learning_rate": 8.698814954575912e-06,
+      "loss": 0.2649,
+      "step": 1318
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 6.574479103088379,
+      "learning_rate": 8.696930350015557e-06,
+      "loss": 0.0986,
+      "step": 1319
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 11.383999824523926,
+      "learning_rate": 8.695044586103297e-06,
+      "loss": 0.2407,
+      "step": 1320
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 9.722286224365234,
+      "learning_rate": 8.693157663430503e-06,
+      "loss": 0.2559,
+      "step": 1321
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 15.406076431274414,
+      "learning_rate": 8.691269582588915e-06,
+      "loss": 0.3262,
+      "step": 1322
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 10.260701179504395,
+      "learning_rate": 8.689380344170629e-06,
+      "loss": 0.3762,
+      "step": 1323
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 6.789151191711426,
+      "learning_rate": 8.687489948768107e-06,
+      "loss": 0.18,
+      "step": 1324
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 9.34760856628418,
+      "learning_rate": 8.685598396974178e-06,
+      "loss": 0.3377,
+      "step": 1325
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 9.871170997619629,
+      "learning_rate": 8.683705689382025e-06,
+      "loss": 0.2278,
+      "step": 1326
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 11.339253425598145,
+      "learning_rate": 8.681811826585203e-06,
+      "loss": 0.2163,
+      "step": 1327
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 10.449474334716797,
+      "learning_rate": 8.67991680917762e-06,
+      "loss": 0.3655,
+      "step": 1328
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 12.968549728393555,
+      "learning_rate": 8.678020637753554e-06,
+      "loss": 0.3818,
+      "step": 1329
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 10.195992469787598,
+      "learning_rate": 8.676123312907641e-06,
+      "loss": 0.1506,
+      "step": 1330
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 10.269356727600098,
+      "learning_rate": 8.674224835234879e-06,
+      "loss": 0.3025,
+      "step": 1331
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 8.058333396911621,
+      "learning_rate": 8.672325205330627e-06,
+      "loss": 0.2298,
+      "step": 1332
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 10.60093879699707,
+      "learning_rate": 8.670424423790607e-06,
+      "loss": 0.2665,
+      "step": 1333
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 7.7602057456970215,
+      "learning_rate": 8.668522491210902e-06,
+      "loss": 0.1495,
+      "step": 1334
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 15.683218002319336,
+      "learning_rate": 8.666619408187953e-06,
+      "loss": 0.1941,
+      "step": 1335
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 13.985320091247559,
+      "learning_rate": 8.664715175318566e-06,
+      "loss": 0.1901,
+      "step": 1336
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 8.808381080627441,
+      "learning_rate": 8.662809793199905e-06,
+      "loss": 0.1463,
+      "step": 1337
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 13.479080200195312,
+      "learning_rate": 8.660903262429495e-06,
+      "loss": 0.2709,
+      "step": 1338
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 10.857939720153809,
+      "learning_rate": 8.65899558360522e-06,
+      "loss": 0.2763,
+      "step": 1339
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 7.575634002685547,
+      "learning_rate": 8.657086757325328e-06,
+      "loss": 0.1106,
+      "step": 1340
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 9.689746856689453,
+      "learning_rate": 8.655176784188421e-06,
+      "loss": 0.1039,
+      "step": 1341
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 14.836522102355957,
+      "learning_rate": 8.653265664793466e-06,
+      "loss": 0.2854,
+      "step": 1342
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 8.706685066223145,
+      "learning_rate": 8.651353399739787e-06,
+      "loss": 0.1102,
+      "step": 1343
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 4.461408615112305,
+      "learning_rate": 8.649439989627067e-06,
+      "loss": 0.1182,
+      "step": 1344
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 9.541738510131836,
+      "learning_rate": 8.64752543505535e-06,
+      "loss": 0.1942,
+      "step": 1345
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 9.528083801269531,
+      "learning_rate": 8.645609736625034e-06,
+      "loss": 0.1365,
+      "step": 1346
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 13.924569129943848,
+      "learning_rate": 8.643692894936882e-06,
+      "loss": 0.1147,
+      "step": 1347
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 10.60889720916748,
+      "learning_rate": 8.64177491059201e-06,
+      "loss": 0.2966,
+      "step": 1348
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 6.5583271980285645,
+      "learning_rate": 8.6398557841919e-06,
+      "loss": 0.1935,
+      "step": 1349
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 9.126585006713867,
+      "learning_rate": 8.637935516338384e-06,
+      "loss": 0.2149,
+      "step": 1350
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 9.94703483581543,
+      "learning_rate": 8.636014107633656e-06,
+      "loss": 0.1974,
+      "step": 1351
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 14.567329406738281,
+      "learning_rate": 8.634091558680265e-06,
+      "loss": 0.4218,
+      "step": 1352
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 10.581974029541016,
+      "learning_rate": 8.632167870081122e-06,
+      "loss": 0.1489,
+      "step": 1353
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 11.252285957336426,
+      "learning_rate": 8.63024304243949e-06,
+      "loss": 0.2567,
+      "step": 1354
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 12.980088233947754,
+      "learning_rate": 8.628317076358997e-06,
+      "loss": 0.3125,
+      "step": 1355
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 9.447537422180176,
+      "learning_rate": 8.626389972443618e-06,
+      "loss": 0.2718,
+      "step": 1356
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 11.60258674621582,
+      "learning_rate": 8.624461731297693e-06,
+      "loss": 0.2514,
+      "step": 1357
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 12.035585403442383,
+      "learning_rate": 8.622532353525913e-06,
+      "loss": 0.2043,
+      "step": 1358
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 8.992303848266602,
+      "learning_rate": 8.620601839733332e-06,
+      "loss": 0.1775,
+      "step": 1359
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 11.451448440551758,
+      "learning_rate": 8.61867019052535e-06,
+      "loss": 0.2562,
+      "step": 1360
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 7.631967544555664,
+      "learning_rate": 8.616737406507738e-06,
+      "loss": 0.1844,
+      "step": 1361
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 11.675711631774902,
+      "learning_rate": 8.614803488286607e-06,
+      "loss": 0.243,
+      "step": 1362
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 10.317102432250977,
+      "learning_rate": 8.612868436468433e-06,
+      "loss": 0.1908,
+      "step": 1363
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 6.427473545074463,
+      "learning_rate": 8.610932251660046e-06,
+      "loss": 0.3112,
+      "step": 1364
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 12.427193641662598,
+      "learning_rate": 8.608994934468633e-06,
+      "loss": 0.3845,
+      "step": 1365
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 9.111933708190918,
+      "learning_rate": 8.607056485501729e-06,
+      "loss": 0.2167,
+      "step": 1366
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 13.258217811584473,
+      "learning_rate": 8.605116905367233e-06,
+      "loss": 0.177,
+      "step": 1367
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 9.499232292175293,
+      "learning_rate": 8.603176194673391e-06,
+      "loss": 0.1997,
+      "step": 1368
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 16.428926467895508,
+      "learning_rate": 8.601234354028807e-06,
+      "loss": 0.2795,
+      "step": 1369
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 9.437528610229492,
+      "learning_rate": 8.599291384042442e-06,
+      "loss": 0.1686,
+      "step": 1370
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 9.590563774108887,
+      "learning_rate": 8.597347285323609e-06,
+      "loss": 0.2797,
+      "step": 1371
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 12.041945457458496,
+      "learning_rate": 8.595402058481971e-06,
+      "loss": 0.2552,
+      "step": 1372
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 19.30535888671875,
+      "learning_rate": 8.59345570412755e-06,
+      "loss": 0.3224,
+      "step": 1373
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 7.823714733123779,
+      "learning_rate": 8.591508222870722e-06,
+      "loss": 0.1984,
+      "step": 1374
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 15.401571273803711,
+      "learning_rate": 8.58955961532221e-06,
+      "loss": 0.3611,
+      "step": 1375
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 12.814584732055664,
+      "learning_rate": 8.587609882093098e-06,
+      "loss": 0.2425,
+      "step": 1376
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 8.597311973571777,
+      "learning_rate": 8.585659023794818e-06,
+      "loss": 0.1827,
+      "step": 1377
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 13.4268217086792,
+      "learning_rate": 8.583707041039158e-06,
+      "loss": 0.299,
+      "step": 1378
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 11.954168319702148,
+      "learning_rate": 8.581753934438253e-06,
+      "loss": 0.2052,
+      "step": 1379
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 7.417294979095459,
+      "learning_rate": 8.579799704604597e-06,
+      "loss": 0.2446,
+      "step": 1380
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 12.388020515441895,
+      "learning_rate": 8.577844352151034e-06,
+      "loss": 0.3138,
+      "step": 1381
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 11.864323616027832,
+      "learning_rate": 8.575887877690759e-06,
+      "loss": 0.4154,
+      "step": 1382
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 6.3389482498168945,
+      "learning_rate": 8.57393028183732e-06,
+      "loss": 0.0846,
+      "step": 1383
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 8.494457244873047,
+      "learning_rate": 8.571971565204614e-06,
+      "loss": 0.15,
+      "step": 1384
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 16.45180320739746,
+      "learning_rate": 8.570011728406895e-06,
+      "loss": 0.397,
+      "step": 1385
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 11.786694526672363,
+      "learning_rate": 8.568050772058763e-06,
+      "loss": 0.3138,
+      "step": 1386
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 6.9836225509643555,
+      "learning_rate": 8.56608869677517e-06,
+      "loss": 0.1611,
+      "step": 1387
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 8.518841743469238,
+      "learning_rate": 8.564125503171421e-06,
+      "loss": 0.2563,
+      "step": 1388
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 6.056849956512451,
+      "learning_rate": 8.562161191863173e-06,
+      "loss": 0.1398,
+      "step": 1389
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 4.888199329376221,
+      "learning_rate": 8.560195763466428e-06,
+      "loss": 0.1382,
+      "step": 1390
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 9.685370445251465,
+      "learning_rate": 8.558229218597541e-06,
+      "loss": 0.1892,
+      "step": 1391
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 17.890581130981445,
+      "learning_rate": 8.55626155787322e-06,
+      "loss": 0.371,
+      "step": 1392
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 8.570674896240234,
+      "learning_rate": 8.55429278191052e-06,
+      "loss": 0.173,
+      "step": 1393
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 11.694134712219238,
+      "learning_rate": 8.552322891326846e-06,
+      "loss": 0.2107,
+      "step": 1394
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 8.359179496765137,
+      "learning_rate": 8.550351886739949e-06,
+      "loss": 0.2025,
+      "step": 1395
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 10.396323204040527,
+      "learning_rate": 8.54837976876794e-06,
+      "loss": 0.1277,
+      "step": 1396
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 7.637752532958984,
+      "learning_rate": 8.546406538029268e-06,
+      "loss": 0.2522,
+      "step": 1397
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 8.991283416748047,
+      "learning_rate": 8.544432195142735e-06,
+      "loss": 0.2791,
+      "step": 1398
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 8.786968231201172,
+      "learning_rate": 8.542456740727496e-06,
+      "loss": 0.2009,
+      "step": 1399
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 11.607145309448242,
+      "learning_rate": 8.540480175403045e-06,
+      "loss": 0.3502,
+      "step": 1400
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 6.047530651092529,
+      "learning_rate": 8.538502499789234e-06,
+      "loss": 0.1878,
+      "step": 1401
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 13.734234809875488,
+      "learning_rate": 8.536523714506257e-06,
+      "loss": 0.2936,
+      "step": 1402
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 11.23990535736084,
+      "learning_rate": 8.53454382017466e-06,
+      "loss": 0.2108,
+      "step": 1403
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 11.387397766113281,
+      "learning_rate": 8.532562817415334e-06,
+      "loss": 0.092,
+      "step": 1404
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 8.523534774780273,
+      "learning_rate": 8.530580706849518e-06,
+      "loss": 0.2708,
+      "step": 1405
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 10.470612525939941,
+      "learning_rate": 8.528597489098798e-06,
+      "loss": 0.1908,
+      "step": 1406
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 8.156074523925781,
+      "learning_rate": 8.526613164785111e-06,
+      "loss": 0.2104,
+      "step": 1407
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 11.000042915344238,
+      "learning_rate": 8.524627734530738e-06,
+      "loss": 0.3268,
+      "step": 1408
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 11.629241943359375,
+      "learning_rate": 8.522641198958304e-06,
+      "loss": 0.2093,
+      "step": 1409
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 8.154757499694824,
+      "learning_rate": 8.520653558690785e-06,
+      "loss": 0.1427,
+      "step": 1410
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 11.115401268005371,
+      "learning_rate": 8.518664814351502e-06,
+      "loss": 0.4652,
+      "step": 1411
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 8.11341381072998,
+      "learning_rate": 8.516674966564124e-06,
+      "loss": 0.175,
+      "step": 1412
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 14.09422779083252,
+      "learning_rate": 8.514684015952664e-06,
+      "loss": 0.2774,
+      "step": 1413
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 10.62497329711914,
+      "learning_rate": 8.51269196314148e-06,
+      "loss": 0.1699,
+      "step": 1414
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 11.167736053466797,
+      "learning_rate": 8.510698808755275e-06,
+      "loss": 0.1938,
+      "step": 1415
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 5.94974946975708,
+      "learning_rate": 8.508704553419105e-06,
+      "loss": 0.1675,
+      "step": 1416
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 8.333573341369629,
+      "learning_rate": 8.506709197758358e-06,
+      "loss": 0.1878,
+      "step": 1417
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 7.065437316894531,
+      "learning_rate": 8.50471274239878e-06,
+      "loss": 0.0838,
+      "step": 1418
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 12.852767944335938,
+      "learning_rate": 8.502715187966455e-06,
+      "loss": 0.2352,
+      "step": 1419
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 8.397226333618164,
+      "learning_rate": 8.500716535087815e-06,
+      "loss": 0.2706,
+      "step": 1420
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 11.840666770935059,
+      "learning_rate": 8.498716784389629e-06,
+      "loss": 0.2559,
+      "step": 1421
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 9.849198341369629,
+      "learning_rate": 8.49671593649902e-06,
+      "loss": 0.2793,
+      "step": 1422
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 8.145561218261719,
+      "learning_rate": 8.494713992043454e-06,
+      "loss": 0.2128,
+      "step": 1423
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 10.586185455322266,
+      "learning_rate": 8.492710951650732e-06,
+      "loss": 0.2905,
+      "step": 1424
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 9.9353666305542,
+      "learning_rate": 8.490706815949006e-06,
+      "loss": 0.2256,
+      "step": 1425
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 8.38582706451416,
+      "learning_rate": 8.488701585566772e-06,
+      "loss": 0.1354,
+      "step": 1426
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 8.140827178955078,
+      "learning_rate": 8.486695261132866e-06,
+      "loss": 0.1929,
+      "step": 1427
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 12.00531005859375,
+      "learning_rate": 8.48468784327647e-06,
+      "loss": 0.3025,
+      "step": 1428
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 12.363675117492676,
+      "learning_rate": 8.482679332627101e-06,
+      "loss": 0.3697,
+      "step": 1429
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 8.129680633544922,
+      "learning_rate": 8.480669729814635e-06,
+      "loss": 0.1476,
+      "step": 1430
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 12.366854667663574,
+      "learning_rate": 8.478659035469273e-06,
+      "loss": 0.3002,
+      "step": 1431
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 8.009378433227539,
+      "learning_rate": 8.476647250221567e-06,
+      "loss": 0.1736,
+      "step": 1432
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 6.676246166229248,
+      "learning_rate": 8.474634374702412e-06,
+      "loss": 0.117,
+      "step": 1433
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 8.843733787536621,
+      "learning_rate": 8.472620409543041e-06,
+      "loss": 0.1762,
+      "step": 1434
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 8.643060684204102,
+      "learning_rate": 8.470605355375033e-06,
+      "loss": 0.326,
+      "step": 1435
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 12.181198120117188,
+      "learning_rate": 8.468589212830301e-06,
+      "loss": 0.4123,
+      "step": 1436
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 9.66492748260498,
+      "learning_rate": 8.466571982541109e-06,
+      "loss": 0.3834,
+      "step": 1437
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 9.26583480834961,
+      "learning_rate": 8.464553665140059e-06,
+      "loss": 0.2843,
+      "step": 1438
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 8.880125045776367,
+      "learning_rate": 8.462534261260084e-06,
+      "loss": 0.188,
+      "step": 1439
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 5.876025199890137,
+      "learning_rate": 8.460513771534475e-06,
+      "loss": 0.0791,
+      "step": 1440
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 9.074946403503418,
+      "learning_rate": 8.458492196596852e-06,
+      "loss": 0.2115,
+      "step": 1441
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 9.121976852416992,
+      "learning_rate": 8.456469537081174e-06,
+      "loss": 0.1774,
+      "step": 1442
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 11.513020515441895,
+      "learning_rate": 8.454445793621748e-06,
+      "loss": 0.2498,
+      "step": 1443
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 15.415754318237305,
+      "learning_rate": 8.452420966853217e-06,
+      "loss": 0.244,
+      "step": 1444
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 10.95475959777832,
+      "learning_rate": 8.450395057410561e-06,
+      "loss": 0.2873,
+      "step": 1445
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 9.540244102478027,
+      "learning_rate": 8.448368065929105e-06,
+      "loss": 0.268,
+      "step": 1446
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 11.70332145690918,
+      "learning_rate": 8.446339993044508e-06,
+      "loss": 0.4026,
+      "step": 1447
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 14.210153579711914,
+      "learning_rate": 8.444310839392773e-06,
+      "loss": 0.282,
+      "step": 1448
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 10.124107360839844,
+      "learning_rate": 8.442280605610238e-06,
+      "loss": 0.1785,
+      "step": 1449
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 6.499838829040527,
+      "learning_rate": 8.440249292333583e-06,
+      "loss": 0.1236,
+      "step": 1450
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 9.87684440612793,
+      "learning_rate": 8.438216900199823e-06,
+      "loss": 0.2542,
+      "step": 1451
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 11.262269973754883,
+      "learning_rate": 8.436183429846314e-06,
+      "loss": 0.211,
+      "step": 1452
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 9.594480514526367,
+      "learning_rate": 8.43414888191075e-06,
+      "loss": 0.2336,
+      "step": 1453
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 12.905770301818848,
+      "learning_rate": 8.432113257031161e-06,
+      "loss": 0.2738,
+      "step": 1454
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 16.228429794311523,
+      "learning_rate": 8.430076555845917e-06,
+      "loss": 0.3771,
+      "step": 1455
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 12.277434349060059,
+      "learning_rate": 8.428038778993723e-06,
+      "loss": 0.3784,
+      "step": 1456
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 6.526297569274902,
+      "learning_rate": 8.425999927113623e-06,
+      "loss": 0.1002,
+      "step": 1457
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 10.36428165435791,
+      "learning_rate": 8.423960000845002e-06,
+      "loss": 0.2326,
+      "step": 1458
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 12.772688865661621,
+      "learning_rate": 8.421919000827572e-06,
+      "loss": 0.3107,
+      "step": 1459
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 9.325165748596191,
+      "learning_rate": 8.41987692770139e-06,
+      "loss": 0.1915,
+      "step": 1460
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 7.714015960693359,
+      "learning_rate": 8.417833782106848e-06,
+      "loss": 0.1777,
+      "step": 1461
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 11.979935646057129,
+      "learning_rate": 8.415789564684673e-06,
+      "loss": 0.1814,
+      "step": 1462
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 10.821907043457031,
+      "learning_rate": 8.413744276075928e-06,
+      "loss": 0.2791,
+      "step": 1463
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 14.235872268676758,
+      "learning_rate": 8.411697916922013e-06,
+      "loss": 0.4187,
+      "step": 1464
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 9.029800415039062,
+      "learning_rate": 8.409650487864662e-06,
+      "loss": 0.3018,
+      "step": 1465
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 12.755525588989258,
+      "learning_rate": 8.407601989545948e-06,
+      "loss": 0.3983,
+      "step": 1466
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 5.491318225860596,
+      "learning_rate": 8.405552422608276e-06,
+      "loss": 0.075,
+      "step": 1467
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 8.981986999511719,
+      "learning_rate": 8.403501787694386e-06,
+      "loss": 0.2018,
+      "step": 1468
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 16.344758987426758,
+      "learning_rate": 8.401450085447356e-06,
+      "loss": 0.2911,
+      "step": 1469
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 8.874183654785156,
+      "learning_rate": 8.399397316510596e-06,
+      "loss": 0.275,
+      "step": 1470
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 12.085587501525879,
+      "learning_rate": 8.39734348152785e-06,
+      "loss": 0.2095,
+      "step": 1471
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 12.409998893737793,
+      "learning_rate": 8.3952885811432e-06,
+      "loss": 0.2578,
+      "step": 1472
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 8.950498580932617,
+      "learning_rate": 8.393232616001057e-06,
+      "loss": 0.2243,
+      "step": 1473
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 10.206308364868164,
+      "learning_rate": 8.39117558674617e-06,
+      "loss": 0.2584,
+      "step": 1474
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 9.8655424118042,
+      "learning_rate": 8.389117494023622e-06,
+      "loss": 0.1168,
+      "step": 1475
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 9.553979873657227,
+      "learning_rate": 8.387058338478826e-06,
+      "loss": 0.2734,
+      "step": 1476
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 14.387800216674805,
+      "learning_rate": 8.38499812075753e-06,
+      "loss": 0.4015,
+      "step": 1477
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 13.732446670532227,
+      "learning_rate": 8.382936841505816e-06,
+      "loss": 0.2158,
+      "step": 1478
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 11.669587135314941,
+      "learning_rate": 8.380874501370098e-06,
+      "loss": 0.2417,
+      "step": 1479
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 16.581207275390625,
+      "learning_rate": 8.378811100997122e-06,
+      "loss": 0.4518,
+      "step": 1480
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 11.113801002502441,
+      "learning_rate": 8.376746641033968e-06,
+      "loss": 0.2197,
+      "step": 1481
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 9.504670143127441,
+      "learning_rate": 8.37468112212805e-06,
+      "loss": 0.1395,
+      "step": 1482
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 15.624802589416504,
+      "learning_rate": 8.372614544927107e-06,
+      "loss": 0.4796,
+      "step": 1483
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 12.477916717529297,
+      "learning_rate": 8.370546910079215e-06,
+      "loss": 0.4118,
+      "step": 1484
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 10.72073745727539,
+      "learning_rate": 8.368478218232787e-06,
+      "loss": 0.327,
+      "step": 1485
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 11.402993202209473,
+      "learning_rate": 8.366408470036555e-06,
+      "loss": 0.216,
+      "step": 1486
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 9.209574699401855,
+      "learning_rate": 8.364337666139593e-06,
+      "loss": 0.1629,
+      "step": 1487
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 8.262737274169922,
+      "learning_rate": 8.362265807191301e-06,
+      "loss": 0.3231,
+      "step": 1488
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 10.550108909606934,
+      "learning_rate": 8.36019289384141e-06,
+      "loss": 0.321,
+      "step": 1489
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 11.74404239654541,
+      "learning_rate": 8.358118926739984e-06,
+      "loss": 0.3464,
+      "step": 1490
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 8.824837684631348,
+      "learning_rate": 8.356043906537415e-06,
+      "loss": 0.1367,
+      "step": 1491
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 8.018375396728516,
+      "learning_rate": 8.353967833884426e-06,
+      "loss": 0.2304,
+      "step": 1492
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 11.33295726776123,
+      "learning_rate": 8.351890709432074e-06,
+      "loss": 0.2296,
+      "step": 1493
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 6.454250335693359,
+      "learning_rate": 8.349812533831734e-06,
+      "loss": 0.1339,
+      "step": 1494
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 11.622160911560059,
+      "learning_rate": 8.347733307735127e-06,
+      "loss": 0.1661,
+      "step": 1495
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 7.924144268035889,
+      "learning_rate": 8.345653031794292e-06,
+      "loss": 0.1674,
+      "step": 1496
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 5.227413177490234,
+      "learning_rate": 8.3435717066616e-06,
+      "loss": 0.15,
+      "step": 1497
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 9.205004692077637,
+      "learning_rate": 8.341489332989752e-06,
+      "loss": 0.2291,
+      "step": 1498
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 5.755086421966553,
+      "learning_rate": 8.339405911431777e-06,
+      "loss": 0.1635,
+      "step": 1499
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 14.380725860595703,
+      "learning_rate": 8.337321442641036e-06,
+      "loss": 0.2557,
+      "step": 1500
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 8.642199516296387,
+      "learning_rate": 8.33523592727121e-06,
+      "loss": 0.1151,
+      "step": 1501
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 9.880890846252441,
+      "learning_rate": 8.333149365976317e-06,
+      "loss": 0.2503,
+      "step": 1502
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 9.447900772094727,
+      "learning_rate": 8.331061759410696e-06,
+      "loss": 0.1467,
+      "step": 1503
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 14.108247756958008,
+      "learning_rate": 8.328973108229026e-06,
+      "loss": 0.1946,
+      "step": 1504
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 13.432592391967773,
+      "learning_rate": 8.326883413086295e-06,
+      "loss": 0.3231,
+      "step": 1505
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 8.821404457092285,
+      "learning_rate": 8.324792674637831e-06,
+      "loss": 0.1733,
+      "step": 1506
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 7.36824893951416,
+      "learning_rate": 8.32270089353929e-06,
+      "loss": 0.1161,
+      "step": 1507
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 8.366787910461426,
+      "learning_rate": 8.32060807044665e-06,
+      "loss": 0.1463,
+      "step": 1508
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 17.965736389160156,
+      "learning_rate": 8.318514206016216e-06,
+      "loss": 0.2539,
+      "step": 1509
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 4.402562618255615,
+      "learning_rate": 8.316419300904622e-06,
+      "loss": 0.112,
+      "step": 1510
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 10.279218673706055,
+      "learning_rate": 8.314323355768822e-06,
+      "loss": 0.3877,
+      "step": 1511
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 13.595497131347656,
+      "learning_rate": 8.31222637126611e-06,
+      "loss": 0.2777,
+      "step": 1512
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 15.896930694580078,
+      "learning_rate": 8.310128348054093e-06,
+      "loss": 0.2185,
+      "step": 1513
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 16.229001998901367,
+      "learning_rate": 8.308029286790707e-06,
+      "loss": 0.4759,
+      "step": 1514
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 8.718769073486328,
+      "learning_rate": 8.305929188134216e-06,
+      "loss": 0.1543,
+      "step": 1515
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 7.8127593994140625,
+      "learning_rate": 8.303828052743208e-06,
+      "loss": 0.1366,
+      "step": 1516
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 11.643912315368652,
+      "learning_rate": 8.301725881276594e-06,
+      "loss": 0.2127,
+      "step": 1517
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 8.178140640258789,
+      "learning_rate": 8.299622674393615e-06,
+      "loss": 0.2773,
+      "step": 1518
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 11.938106536865234,
+      "learning_rate": 8.29751843275383e-06,
+      "loss": 0.2232,
+      "step": 1519
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 10.227789878845215,
+      "learning_rate": 8.295413157017127e-06,
+      "loss": 0.1444,
+      "step": 1520
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 8.718910217285156,
+      "learning_rate": 8.293306847843721e-06,
+      "loss": 0.1833,
+      "step": 1521
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 9.718162536621094,
+      "learning_rate": 8.291199505894143e-06,
+      "loss": 0.3573,
+      "step": 1522
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 8.973173141479492,
+      "learning_rate": 8.289091131829254e-06,
+      "loss": 0.1908,
+      "step": 1523
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 8.066110610961914,
+      "learning_rate": 8.286981726310237e-06,
+      "loss": 0.1579,
+      "step": 1524
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 15.239232063293457,
+      "learning_rate": 8.284871289998599e-06,
+      "loss": 0.3475,
+      "step": 1525
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 10.371790885925293,
+      "learning_rate": 8.282759823556168e-06,
+      "loss": 0.375,
+      "step": 1526
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 9.69017505645752,
+      "learning_rate": 8.280647327645099e-06,
+      "loss": 0.2998,
+      "step": 1527
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 17.301774978637695,
+      "learning_rate": 8.278533802927866e-06,
+      "loss": 0.3761,
+      "step": 1528
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 13.657183647155762,
+      "learning_rate": 8.27641925006727e-06,
+      "loss": 0.2225,
+      "step": 1529
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 8.001036643981934,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 0.2256,
+      "step": 1530
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 12.63283634185791,
+      "learning_rate": 8.272187062568782e-06,
+      "loss": 0.3001,
+      "step": 1531
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 7.194914817810059,
+      "learning_rate": 8.270069429258101e-06,
+      "loss": 0.1368,
+      "step": 1532
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 10.89219856262207,
+      "learning_rate": 8.267950770458471e-06,
+      "loss": 0.225,
+      "step": 1533
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 8.241290092468262,
+      "learning_rate": 8.265831086834297e-06,
+      "loss": 0.2297,
+      "step": 1534
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 6.2861809730529785,
+      "learning_rate": 8.263710379050311e-06,
+      "loss": 0.2982,
+      "step": 1535
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 9.264528274536133,
+      "learning_rate": 8.261588647771566e-06,
+      "loss": 0.216,
+      "step": 1536
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 10.993940353393555,
+      "learning_rate": 8.25946589366343e-06,
+      "loss": 0.2667,
+      "step": 1537
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 13.918134689331055,
+      "learning_rate": 8.257342117391595e-06,
+      "loss": 0.22,
+      "step": 1538
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 8.253373146057129,
+      "learning_rate": 8.255217319622079e-06,
+      "loss": 0.3522,
+      "step": 1539
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 11.742491722106934,
+      "learning_rate": 8.25309150102121e-06,
+      "loss": 0.2004,
+      "step": 1540
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 11.605565071105957,
+      "learning_rate": 8.250964662255647e-06,
+      "loss": 0.5273,
+      "step": 1541
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 10.37311840057373,
+      "learning_rate": 8.248836803992359e-06,
+      "loss": 0.2116,
+      "step": 1542
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 7.8454084396362305,
+      "learning_rate": 8.246707926898641e-06,
+      "loss": 0.2083,
+      "step": 1543
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 9.906599044799805,
+      "learning_rate": 8.244578031642105e-06,
+      "loss": 0.3045,
+      "step": 1544
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 10.094032287597656,
+      "learning_rate": 8.242447118890686e-06,
+      "loss": 0.3945,
+      "step": 1545
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 14.51402473449707,
+      "learning_rate": 8.24031518931263e-06,
+      "loss": 0.4044,
+      "step": 1546
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 16.72068977355957,
+      "learning_rate": 8.238182243576512e-06,
+      "loss": 0.1728,
+      "step": 1547
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 6.506950378417969,
+      "learning_rate": 8.236048282351215e-06,
+      "loss": 0.1766,
+      "step": 1548
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 12.43808650970459,
+      "learning_rate": 8.233913306305952e-06,
+      "loss": 0.3335,
+      "step": 1549
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 6.460045337677002,
+      "learning_rate": 8.231777316110245e-06,
+      "loss": 0.129,
+      "step": 1550
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 10.985332489013672,
+      "learning_rate": 8.229640312433938e-06,
+      "loss": 0.1732,
+      "step": 1551
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 10.359841346740723,
+      "learning_rate": 8.227502295947192e-06,
+      "loss": 0.308,
+      "step": 1552
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 13.781259536743164,
+      "learning_rate": 8.225363267320486e-06,
+      "loss": 0.2319,
+      "step": 1553
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 11.385099411010742,
+      "learning_rate": 8.223223227224617e-06,
+      "loss": 0.5422,
+      "step": 1554
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 9.180119514465332,
+      "learning_rate": 8.221082176330697e-06,
+      "loss": 0.1863,
+      "step": 1555
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 11.483895301818848,
+      "learning_rate": 8.218940115310157e-06,
+      "loss": 0.112,
+      "step": 1556
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 12.00844955444336,
+      "learning_rate": 8.216797044834746e-06,
+      "loss": 0.2052,
+      "step": 1557
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 10.490771293640137,
+      "learning_rate": 8.214652965576524e-06,
+      "loss": 0.1302,
+      "step": 1558
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 9.999978065490723,
+      "learning_rate": 8.212507878207873e-06,
+      "loss": 0.2205,
+      "step": 1559
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 7.9145073890686035,
+      "learning_rate": 8.210361783401491e-06,
+      "loss": 0.2457,
+      "step": 1560
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 14.882489204406738,
+      "learning_rate": 8.20821468183039e-06,
+      "loss": 0.397,
+      "step": 1561
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 12.082924842834473,
+      "learning_rate": 8.206066574167893e-06,
+      "loss": 0.1636,
+      "step": 1562
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 7.291049480438232,
+      "learning_rate": 8.203917461087648e-06,
+      "loss": 0.1304,
+      "step": 1563
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 9.953224182128906,
+      "learning_rate": 8.201767343263612e-06,
+      "loss": 0.2151,
+      "step": 1564
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 11.155455589294434,
+      "learning_rate": 8.19961622137006e-06,
+      "loss": 0.3044,
+      "step": 1565
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 7.143294811248779,
+      "learning_rate": 8.19746409608158e-06,
+      "loss": 0.233,
+      "step": 1566
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 11.715970993041992,
+      "learning_rate": 8.195310968073076e-06,
+      "loss": 0.2946,
+      "step": 1567
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 10.829318046569824,
+      "learning_rate": 8.193156838019764e-06,
+      "loss": 0.2127,
+      "step": 1568
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 14.896932601928711,
+      "learning_rate": 8.191001706597178e-06,
+      "loss": 0.2431,
+      "step": 1569
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 7.188117504119873,
+      "learning_rate": 8.188845574481162e-06,
+      "loss": 0.3061,
+      "step": 1570
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 8.160257339477539,
+      "learning_rate": 8.186688442347876e-06,
+      "loss": 0.1992,
+      "step": 1571
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 11.449005126953125,
+      "learning_rate": 8.184530310873797e-06,
+      "loss": 0.301,
+      "step": 1572
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 12.320124626159668,
+      "learning_rate": 8.182371180735708e-06,
+      "loss": 0.3014,
+      "step": 1573
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 7.88347053527832,
+      "learning_rate": 8.180211052610709e-06,
+      "loss": 0.1411,
+      "step": 1574
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 9.334169387817383,
+      "learning_rate": 8.178049927176217e-06,
+      "loss": 0.2175,
+      "step": 1575
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 7.805450439453125,
+      "learning_rate": 8.175887805109953e-06,
+      "loss": 0.1032,
+      "step": 1576
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 10.2393798828125,
+      "learning_rate": 8.173724687089957e-06,
+      "loss": 0.2684,
+      "step": 1577
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 6.523881435394287,
+      "learning_rate": 8.171560573794581e-06,
+      "loss": 0.1122,
+      "step": 1578
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 11.751117706298828,
+      "learning_rate": 8.169395465902488e-06,
+      "loss": 0.2936,
+      "step": 1579
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 6.747082233428955,
+      "learning_rate": 8.167229364092648e-06,
+      "loss": 0.1729,
+      "step": 1580
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 9.153532981872559,
+      "learning_rate": 8.165062269044353e-06,
+      "loss": 0.1745,
+      "step": 1581
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 9.072455406188965,
+      "learning_rate": 8.162894181437198e-06,
+      "loss": 0.2978,
+      "step": 1582
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 12.090812683105469,
+      "learning_rate": 8.160725101951093e-06,
+      "loss": 0.168,
+      "step": 1583
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 11.82819652557373,
+      "learning_rate": 8.158555031266255e-06,
+      "loss": 0.1835,
+      "step": 1584
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 14.153754234313965,
+      "learning_rate": 8.15638397006322e-06,
+      "loss": 0.3288,
+      "step": 1585
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 5.828611850738525,
+      "learning_rate": 8.154211919022827e-06,
+      "loss": 0.1065,
+      "step": 1586
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 13.306476593017578,
+      "learning_rate": 8.152038878826227e-06,
+      "loss": 0.4128,
+      "step": 1587
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 5.285043716430664,
+      "learning_rate": 8.149864850154885e-06,
+      "loss": 0.0834,
+      "step": 1588
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 12.241046905517578,
+      "learning_rate": 8.14768983369057e-06,
+      "loss": 0.2111,
+      "step": 1589
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 9.584244728088379,
+      "learning_rate": 8.145513830115367e-06,
+      "loss": 0.1824,
+      "step": 1590
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 14.109899520874023,
+      "learning_rate": 8.143336840111667e-06,
+      "loss": 0.3323,
+      "step": 1591
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 5.801514625549316,
+      "learning_rate": 8.141158864362168e-06,
+      "loss": 0.1873,
+      "step": 1592
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 12.577149391174316,
+      "learning_rate": 8.138979903549881e-06,
+      "loss": 0.1596,
+      "step": 1593
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 14.648880004882812,
+      "learning_rate": 8.136799958358129e-06,
+      "loss": 0.3276,
+      "step": 1594
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 12.901571273803711,
+      "learning_rate": 8.134619029470535e-06,
+      "loss": 0.3004,
+      "step": 1595
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 11.602432250976562,
+      "learning_rate": 8.132437117571034e-06,
+      "loss": 0.2821,
+      "step": 1596
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 19.047420501708984,
+      "learning_rate": 8.130254223343874e-06,
+      "loss": 0.3642,
+      "step": 1597
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 8.796975135803223,
+      "learning_rate": 8.128070347473609e-06,
+      "loss": 0.253,
+      "step": 1598
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 11.138106346130371,
+      "learning_rate": 8.125885490645094e-06,
+      "loss": 0.2418,
+      "step": 1599
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 12.982224464416504,
+      "learning_rate": 8.1236996535435e-06,
+      "loss": 0.2354,
+      "step": 1600
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 14.292930603027344,
+      "learning_rate": 8.121512836854301e-06,
+      "loss": 0.3688,
+      "step": 1601
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 15.489351272583008,
+      "learning_rate": 8.11932504126328e-06,
+      "loss": 0.2526,
+      "step": 1602
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 11.492733001708984,
+      "learning_rate": 8.117136267456527e-06,
+      "loss": 0.2334,
+      "step": 1603
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 13.331953048706055,
+      "learning_rate": 8.11494651612044e-06,
+      "loss": 0.1747,
+      "step": 1604
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 3.8955130577087402,
+      "learning_rate": 8.112755787941718e-06,
+      "loss": 0.0652,
+      "step": 1605
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 11.033143043518066,
+      "learning_rate": 8.110564083607371e-06,
+      "loss": 0.1371,
+      "step": 1606
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 11.704679489135742,
+      "learning_rate": 8.108371403804715e-06,
+      "loss": 0.2166,
+      "step": 1607
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 13.181831359863281,
+      "learning_rate": 8.106177749221371e-06,
+      "loss": 0.2546,
+      "step": 1608
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 15.890047073364258,
+      "learning_rate": 8.103983120545266e-06,
+      "loss": 0.4657,
+      "step": 1609
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 9.194727897644043,
+      "learning_rate": 8.101787518464634e-06,
+      "loss": 0.0659,
+      "step": 1610
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 15.165385246276855,
+      "learning_rate": 8.09959094366801e-06,
+      "loss": 0.317,
+      "step": 1611
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 5.944538593292236,
+      "learning_rate": 8.097393396844237e-06,
+      "loss": 0.1192,
+      "step": 1612
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 7.57356071472168,
+      "learning_rate": 8.095194878682464e-06,
+      "loss": 0.1662,
+      "step": 1613
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 14.427013397216797,
+      "learning_rate": 8.092995389872142e-06,
+      "loss": 0.3154,
+      "step": 1614
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 28.98470115661621,
+      "learning_rate": 8.090794931103026e-06,
+      "loss": 0.2516,
+      "step": 1615
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 12.864948272705078,
+      "learning_rate": 8.08859350306518e-06,
+      "loss": 0.4232,
+      "step": 1616
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 9.191838264465332,
+      "learning_rate": 8.086391106448965e-06,
+      "loss": 0.3066,
+      "step": 1617
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 11.798999786376953,
+      "learning_rate": 8.084187741945053e-06,
+      "loss": 0.2467,
+      "step": 1618
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 12.892189979553223,
+      "learning_rate": 8.081983410244411e-06,
+      "loss": 0.4092,
+      "step": 1619
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 10.776820182800293,
+      "learning_rate": 8.079778112038318e-06,
+      "loss": 0.307,
+      "step": 1620
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 6.5920023918151855,
+      "learning_rate": 8.077571848018353e-06,
+      "loss": 0.1902,
+      "step": 1621
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 13.41633129119873,
+      "learning_rate": 8.075364618876395e-06,
+      "loss": 0.3795,
+      "step": 1622
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 14.219651222229004,
+      "learning_rate": 8.073156425304627e-06,
+      "loss": 0.2816,
+      "step": 1623
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 7.946049213409424,
+      "learning_rate": 8.070947267995537e-06,
+      "loss": 0.1703,
+      "step": 1624
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 11.483860969543457,
+      "learning_rate": 8.068737147641913e-06,
+      "loss": 0.3815,
+      "step": 1625
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 8.83636474609375,
+      "learning_rate": 8.066526064936847e-06,
+      "loss": 0.1544,
+      "step": 1626
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 9.180328369140625,
+      "learning_rate": 8.064314020573727e-06,
+      "loss": 0.3528,
+      "step": 1627
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 12.852416038513184,
+      "learning_rate": 8.06210101524625e-06,
+      "loss": 0.3898,
+      "step": 1628
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 14.532186508178711,
+      "learning_rate": 8.059887049648413e-06,
+      "loss": 0.2884,
+      "step": 1629
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 14.745027542114258,
+      "learning_rate": 8.057672124474508e-06,
+      "loss": 0.295,
+      "step": 1630
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 16.46725082397461,
+      "learning_rate": 8.055456240419135e-06,
+      "loss": 0.276,
+      "step": 1631
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 15.425137519836426,
+      "learning_rate": 8.053239398177191e-06,
+      "loss": 0.2127,
+      "step": 1632
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 8.505598068237305,
+      "learning_rate": 8.051021598443875e-06,
+      "loss": 0.1836,
+      "step": 1633
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 13.276931762695312,
+      "learning_rate": 8.048802841914687e-06,
+      "loss": 0.3345,
+      "step": 1634
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 17.97151756286621,
+      "learning_rate": 8.046583129285422e-06,
+      "loss": 0.5191,
+      "step": 1635
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 10.429954528808594,
+      "learning_rate": 8.044362461252183e-06,
+      "loss": 0.2206,
+      "step": 1636
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 13.042590141296387,
+      "learning_rate": 8.042140838511365e-06,
+      "loss": 0.3459,
+      "step": 1637
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 13.404379844665527,
+      "learning_rate": 8.039918261759667e-06,
+      "loss": 0.2897,
+      "step": 1638
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 14.955395698547363,
+      "learning_rate": 8.037694731694085e-06,
+      "loss": 0.2448,
+      "step": 1639
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 9.675850868225098,
+      "learning_rate": 8.035470249011916e-06,
+      "loss": 0.161,
+      "step": 1640
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 6.97806453704834,
+      "learning_rate": 8.033244814410754e-06,
+      "loss": 0.1381,
+      "step": 1641
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 9.638113021850586,
+      "learning_rate": 8.031018428588492e-06,
+      "loss": 0.2263,
+      "step": 1642
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 8.244622230529785,
+      "learning_rate": 8.02879109224332e-06,
+      "loss": 0.1316,
+      "step": 1643
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 12.371101379394531,
+      "learning_rate": 8.026562806073729e-06,
+      "loss": 0.29,
+      "step": 1644
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 6.359392166137695,
+      "learning_rate": 8.024333570778507e-06,
+      "loss": 0.1871,
+      "step": 1645
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 9.152475357055664,
+      "learning_rate": 8.022103387056734e-06,
+      "loss": 0.1462,
+      "step": 1646
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 7.387409687042236,
+      "learning_rate": 8.019872255607799e-06,
+      "loss": 0.2197,
+      "step": 1647
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 7.848971366882324,
+      "learning_rate": 8.017640177131378e-06,
+      "loss": 0.1226,
+      "step": 1648
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 8.904796600341797,
+      "learning_rate": 8.015407152327448e-06,
+      "loss": 0.1652,
+      "step": 1649
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 14.934941291809082,
+      "learning_rate": 8.013173181896283e-06,
+      "loss": 0.3014,
+      "step": 1650
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 11.010063171386719,
+      "learning_rate": 8.01093826653845e-06,
+      "loss": 0.3509,
+      "step": 1651
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 10.057924270629883,
+      "learning_rate": 8.00870240695482e-06,
+      "loss": 0.2244,
+      "step": 1652
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 8.426878929138184,
+      "learning_rate": 8.006465603846552e-06,
+      "loss": 0.2468,
+      "step": 1653
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 19.0703067779541,
+      "learning_rate": 8.004227857915104e-06,
+      "loss": 0.3134,
+      "step": 1654
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 9.693198204040527,
+      "learning_rate": 8.00198916986223e-06,
+      "loss": 0.2856,
+      "step": 1655
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 10.663041114807129,
+      "learning_rate": 7.999749540389981e-06,
+      "loss": 0.2706,
+      "step": 1656
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 8.400293350219727,
+      "learning_rate": 7.9975089702007e-06,
+      "loss": 0.1299,
+      "step": 1657
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 11.346268653869629,
+      "learning_rate": 7.995267459997027e-06,
+      "loss": 0.2599,
+      "step": 1658
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 5.842451572418213,
+      "learning_rate": 7.993025010481895e-06,
+      "loss": 0.0766,
+      "step": 1659
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 8.552811622619629,
+      "learning_rate": 7.990781622358535e-06,
+      "loss": 0.1894,
+      "step": 1660
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 8.592625617980957,
+      "learning_rate": 7.988537296330468e-06,
+      "loss": 0.1249,
+      "step": 1661
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 13.246793746948242,
+      "learning_rate": 7.986292033101516e-06,
+      "loss": 0.3097,
+      "step": 1662
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 7.13145637512207,
+      "learning_rate": 7.984045833375783e-06,
+      "loss": 0.0844,
+      "step": 1663
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 6.8770575523376465,
+      "learning_rate": 7.981798697857679e-06,
+      "loss": 0.1825,
+      "step": 1664
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 10.966218948364258,
+      "learning_rate": 7.979550627251901e-06,
+      "loss": 0.2699,
+      "step": 1665
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 11.045105934143066,
+      "learning_rate": 7.97730162226344e-06,
+      "loss": 0.3518,
+      "step": 1666
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 13.319138526916504,
+      "learning_rate": 7.975051683597583e-06,
+      "loss": 0.2802,
+      "step": 1667
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 12.945463180541992,
+      "learning_rate": 7.972800811959906e-06,
+      "loss": 0.2447,
+      "step": 1668
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 12.985308647155762,
+      "learning_rate": 7.970549008056276e-06,
+      "loss": 0.2962,
+      "step": 1669
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 8.034134864807129,
+      "learning_rate": 7.968296272592862e-06,
+      "loss": 0.1672,
+      "step": 1670
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 8.849628448486328,
+      "learning_rate": 7.966042606276112e-06,
+      "loss": 0.1848,
+      "step": 1671
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 6.787851333618164,
+      "learning_rate": 7.963788009812775e-06,
+      "loss": 0.1938,
+      "step": 1672
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 11.477965354919434,
+      "learning_rate": 7.961532483909891e-06,
+      "loss": 0.3044,
+      "step": 1673
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 13.103387832641602,
+      "learning_rate": 7.959276029274788e-06,
+      "loss": 0.2708,
+      "step": 1674
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 11.375099182128906,
+      "learning_rate": 7.957018646615085e-06,
+      "loss": 0.4831,
+      "step": 1675
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 7.0524115562438965,
+      "learning_rate": 7.954760336638696e-06,
+      "loss": 0.117,
+      "step": 1676
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 7.160011291503906,
+      "learning_rate": 7.952501100053824e-06,
+      "loss": 0.2021,
+      "step": 1677
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 9.239439010620117,
+      "learning_rate": 7.95024093756896e-06,
+      "loss": 0.1403,
+      "step": 1678
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 9.812825202941895,
+      "learning_rate": 7.947979849892891e-06,
+      "loss": 0.1509,
+      "step": 1679
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 8.88759994506836,
+      "learning_rate": 7.945717837734688e-06,
+      "loss": 0.1104,
+      "step": 1680
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 8.07174301147461,
+      "learning_rate": 7.943454901803717e-06,
+      "loss": 0.2833,
+      "step": 1681
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 9.000137329101562,
+      "learning_rate": 7.941191042809628e-06,
+      "loss": 0.2401,
+      "step": 1682
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 11.587701797485352,
+      "learning_rate": 7.938926261462366e-06,
+      "loss": 0.2327,
+      "step": 1683
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 15.040950775146484,
+      "learning_rate": 7.936660558472165e-06,
+      "loss": 0.2578,
+      "step": 1684
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 10.90065860748291,
+      "learning_rate": 7.934393934549542e-06,
+      "loss": 0.1423,
+      "step": 1685
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8526148969889065,
+      "eval_loss": 0.467151403427124,
+      "eval_runtime": 11.4201,
+      "eval_samples_per_second": 165.76,
+      "eval_steps_per_second": 20.753,
+      "step": 1685
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 5.472853660583496,
+      "learning_rate": 7.932126390405308e-06,
+      "loss": 0.1189,
+      "step": 1686
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 7.880744934082031,
+      "learning_rate": 7.929857926750565e-06,
+      "loss": 0.1523,
+      "step": 1687
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 12.172412872314453,
+      "learning_rate": 7.927588544296696e-06,
+      "loss": 0.2774,
+      "step": 1688
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 5.3145036697387695,
+      "learning_rate": 7.925318243755378e-06,
+      "loss": 0.1462,
+      "step": 1689
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 4.823976993560791,
+      "learning_rate": 7.923047025838573e-06,
+      "loss": 0.0619,
+      "step": 1690
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 9.499890327453613,
+      "learning_rate": 7.92077489125853e-06,
+      "loss": 0.254,
+      "step": 1691
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 5.764005661010742,
+      "learning_rate": 7.918501840727791e-06,
+      "loss": 0.1125,
+      "step": 1692
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 10.792818069458008,
+      "learning_rate": 7.916227874959179e-06,
+      "loss": 0.292,
+      "step": 1693
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 7.254709720611572,
+      "learning_rate": 7.913952994665805e-06,
+      "loss": 0.0912,
+      "step": 1694
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 7.8227643966674805,
+      "learning_rate": 7.91167720056107e-06,
+      "loss": 0.2016,
+      "step": 1695
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 6.892558574676514,
+      "learning_rate": 7.909400493358661e-06,
+      "loss": 0.088,
+      "step": 1696
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 5.473608493804932,
+      "learning_rate": 7.907122873772545e-06,
+      "loss": 0.0698,
+      "step": 1697
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 6.397436618804932,
+      "learning_rate": 7.904844342516986e-06,
+      "loss": 0.2197,
+      "step": 1698
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 10.40845012664795,
+      "learning_rate": 7.902564900306525e-06,
+      "loss": 0.1735,
+      "step": 1699
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 3.3476366996765137,
+      "learning_rate": 7.900284547855992e-06,
+      "loss": 0.0538,
+      "step": 1700
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 7.5931596755981445,
+      "learning_rate": 7.898003285880502e-06,
+      "loss": 0.1396,
+      "step": 1701
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 10.305922508239746,
+      "learning_rate": 7.895721115095456e-06,
+      "loss": 0.1576,
+      "step": 1702
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 8.449909210205078,
+      "learning_rate": 7.893438036216539e-06,
+      "loss": 0.2015,
+      "step": 1703
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 5.460334300994873,
+      "learning_rate": 7.89115404995972e-06,
+      "loss": 0.067,
+      "step": 1704
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 6.984951972961426,
+      "learning_rate": 7.888869157041257e-06,
+      "loss": 0.1072,
+      "step": 1705
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 7.931661605834961,
+      "learning_rate": 7.886583358177685e-06,
+      "loss": 0.1363,
+      "step": 1706
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 6.6865763664245605,
+      "learning_rate": 7.88429665408583e-06,
+      "loss": 0.1539,
+      "step": 1707
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 7.8935980796813965,
+      "learning_rate": 7.8820090454828e-06,
+      "loss": 0.1878,
+      "step": 1708
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 9.355162620544434,
+      "learning_rate": 7.879720533085982e-06,
+      "loss": 0.0766,
+      "step": 1709
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 4.104711055755615,
+      "learning_rate": 7.87743111761305e-06,
+      "loss": 0.0769,
+      "step": 1710
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 6.932102203369141,
+      "learning_rate": 7.875140799781966e-06,
+      "loss": 0.1095,
+      "step": 1711
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 8.628235816955566,
+      "learning_rate": 7.872849580310969e-06,
+      "loss": 0.2044,
+      "step": 1712
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 7.872895240783691,
+      "learning_rate": 7.870557459918578e-06,
+      "loss": 0.1317,
+      "step": 1713
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 10.349996566772461,
+      "learning_rate": 7.868264439323601e-06,
+      "loss": 0.1044,
+      "step": 1714
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 20.98219108581543,
+      "learning_rate": 7.865970519245129e-06,
+      "loss": 0.1501,
+      "step": 1715
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 14.068634033203125,
+      "learning_rate": 7.863675700402527e-06,
+      "loss": 0.189,
+      "step": 1716
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 10.971488952636719,
+      "learning_rate": 7.861379983515449e-06,
+      "loss": 0.1963,
+      "step": 1717
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 9.381933212280273,
+      "learning_rate": 7.859083369303828e-06,
+      "loss": 0.1522,
+      "step": 1718
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 8.87316608428955,
+      "learning_rate": 7.856785858487882e-06,
+      "loss": 0.101,
+      "step": 1719
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 9.519981384277344,
+      "learning_rate": 7.8544874517881e-06,
+      "loss": 0.1414,
+      "step": 1720
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 13.593430519104004,
+      "learning_rate": 7.852188149925263e-06,
+      "loss": 0.1521,
+      "step": 1721
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 8.644328117370605,
+      "learning_rate": 7.849887953620433e-06,
+      "loss": 0.1313,
+      "step": 1722
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 3.1654157638549805,
+      "learning_rate": 7.847586863594941e-06,
+      "loss": 0.0438,
+      "step": 1723
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 12.764348030090332,
+      "learning_rate": 7.845284880570408e-06,
+      "loss": 0.2143,
+      "step": 1724
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 6.600203037261963,
+      "learning_rate": 7.842982005268733e-06,
+      "loss": 0.104,
+      "step": 1725
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 8.273323059082031,
+      "learning_rate": 7.840678238412096e-06,
+      "loss": 0.1691,
+      "step": 1726
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 5.351938247680664,
+      "learning_rate": 7.838373580722952e-06,
+      "loss": 0.3137,
+      "step": 1727
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 4.8910627365112305,
+      "learning_rate": 7.836068032924039e-06,
+      "loss": 0.0832,
+      "step": 1728
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 9.391779899597168,
+      "learning_rate": 7.833761595738375e-06,
+      "loss": 0.1327,
+      "step": 1729
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 8.333192825317383,
+      "learning_rate": 7.831454269889251e-06,
+      "loss": 0.1763,
+      "step": 1730
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 4.4484453201293945,
+      "learning_rate": 7.829146056100246e-06,
+      "loss": 0.0549,
+      "step": 1731
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 6.503353118896484,
+      "learning_rate": 7.82683695509521e-06,
+      "loss": 0.2499,
+      "step": 1732
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 5.235106468200684,
+      "learning_rate": 7.824526967598274e-06,
+      "loss": 0.0617,
+      "step": 1733
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 8.127753257751465,
+      "learning_rate": 7.822216094333847e-06,
+      "loss": 0.085,
+      "step": 1734
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 6.768527030944824,
+      "learning_rate": 7.819904336026615e-06,
+      "loss": 0.1325,
+      "step": 1735
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 4.215873718261719,
+      "learning_rate": 7.817591693401542e-06,
+      "loss": 0.0677,
+      "step": 1736
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 8.056612968444824,
+      "learning_rate": 7.81527816718387e-06,
+      "loss": 0.1486,
+      "step": 1737
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 14.140884399414062,
+      "learning_rate": 7.812963758099118e-06,
+      "loss": 0.3109,
+      "step": 1738
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 5.586012840270996,
+      "learning_rate": 7.810648466873077e-06,
+      "loss": 0.1001,
+      "step": 1739
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 11.217543601989746,
+      "learning_rate": 7.808332294231824e-06,
+      "loss": 0.1445,
+      "step": 1740
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 7.665545463562012,
+      "learning_rate": 7.806015240901705e-06,
+      "loss": 0.1263,
+      "step": 1741
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 4.549514293670654,
+      "learning_rate": 7.803697307609345e-06,
+      "loss": 0.0889,
+      "step": 1742
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 7.4150614738464355,
+      "learning_rate": 7.801378495081645e-06,
+      "loss": 0.148,
+      "step": 1743
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 4.344112396240234,
+      "learning_rate": 7.799058804045781e-06,
+      "loss": 0.0545,
+      "step": 1744
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 10.150915145874023,
+      "learning_rate": 7.796738235229203e-06,
+      "loss": 0.3222,
+      "step": 1745
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 4.348765850067139,
+      "learning_rate": 7.79441678935964e-06,
+      "loss": 0.06,
+      "step": 1746
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 4.488232612609863,
+      "learning_rate": 7.792094467165092e-06,
+      "loss": 0.0369,
+      "step": 1747
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 8.27028751373291,
+      "learning_rate": 7.789771269373838e-06,
+      "loss": 0.2067,
+      "step": 1748
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 6.904623031616211,
+      "learning_rate": 7.787447196714428e-06,
+      "loss": 0.2381,
+      "step": 1749
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 7.626152038574219,
+      "learning_rate": 7.785122249915688e-06,
+      "loss": 0.1478,
+      "step": 1750
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 5.503763675689697,
+      "learning_rate": 7.782796429706721e-06,
+      "loss": 0.174,
+      "step": 1751
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 8.20252799987793,
+      "learning_rate": 7.780469736816893e-06,
+      "loss": 0.0753,
+      "step": 1752
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 6.0944671630859375,
+      "learning_rate": 7.77814217197586e-06,
+      "loss": 0.0896,
+      "step": 1753
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 5.55197811126709,
+      "learning_rate": 7.775813735913537e-06,
+      "loss": 0.1225,
+      "step": 1754
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 11.008299827575684,
+      "learning_rate": 7.773484429360122e-06,
+      "loss": 0.2342,
+      "step": 1755
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 4.809423446655273,
+      "learning_rate": 7.77115425304608e-06,
+      "loss": 0.053,
+      "step": 1756
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 13.733948707580566,
+      "learning_rate": 7.768823207702152e-06,
+      "loss": 0.3119,
+      "step": 1757
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 5.083228588104248,
+      "learning_rate": 7.766491294059349e-06,
+      "loss": 0.0413,
+      "step": 1758
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 8.167535781860352,
+      "learning_rate": 7.764158512848955e-06,
+      "loss": 0.1725,
+      "step": 1759
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 15.466053009033203,
+      "learning_rate": 7.76182486480253e-06,
+      "loss": 0.0599,
+      "step": 1760
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 6.521265983581543,
+      "learning_rate": 7.7594903506519e-06,
+      "loss": 0.1208,
+      "step": 1761
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 10.546597480773926,
+      "learning_rate": 7.757154971129166e-06,
+      "loss": 0.1398,
+      "step": 1762
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 7.894347190856934,
+      "learning_rate": 7.7548187269667e-06,
+      "loss": 0.175,
+      "step": 1763
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 10.137228965759277,
+      "learning_rate": 7.752481618897142e-06,
+      "loss": 0.2042,
+      "step": 1764
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 6.5191545486450195,
+      "learning_rate": 7.750143647653409e-06,
+      "loss": 0.0726,
+      "step": 1765
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 7.108578205108643,
+      "learning_rate": 7.747804813968684e-06,
+      "loss": 0.1484,
+      "step": 1766
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 8.344608306884766,
+      "learning_rate": 7.745465118576423e-06,
+      "loss": 0.1152,
+      "step": 1767
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 15.176689147949219,
+      "learning_rate": 7.743124562210351e-06,
+      "loss": 0.1822,
+      "step": 1768
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 8.719675064086914,
+      "learning_rate": 7.74078314560446e-06,
+      "loss": 0.1347,
+      "step": 1769
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 11.322617530822754,
+      "learning_rate": 7.738440869493018e-06,
+      "loss": 0.1851,
+      "step": 1770
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 3.826582908630371,
+      "learning_rate": 7.736097734610557e-06,
+      "loss": 0.0283,
+      "step": 1771
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 9.679805755615234,
+      "learning_rate": 7.733753741691881e-06,
+      "loss": 0.1935,
+      "step": 1772
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 11.387218475341797,
+      "learning_rate": 7.731408891472065e-06,
+      "loss": 0.1258,
+      "step": 1773
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 3.9087154865264893,
+      "learning_rate": 7.72906318468645e-06,
+      "loss": 0.0564,
+      "step": 1774
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 7.49425745010376,
+      "learning_rate": 7.726716622070643e-06,
+      "loss": 0.1271,
+      "step": 1775
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 8.774395942687988,
+      "learning_rate": 7.724369204360527e-06,
+      "loss": 0.1836,
+      "step": 1776
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 6.549901485443115,
+      "learning_rate": 7.722020932292244e-06,
+      "loss": 0.1114,
+      "step": 1777
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 12.289834022521973,
+      "learning_rate": 7.719671806602212e-06,
+      "loss": 0.2325,
+      "step": 1778
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 11.895524978637695,
+      "learning_rate": 7.717321828027114e-06,
+      "loss": 0.2158,
+      "step": 1779
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 11.651168823242188,
+      "learning_rate": 7.714970997303898e-06,
+      "loss": 0.088,
+      "step": 1780
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 6.040383338928223,
+      "learning_rate": 7.712619315169782e-06,
+      "loss": 0.1245,
+      "step": 1781
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 9.580984115600586,
+      "learning_rate": 7.710266782362248e-06,
+      "loss": 0.2584,
+      "step": 1782
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 8.650046348571777,
+      "learning_rate": 7.707913399619048e-06,
+      "loss": 0.1506,
+      "step": 1783
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 6.669062614440918,
+      "learning_rate": 7.7055591676782e-06,
+      "loss": 0.1066,
+      "step": 1784
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 13.426026344299316,
+      "learning_rate": 7.703204087277989e-06,
+      "loss": 0.2163,
+      "step": 1785
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 10.157557487487793,
+      "learning_rate": 7.700848159156961e-06,
+      "loss": 0.1244,
+      "step": 1786
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 7.363010406494141,
+      "learning_rate": 7.698491384053935e-06,
+      "loss": 0.1471,
+      "step": 1787
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 4.805739402770996,
+      "learning_rate": 7.696133762707991e-06,
+      "loss": 0.0405,
+      "step": 1788
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 10.679381370544434,
+      "learning_rate": 7.693775295858474e-06,
+      "loss": 0.1527,
+      "step": 1789
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 13.482617378234863,
+      "learning_rate": 7.691415984244998e-06,
+      "loss": 0.0943,
+      "step": 1790
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 13.224325180053711,
+      "learning_rate": 7.689055828607439e-06,
+      "loss": 0.1801,
+      "step": 1791
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 9.897109985351562,
+      "learning_rate": 7.686694829685937e-06,
+      "loss": 0.1222,
+      "step": 1792
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 16.330673217773438,
+      "learning_rate": 7.684332988220901e-06,
+      "loss": 0.2283,
+      "step": 1793
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 3.774388551712036,
+      "learning_rate": 7.681970304952998e-06,
+      "loss": 0.0779,
+      "step": 1794
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 26.960880279541016,
+      "learning_rate": 7.679606780623162e-06,
+      "loss": 0.1233,
+      "step": 1795
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 6.422726631164551,
+      "learning_rate": 7.677242415972593e-06,
+      "loss": 0.0859,
+      "step": 1796
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 5.697916507720947,
+      "learning_rate": 7.67487721174275e-06,
+      "loss": 0.0506,
+      "step": 1797
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 7.472025394439697,
+      "learning_rate": 7.672511168675358e-06,
+      "loss": 0.0769,
+      "step": 1798
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 6.718996047973633,
+      "learning_rate": 7.670144287512407e-06,
+      "loss": 0.1004,
+      "step": 1799
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 16.97801971435547,
+      "learning_rate": 7.667776568996143e-06,
+      "loss": 0.2235,
+      "step": 1800
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 11.739739418029785,
+      "learning_rate": 7.665408013869082e-06,
+      "loss": 0.2575,
+      "step": 1801
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 8.007694244384766,
+      "learning_rate": 7.663038622873999e-06,
+      "loss": 0.3075,
+      "step": 1802
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 15.290051460266113,
+      "learning_rate": 7.660668396753932e-06,
+      "loss": 0.2396,
+      "step": 1803
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 10.207197189331055,
+      "learning_rate": 7.658297336252181e-06,
+      "loss": 0.1653,
+      "step": 1804
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 5.925869941711426,
+      "learning_rate": 7.655925442112303e-06,
+      "loss": 0.1149,
+      "step": 1805
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 10.37567138671875,
+      "learning_rate": 7.653552715078127e-06,
+      "loss": 0.1288,
+      "step": 1806
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 5.54917573928833,
+      "learning_rate": 7.651179155893734e-06,
+      "loss": 0.0833,
+      "step": 1807
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 8.356887817382812,
+      "learning_rate": 7.648804765303463e-06,
+      "loss": 0.0743,
+      "step": 1808
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 6.022553443908691,
+      "learning_rate": 7.646429544051928e-06,
+      "loss": 0.0882,
+      "step": 1809
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 8.277181625366211,
+      "learning_rate": 7.64405349288399e-06,
+      "loss": 0.0828,
+      "step": 1810
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 9.221254348754883,
+      "learning_rate": 7.641676612544775e-06,
+      "loss": 0.0717,
+      "step": 1811
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 13.451896667480469,
+      "learning_rate": 7.63929890377967e-06,
+      "loss": 0.1805,
+      "step": 1812
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 4.225894451141357,
+      "learning_rate": 7.636920367334324e-06,
+      "loss": 0.0394,
+      "step": 1813
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 6.138599872589111,
+      "learning_rate": 7.63454100395464e-06,
+      "loss": 0.0526,
+      "step": 1814
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 3.7959001064300537,
+      "learning_rate": 7.63216081438678e-06,
+      "loss": 0.0388,
+      "step": 1815
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 7.665092468261719,
+      "learning_rate": 7.62977979937717e-06,
+      "loss": 0.1423,
+      "step": 1816
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 6.864563465118408,
+      "learning_rate": 7.627397959672497e-06,
+      "loss": 0.0846,
+      "step": 1817
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 9.641164779663086,
+      "learning_rate": 7.625015296019698e-06,
+      "loss": 0.1123,
+      "step": 1818
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 9.005359649658203,
+      "learning_rate": 7.622631809165972e-06,
+      "loss": 0.1088,
+      "step": 1819
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 7.673786163330078,
+      "learning_rate": 7.62024749985878e-06,
+      "loss": 0.0958,
+      "step": 1820
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 6.59225606918335,
+      "learning_rate": 7.617862368845836e-06,
+      "loss": 0.1174,
+      "step": 1821
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 3.9118568897247314,
+      "learning_rate": 7.615476416875114e-06,
+      "loss": 0.0386,
+      "step": 1822
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 7.735174179077148,
+      "learning_rate": 7.613089644694847e-06,
+      "loss": 0.1273,
+      "step": 1823
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 6.3440775871276855,
+      "learning_rate": 7.610702053053519e-06,
+      "loss": 0.107,
+      "step": 1824
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 8.716217994689941,
+      "learning_rate": 7.6083136426998786e-06,
+      "loss": 0.1251,
+      "step": 1825
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 11.212098121643066,
+      "learning_rate": 7.605924414382926e-06,
+      "loss": 0.1832,
+      "step": 1826
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 9.56491470336914,
+      "learning_rate": 7.6035343688519215e-06,
+      "loss": 0.093,
+      "step": 1827
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 10.128671646118164,
+      "learning_rate": 7.6011435068563766e-06,
+      "loss": 0.1391,
+      "step": 1828
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 8.256789207458496,
+      "learning_rate": 7.598751829146067e-06,
+      "loss": 0.3059,
+      "step": 1829
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 10.06120491027832,
+      "learning_rate": 7.596359336471015e-06,
+      "loss": 0.0922,
+      "step": 1830
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 8.2539644241333,
+      "learning_rate": 7.593966029581505e-06,
+      "loss": 0.0817,
+      "step": 1831
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 10.67087459564209,
+      "learning_rate": 7.5915719092280746e-06,
+      "loss": 0.1859,
+      "step": 1832
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 8.103544235229492,
+      "learning_rate": 7.589176976161514e-06,
+      "loss": 0.0878,
+      "step": 1833
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 11.449383735656738,
+      "learning_rate": 7.586781231132873e-06,
+      "loss": 0.2088,
+      "step": 1834
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 5.808707237243652,
+      "learning_rate": 7.584384674893454e-06,
+      "loss": 0.1064,
+      "step": 1835
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 8.995357513427734,
+      "learning_rate": 7.5819873081948105e-06,
+      "loss": 0.0911,
+      "step": 1836
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 8.965944290161133,
+      "learning_rate": 7.579589131788756e-06,
+      "loss": 0.0703,
+      "step": 1837
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 10.279400825500488,
+      "learning_rate": 7.577190146427356e-06,
+      "loss": 0.1299,
+      "step": 1838
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 8.087530136108398,
+      "learning_rate": 7.574790352862924e-06,
+      "loss": 0.1011,
+      "step": 1839
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 3.0638861656188965,
+      "learning_rate": 7.572389751848037e-06,
+      "loss": 0.0513,
+      "step": 1840
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 8.572397232055664,
+      "learning_rate": 7.569988344135517e-06,
+      "loss": 0.1336,
+      "step": 1841
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 7.73654842376709,
+      "learning_rate": 7.567586130478443e-06,
+      "loss": 0.0527,
+      "step": 1842
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 9.925785064697266,
+      "learning_rate": 7.565183111630145e-06,
+      "loss": 0.1155,
+      "step": 1843
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 6.975909233093262,
+      "learning_rate": 7.562779288344206e-06,
+      "loss": 0.1007,
+      "step": 1844
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 7.92281436920166,
+      "learning_rate": 7.560374661374463e-06,
+      "loss": 0.1923,
+      "step": 1845
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 2.749950647354126,
+      "learning_rate": 7.557969231475002e-06,
+      "loss": 0.0362,
+      "step": 1846
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 7.031035900115967,
+      "learning_rate": 7.555562999400162e-06,
+      "loss": 0.06,
+      "step": 1847
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 8.419447898864746,
+      "learning_rate": 7.553155965904535e-06,
+      "loss": 0.0489,
+      "step": 1848
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 10.56812572479248,
+      "learning_rate": 7.5507481317429645e-06,
+      "loss": 0.1123,
+      "step": 1849
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 6.96979284286499,
+      "learning_rate": 7.548339497670538e-06,
+      "loss": 0.096,
+      "step": 1850
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 19.239830017089844,
+      "learning_rate": 7.545930064442606e-06,
+      "loss": 0.2141,
+      "step": 1851
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 7.007989406585693,
+      "learning_rate": 7.543519832814761e-06,
+      "loss": 0.1078,
+      "step": 1852
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 9.355853080749512,
+      "learning_rate": 7.541108803542846e-06,
+      "loss": 0.1344,
+      "step": 1853
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 12.150227546691895,
+      "learning_rate": 7.538696977382959e-06,
+      "loss": 0.2888,
+      "step": 1854
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 5.291398048400879,
+      "learning_rate": 7.536284355091443e-06,
+      "loss": 0.0454,
+      "step": 1855
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 10.27534294128418,
+      "learning_rate": 7.533870937424896e-06,
+      "loss": 0.1549,
+      "step": 1856
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 7.7099432945251465,
+      "learning_rate": 7.5314567251401605e-06,
+      "loss": 0.0651,
+      "step": 1857
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 15.50121021270752,
+      "learning_rate": 7.529041718994326e-06,
+      "loss": 0.1818,
+      "step": 1858
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 7.44112491607666,
+      "learning_rate": 7.526625919744741e-06,
+      "loss": 0.1381,
+      "step": 1859
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 5.895039081573486,
+      "learning_rate": 7.524209328148995e-06,
+      "loss": 0.0737,
+      "step": 1860
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 17.821828842163086,
+      "learning_rate": 7.521791944964923e-06,
+      "loss": 0.2203,
+      "step": 1861
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 8.662997245788574,
+      "learning_rate": 7.519373770950619e-06,
+      "loss": 0.1165,
+      "step": 1862
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 15.438583374023438,
+      "learning_rate": 7.516954806864416e-06,
+      "loss": 0.133,
+      "step": 1863
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 3.452115774154663,
+      "learning_rate": 7.5145350534648955e-06,
+      "loss": 0.0386,
+      "step": 1864
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 15.005075454711914,
+      "learning_rate": 7.512114511510893e-06,
+      "loss": 0.1262,
+      "step": 1865
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 11.19264030456543,
+      "learning_rate": 7.509693181761485e-06,
+      "loss": 0.2959,
+      "step": 1866
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 8.72252368927002,
+      "learning_rate": 7.507271064975995e-06,
+      "loss": 0.1538,
+      "step": 1867
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 13.720284461975098,
+      "learning_rate": 7.504848161913999e-06,
+      "loss": 0.2581,
+      "step": 1868
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 12.157021522521973,
+      "learning_rate": 7.502424473335311e-06,
+      "loss": 0.1138,
+      "step": 1869
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 13.608020782470703,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.1219,
+      "step": 1870
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 11.077437400817871,
+      "learning_rate": 7.497574742668378e-06,
+      "loss": 0.1651,
+      "step": 1871
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 22.26176643371582,
+      "learning_rate": 7.495148702100998e-06,
+      "loss": 0.2788,
+      "step": 1872
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 5.8257927894592285,
+      "learning_rate": 7.492721879058666e-06,
+      "loss": 0.0509,
+      "step": 1873
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 12.714739799499512,
+      "learning_rate": 7.490294274302428e-06,
+      "loss": 0.2517,
+      "step": 1874
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 5.771032810211182,
+      "learning_rate": 7.487865888593579e-06,
+      "loss": 0.0553,
+      "step": 1875
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 6.254287242889404,
+      "learning_rate": 7.485436722693658e-06,
+      "loss": 0.0823,
+      "step": 1876
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 8.646843910217285,
+      "learning_rate": 7.483006777364446e-06,
+      "loss": 0.1827,
+      "step": 1877
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 14.277413368225098,
+      "learning_rate": 7.480576053367968e-06,
+      "loss": 0.2071,
+      "step": 1878
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 17.362031936645508,
+      "learning_rate": 7.478144551466501e-06,
+      "loss": 0.1412,
+      "step": 1879
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 9.932319641113281,
+      "learning_rate": 7.4757122724225575e-06,
+      "loss": 0.3386,
+      "step": 1880
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 3.624013900756836,
+      "learning_rate": 7.473279216998896e-06,
+      "loss": 0.0383,
+      "step": 1881
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 10.792884826660156,
+      "learning_rate": 7.47084538595852e-06,
+      "loss": 0.2028,
+      "step": 1882
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 7.358582496643066,
+      "learning_rate": 7.468410780064674e-06,
+      "loss": 0.1703,
+      "step": 1883
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 10.806937217712402,
+      "learning_rate": 7.465975400080848e-06,
+      "loss": 0.0842,
+      "step": 1884
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 9.701594352722168,
+      "learning_rate": 7.463539246770775e-06,
+      "loss": 0.0524,
+      "step": 1885
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 8.94115161895752,
+      "learning_rate": 7.461102320898425e-06,
+      "loss": 0.1172,
+      "step": 1886
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 7.358184337615967,
+      "learning_rate": 7.45866462322802e-06,
+      "loss": 0.1264,
+      "step": 1887
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 3.4271559715270996,
+      "learning_rate": 7.456226154524014e-06,
+      "loss": 0.0476,
+      "step": 1888
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 12.198417663574219,
+      "learning_rate": 7.453786915551108e-06,
+      "loss": 0.1373,
+      "step": 1889
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 11.943309783935547,
+      "learning_rate": 7.451346907074245e-06,
+      "loss": 0.2498,
+      "step": 1890
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 9.531939506530762,
+      "learning_rate": 7.448906129858607e-06,
+      "loss": 0.1154,
+      "step": 1891
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 10.394180297851562,
+      "learning_rate": 7.4464645846696186e-06,
+      "loss": 0.1128,
+      "step": 1892
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 9.62152099609375,
+      "learning_rate": 7.444022272272945e-06,
+      "loss": 0.1217,
+      "step": 1893
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 11.14421558380127,
+      "learning_rate": 7.441579193434491e-06,
+      "loss": 0.1423,
+      "step": 1894
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 13.293745040893555,
+      "learning_rate": 7.439135348920403e-06,
+      "loss": 0.2837,
+      "step": 1895
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 7.392568588256836,
+      "learning_rate": 7.436690739497068e-06,
+      "loss": 0.0912,
+      "step": 1896
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 8.390217781066895,
+      "learning_rate": 7.434245365931109e-06,
+      "loss": 0.0951,
+      "step": 1897
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 6.018220901489258,
+      "learning_rate": 7.431799228989394e-06,
+      "loss": 0.1147,
+      "step": 1898
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 9.345630645751953,
+      "learning_rate": 7.429352329439028e-06,
+      "loss": 0.1571,
+      "step": 1899
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 5.458803653717041,
+      "learning_rate": 7.426904668047352e-06,
+      "loss": 0.07,
+      "step": 1900
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 11.389274597167969,
+      "learning_rate": 7.424456245581955e-06,
+      "loss": 0.2003,
+      "step": 1901
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 7.642900466918945,
+      "learning_rate": 7.4220070628106536e-06,
+      "loss": 0.0768,
+      "step": 1902
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 11.529337882995605,
+      "learning_rate": 7.419557120501508e-06,
+      "loss": 0.1042,
+      "step": 1903
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 9.364057540893555,
+      "learning_rate": 7.4171064194228196e-06,
+      "loss": 0.1308,
+      "step": 1904
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 9.944604873657227,
+      "learning_rate": 7.4146549603431225e-06,
+      "loss": 0.1493,
+      "step": 1905
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 11.133593559265137,
+      "learning_rate": 7.412202744031192e-06,
+      "loss": 0.2781,
+      "step": 1906
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 7.128958225250244,
+      "learning_rate": 7.409749771256038e-06,
+      "loss": 0.1175,
+      "step": 1907
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 12.823749542236328,
+      "learning_rate": 7.407296042786909e-06,
+      "loss": 0.2114,
+      "step": 1908
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 8.45661449432373,
+      "learning_rate": 7.404841559393292e-06,
+      "loss": 0.1723,
+      "step": 1909
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 4.854256629943848,
+      "learning_rate": 7.40238632184491e-06,
+      "loss": 0.075,
+      "step": 1910
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 9.63552188873291,
+      "learning_rate": 7.399930330911717e-06,
+      "loss": 0.1672,
+      "step": 1911
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 13.733352661132812,
+      "learning_rate": 7.397473587363916e-06,
+      "loss": 0.1653,
+      "step": 1912
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 7.626954555511475,
+      "learning_rate": 7.395016091971932e-06,
+      "loss": 0.0753,
+      "step": 1913
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 8.17002010345459,
+      "learning_rate": 7.392557845506433e-06,
+      "loss": 0.2993,
+      "step": 1914
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 12.628567695617676,
+      "learning_rate": 7.390098848738324e-06,
+      "loss": 0.1554,
+      "step": 1915
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 16.363197326660156,
+      "learning_rate": 7.387639102438741e-06,
+      "loss": 0.2683,
+      "step": 1916
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 8.35317611694336,
+      "learning_rate": 7.385178607379056e-06,
+      "loss": 0.134,
+      "step": 1917
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 15.369413375854492,
+      "learning_rate": 7.382717364330877e-06,
+      "loss": 0.1876,
+      "step": 1918
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 3.8694303035736084,
+      "learning_rate": 7.380255374066048e-06,
+      "loss": 0.1117,
+      "step": 1919
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 18.648025512695312,
+      "learning_rate": 7.377792637356644e-06,
+      "loss": 0.18,
+      "step": 1920
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 13.308477401733398,
+      "learning_rate": 7.3753291549749764e-06,
+      "loss": 0.0798,
+      "step": 1921
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 10.794960975646973,
+      "learning_rate": 7.372864927693587e-06,
+      "loss": 0.1145,
+      "step": 1922
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 15.443461418151855,
+      "learning_rate": 7.370399956285257e-06,
+      "loss": 0.1763,
+      "step": 1923
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 11.408613204956055,
+      "learning_rate": 7.367934241522998e-06,
+      "loss": 0.1297,
+      "step": 1924
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 10.385476112365723,
+      "learning_rate": 7.365467784180051e-06,
+      "loss": 0.0982,
+      "step": 1925
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 13.274757385253906,
+      "learning_rate": 7.363000585029897e-06,
+      "loss": 0.1228,
+      "step": 1926
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 9.059858322143555,
+      "learning_rate": 7.360532644846245e-06,
+      "loss": 0.1078,
+      "step": 1927
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 13.856675148010254,
+      "learning_rate": 7.358063964403035e-06,
+      "loss": 0.2427,
+      "step": 1928
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 15.18313217163086,
+      "learning_rate": 7.355594544474444e-06,
+      "loss": 0.1191,
+      "step": 1929
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 10.977679252624512,
+      "learning_rate": 7.35312438583488e-06,
+      "loss": 0.1802,
+      "step": 1930
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 12.373780250549316,
+      "learning_rate": 7.3506534892589786e-06,
+      "loss": 0.146,
+      "step": 1931
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 5.720775604248047,
+      "learning_rate": 7.34818185552161e-06,
+      "loss": 0.0603,
+      "step": 1932
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 5.456993103027344,
+      "learning_rate": 7.345709485397873e-06,
+      "loss": 0.0482,
+      "step": 1933
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 7.471248149871826,
+      "learning_rate": 7.3432363796631025e-06,
+      "loss": 0.1554,
+      "step": 1934
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 8.800732612609863,
+      "learning_rate": 7.340762539092858e-06,
+      "loss": 0.1084,
+      "step": 1935
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 10.85830307006836,
+      "learning_rate": 7.3382879644629345e-06,
+      "loss": 0.2723,
+      "step": 1936
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 6.854144096374512,
+      "learning_rate": 7.3358126565493545e-06,
+      "loss": 0.0936,
+      "step": 1937
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 6.380908012390137,
+      "learning_rate": 7.333336616128369e-06,
+      "loss": 0.0915,
+      "step": 1938
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 16.320262908935547,
+      "learning_rate": 7.330859843976462e-06,
+      "loss": 0.1962,
+      "step": 1939
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 11.5984468460083,
+      "learning_rate": 7.3283823408703466e-06,
+      "loss": 0.1891,
+      "step": 1940
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 7.462270736694336,
+      "learning_rate": 7.325904107586964e-06,
+      "loss": 0.1061,
+      "step": 1941
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 8.484504699707031,
+      "learning_rate": 7.323425144903482e-06,
+      "loss": 0.136,
+      "step": 1942
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 13.491326332092285,
+      "learning_rate": 7.3209454535973015e-06,
+      "loss": 0.1619,
+      "step": 1943
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 11.746870994567871,
+      "learning_rate": 7.31846503444605e-06,
+      "loss": 0.2642,
+      "step": 1944
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 7.529642105102539,
+      "learning_rate": 7.315983888227583e-06,
+      "loss": 0.1239,
+      "step": 1945
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 6.648303985595703,
+      "learning_rate": 7.313502015719985e-06,
+      "loss": 0.0572,
+      "step": 1946
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 9.567227363586426,
+      "learning_rate": 7.311019417701567e-06,
+      "loss": 0.0866,
+      "step": 1947
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 8.908608436584473,
+      "learning_rate": 7.308536094950868e-06,
+      "loss": 0.0595,
+      "step": 1948
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 8.513456344604492,
+      "learning_rate": 7.306052048246653e-06,
+      "loss": 0.2112,
+      "step": 1949
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 4.941246032714844,
+      "learning_rate": 7.303567278367918e-06,
+      "loss": 0.0454,
+      "step": 1950
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 4.626683712005615,
+      "learning_rate": 7.301081786093882e-06,
+      "loss": 0.0612,
+      "step": 1951
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 5.85500955581665,
+      "learning_rate": 7.298595572203991e-06,
+      "loss": 0.194,
+      "step": 1952
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 17.181062698364258,
+      "learning_rate": 7.296108637477919e-06,
+      "loss": 0.1944,
+      "step": 1953
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 5.977094650268555,
+      "learning_rate": 7.293620982695564e-06,
+      "loss": 0.0957,
+      "step": 1954
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 7.976542949676514,
+      "learning_rate": 7.291132608637053e-06,
+      "loss": 0.0965,
+      "step": 1955
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 10.394746780395508,
+      "learning_rate": 7.288643516082734e-06,
+      "loss": 0.1769,
+      "step": 1956
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 14.247730255126953,
+      "learning_rate": 7.2861537058131846e-06,
+      "loss": 0.2371,
+      "step": 1957
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 12.47955322265625,
+      "learning_rate": 7.283663178609204e-06,
+      "loss": 0.3659,
+      "step": 1958
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 8.302145004272461,
+      "learning_rate": 7.281171935251818e-06,
+      "loss": 0.0931,
+      "step": 1959
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 8.103628158569336,
+      "learning_rate": 7.278679976522279e-06,
+      "loss": 0.1146,
+      "step": 1960
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 15.14854621887207,
+      "learning_rate": 7.276187303202059e-06,
+      "loss": 0.1728,
+      "step": 1961
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 5.338321685791016,
+      "learning_rate": 7.273693916072859e-06,
+      "loss": 0.0574,
+      "step": 1962
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 10.998879432678223,
+      "learning_rate": 7.2711998159165996e-06,
+      "loss": 0.2045,
+      "step": 1963
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 11.1698637008667,
+      "learning_rate": 7.268705003515427e-06,
+      "loss": 0.0871,
+      "step": 1964
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 11.50326919555664,
+      "learning_rate": 7.266209479651712e-06,
+      "loss": 0.2162,
+      "step": 1965
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 9.136773109436035,
+      "learning_rate": 7.263713245108048e-06,
+      "loss": 0.1027,
+      "step": 1966
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 12.213746070861816,
+      "learning_rate": 7.261216300667246e-06,
+      "loss": 0.0653,
+      "step": 1967
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 7.0609259605407715,
+      "learning_rate": 7.258718647112348e-06,
+      "loss": 0.054,
+      "step": 1968
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 7.557986259460449,
+      "learning_rate": 7.256220285226615e-06,
+      "loss": 0.062,
+      "step": 1969
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 9.677338600158691,
+      "learning_rate": 7.253721215793528e-06,
+      "loss": 0.1839,
+      "step": 1970
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 12.362751007080078,
+      "learning_rate": 7.25122143959679e-06,
+      "loss": 0.1849,
+      "step": 1971
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 8.19864559173584,
+      "learning_rate": 7.24872095742033e-06,
+      "loss": 0.338,
+      "step": 1972
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 7.282526969909668,
+      "learning_rate": 7.246219770048294e-06,
+      "loss": 0.1544,
+      "step": 1973
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 8.207942962646484,
+      "learning_rate": 7.243717878265051e-06,
+      "loss": 0.1148,
+      "step": 1974
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 5.761040687561035,
+      "learning_rate": 7.241215282855189e-06,
+      "loss": 0.0876,
+      "step": 1975
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 6.369480609893799,
+      "learning_rate": 7.238711984603521e-06,
+      "loss": 0.0688,
+      "step": 1976
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 8.323854446411133,
+      "learning_rate": 7.236207984295078e-06,
+      "loss": 0.0746,
+      "step": 1977
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 13.591177940368652,
+      "learning_rate": 7.2337032827151064e-06,
+      "loss": 0.1182,
+      "step": 1978
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 3.7483909130096436,
+      "learning_rate": 7.231197880649082e-06,
+      "loss": 0.0932,
+      "step": 1979
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 9.281496047973633,
+      "learning_rate": 7.2286917788826926e-06,
+      "loss": 0.1205,
+      "step": 1980
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 8.590771675109863,
+      "learning_rate": 7.226184978201847e-06,
+      "loss": 0.1057,
+      "step": 1981
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 5.359656810760498,
+      "learning_rate": 7.223677479392677e-06,
+      "loss": 0.0426,
+      "step": 1982
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 11.260873794555664,
+      "learning_rate": 7.221169283241528e-06,
+      "loss": 0.2316,
+      "step": 1983
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 4.898187160491943,
+      "learning_rate": 7.218660390534969e-06,
+      "loss": 0.0614,
+      "step": 1984
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 6.83081579208374,
+      "learning_rate": 7.216150802059782e-06,
+      "loss": 0.0904,
+      "step": 1985
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 9.100411415100098,
+      "learning_rate": 7.213640518602972e-06,
+      "loss": 0.1154,
+      "step": 1986
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 8.85955810546875,
+      "learning_rate": 7.211129540951762e-06,
+      "loss": 0.1508,
+      "step": 1987
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 16.003326416015625,
+      "learning_rate": 7.208617869893589e-06,
+      "loss": 0.3097,
+      "step": 1988
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 13.870522499084473,
+      "learning_rate": 7.206105506216107e-06,
+      "loss": 0.1744,
+      "step": 1989
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 15.919938087463379,
+      "learning_rate": 7.203592450707193e-06,
+      "loss": 0.1214,
+      "step": 1990
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 11.20141315460205,
+      "learning_rate": 7.201078704154938e-06,
+      "loss": 0.1063,
+      "step": 1991
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 14.137307167053223,
+      "learning_rate": 7.198564267347645e-06,
+      "loss": 0.1595,
+      "step": 1992
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 8.890684127807617,
+      "learning_rate": 7.196049141073842e-06,
+      "loss": 0.1567,
+      "step": 1993
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 11.65476131439209,
+      "learning_rate": 7.193533326122268e-06,
+      "loss": 0.0905,
+      "step": 1994
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 16.40561866760254,
+      "learning_rate": 7.1910168232818765e-06,
+      "loss": 0.0965,
+      "step": 1995
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 13.461186408996582,
+      "learning_rate": 7.188499633341842e-06,
+      "loss": 0.2659,
+      "step": 1996
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 9.509493827819824,
+      "learning_rate": 7.185981757091549e-06,
+      "loss": 0.0852,
+      "step": 1997
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 8.105579376220703,
+      "learning_rate": 7.1834631953206015e-06,
+      "loss": 0.2161,
+      "step": 1998
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 11.3521728515625,
+      "learning_rate": 7.180943948818817e-06,
+      "loss": 0.123,
+      "step": 1999
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 18.93787384033203,
+      "learning_rate": 7.178424018376224e-06,
+      "loss": 0.2381,
+      "step": 2000
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 8.154054641723633,
+      "learning_rate": 7.175903404783075e-06,
+      "loss": 0.1154,
+      "step": 2001
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 7.288686752319336,
+      "learning_rate": 7.173382108829826e-06,
+      "loss": 0.0874,
+      "step": 2002
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 8.670666694641113,
+      "learning_rate": 7.1708601313071525e-06,
+      "loss": 0.0814,
+      "step": 2003
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 10.998640060424805,
+      "learning_rate": 7.168337473005944e-06,
+      "loss": 0.2025,
+      "step": 2004
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 12.384292602539062,
+      "learning_rate": 7.165814134717303e-06,
+      "loss": 0.1011,
+      "step": 2005
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 10.838531494140625,
+      "learning_rate": 7.163290117232542e-06,
+      "loss": 0.2076,
+      "step": 2006
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 10.186249732971191,
+      "learning_rate": 7.160765421343192e-06,
+      "loss": 0.1042,
+      "step": 2007
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 8.326089859008789,
+      "learning_rate": 7.1582400478409915e-06,
+      "loss": 0.0832,
+      "step": 2008
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 8.059404373168945,
+      "learning_rate": 7.155713997517897e-06,
+      "loss": 0.0556,
+      "step": 2009
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 6.703413963317871,
+      "learning_rate": 7.153187271166071e-06,
+      "loss": 0.0613,
+      "step": 2010
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 11.536551475524902,
+      "learning_rate": 7.150659869577892e-06,
+      "loss": 0.0708,
+      "step": 2011
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 8.333263397216797,
+      "learning_rate": 7.148131793545949e-06,
+      "loss": 0.0649,
+      "step": 2012
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 4.738850116729736,
+      "learning_rate": 7.145603043863045e-06,
+      "loss": 0.1478,
+      "step": 2013
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 5.634189605712891,
+      "learning_rate": 7.14307362132219e-06,
+      "loss": 0.0925,
+      "step": 2014
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 9.243416786193848,
+      "learning_rate": 7.14054352671661e-06,
+      "loss": 0.0778,
+      "step": 2015
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 10.151582717895508,
+      "learning_rate": 7.138012760839736e-06,
+      "loss": 0.1712,
+      "step": 2016
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 3.390695333480835,
+      "learning_rate": 7.135481324485214e-06,
+      "loss": 0.1043,
+      "step": 2017
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 9.471257209777832,
+      "learning_rate": 7.1329492184468975e-06,
+      "loss": 0.0968,
+      "step": 2018
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 4.9878129959106445,
+      "learning_rate": 7.130416443518852e-06,
+      "loss": 0.0387,
+      "step": 2019
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 15.742303848266602,
+      "learning_rate": 7.127883000495353e-06,
+      "loss": 0.1265,
+      "step": 2020
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 15.836478233337402,
+      "learning_rate": 7.125348890170882e-06,
+      "loss": 0.303,
+      "step": 2021
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 7.334164142608643,
+      "learning_rate": 7.122814113340133e-06,
+      "loss": 0.082,
+      "step": 2022
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 11.46505355834961,
+      "learning_rate": 7.12027867079801e-06,
+      "loss": 0.1104,
+      "step": 2023
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 15.88946533203125,
+      "learning_rate": 7.117742563339622e-06,
+      "loss": 0.1781,
+      "step": 2024
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 18.798799514770508,
+      "learning_rate": 7.1152057917602904e-06,
+      "loss": 0.1044,
+      "step": 2025
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 10.163244247436523,
+      "learning_rate": 7.112668356855542e-06,
+      "loss": 0.0702,
+      "step": 2026
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 8.124094009399414,
+      "learning_rate": 7.110130259421113e-06,
+      "loss": 0.0673,
+      "step": 2027
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 8.489636421203613,
+      "learning_rate": 7.1075915002529465e-06,
+      "loss": 0.1237,
+      "step": 2028
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 10.574942588806152,
+      "learning_rate": 7.105052080147194e-06,
+      "loss": 0.1249,
+      "step": 2029
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 13.973773956298828,
+      "learning_rate": 7.102511999900213e-06,
+      "loss": 0.0936,
+      "step": 2030
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 5.9630842208862305,
+      "learning_rate": 7.09997126030857e-06,
+      "loss": 0.0565,
+      "step": 2031
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 7.001921653747559,
+      "learning_rate": 7.097429862169038e-06,
+      "loss": 0.0693,
+      "step": 2032
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 20.590171813964844,
+      "learning_rate": 7.094887806278594e-06,
+      "loss": 0.136,
+      "step": 2033
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 13.059353828430176,
+      "learning_rate": 7.092345093434424e-06,
+      "loss": 0.1583,
+      "step": 2034
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 11.161260604858398,
+      "learning_rate": 7.089801724433918e-06,
+      "loss": 0.1671,
+      "step": 2035
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 12.960087776184082,
+      "learning_rate": 7.087257700074674e-06,
+      "loss": 0.1542,
+      "step": 2036
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 10.72169017791748,
+      "learning_rate": 7.084713021154493e-06,
+      "loss": 0.0937,
+      "step": 2037
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 7.292314529418945,
+      "learning_rate": 7.082167688471384e-06,
+      "loss": 0.077,
+      "step": 2038
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 10.922294616699219,
+      "learning_rate": 7.079621702823557e-06,
+      "loss": 0.1529,
+      "step": 2039
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 13.563387870788574,
+      "learning_rate": 7.0770750650094335e-06,
+      "loss": 0.1051,
+      "step": 2040
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 11.582254409790039,
+      "learning_rate": 7.074527775827631e-06,
+      "loss": 0.1423,
+      "step": 2041
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 14.434452056884766,
+      "learning_rate": 7.071979836076979e-06,
+      "loss": 0.2192,
+      "step": 2042
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 11.442138671875,
+      "learning_rate": 7.069431246556505e-06,
+      "loss": 0.08,
+      "step": 2043
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 3.639971971511841,
+      "learning_rate": 7.066882008065445e-06,
+      "loss": 0.0361,
+      "step": 2044
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 8.988595962524414,
+      "learning_rate": 7.064332121403237e-06,
+      "loss": 0.1059,
+      "step": 2045
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 7.334170818328857,
+      "learning_rate": 7.061781587369518e-06,
+      "loss": 0.0905,
+      "step": 2046
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 9.073087692260742,
+      "learning_rate": 7.059230406764135e-06,
+      "loss": 0.2694,
+      "step": 2047
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 15.438549041748047,
+      "learning_rate": 7.056678580387134e-06,
+      "loss": 0.0674,
+      "step": 2048
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 5.677149295806885,
+      "learning_rate": 7.054126109038763e-06,
+      "loss": 0.0734,
+      "step": 2049
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 12.678391456604004,
+      "learning_rate": 7.051572993519474e-06,
+      "loss": 0.0936,
+      "step": 2050
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 7.4557013511657715,
+      "learning_rate": 7.049019234629919e-06,
+      "loss": 0.0842,
+      "step": 2051
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 18.09970474243164,
+      "learning_rate": 7.046464833170955e-06,
+      "loss": 0.0828,
+      "step": 2052
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 9.901593208312988,
+      "learning_rate": 7.043909789943637e-06,
+      "loss": 0.2605,
+      "step": 2053
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 3.462232828140259,
+      "learning_rate": 7.041354105749224e-06,
+      "loss": 0.0368,
+      "step": 2054
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 8.517525672912598,
+      "learning_rate": 7.038797781389174e-06,
+      "loss": 0.0714,
+      "step": 2055
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 11.246149063110352,
+      "learning_rate": 7.036240817665145e-06,
+      "loss": 0.1442,
+      "step": 2056
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 8.667082786560059,
+      "learning_rate": 7.033683215379002e-06,
+      "loss": 0.0829,
+      "step": 2057
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 5.802610397338867,
+      "learning_rate": 7.0311249753328e-06,
+      "loss": 0.15,
+      "step": 2058
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 20.464017868041992,
+      "learning_rate": 7.028566098328802e-06,
+      "loss": 0.2684,
+      "step": 2059
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 4.815258979797363,
+      "learning_rate": 7.026006585169467e-06,
+      "loss": 0.049,
+      "step": 2060
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 13.894495964050293,
+      "learning_rate": 7.023446436657455e-06,
+      "loss": 0.2327,
+      "step": 2061
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 7.54653263092041,
+      "learning_rate": 7.020885653595625e-06,
+      "loss": 0.1102,
+      "step": 2062
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 12.344463348388672,
+      "learning_rate": 7.018324236787034e-06,
+      "loss": 0.1591,
+      "step": 2063
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 19.892738342285156,
+      "learning_rate": 7.015762187034939e-06,
+      "loss": 0.1813,
+      "step": 2064
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 15.647858619689941,
+      "learning_rate": 7.013199505142796e-06,
+      "loss": 0.1977,
+      "step": 2065
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 7.887336254119873,
+      "learning_rate": 7.010636191914258e-06,
+      "loss": 0.0988,
+      "step": 2066
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 8.507963180541992,
+      "learning_rate": 7.008072248153172e-06,
+      "loss": 0.1883,
+      "step": 2067
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 16.05280876159668,
+      "learning_rate": 7.005507674663594e-06,
+      "loss": 0.1706,
+      "step": 2068
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 9.852394104003906,
+      "learning_rate": 7.002942472249767e-06,
+      "loss": 0.1953,
+      "step": 2069
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 10.603446006774902,
+      "learning_rate": 7.0003766417161335e-06,
+      "loss": 0.2557,
+      "step": 2070
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 6.354084491729736,
+      "learning_rate": 6.9978101838673375e-06,
+      "loss": 0.0905,
+      "step": 2071
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 8.485651016235352,
+      "learning_rate": 6.9952430995082114e-06,
+      "loss": 0.0518,
+      "step": 2072
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 4.823178291320801,
+      "learning_rate": 6.992675389443792e-06,
+      "loss": 0.0529,
+      "step": 2073
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 2.0406432151794434,
+      "learning_rate": 6.990107054479313e-06,
+      "loss": 0.0892,
+      "step": 2074
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 9.512364387512207,
+      "learning_rate": 6.987538095420193e-06,
+      "loss": 0.114,
+      "step": 2075
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 7.043548583984375,
+      "learning_rate": 6.984968513072059e-06,
+      "loss": 0.1345,
+      "step": 2076
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 7.650743007659912,
+      "learning_rate": 6.982398308240727e-06,
+      "loss": 0.0654,
+      "step": 2077
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 6.3703389167785645,
+      "learning_rate": 6.979827481732209e-06,
+      "loss": 0.0632,
+      "step": 2078
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 9.077935218811035,
+      "learning_rate": 6.977256034352713e-06,
+      "loss": 0.1205,
+      "step": 2079
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 12.779865264892578,
+      "learning_rate": 6.974683966908642e-06,
+      "loss": 0.1899,
+      "step": 2080
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 12.632820129394531,
+      "learning_rate": 6.972111280206588e-06,
+      "loss": 0.0898,
+      "step": 2081
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 12.215779304504395,
+      "learning_rate": 6.969537975053349e-06,
+      "loss": 0.2164,
+      "step": 2082
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 11.016061782836914,
+      "learning_rate": 6.966964052255903e-06,
+      "loss": 0.1431,
+      "step": 2083
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 9.081803321838379,
+      "learning_rate": 6.964389512621432e-06,
+      "loss": 0.1487,
+      "step": 2084
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 9.90294361114502,
+      "learning_rate": 6.961814356957308e-06,
+      "loss": 0.1537,
+      "step": 2085
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 9.585633277893066,
+      "learning_rate": 6.959238586071094e-06,
+      "loss": 0.0679,
+      "step": 2086
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 15.61821174621582,
+      "learning_rate": 6.95666220077055e-06,
+      "loss": 0.2084,
+      "step": 2087
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 9.232343673706055,
+      "learning_rate": 6.9540852018636255e-06,
+      "loss": 0.1259,
+      "step": 2088
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 13.168516159057617,
+      "learning_rate": 6.951507590158461e-06,
+      "loss": 0.1245,
+      "step": 2089
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 9.528916358947754,
+      "learning_rate": 6.948929366463397e-06,
+      "loss": 0.061,
+      "step": 2090
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 6.790521621704102,
+      "learning_rate": 6.946350531586959e-06,
+      "loss": 0.133,
+      "step": 2091
+    },
+    {
+      "epoch": 3.72,
+      "grad_norm": 10.879219055175781,
+      "learning_rate": 6.943771086337863e-06,
+      "loss": 0.1552,
+      "step": 2092
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 7.943449020385742,
+      "learning_rate": 6.9411910315250206e-06,
+      "loss": 0.1656,
+      "step": 2093
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 17.995393753051758,
+      "learning_rate": 6.938610367957536e-06,
+      "loss": 0.3659,
+      "step": 2094
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 8.700067520141602,
+      "learning_rate": 6.936029096444697e-06,
+      "loss": 0.2329,
+      "step": 2095
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 7.176606178283691,
+      "learning_rate": 6.93344721779599e-06,
+      "loss": 0.0665,
+      "step": 2096
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 11.657209396362305,
+      "learning_rate": 6.930864732821086e-06,
+      "loss": 0.1645,
+      "step": 2097
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 6.17781925201416,
+      "learning_rate": 6.92828164232985e-06,
+      "loss": 0.1748,
+      "step": 2098
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 12.678108215332031,
+      "learning_rate": 6.925697947132336e-06,
+      "loss": 0.0995,
+      "step": 2099
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 11.326314926147461,
+      "learning_rate": 6.923113648038784e-06,
+      "loss": 0.1808,
+      "step": 2100
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 11.189144134521484,
+      "learning_rate": 6.9205287458596305e-06,
+      "loss": 0.1521,
+      "step": 2101
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 10.936103820800781,
+      "learning_rate": 6.917943241405494e-06,
+      "loss": 0.1443,
+      "step": 2102
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 9.6465425491333,
+      "learning_rate": 6.915357135487184e-06,
+      "loss": 0.1108,
+      "step": 2103
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 8.065037727355957,
+      "learning_rate": 6.912770428915703e-06,
+      "loss": 0.1702,
+      "step": 2104
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 6.64470911026001,
+      "learning_rate": 6.910183122502236e-06,
+      "loss": 0.1058,
+      "step": 2105
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 13.578435897827148,
+      "learning_rate": 6.907595217058159e-06,
+      "loss": 0.1711,
+      "step": 2106
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 5.5899786949157715,
+      "learning_rate": 6.905006713395036e-06,
+      "loss": 0.1602,
+      "step": 2107
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 18.449058532714844,
+      "learning_rate": 6.902417612324615e-06,
+      "loss": 0.114,
+      "step": 2108
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 9.409819602966309,
+      "learning_rate": 6.89982791465884e-06,
+      "loss": 0.0542,
+      "step": 2109
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 3.935387134552002,
+      "learning_rate": 6.897237621209831e-06,
+      "loss": 0.027,
+      "step": 2110
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 11.534008979797363,
+      "learning_rate": 6.894646732789901e-06,
+      "loss": 0.1456,
+      "step": 2111
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 11.699277877807617,
+      "learning_rate": 6.892055250211552e-06,
+      "loss": 0.208,
+      "step": 2112
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 8.532713890075684,
+      "learning_rate": 6.889463174287467e-06,
+      "loss": 0.1014,
+      "step": 2113
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 11.294116020202637,
+      "learning_rate": 6.886870505830516e-06,
+      "loss": 0.0846,
+      "step": 2114
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 9.278223037719727,
+      "learning_rate": 6.884277245653758e-06,
+      "loss": 0.1243,
+      "step": 2115
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 6.128586292266846,
+      "learning_rate": 6.8816833945704346e-06,
+      "loss": 0.0653,
+      "step": 2116
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 5.258224010467529,
+      "learning_rate": 6.879088953393973e-06,
+      "loss": 0.132,
+      "step": 2117
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 13.543342590332031,
+      "learning_rate": 6.876493922937988e-06,
+      "loss": 0.324,
+      "step": 2118
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 8.353715896606445,
+      "learning_rate": 6.873898304016277e-06,
+      "loss": 0.0804,
+      "step": 2119
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 10.511133193969727,
+      "learning_rate": 6.87130209744282e-06,
+      "loss": 0.0985,
+      "step": 2120
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 6.096007347106934,
+      "learning_rate": 6.868705304031786e-06,
+      "loss": 0.0757,
+      "step": 2121
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 6.6087141036987305,
+      "learning_rate": 6.866107924597525e-06,
+      "loss": 0.0933,
+      "step": 2122
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 3.3812384605407715,
+      "learning_rate": 6.86350995995457e-06,
+      "loss": 0.0389,
+      "step": 2123
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 7.996529579162598,
+      "learning_rate": 6.860911410917642e-06,
+      "loss": 0.0832,
+      "step": 2124
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 10.237898826599121,
+      "learning_rate": 6.858312278301638e-06,
+      "loss": 0.1835,
+      "step": 2125
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 8.804056167602539,
+      "learning_rate": 6.855712562921646e-06,
+      "loss": 0.1185,
+      "step": 2126
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 10.697150230407715,
+      "learning_rate": 6.853112265592929e-06,
+      "loss": 0.1151,
+      "step": 2127
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 10.155056953430176,
+      "learning_rate": 6.85051138713094e-06,
+      "loss": 0.2764,
+      "step": 2128
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 6.843580722808838,
+      "learning_rate": 6.847909928351309e-06,
+      "loss": 0.0958,
+      "step": 2129
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 8.61191177368164,
+      "learning_rate": 6.845307890069851e-06,
+      "loss": 0.0834,
+      "step": 2130
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 11.922743797302246,
+      "learning_rate": 6.842705273102558e-06,
+      "loss": 0.1861,
+      "step": 2131
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 10.70093822479248,
+      "learning_rate": 6.840102078265611e-06,
+      "loss": 0.1716,
+      "step": 2132
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 12.016267776489258,
+      "learning_rate": 6.837498306375365e-06,
+      "loss": 0.0973,
+      "step": 2133
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 5.729307174682617,
+      "learning_rate": 6.834893958248361e-06,
+      "loss": 0.0994,
+      "step": 2134
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 8.294934272766113,
+      "learning_rate": 6.832289034701318e-06,
+      "loss": 0.0779,
+      "step": 2135
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 11.573381423950195,
+      "learning_rate": 6.829683536551135e-06,
+      "loss": 0.1485,
+      "step": 2136
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 12.28051471710205,
+      "learning_rate": 6.827077464614894e-06,
+      "loss": 0.3189,
+      "step": 2137
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 6.619340419769287,
+      "learning_rate": 6.824470819709855e-06,
+      "loss": 0.1076,
+      "step": 2138
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 16.39983558654785,
+      "learning_rate": 6.821863602653457e-06,
+      "loss": 0.21,
+      "step": 2139
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 5.833128929138184,
+      "learning_rate": 6.8192558142633215e-06,
+      "loss": 0.0547,
+      "step": 2140
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 4.53016996383667,
+      "learning_rate": 6.816647455357245e-06,
+      "loss": 0.0236,
+      "step": 2141
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 15.005500793457031,
+      "learning_rate": 6.814038526753205e-06,
+      "loss": 0.1959,
+      "step": 2142
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 5.906820774078369,
+      "learning_rate": 6.81142902926936e-06,
+      "loss": 0.0607,
+      "step": 2143
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 6.030914306640625,
+      "learning_rate": 6.808818963724042e-06,
+      "loss": 0.0411,
+      "step": 2144
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 4.453580379486084,
+      "learning_rate": 6.806208330935766e-06,
+      "loss": 0.0407,
+      "step": 2145
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 7.230526924133301,
+      "learning_rate": 6.803597131723221e-06,
+      "loss": 0.2018,
+      "step": 2146
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 13.467733383178711,
+      "learning_rate": 6.8009853669052735e-06,
+      "loss": 0.1317,
+      "step": 2147
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 12.02500057220459,
+      "learning_rate": 6.798373037300972e-06,
+      "loss": 0.253,
+      "step": 2148
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 5.369839191436768,
+      "learning_rate": 6.7957601437295395e-06,
+      "loss": 0.1049,
+      "step": 2149
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 9.815206527709961,
+      "learning_rate": 6.7931466870103735e-06,
+      "loss": 0.2355,
+      "step": 2150
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 9.845855712890625,
+      "learning_rate": 6.790532667963052e-06,
+      "loss": 0.1269,
+      "step": 2151
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 6.96058464050293,
+      "learning_rate": 6.787918087407326e-06,
+      "loss": 0.218,
+      "step": 2152
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 12.927353858947754,
+      "learning_rate": 6.785302946163124e-06,
+      "loss": 0.1539,
+      "step": 2153
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 6.627811431884766,
+      "learning_rate": 6.782687245050552e-06,
+      "loss": 0.207,
+      "step": 2154
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 15.487345695495605,
+      "learning_rate": 6.78007098488989e-06,
+      "loss": 0.166,
+      "step": 2155
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 10.391948699951172,
+      "learning_rate": 6.77745416650159e-06,
+      "loss": 0.2767,
+      "step": 2156
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 7.349318027496338,
+      "learning_rate": 6.774836790706287e-06,
+      "loss": 0.072,
+      "step": 2157
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 13.063915252685547,
+      "learning_rate": 6.7722188583247835e-06,
+      "loss": 0.156,
+      "step": 2158
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 8.631080627441406,
+      "learning_rate": 6.76960037017806e-06,
+      "loss": 0.0932,
+      "step": 2159
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 8.894039154052734,
+      "learning_rate": 6.766981327087271e-06,
+      "loss": 0.1145,
+      "step": 2160
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 4.650251865386963,
+      "learning_rate": 6.764361729873742e-06,
+      "loss": 0.0723,
+      "step": 2161
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 6.901477813720703,
+      "learning_rate": 6.761741579358978e-06,
+      "loss": 0.1136,
+      "step": 2162
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 9.112944602966309,
+      "learning_rate": 6.759120876364653e-06,
+      "loss": 0.072,
+      "step": 2163
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 5.141025543212891,
+      "learning_rate": 6.7564996217126146e-06,
+      "loss": 0.089,
+      "step": 2164
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 11.851815223693848,
+      "learning_rate": 6.753877816224886e-06,
+      "loss": 0.2308,
+      "step": 2165
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 5.979508399963379,
+      "learning_rate": 6.751255460723662e-06,
+      "loss": 0.0438,
+      "step": 2166
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 7.343383312225342,
+      "learning_rate": 6.748632556031306e-06,
+      "loss": 0.1168,
+      "step": 2167
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 9.727837562561035,
+      "learning_rate": 6.7460091029703614e-06,
+      "loss": 0.0753,
+      "step": 2168
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 9.343822479248047,
+      "learning_rate": 6.7433851023635364e-06,
+      "loss": 0.1404,
+      "step": 2169
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 8.908103942871094,
+      "learning_rate": 6.740760555033715e-06,
+      "loss": 0.0515,
+      "step": 2170
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 15.48489761352539,
+      "learning_rate": 6.738135461803952e-06,
+      "loss": 0.1807,
+      "step": 2171
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 5.083133220672607,
+      "learning_rate": 6.735509823497469e-06,
+      "loss": 0.0327,
+      "step": 2172
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 7.682764053344727,
+      "learning_rate": 6.732883640937667e-06,
+      "loss": 0.0732,
+      "step": 2173
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 17.20330238342285,
+      "learning_rate": 6.73025691494811e-06,
+      "loss": 0.1389,
+      "step": 2174
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 7.45797061920166,
+      "learning_rate": 6.727629646352536e-06,
+      "loss": 0.0794,
+      "step": 2175
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 6.482460021972656,
+      "learning_rate": 6.725001835974854e-06,
+      "loss": 0.1234,
+      "step": 2176
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 8.01945972442627,
+      "learning_rate": 6.72237348463914e-06,
+      "loss": 0.0639,
+      "step": 2177
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 63.530799865722656,
+      "learning_rate": 6.719744593169642e-06,
+      "loss": 0.2577,
+      "step": 2178
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 6.339193344116211,
+      "learning_rate": 6.717115162390775e-06,
+      "loss": 0.1274,
+      "step": 2179
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 9.836102485656738,
+      "learning_rate": 6.714485193127126e-06,
+      "loss": 0.1905,
+      "step": 2180
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 8.61129379272461,
+      "learning_rate": 6.711854686203448e-06,
+      "loss": 0.1833,
+      "step": 2181
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 8.83160400390625,
+      "learning_rate": 6.709223642444667e-06,
+      "loss": 0.1226,
+      "step": 2182
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 6.071281433105469,
+      "learning_rate": 6.706592062675872e-06,
+      "loss": 0.0894,
+      "step": 2183
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 9.98853874206543,
+      "learning_rate": 6.703959947722323e-06,
+      "loss": 0.1124,
+      "step": 2184
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 5.988928318023682,
+      "learning_rate": 6.701327298409448e-06,
+      "loss": 0.0454,
+      "step": 2185
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 17.088123321533203,
+      "learning_rate": 6.69869411556284e-06,
+      "loss": 0.1895,
+      "step": 2186
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 8.75080394744873,
+      "learning_rate": 6.696060400008263e-06,
+      "loss": 0.1079,
+      "step": 2187
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 13.845464706420898,
+      "learning_rate": 6.6934261525716474e-06,
+      "loss": 0.2254,
+      "step": 2188
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 7.18847131729126,
+      "learning_rate": 6.690791374079086e-06,
+      "loss": 0.2187,
+      "step": 2189
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 10.83674144744873,
+      "learning_rate": 6.688156065356845e-06,
+      "loss": 0.1498,
+      "step": 2190
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 18.808818817138672,
+      "learning_rate": 6.6855202272313505e-06,
+      "loss": 0.2152,
+      "step": 2191
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 4.712621688842773,
+      "learning_rate": 6.682883860529199e-06,
+      "loss": 0.0363,
+      "step": 2192
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 5.894738674163818,
+      "learning_rate": 6.680246966077151e-06,
+      "loss": 0.0812,
+      "step": 2193
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 13.597196578979492,
+      "learning_rate": 6.677609544702133e-06,
+      "loss": 0.0894,
+      "step": 2194
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 5.739461421966553,
+      "learning_rate": 6.674971597231236e-06,
+      "loss": 0.0776,
+      "step": 2195
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 9.959057807922363,
+      "learning_rate": 6.672333124491717e-06,
+      "loss": 0.1191,
+      "step": 2196
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 5.5744242668151855,
+      "learning_rate": 6.6696941273109935e-06,
+      "loss": 0.0655,
+      "step": 2197
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 18.685091018676758,
+      "learning_rate": 6.667054606516658e-06,
+      "loss": 0.2534,
+      "step": 2198
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 11.992645263671875,
+      "learning_rate": 6.664414562936454e-06,
+      "loss": 0.2303,
+      "step": 2199
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 13.216694831848145,
+      "learning_rate": 6.6617739973982985e-06,
+      "loss": 0.1965,
+      "step": 2200
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 17.05311393737793,
+      "learning_rate": 6.659132910730269e-06,
+      "loss": 0.1887,
+      "step": 2201
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 14.131758689880371,
+      "learning_rate": 6.656491303760606e-06,
+      "loss": 0.2303,
+      "step": 2202
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 15.661176681518555,
+      "learning_rate": 6.653849177317712e-06,
+      "loss": 0.0586,
+      "step": 2203
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 4.0805888175964355,
+      "learning_rate": 6.651206532230155e-06,
+      "loss": 0.0596,
+      "step": 2204
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 8.620298385620117,
+      "learning_rate": 6.648563369326666e-06,
+      "loss": 0.0682,
+      "step": 2205
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 8.31601619720459,
+      "learning_rate": 6.645919689436135e-06,
+      "loss": 0.1108,
+      "step": 2206
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 8.118610382080078,
+      "learning_rate": 6.643275493387618e-06,
+      "loss": 0.2258,
+      "step": 2207
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 17.5425968170166,
+      "learning_rate": 6.640630782010329e-06,
+      "loss": 0.3236,
+      "step": 2208
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 11.357842445373535,
+      "learning_rate": 6.637985556133646e-06,
+      "loss": 0.1683,
+      "step": 2209
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 12.794878005981445,
+      "learning_rate": 6.635339816587109e-06,
+      "loss": 0.1252,
+      "step": 2210
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 4.031367301940918,
+      "learning_rate": 6.6326935642004165e-06,
+      "loss": 0.0395,
+      "step": 2211
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 14.40556526184082,
+      "learning_rate": 6.630046799803431e-06,
+      "loss": 0.3403,
+      "step": 2212
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 6.6147541999816895,
+      "learning_rate": 6.6273995242261746e-06,
+      "loss": 0.0568,
+      "step": 2213
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 5.34834098815918,
+      "learning_rate": 6.6247517382988245e-06,
+      "loss": 0.0781,
+      "step": 2214
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 18.36115074157715,
+      "learning_rate": 6.622103442851728e-06,
+      "loss": 0.1672,
+      "step": 2215
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 6.102049350738525,
+      "learning_rate": 6.619454638715384e-06,
+      "loss": 0.1017,
+      "step": 2216
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 9.813507080078125,
+      "learning_rate": 6.616805326720452e-06,
+      "loss": 0.0788,
+      "step": 2217
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 6.9955363273620605,
+      "learning_rate": 6.614155507697756e-06,
+      "loss": 0.0556,
+      "step": 2218
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 11.743613243103027,
+      "learning_rate": 6.611505182478273e-06,
+      "loss": 0.1827,
+      "step": 2219
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 9.29969596862793,
+      "learning_rate": 6.60885435189314e-06,
+      "loss": 0.1429,
+      "step": 2220
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 8.864079475402832,
+      "learning_rate": 6.606203016773658e-06,
+      "loss": 0.0929,
+      "step": 2221
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 11.352057456970215,
+      "learning_rate": 6.6035511779512764e-06,
+      "loss": 0.1803,
+      "step": 2222
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 14.892663955688477,
+      "learning_rate": 6.600898836257612e-06,
+      "loss": 0.1394,
+      "step": 2223
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 16.513500213623047,
+      "learning_rate": 6.5982459925244326e-06,
+      "loss": 0.2821,
+      "step": 2224
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 5.189689636230469,
+      "learning_rate": 6.595592647583666e-06,
+      "loss": 0.0549,
+      "step": 2225
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 13.391901969909668,
+      "learning_rate": 6.5929388022674e-06,
+      "loss": 0.0844,
+      "step": 2226
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 10.096096992492676,
+      "learning_rate": 6.590284457407876e-06,
+      "loss": 0.2227,
+      "step": 2227
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 8.945037841796875,
+      "learning_rate": 6.587629613837489e-06,
+      "loss": 0.0717,
+      "step": 2228
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 7.82534646987915,
+      "learning_rate": 6.584974272388798e-06,
+      "loss": 0.0602,
+      "step": 2229
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 8.225728988647461,
+      "learning_rate": 6.582318433894513e-06,
+      "loss": 0.2132,
+      "step": 2230
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 5.012326717376709,
+      "learning_rate": 6.579662099187501e-06,
+      "loss": 0.0523,
+      "step": 2231
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 5.504785537719727,
+      "learning_rate": 6.577005269100784e-06,
+      "loss": 0.1074,
+      "step": 2232
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 13.99877643585205,
+      "learning_rate": 6.57434794446754e-06,
+      "loss": 0.0694,
+      "step": 2233
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 7.2316460609436035,
+      "learning_rate": 6.571690126121103e-06,
+      "loss": 0.1776,
+      "step": 2234
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 5.969881057739258,
+      "learning_rate": 6.569031814894962e-06,
+      "loss": 0.0395,
+      "step": 2235
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 10.323885917663574,
+      "learning_rate": 6.5663730116227555e-06,
+      "loss": 0.1181,
+      "step": 2236
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 8.310761451721191,
+      "learning_rate": 6.563713717138283e-06,
+      "loss": 0.0942,
+      "step": 2237
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 10.670938491821289,
+      "learning_rate": 6.5610539322754965e-06,
+      "loss": 0.1623,
+      "step": 2238
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 8.428224563598633,
+      "learning_rate": 6.558393657868497e-06,
+      "loss": 0.1727,
+      "step": 2239
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 7.2235283851623535,
+      "learning_rate": 6.555732894751548e-06,
+      "loss": 0.0913,
+      "step": 2240
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 9.543944358825684,
+      "learning_rate": 6.553071643759055e-06,
+      "loss": 0.1655,
+      "step": 2241
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 17.21744728088379,
+      "learning_rate": 6.550409905725585e-06,
+      "loss": 0.181,
+      "step": 2242
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 12.771845817565918,
+      "learning_rate": 6.547747681485857e-06,
+      "loss": 0.1421,
+      "step": 2243
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 12.18629264831543,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.2305,
+      "step": 2244
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 11.492579460144043,
+      "learning_rate": 6.5424217777272506e-06,
+      "loss": 0.0538,
+      "step": 2245
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 14.022279739379883,
+      "learning_rate": 6.539758099878569e-06,
+      "loss": 0.2763,
+      "step": 2246
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 12.583410263061523,
+      "learning_rate": 6.537093939164017e-06,
+      "loss": 0.0988,
+      "step": 2247
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.8621236133122029,
+      "eval_loss": 0.5062664151191711,
+      "eval_runtime": 11.2234,
+      "eval_samples_per_second": 168.665,
+      "eval_steps_per_second": 21.117,
+      "step": 2247
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 5.805846214294434,
+      "learning_rate": 6.534429296419074e-06,
+      "loss": 0.0594,
+      "step": 2248
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 8.747864723205566,
+      "learning_rate": 6.5317641724793655e-06,
+      "loss": 0.071,
+      "step": 2249
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 5.278066158294678,
+      "learning_rate": 6.529098568180672e-06,
+      "loss": 0.0717,
+      "step": 2250
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 9.778899192810059,
+      "learning_rate": 6.526432484358923e-06,
+      "loss": 0.1551,
+      "step": 2251
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 3.976569652557373,
+      "learning_rate": 6.523765921850198e-06,
+      "loss": 0.0324,
+      "step": 2252
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 4.234729766845703,
+      "learning_rate": 6.5210988814907275e-06,
+      "loss": 0.074,
+      "step": 2253
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 3.073512077331543,
+      "learning_rate": 6.5184313641168905e-06,
+      "loss": 0.0224,
+      "step": 2254
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 11.034455299377441,
+      "learning_rate": 6.515763370565218e-06,
+      "loss": 0.0446,
+      "step": 2255
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 4.331269264221191,
+      "learning_rate": 6.513094901672388e-06,
+      "loss": 0.0339,
+      "step": 2256
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 4.928471565246582,
+      "learning_rate": 6.510425958275228e-06,
+      "loss": 0.0674,
+      "step": 2257
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 1.2260003089904785,
+      "learning_rate": 6.507756541210716e-06,
+      "loss": 0.0218,
+      "step": 2258
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 11.655609130859375,
+      "learning_rate": 6.505086651315975e-06,
+      "loss": 0.1162,
+      "step": 2259
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 9.695042610168457,
+      "learning_rate": 6.502416289428282e-06,
+      "loss": 0.1634,
+      "step": 2260
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 6.469884395599365,
+      "learning_rate": 6.499745456385054e-06,
+      "loss": 0.0439,
+      "step": 2261
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 2.3879811763763428,
+      "learning_rate": 6.497074153023863e-06,
+      "loss": 0.0252,
+      "step": 2262
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 8.295581817626953,
+      "learning_rate": 6.494402380182428e-06,
+      "loss": 0.044,
+      "step": 2263
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 1.9467341899871826,
+      "learning_rate": 6.491730138698609e-06,
+      "loss": 0.0225,
+      "step": 2264
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 5.426387310028076,
+      "learning_rate": 6.489057429410418e-06,
+      "loss": 0.0292,
+      "step": 2265
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 8.834466934204102,
+      "learning_rate": 6.486384253156014e-06,
+      "loss": 0.1133,
+      "step": 2266
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 2.5623581409454346,
+      "learning_rate": 6.4837106107737e-06,
+      "loss": 0.028,
+      "step": 2267
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 8.177447319030762,
+      "learning_rate": 6.481036503101927e-06,
+      "loss": 0.0938,
+      "step": 2268
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 4.6704583168029785,
+      "learning_rate": 6.4783619309792915e-06,
+      "loss": 0.0502,
+      "step": 2269
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 1.0883411169052124,
+      "learning_rate": 6.475686895244534e-06,
+      "loss": 0.0142,
+      "step": 2270
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 2.778691053390503,
+      "learning_rate": 6.473011396736543e-06,
+      "loss": 0.0237,
+      "step": 2271
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.9980936646461487,
+      "learning_rate": 6.470335436294351e-06,
+      "loss": 0.0163,
+      "step": 2272
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 8.064085006713867,
+      "learning_rate": 6.467659014757134e-06,
+      "loss": 0.036,
+      "step": 2273
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 5.327763557434082,
+      "learning_rate": 6.464982132964215e-06,
+      "loss": 0.1641,
+      "step": 2274
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 6.040185928344727,
+      "learning_rate": 6.462304791755059e-06,
+      "loss": 0.0918,
+      "step": 2275
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 14.693153381347656,
+      "learning_rate": 6.459626991969278e-06,
+      "loss": 0.1294,
+      "step": 2276
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 5.633932590484619,
+      "learning_rate": 6.456948734446624e-06,
+      "loss": 0.0549,
+      "step": 2277
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 3.8117315769195557,
+      "learning_rate": 6.454270020026996e-06,
+      "loss": 0.0273,
+      "step": 2278
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 9.114907264709473,
+      "learning_rate": 6.4515908495504346e-06,
+      "loss": 0.0853,
+      "step": 2279
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 11.79755687713623,
+      "learning_rate": 6.448911223857124e-06,
+      "loss": 0.1652,
+      "step": 2280
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 11.660473823547363,
+      "learning_rate": 6.446231143787389e-06,
+      "loss": 0.1758,
+      "step": 2281
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 2.459082841873169,
+      "learning_rate": 6.443550610181703e-06,
+      "loss": 0.0268,
+      "step": 2282
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 8.10896110534668,
+      "learning_rate": 6.440869623880674e-06,
+      "loss": 0.0487,
+      "step": 2283
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 3.1131513118743896,
+      "learning_rate": 6.438188185725055e-06,
+      "loss": 0.0541,
+      "step": 2284
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 5.365321636199951,
+      "learning_rate": 6.435506296555742e-06,
+      "loss": 0.0899,
+      "step": 2285
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 7.962471008300781,
+      "learning_rate": 6.432823957213772e-06,
+      "loss": 0.0835,
+      "step": 2286
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 3.1886181831359863,
+      "learning_rate": 6.430141168540322e-06,
+      "loss": 0.0258,
+      "step": 2287
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 9.396906852722168,
+      "learning_rate": 6.427457931376712e-06,
+      "loss": 0.074,
+      "step": 2288
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 4.025094032287598,
+      "learning_rate": 6.424774246564398e-06,
+      "loss": 0.0335,
+      "step": 2289
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 10.876994132995605,
+      "learning_rate": 6.422090114944982e-06,
+      "loss": 0.1373,
+      "step": 2290
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 6.628119945526123,
+      "learning_rate": 6.419405537360205e-06,
+      "loss": 0.0359,
+      "step": 2291
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 2.938720703125,
+      "learning_rate": 6.416720514651943e-06,
+      "loss": 0.0228,
+      "step": 2292
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 10.269327163696289,
+      "learning_rate": 6.4140350476622185e-06,
+      "loss": 0.0278,
+      "step": 2293
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 1.0717568397521973,
+      "learning_rate": 6.411349137233187e-06,
+      "loss": 0.0097,
+      "step": 2294
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 3.759758710861206,
+      "learning_rate": 6.408662784207149e-06,
+      "loss": 0.0235,
+      "step": 2295
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 2.3062937259674072,
+      "learning_rate": 6.405975989426541e-06,
+      "loss": 0.0207,
+      "step": 2296
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 6.465508937835693,
+      "learning_rate": 6.403288753733932e-06,
+      "loss": 0.0379,
+      "step": 2297
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 4.000199794769287,
+      "learning_rate": 6.400601077972043e-06,
+      "loss": 0.0341,
+      "step": 2298
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 9.041462898254395,
+      "learning_rate": 6.39791296298372e-06,
+      "loss": 0.1399,
+      "step": 2299
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 11.293344497680664,
+      "learning_rate": 6.3952244096119535e-06,
+      "loss": 0.0983,
+      "step": 2300
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 1.1056654453277588,
+      "learning_rate": 6.39253541869987e-06,
+      "loss": 0.0159,
+      "step": 2301
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 7.504707336425781,
+      "learning_rate": 6.389845991090734e-06,
+      "loss": 0.0719,
+      "step": 2302
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 6.796584606170654,
+      "learning_rate": 6.387156127627944e-06,
+      "loss": 0.0524,
+      "step": 2303
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 1.6993805170059204,
+      "learning_rate": 6.3844658291550384e-06,
+      "loss": 0.0204,
+      "step": 2304
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 9.506914138793945,
+      "learning_rate": 6.381775096515692e-06,
+      "loss": 0.0716,
+      "step": 2305
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 1.6772551536560059,
+      "learning_rate": 6.379083930553711e-06,
+      "loss": 0.02,
+      "step": 2306
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 8.696874618530273,
+      "learning_rate": 6.376392332113045e-06,
+      "loss": 0.1266,
+      "step": 2307
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 1.1623690128326416,
+      "learning_rate": 6.373700302037773e-06,
+      "loss": 0.0144,
+      "step": 2308
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 10.718400955200195,
+      "learning_rate": 6.371007841172113e-06,
+      "loss": 0.1548,
+      "step": 2309
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 12.784758567810059,
+      "learning_rate": 6.368314950360416e-06,
+      "loss": 0.2081,
+      "step": 2310
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 2.1475980281829834,
+      "learning_rate": 6.365621630447168e-06,
+      "loss": 0.0238,
+      "step": 2311
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 24.984399795532227,
+      "learning_rate": 6.362927882276991e-06,
+      "loss": 0.3198,
+      "step": 2312
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 6.751765727996826,
+      "learning_rate": 6.36023370669464e-06,
+      "loss": 0.0462,
+      "step": 2313
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 6.90775203704834,
+      "learning_rate": 6.357539104545003e-06,
+      "loss": 0.0316,
+      "step": 2314
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 8.1048583984375,
+      "learning_rate": 6.354844076673108e-06,
+      "loss": 0.0681,
+      "step": 2315
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 2.648008346557617,
+      "learning_rate": 6.352148623924108e-06,
+      "loss": 0.0154,
+      "step": 2316
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 4.5985846519470215,
+      "learning_rate": 6.349452747143292e-06,
+      "loss": 0.0274,
+      "step": 2317
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 6.927505016326904,
+      "learning_rate": 6.346756447176086e-06,
+      "loss": 0.0592,
+      "step": 2318
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 6.86525821685791,
+      "learning_rate": 6.344059724868046e-06,
+      "loss": 0.0671,
+      "step": 2319
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 8.962815284729004,
+      "learning_rate": 6.341362581064856e-06,
+      "loss": 0.0263,
+      "step": 2320
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 1.1991780996322632,
+      "learning_rate": 6.3386650166123406e-06,
+      "loss": 0.0103,
+      "step": 2321
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 7.586130142211914,
+      "learning_rate": 6.335967032356449e-06,
+      "loss": 0.0298,
+      "step": 2322
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 9.278162002563477,
+      "learning_rate": 6.333268629143268e-06,
+      "loss": 0.1006,
+      "step": 2323
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 9.720531463623047,
+      "learning_rate": 6.33056980781901e-06,
+      "loss": 0.0779,
+      "step": 2324
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 9.90295124053955,
+      "learning_rate": 6.327870569230022e-06,
+      "loss": 0.2169,
+      "step": 2325
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 14.937568664550781,
+      "learning_rate": 6.325170914222784e-06,
+      "loss": 0.1599,
+      "step": 2326
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 8.169679641723633,
+      "learning_rate": 6.322470843643901e-06,
+      "loss": 0.1831,
+      "step": 2327
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 4.50398063659668,
+      "learning_rate": 6.319770358340111e-06,
+      "loss": 0.0604,
+      "step": 2328
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 16.808855056762695,
+      "learning_rate": 6.317069459158284e-06,
+      "loss": 0.0723,
+      "step": 2329
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 8.388458251953125,
+      "learning_rate": 6.314368146945418e-06,
+      "loss": 0.1395,
+      "step": 2330
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 8.1686429977417,
+      "learning_rate": 6.311666422548637e-06,
+      "loss": 0.0772,
+      "step": 2331
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 3.305044651031494,
+      "learning_rate": 6.308964286815203e-06,
+      "loss": 0.0223,
+      "step": 2332
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 21.832189559936523,
+      "learning_rate": 6.306261740592498e-06,
+      "loss": 0.0834,
+      "step": 2333
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 7.419219493865967,
+      "learning_rate": 6.303558784728036e-06,
+      "loss": 0.0536,
+      "step": 2334
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 8.077475547790527,
+      "learning_rate": 6.300855420069465e-06,
+      "loss": 0.0534,
+      "step": 2335
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 4.798830032348633,
+      "learning_rate": 6.29815164746455e-06,
+      "loss": 0.0686,
+      "step": 2336
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 10.13785171508789,
+      "learning_rate": 6.295447467761196e-06,
+      "loss": 0.1136,
+      "step": 2337
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 8.739697456359863,
+      "learning_rate": 6.292742881807424e-06,
+      "loss": 0.1775,
+      "step": 2338
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 16.619060516357422,
+      "learning_rate": 6.290037890451391e-06,
+      "loss": 0.1146,
+      "step": 2339
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 3.2100861072540283,
+      "learning_rate": 6.28733249454138e-06,
+      "loss": 0.0192,
+      "step": 2340
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 6.654782295227051,
+      "learning_rate": 6.284626694925796e-06,
+      "loss": 0.0745,
+      "step": 2341
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 11.793569564819336,
+      "learning_rate": 6.2819204924531755e-06,
+      "loss": 0.1033,
+      "step": 2342
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 17.97243309020996,
+      "learning_rate": 6.279213887972179e-06,
+      "loss": 0.1951,
+      "step": 2343
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 12.016203880310059,
+      "learning_rate": 6.276506882331594e-06,
+      "loss": 0.1467,
+      "step": 2344
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 8.294089317321777,
+      "learning_rate": 6.273799476380332e-06,
+      "loss": 0.0539,
+      "step": 2345
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 8.662949562072754,
+      "learning_rate": 6.271091670967437e-06,
+      "loss": 0.0467,
+      "step": 2346
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 14.9105224609375,
+      "learning_rate": 6.268383466942064e-06,
+      "loss": 0.0807,
+      "step": 2347
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 2.4065921306610107,
+      "learning_rate": 6.265674865153508e-06,
+      "loss": 0.012,
+      "step": 2348
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 17.48012351989746,
+      "learning_rate": 6.262965866451183e-06,
+      "loss": 0.2108,
+      "step": 2349
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 10.257842063903809,
+      "learning_rate": 6.260256471684622e-06,
+      "loss": 0.0975,
+      "step": 2350
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 3.991021156311035,
+      "learning_rate": 6.257546681703492e-06,
+      "loss": 0.0277,
+      "step": 2351
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 13.818306922912598,
+      "learning_rate": 6.254836497357577e-06,
+      "loss": 0.0378,
+      "step": 2352
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 7.711850166320801,
+      "learning_rate": 6.252125919496785e-06,
+      "loss": 0.1059,
+      "step": 2353
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 6.257598400115967,
+      "learning_rate": 6.249414948971154e-06,
+      "loss": 0.0686,
+      "step": 2354
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 7.449881553649902,
+      "learning_rate": 6.246703586630838e-06,
+      "loss": 0.0741,
+      "step": 2355
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 6.555480480194092,
+      "learning_rate": 6.243991833326114e-06,
+      "loss": 0.0401,
+      "step": 2356
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 6.086026668548584,
+      "learning_rate": 6.241279689907387e-06,
+      "loss": 0.057,
+      "step": 2357
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 5.498207092285156,
+      "learning_rate": 6.23856715722518e-06,
+      "loss": 0.0417,
+      "step": 2358
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 11.321758270263672,
+      "learning_rate": 6.2358542361301385e-06,
+      "loss": 0.0487,
+      "step": 2359
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 6.801464080810547,
+      "learning_rate": 6.233140927473033e-06,
+      "loss": 0.0252,
+      "step": 2360
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 8.722494125366211,
+      "learning_rate": 6.2304272321047485e-06,
+      "loss": 0.0971,
+      "step": 2361
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 3.894118547439575,
+      "learning_rate": 6.2277131508763015e-06,
+      "loss": 0.0223,
+      "step": 2362
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 9.350339889526367,
+      "learning_rate": 6.22499868463882e-06,
+      "loss": 0.0833,
+      "step": 2363
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 9.553706169128418,
+      "learning_rate": 6.222283834243557e-06,
+      "loss": 0.097,
+      "step": 2364
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 16.266565322875977,
+      "learning_rate": 6.219568600541886e-06,
+      "loss": 0.1809,
+      "step": 2365
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 7.050131320953369,
+      "learning_rate": 6.216852984385302e-06,
+      "loss": 0.0427,
+      "step": 2366
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 9.085468292236328,
+      "learning_rate": 6.214136986625416e-06,
+      "loss": 0.0917,
+      "step": 2367
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 4.1881256103515625,
+      "learning_rate": 6.211420608113961e-06,
+      "loss": 0.0847,
+      "step": 2368
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 7.193945407867432,
+      "learning_rate": 6.208703849702792e-06,
+      "loss": 0.0436,
+      "step": 2369
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 30.422250747680664,
+      "learning_rate": 6.205986712243876e-06,
+      "loss": 0.1187,
+      "step": 2370
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 6.978949546813965,
+      "learning_rate": 6.203269196589306e-06,
+      "loss": 0.0315,
+      "step": 2371
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 7.890665054321289,
+      "learning_rate": 6.200551303591294e-06,
+      "loss": 0.0295,
+      "step": 2372
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 9.643845558166504,
+      "learning_rate": 6.19783303410216e-06,
+      "loss": 0.0489,
+      "step": 2373
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 5.611057758331299,
+      "learning_rate": 6.195114388974355e-06,
+      "loss": 0.1456,
+      "step": 2374
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 3.6160762310028076,
+      "learning_rate": 6.192395369060439e-06,
+      "loss": 0.0183,
+      "step": 2375
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 12.080123901367188,
+      "learning_rate": 6.189675975213094e-06,
+      "loss": 0.0909,
+      "step": 2376
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 4.015498638153076,
+      "learning_rate": 6.1869562082851195e-06,
+      "loss": 0.0257,
+      "step": 2377
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 34.829627990722656,
+      "learning_rate": 6.184236069129427e-06,
+      "loss": 0.1424,
+      "step": 2378
+    },
+    {
+      "epoch": 4.23,
+      "grad_norm": 1.9089000225067139,
+      "learning_rate": 6.1815155585990496e-06,
+      "loss": 0.0192,
+      "step": 2379
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 20.361480712890625,
+      "learning_rate": 6.178794677547138e-06,
+      "loss": 0.0553,
+      "step": 2380
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 4.754912853240967,
+      "learning_rate": 6.176073426826951e-06,
+      "loss": 0.1068,
+      "step": 2381
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 1.0552390813827515,
+      "learning_rate": 6.173351807291873e-06,
+      "loss": 0.0105,
+      "step": 2382
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 6.130521774291992,
+      "learning_rate": 6.170629819795399e-06,
+      "loss": 0.042,
+      "step": 2383
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 2.9908390045166016,
+      "learning_rate": 6.167907465191141e-06,
+      "loss": 0.0215,
+      "step": 2384
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 10.758929252624512,
+      "learning_rate": 6.165184744332824e-06,
+      "loss": 0.0809,
+      "step": 2385
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 18.373193740844727,
+      "learning_rate": 6.162461658074289e-06,
+      "loss": 0.141,
+      "step": 2386
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 7.260679721832275,
+      "learning_rate": 6.159738207269491e-06,
+      "loss": 0.0483,
+      "step": 2387
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 5.219804763793945,
+      "learning_rate": 6.157014392772504e-06,
+      "loss": 0.1465,
+      "step": 2388
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 3.575394868850708,
+      "learning_rate": 6.154290215437508e-06,
+      "loss": 0.0354,
+      "step": 2389
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 6.238715648651123,
+      "learning_rate": 6.151565676118805e-06,
+      "loss": 0.0342,
+      "step": 2390
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 6.119637966156006,
+      "learning_rate": 6.148840775670803e-06,
+      "loss": 0.031,
+      "step": 2391
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 1.976996898651123,
+      "learning_rate": 6.146115514948028e-06,
+      "loss": 0.0131,
+      "step": 2392
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 6.343568801879883,
+      "learning_rate": 6.143389894805119e-06,
+      "loss": 0.179,
+      "step": 2393
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 13.347746849060059,
+      "learning_rate": 6.140663916096824e-06,
+      "loss": 0.0363,
+      "step": 2394
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 10.63392162322998,
+      "learning_rate": 6.137937579678007e-06,
+      "loss": 0.0516,
+      "step": 2395
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 5.495789527893066,
+      "learning_rate": 6.135210886403644e-06,
+      "loss": 0.0766,
+      "step": 2396
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 10.790604591369629,
+      "learning_rate": 6.132483837128823e-06,
+      "loss": 0.0834,
+      "step": 2397
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 4.208777904510498,
+      "learning_rate": 6.129756432708739e-06,
+      "loss": 0.0366,
+      "step": 2398
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 7.675404071807861,
+      "learning_rate": 6.127028673998705e-06,
+      "loss": 0.065,
+      "step": 2399
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 9.232346534729004,
+      "learning_rate": 6.124300561854139e-06,
+      "loss": 0.0538,
+      "step": 2400
+    },
+    {
+      "epoch": 4.27,
+      "grad_norm": 3.478067398071289,
+      "learning_rate": 6.121572097130577e-06,
+      "loss": 0.0287,
+      "step": 2401
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 2.8621883392333984,
+      "learning_rate": 6.11884328068366e-06,
+      "loss": 0.0209,
+      "step": 2402
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 17.33451271057129,
+      "learning_rate": 6.116114113369137e-06,
+      "loss": 0.1905,
+      "step": 2403
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 7.681353569030762,
+      "learning_rate": 6.113384596042877e-06,
+      "loss": 0.0904,
+      "step": 2404
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 8.07400894165039,
+      "learning_rate": 6.1106547295608495e-06,
+      "loss": 0.0493,
+      "step": 2405
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 16.362756729125977,
+      "learning_rate": 6.107924514779136e-06,
+      "loss": 0.1114,
+      "step": 2406
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 9.001738548278809,
+      "learning_rate": 6.1051939525539296e-06,
+      "loss": 0.0603,
+      "step": 2407
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 5.426167964935303,
+      "learning_rate": 6.1024630437415285e-06,
+      "loss": 0.029,
+      "step": 2408
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 6.244989395141602,
+      "learning_rate": 6.099731789198344e-06,
+      "loss": 0.0902,
+      "step": 2409
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 6.454914569854736,
+      "learning_rate": 6.097000189780893e-06,
+      "loss": 0.0422,
+      "step": 2410
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 4.148303031921387,
+      "learning_rate": 6.094268246345799e-06,
+      "loss": 0.0243,
+      "step": 2411
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 10.426963806152344,
+      "learning_rate": 6.091535959749796e-06,
+      "loss": 0.1333,
+      "step": 2412
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 7.521871566772461,
+      "learning_rate": 6.088803330849727e-06,
+      "loss": 0.0659,
+      "step": 2413
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 11.660032272338867,
+      "learning_rate": 6.08607036050254e-06,
+      "loss": 0.065,
+      "step": 2414
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 7.871161460876465,
+      "learning_rate": 6.0833370495652885e-06,
+      "loss": 0.0412,
+      "step": 2415
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 11.872117042541504,
+      "learning_rate": 6.080603398895137e-06,
+      "loss": 0.1052,
+      "step": 2416
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 4.864875316619873,
+      "learning_rate": 6.077869409349352e-06,
+      "loss": 0.0367,
+      "step": 2417
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 3.5103936195373535,
+      "learning_rate": 6.075135081785311e-06,
+      "loss": 0.0248,
+      "step": 2418
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 13.099506378173828,
+      "learning_rate": 6.072400417060494e-06,
+      "loss": 0.1021,
+      "step": 2419
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 1.618504285812378,
+      "learning_rate": 6.0696654160324875e-06,
+      "loss": 0.0161,
+      "step": 2420
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 4.578799724578857,
+      "learning_rate": 6.066930079558985e-06,
+      "loss": 0.033,
+      "step": 2421
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 7.834613800048828,
+      "learning_rate": 6.064194408497783e-06,
+      "loss": 0.0846,
+      "step": 2422
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 4.570634841918945,
+      "learning_rate": 6.061458403706782e-06,
+      "loss": 0.0329,
+      "step": 2423
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 4.334469795227051,
+      "learning_rate": 6.0587220660439935e-06,
+      "loss": 0.0297,
+      "step": 2424
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 1.9516427516937256,
+      "learning_rate": 6.055985396367526e-06,
+      "loss": 0.0181,
+      "step": 2425
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 3.181553363800049,
+      "learning_rate": 6.053248395535594e-06,
+      "loss": 0.0164,
+      "step": 2426
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 3.2519352436065674,
+      "learning_rate": 6.050511064406521e-06,
+      "loss": 0.0263,
+      "step": 2427
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 10.431090354919434,
+      "learning_rate": 6.047773403838726e-06,
+      "loss": 0.1106,
+      "step": 2428
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 14.224845886230469,
+      "learning_rate": 6.045035414690736e-06,
+      "loss": 0.0652,
+      "step": 2429
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 11.30257511138916,
+      "learning_rate": 6.042297097821184e-06,
+      "loss": 0.06,
+      "step": 2430
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 1.944788932800293,
+      "learning_rate": 6.039558454088796e-06,
+      "loss": 0.0147,
+      "step": 2431
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 0.8385435938835144,
+      "learning_rate": 6.0368194843524135e-06,
+      "loss": 0.0102,
+      "step": 2432
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 7.530121326446533,
+      "learning_rate": 6.034080189470968e-06,
+      "loss": 0.1499,
+      "step": 2433
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 1.0272918939590454,
+      "learning_rate": 6.031340570303501e-06,
+      "loss": 0.011,
+      "step": 2434
+    },
+    {
+      "epoch": 4.33,
+      "grad_norm": 6.437071323394775,
+      "learning_rate": 6.028600627709151e-06,
+      "loss": 0.0374,
+      "step": 2435
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 18.063796997070312,
+      "learning_rate": 6.025860362547162e-06,
+      "loss": 0.1282,
+      "step": 2436
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 4.4657416343688965,
+      "learning_rate": 6.023119775676875e-06,
+      "loss": 0.0308,
+      "step": 2437
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 9.124656677246094,
+      "learning_rate": 6.020378867957738e-06,
+      "loss": 0.1132,
+      "step": 2438
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 1.2656121253967285,
+      "learning_rate": 6.017637640249291e-06,
+      "loss": 0.0115,
+      "step": 2439
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 11.219545364379883,
+      "learning_rate": 6.014896093411181e-06,
+      "loss": 0.1512,
+      "step": 2440
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 4.664861679077148,
+      "learning_rate": 6.012154228303153e-06,
+      "loss": 0.0622,
+      "step": 2441
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 4.531983375549316,
+      "learning_rate": 6.009412045785051e-06,
+      "loss": 0.0224,
+      "step": 2442
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 5.476521968841553,
+      "learning_rate": 6.006669546716821e-06,
+      "loss": 0.0193,
+      "step": 2443
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 10.46080493927002,
+      "learning_rate": 6.0039267319585075e-06,
+      "loss": 0.0818,
+      "step": 2444
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 14.179156303405762,
+      "learning_rate": 6.001183602370249e-06,
+      "loss": 0.0763,
+      "step": 2445
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 9.602608680725098,
+      "learning_rate": 5.998440158812291e-06,
+      "loss": 0.0824,
+      "step": 2446
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 4.210625648498535,
+      "learning_rate": 5.9956964021449726e-06,
+      "loss": 0.0255,
+      "step": 2447
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 7.735289096832275,
+      "learning_rate": 5.9929523332287275e-06,
+      "loss": 0.1101,
+      "step": 2448
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 6.355183124542236,
+      "learning_rate": 5.9902079529240984e-06,
+      "loss": 0.0809,
+      "step": 2449
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 23.363840103149414,
+      "learning_rate": 5.987463262091715e-06,
+      "loss": 0.0637,
+      "step": 2450
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 2.5800158977508545,
+      "learning_rate": 5.984718261592307e-06,
+      "loss": 0.014,
+      "step": 2451
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 15.983451843261719,
+      "learning_rate": 5.981972952286707e-06,
+      "loss": 0.0583,
+      "step": 2452
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 4.907855033874512,
+      "learning_rate": 5.9792273350358354e-06,
+      "loss": 0.0363,
+      "step": 2453
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 20.51796531677246,
+      "learning_rate": 5.976481410700716e-06,
+      "loss": 0.197,
+      "step": 2454
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 13.064391136169434,
+      "learning_rate": 5.973735180142468e-06,
+      "loss": 0.2124,
+      "step": 2455
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 8.090473175048828,
+      "learning_rate": 5.970988644222302e-06,
+      "loss": 0.0968,
+      "step": 2456
+    },
+    {
+      "epoch": 4.37,
+      "grad_norm": 4.694125652313232,
+      "learning_rate": 5.96824180380153e-06,
+      "loss": 0.0525,
+      "step": 2457
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 5.388000011444092,
+      "learning_rate": 5.965494659741556e-06,
+      "loss": 0.0507,
+      "step": 2458
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 14.221235275268555,
+      "learning_rate": 5.962747212903881e-06,
+      "loss": 0.2003,
+      "step": 2459
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 8.771271705627441,
+      "learning_rate": 5.959999464150101e-06,
+      "loss": 0.1077,
+      "step": 2460
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 14.621861457824707,
+      "learning_rate": 5.957251414341905e-06,
+      "loss": 0.0881,
+      "step": 2461
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 3.0287599563598633,
+      "learning_rate": 5.954503064341075e-06,
+      "loss": 0.127,
+      "step": 2462
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 2.421968936920166,
+      "learning_rate": 5.951754415009493e-06,
+      "loss": 0.0198,
+      "step": 2463
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 2.266444683074951,
+      "learning_rate": 5.9490054672091305e-06,
+      "loss": 0.0148,
+      "step": 2464
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 4.0825018882751465,
+      "learning_rate": 5.946256221802052e-06,
+      "loss": 0.0249,
+      "step": 2465
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 1.1941858530044556,
+      "learning_rate": 5.943506679650418e-06,
+      "loss": 0.0093,
+      "step": 2466
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 3.2386770248413086,
+      "learning_rate": 5.94075684161648e-06,
+      "loss": 0.0178,
+      "step": 2467
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 8.822221755981445,
+      "learning_rate": 5.938006708562587e-06,
+      "loss": 0.1043,
+      "step": 2468
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 3.499068260192871,
+      "learning_rate": 5.935256281351171e-06,
+      "loss": 0.0207,
+      "step": 2469
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 10.456811904907227,
+      "learning_rate": 5.932505560844766e-06,
+      "loss": 0.0773,
+      "step": 2470
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 6.2233991622924805,
+      "learning_rate": 5.9297545479059915e-06,
+      "loss": 0.043,
+      "step": 2471
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 2.122048854827881,
+      "learning_rate": 5.927003243397564e-06,
+      "loss": 0.0169,
+      "step": 2472
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 11.568830490112305,
+      "learning_rate": 5.924251648182284e-06,
+      "loss": 0.1826,
+      "step": 2473
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 13.574807167053223,
+      "learning_rate": 5.921499763123052e-06,
+      "loss": 0.1012,
+      "step": 2474
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 9.679203987121582,
+      "learning_rate": 5.918747589082853e-06,
+      "loss": 0.0354,
+      "step": 2475
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 6.873231410980225,
+      "learning_rate": 5.9159951269247636e-06,
+      "loss": 0.0407,
+      "step": 2476
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 9.451889991760254,
+      "learning_rate": 5.9132423775119555e-06,
+      "loss": 0.0297,
+      "step": 2477
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 4.155418395996094,
+      "learning_rate": 5.910489341707684e-06,
+      "loss": 0.0153,
+      "step": 2478
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 18.543643951416016,
+      "learning_rate": 5.907736020375297e-06,
+      "loss": 0.0808,
+      "step": 2479
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 12.689310073852539,
+      "learning_rate": 5.904982414378233e-06,
+      "loss": 0.0318,
+      "step": 2480
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 5.997331142425537,
+      "learning_rate": 5.902228524580018e-06,
+      "loss": 0.0287,
+      "step": 2481
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 12.747438430786133,
+      "learning_rate": 5.89947435184427e-06,
+      "loss": 0.154,
+      "step": 2482
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 1.533725619316101,
+      "learning_rate": 5.8967198970346905e-06,
+      "loss": 0.0162,
+      "step": 2483
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 3.8231265544891357,
+      "learning_rate": 5.893965161015074e-06,
+      "loss": 0.0175,
+      "step": 2484
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 3.9796805381774902,
+      "learning_rate": 5.891210144649303e-06,
+      "loss": 0.0207,
+      "step": 2485
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 10.854273796081543,
+      "learning_rate": 5.888454848801345e-06,
+      "loss": 0.068,
+      "step": 2486
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 8.19113826751709,
+      "learning_rate": 5.885699274335255e-06,
+      "loss": 0.0371,
+      "step": 2487
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 7.531518936157227,
+      "learning_rate": 5.882943422115179e-06,
+      "loss": 0.0289,
+      "step": 2488
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 25.16332244873047,
+      "learning_rate": 5.8801872930053494e-06,
+      "loss": 0.1747,
+      "step": 2489
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 15.811159133911133,
+      "learning_rate": 5.877430887870081e-06,
+      "loss": 0.1325,
+      "step": 2490
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 18.80916404724121,
+      "learning_rate": 5.874674207573782e-06,
+      "loss": 0.1095,
+      "step": 2491
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 9.724221229553223,
+      "learning_rate": 5.871917252980941e-06,
+      "loss": 0.1027,
+      "step": 2492
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 9.214253425598145,
+      "learning_rate": 5.8691600249561345e-06,
+      "loss": 0.0559,
+      "step": 2493
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 4.514297962188721,
+      "learning_rate": 5.8664025243640265e-06,
+      "loss": 0.0325,
+      "step": 2494
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 10.078828811645508,
+      "learning_rate": 5.863644752069364e-06,
+      "loss": 0.1383,
+      "step": 2495
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 10.932859420776367,
+      "learning_rate": 5.8608867089369825e-06,
+      "loss": 0.1063,
+      "step": 2496
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 4.866419315338135,
+      "learning_rate": 5.8581283958317995e-06,
+      "loss": 0.0309,
+      "step": 2497
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 1.7066974639892578,
+      "learning_rate": 5.8553698136188155e-06,
+      "loss": 0.0144,
+      "step": 2498
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 6.9340500831604,
+      "learning_rate": 5.85261096316312e-06,
+      "loss": 0.1311,
+      "step": 2499
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 5.031786918640137,
+      "learning_rate": 5.849851845329884e-06,
+      "loss": 0.0289,
+      "step": 2500
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 9.146328926086426,
+      "learning_rate": 5.847092460984361e-06,
+      "loss": 0.1598,
+      "step": 2501
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 11.521110534667969,
+      "learning_rate": 5.844332810991894e-06,
+      "loss": 0.0599,
+      "step": 2502
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 6.217971324920654,
+      "learning_rate": 5.841572896217903e-06,
+      "loss": 0.0345,
+      "step": 2503
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 9.138691902160645,
+      "learning_rate": 5.838812717527891e-06,
+      "loss": 0.0332,
+      "step": 2504
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 7.380448818206787,
+      "learning_rate": 5.836052275787448e-06,
+      "loss": 0.1067,
+      "step": 2505
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 9.53880500793457,
+      "learning_rate": 5.833291571862245e-06,
+      "loss": 0.0512,
+      "step": 2506
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 9.868330955505371,
+      "learning_rate": 5.830530606618035e-06,
+      "loss": 0.064,
+      "step": 2507
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 4.733244895935059,
+      "learning_rate": 5.82776938092065e-06,
+      "loss": 0.0268,
+      "step": 2508
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 9.090733528137207,
+      "learning_rate": 5.825007895636008e-06,
+      "loss": 0.0527,
+      "step": 2509
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 4.855319976806641,
+      "learning_rate": 5.822246151630109e-06,
+      "loss": 0.0378,
+      "step": 2510
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 9.772489547729492,
+      "learning_rate": 5.819484149769029e-06,
+      "loss": 0.0409,
+      "step": 2511
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 7.024757385253906,
+      "learning_rate": 5.816721890918926e-06,
+      "loss": 0.0255,
+      "step": 2512
+    },
+    {
+      "epoch": 4.47,
+      "grad_norm": 14.35616683959961,
+      "learning_rate": 5.813959375946045e-06,
+      "loss": 0.081,
+      "step": 2513
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 2.0977540016174316,
+      "learning_rate": 5.811196605716702e-06,
+      "loss": 0.0137,
+      "step": 2514
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 19.764875411987305,
+      "learning_rate": 5.808433581097301e-06,
+      "loss": 0.2271,
+      "step": 2515
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 3.847980499267578,
+      "learning_rate": 5.805670302954322e-06,
+      "loss": 0.0165,
+      "step": 2516
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 25.143722534179688,
+      "learning_rate": 5.802906772154323e-06,
+      "loss": 0.3161,
+      "step": 2517
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 13.414109230041504,
+      "learning_rate": 5.800142989563944e-06,
+      "loss": 0.0773,
+      "step": 2518
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 8.438363075256348,
+      "learning_rate": 5.797378956049905e-06,
+      "loss": 0.028,
+      "step": 2519
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 10.87607192993164,
+      "learning_rate": 5.794614672479e-06,
+      "loss": 0.1294,
+      "step": 2520
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 1.7871644496917725,
+      "learning_rate": 5.791850139718107e-06,
+      "loss": 0.0106,
+      "step": 2521
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 8.784443855285645,
+      "learning_rate": 5.7890853586341785e-06,
+      "loss": 0.0567,
+      "step": 2522
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 7.204719066619873,
+      "learning_rate": 5.786320330094244e-06,
+      "loss": 0.0503,
+      "step": 2523
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 14.490324974060059,
+      "learning_rate": 5.783555054965414e-06,
+      "loss": 0.2891,
+      "step": 2524
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 7.0030131340026855,
+      "learning_rate": 5.780789534114875e-06,
+      "loss": 0.0473,
+      "step": 2525
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 6.5853729248046875,
+      "learning_rate": 5.778023768409888e-06,
+      "loss": 0.0588,
+      "step": 2526
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 13.251618385314941,
+      "learning_rate": 5.775257758717796e-06,
+      "loss": 0.1273,
+      "step": 2527
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 10.428656578063965,
+      "learning_rate": 5.772491505906014e-06,
+      "loss": 0.2041,
+      "step": 2528
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 7.841995716094971,
+      "learning_rate": 5.7697250108420345e-06,
+      "loss": 0.1269,
+      "step": 2529
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 8.670536994934082,
+      "learning_rate": 5.766958274393428e-06,
+      "loss": 0.0952,
+      "step": 2530
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 3.271763324737549,
+      "learning_rate": 5.764191297427837e-06,
+      "loss": 0.0221,
+      "step": 2531
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 7.023367881774902,
+      "learning_rate": 5.761424080812982e-06,
+      "loss": 0.0296,
+      "step": 2532
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 16.9525146484375,
+      "learning_rate": 5.758656625416659e-06,
+      "loss": 0.096,
+      "step": 2533
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 4.163442611694336,
+      "learning_rate": 5.755888932106736e-06,
+      "loss": 0.0195,
+      "step": 2534
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 12.459059715270996,
+      "learning_rate": 5.753121001751161e-06,
+      "loss": 0.175,
+      "step": 2535
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 8.545014381408691,
+      "learning_rate": 5.750352835217949e-06,
+      "loss": 0.0785,
+      "step": 2536
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 6.57154655456543,
+      "learning_rate": 5.747584433375194e-06,
+      "loss": 0.0299,
+      "step": 2537
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 14.866918563842773,
+      "learning_rate": 5.7448157970910646e-06,
+      "loss": 0.0719,
+      "step": 2538
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 3.107337713241577,
+      "learning_rate": 5.7420469272338e-06,
+      "loss": 0.0185,
+      "step": 2539
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 9.341525077819824,
+      "learning_rate": 5.739277824671711e-06,
+      "loss": 0.0835,
+      "step": 2540
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 8.494711875915527,
+      "learning_rate": 5.736508490273189e-06,
+      "loss": 0.0466,
+      "step": 2541
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 6.9288201332092285,
+      "learning_rate": 5.73373892490669e-06,
+      "loss": 0.1858,
+      "step": 2542
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 8.124451637268066,
+      "learning_rate": 5.730969129440744e-06,
+      "loss": 0.036,
+      "step": 2543
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 6.833055019378662,
+      "learning_rate": 5.72819910474396e-06,
+      "loss": 0.0814,
+      "step": 2544
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 27.8125057220459,
+      "learning_rate": 5.725428851685011e-06,
+      "loss": 0.1553,
+      "step": 2545
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 3.7269723415374756,
+      "learning_rate": 5.722658371132641e-06,
+      "loss": 0.0304,
+      "step": 2546
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 7.968837261199951,
+      "learning_rate": 5.719887663955674e-06,
+      "loss": 0.0724,
+      "step": 2547
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 5.496086597442627,
+      "learning_rate": 5.717116731022998e-06,
+      "loss": 0.051,
+      "step": 2548
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 1.8676257133483887,
+      "learning_rate": 5.714345573203571e-06,
+      "loss": 0.0154,
+      "step": 2549
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 3.7054901123046875,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 0.084,
+      "step": 2550
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 7.717898845672607,
+      "learning_rate": 5.708802586380665e-06,
+      "loss": 0.034,
+      "step": 2551
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 4.147486686706543,
+      "learning_rate": 5.706030759115458e-06,
+      "loss": 0.0342,
+      "step": 2552
+    },
+    {
+      "epoch": 4.54,
+      "grad_norm": 7.372539520263672,
+      "learning_rate": 5.703258710440046e-06,
+      "loss": 0.0317,
+      "step": 2553
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 7.560828685760498,
+      "learning_rate": 5.70048644122374e-06,
+      "loss": 0.1395,
+      "step": 2554
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 15.453147888183594,
+      "learning_rate": 5.697713952335918e-06,
+      "loss": 0.1177,
+      "step": 2555
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 11.774754524230957,
+      "learning_rate": 5.694941244646031e-06,
+      "loss": 0.041,
+      "step": 2556
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 6.25853967666626,
+      "learning_rate": 5.6921683190235935e-06,
+      "loss": 0.0429,
+      "step": 2557
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 2.417999505996704,
+      "learning_rate": 5.689395176338193e-06,
+      "loss": 0.0183,
+      "step": 2558
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 11.772210121154785,
+      "learning_rate": 5.68662181745948e-06,
+      "loss": 0.1342,
+      "step": 2559
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 7.523478031158447,
+      "learning_rate": 5.683848243257181e-06,
+      "loss": 0.0445,
+      "step": 2560
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 10.314644813537598,
+      "learning_rate": 5.6810744546010796e-06,
+      "loss": 0.0311,
+      "step": 2561
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 10.055065155029297,
+      "learning_rate": 5.678300452361033e-06,
+      "loss": 0.0674,
+      "step": 2562
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 37.70061111450195,
+      "learning_rate": 5.675526237406965e-06,
+      "loss": 0.0985,
+      "step": 2563
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 9.104837417602539,
+      "learning_rate": 5.672751810608867e-06,
+      "loss": 0.0331,
+      "step": 2564
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 7.513531684875488,
+      "learning_rate": 5.669977172836791e-06,
+      "loss": 0.0973,
+      "step": 2565
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 2.26885986328125,
+      "learning_rate": 5.667202324960864e-06,
+      "loss": 0.0119,
+      "step": 2566
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 5.3700175285339355,
+      "learning_rate": 5.664427267851271e-06,
+      "loss": 0.0743,
+      "step": 2567
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 5.290533542633057,
+      "learning_rate": 5.661652002378266e-06,
+      "loss": 0.0677,
+      "step": 2568
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 6.578613758087158,
+      "learning_rate": 5.6588765294121705e-06,
+      "loss": 0.0817,
+      "step": 2569
+    },
+    {
+      "epoch": 4.57,
+      "grad_norm": 11.631264686584473,
+      "learning_rate": 5.656100849823366e-06,
+      "loss": 0.1327,
+      "step": 2570
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 10.633448600769043,
+      "learning_rate": 5.653324964482303e-06,
+      "loss": 0.1093,
+      "step": 2571
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 2.965433120727539,
+      "learning_rate": 5.650548874259494e-06,
+      "loss": 0.0169,
+      "step": 2572
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 9.969054222106934,
+      "learning_rate": 5.647772580025519e-06,
+      "loss": 0.1507,
+      "step": 2573
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 2.85343074798584,
+      "learning_rate": 5.644996082651018e-06,
+      "loss": 0.0204,
+      "step": 2574
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 3.34655499458313,
+      "learning_rate": 5.642219383006696e-06,
+      "loss": 0.0208,
+      "step": 2575
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 12.270248413085938,
+      "learning_rate": 5.639442481963322e-06,
+      "loss": 0.1083,
+      "step": 2576
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 12.202247619628906,
+      "learning_rate": 5.636665380391731e-06,
+      "loss": 0.0835,
+      "step": 2577
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 5.8492841720581055,
+      "learning_rate": 5.633888079162813e-06,
+      "loss": 0.0304,
+      "step": 2578
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 3.960362195968628,
+      "learning_rate": 5.63111057914753e-06,
+      "loss": 0.0197,
+      "step": 2579
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 7.117588520050049,
+      "learning_rate": 5.628332881216899e-06,
+      "loss": 0.0512,
+      "step": 2580
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 10.851568222045898,
+      "learning_rate": 5.625554986242005e-06,
+      "loss": 0.0749,
+      "step": 2581
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 6.737936019897461,
+      "learning_rate": 5.622776895093988e-06,
+      "loss": 0.0555,
+      "step": 2582
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 5.061647415161133,
+      "learning_rate": 5.619998608644055e-06,
+      "loss": 0.0769,
+      "step": 2583
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 6.231665134429932,
+      "learning_rate": 5.617220127763474e-06,
+      "loss": 0.0958,
+      "step": 2584
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 5.929920673370361,
+      "learning_rate": 5.614441453323571e-06,
+      "loss": 0.0328,
+      "step": 2585
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 36.00010681152344,
+      "learning_rate": 5.611662586195735e-06,
+      "loss": 0.3419,
+      "step": 2586
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 2.3000612258911133,
+      "learning_rate": 5.608883527251413e-06,
+      "loss": 0.0196,
+      "step": 2587
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 8.063896179199219,
+      "learning_rate": 5.606104277362116e-06,
+      "loss": 0.0387,
+      "step": 2588
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 10.224661827087402,
+      "learning_rate": 5.603324837399411e-06,
+      "loss": 0.1192,
+      "step": 2589
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 5.370235919952393,
+      "learning_rate": 5.600545208234927e-06,
+      "loss": 0.0918,
+      "step": 2590
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 16.218793869018555,
+      "learning_rate": 5.597765390740351e-06,
+      "loss": 0.105,
+      "step": 2591
+    },
+    {
+      "epoch": 4.61,
+      "grad_norm": 8.290640830993652,
+      "learning_rate": 5.59498538578743e-06,
+      "loss": 0.0856,
+      "step": 2592
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 10.509263038635254,
+      "learning_rate": 5.592205194247969e-06,
+      "loss": 0.1032,
+      "step": 2593
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 20.683856964111328,
+      "learning_rate": 5.5894248169938335e-06,
+      "loss": 0.1636,
+      "step": 2594
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 6.784024715423584,
+      "learning_rate": 5.586644254896945e-06,
+      "loss": 0.1277,
+      "step": 2595
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 3.5502591133117676,
+      "learning_rate": 5.583863508829281e-06,
+      "loss": 0.0153,
+      "step": 2596
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 2.7197062969207764,
+      "learning_rate": 5.581082579662884e-06,
+      "loss": 0.0185,
+      "step": 2597
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 3.791726589202881,
+      "learning_rate": 5.578301468269848e-06,
+      "loss": 0.0363,
+      "step": 2598
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 9.701123237609863,
+      "learning_rate": 5.575520175522323e-06,
+      "loss": 0.0488,
+      "step": 2599
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 7.783981800079346,
+      "learning_rate": 5.57273870229252e-06,
+      "loss": 0.045,
+      "step": 2600
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 5.940560817718506,
+      "learning_rate": 5.569957049452703e-06,
+      "loss": 0.0448,
+      "step": 2601
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 6.617189407348633,
+      "learning_rate": 5.5671752178751985e-06,
+      "loss": 0.0742,
+      "step": 2602
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 6.231395244598389,
+      "learning_rate": 5.564393208432382e-06,
+      "loss": 0.0443,
+      "step": 2603
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 8.922979354858398,
+      "learning_rate": 5.561611021996685e-06,
+      "loss": 0.0583,
+      "step": 2604
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 5.715095043182373,
+      "learning_rate": 5.558828659440603e-06,
+      "loss": 0.1402,
+      "step": 2605
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 1.9688180685043335,
+      "learning_rate": 5.5560461216366754e-06,
+      "loss": 0.012,
+      "step": 2606
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 5.521790504455566,
+      "learning_rate": 5.553263409457504e-06,
+      "loss": 0.0449,
+      "step": 2607
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 14.031998634338379,
+      "learning_rate": 5.550480523775745e-06,
+      "loss": 0.2721,
+      "step": 2608
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 4.592979907989502,
+      "learning_rate": 5.547697465464106e-06,
+      "loss": 0.0283,
+      "step": 2609
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 5.278326034545898,
+      "learning_rate": 5.544914235395347e-06,
+      "loss": 0.0279,
+      "step": 2610
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 8.075479507446289,
+      "learning_rate": 5.542130834442291e-06,
+      "loss": 0.0462,
+      "step": 2611
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 14.621414184570312,
+      "learning_rate": 5.539347263477801e-06,
+      "loss": 0.0754,
+      "step": 2612
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 3.9016671180725098,
+      "learning_rate": 5.536563523374807e-06,
+      "loss": 0.0235,
+      "step": 2613
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 8.440091133117676,
+      "learning_rate": 5.533779615006284e-06,
+      "loss": 0.0755,
+      "step": 2614
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 5.51993989944458,
+      "learning_rate": 5.5309955392452585e-06,
+      "loss": 0.1356,
+      "step": 2615
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 8.146055221557617,
+      "learning_rate": 5.528211296964817e-06,
+      "loss": 0.0379,
+      "step": 2616
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 8.272150039672852,
+      "learning_rate": 5.525426889038091e-06,
+      "loss": 0.0283,
+      "step": 2617
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 7.395650863647461,
+      "learning_rate": 5.522642316338268e-06,
+      "loss": 0.1782,
+      "step": 2618
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 3.728787422180176,
+      "learning_rate": 5.5198575797385855e-06,
+      "loss": 0.0191,
+      "step": 2619
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 14.346944808959961,
+      "learning_rate": 5.517072680112332e-06,
+      "loss": 0.1206,
+      "step": 2620
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 5.973649501800537,
+      "learning_rate": 5.514287618332849e-06,
+      "loss": 0.0962,
+      "step": 2621
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 12.217443466186523,
+      "learning_rate": 5.511502395273527e-06,
+      "loss": 0.062,
+      "step": 2622
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 10.802512168884277,
+      "learning_rate": 5.50871701180781e-06,
+      "loss": 0.1609,
+      "step": 2623
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 21.31576919555664,
+      "learning_rate": 5.505931468809187e-06,
+      "loss": 0.0579,
+      "step": 2624
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 3.8688719272613525,
+      "learning_rate": 5.503145767151201e-06,
+      "loss": 0.0221,
+      "step": 2625
+    },
+    {
+      "epoch": 4.67,
+      "grad_norm": 3.5676674842834473,
+      "learning_rate": 5.5003599077074456e-06,
+      "loss": 0.0203,
+      "step": 2626
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 12.691487312316895,
+      "learning_rate": 5.497573891351561e-06,
+      "loss": 0.1381,
+      "step": 2627
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 9.089301109313965,
+      "learning_rate": 5.494787718957237e-06,
+      "loss": 0.0326,
+      "step": 2628
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 6.882167816162109,
+      "learning_rate": 5.492001391398214e-06,
+      "loss": 0.0527,
+      "step": 2629
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 5.759741306304932,
+      "learning_rate": 5.4892149095482815e-06,
+      "loss": 0.034,
+      "step": 2630
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 6.371630668640137,
+      "learning_rate": 5.486428274281275e-06,
+      "loss": 0.0335,
+      "step": 2631
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 16.668996810913086,
+      "learning_rate": 5.483641486471076e-06,
+      "loss": 0.1175,
+      "step": 2632
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 6.310295104980469,
+      "learning_rate": 5.480854546991624e-06,
+      "loss": 0.1453,
+      "step": 2633
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 18.27882957458496,
+      "learning_rate": 5.478067456716894e-06,
+      "loss": 0.2163,
+      "step": 2634
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 8.908318519592285,
+      "learning_rate": 5.475280216520913e-06,
+      "loss": 0.1097,
+      "step": 2635
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 11.915366172790527,
+      "learning_rate": 5.47249282727776e-06,
+      "loss": 0.0808,
+      "step": 2636
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 2.6465559005737305,
+      "learning_rate": 5.4697052898615514e-06,
+      "loss": 0.0176,
+      "step": 2637
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 14.156755447387695,
+      "learning_rate": 5.466917605146457e-06,
+      "loss": 0.0299,
+      "step": 2638
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 3.3986592292785645,
+      "learning_rate": 5.464129774006691e-06,
+      "loss": 0.0176,
+      "step": 2639
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 10.703646659851074,
+      "learning_rate": 5.46134179731651e-06,
+      "loss": 0.0275,
+      "step": 2640
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 19.696365356445312,
+      "learning_rate": 5.458553675950225e-06,
+      "loss": 0.1207,
+      "step": 2641
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 6.819868564605713,
+      "learning_rate": 5.455765410782181e-06,
+      "loss": 0.1201,
+      "step": 2642
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 13.879584312438965,
+      "learning_rate": 5.452977002686776e-06,
+      "loss": 0.0594,
+      "step": 2643
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 9.226188659667969,
+      "learning_rate": 5.450188452538454e-06,
+      "loss": 0.0668,
+      "step": 2644
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 4.56365442276001,
+      "learning_rate": 5.4473997612116956e-06,
+      "loss": 0.0569,
+      "step": 2645
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 5.029369831085205,
+      "learning_rate": 5.444610929581031e-06,
+      "loss": 0.0757,
+      "step": 2646
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 1.8957719802856445,
+      "learning_rate": 5.441821958521037e-06,
+      "loss": 0.0172,
+      "step": 2647
+    },
+    {
+      "epoch": 4.71,
+      "grad_norm": 14.564288139343262,
+      "learning_rate": 5.439032848906326e-06,
+      "loss": 0.0534,
+      "step": 2648
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 11.97154426574707,
+      "learning_rate": 5.436243601611563e-06,
+      "loss": 0.0612,
+      "step": 2649
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 9.522682189941406,
+      "learning_rate": 5.4334542175114495e-06,
+      "loss": 0.1259,
+      "step": 2650
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 27.065189361572266,
+      "learning_rate": 5.430664697480731e-06,
+      "loss": 0.047,
+      "step": 2651
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 4.405757904052734,
+      "learning_rate": 5.4278750423942e-06,
+      "loss": 0.0742,
+      "step": 2652
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 7.674710750579834,
+      "learning_rate": 5.425085253126686e-06,
+      "loss": 0.0421,
+      "step": 2653
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 6.950433731079102,
+      "learning_rate": 5.422295330553064e-06,
+      "loss": 0.0476,
+      "step": 2654
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 8.767231941223145,
+      "learning_rate": 5.419505275548249e-06,
+      "loss": 0.0644,
+      "step": 2655
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 11.383990287780762,
+      "learning_rate": 5.416715088987199e-06,
+      "loss": 0.0924,
+      "step": 2656
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 5.96630859375,
+      "learning_rate": 5.4139247717449105e-06,
+      "loss": 0.041,
+      "step": 2657
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 6.308305740356445,
+      "learning_rate": 5.411134324696425e-06,
+      "loss": 0.0688,
+      "step": 2658
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 5.560367584228516,
+      "learning_rate": 5.408343748716821e-06,
+      "loss": 0.0509,
+      "step": 2659
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 17.08702850341797,
+      "learning_rate": 5.40555304468122e-06,
+      "loss": 0.0944,
+      "step": 2660
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 4.369718074798584,
+      "learning_rate": 5.402762213464784e-06,
+      "loss": 0.0507,
+      "step": 2661
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 4.786932468414307,
+      "learning_rate": 5.399971255942708e-06,
+      "loss": 0.0355,
+      "step": 2662
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 9.56259536743164,
+      "learning_rate": 5.397180172990239e-06,
+      "loss": 0.0344,
+      "step": 2663
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 8.795884132385254,
+      "learning_rate": 5.394388965482653e-06,
+      "loss": 0.0298,
+      "step": 2664
+    },
+    {
+      "epoch": 4.74,
+      "grad_norm": 4.550373077392578,
+      "learning_rate": 5.391597634295269e-06,
+      "loss": 0.0779,
+      "step": 2665
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 10.376954078674316,
+      "learning_rate": 5.388806180303444e-06,
+      "loss": 0.0814,
+      "step": 2666
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 12.72496509552002,
+      "learning_rate": 5.386014604382576e-06,
+      "loss": 0.1003,
+      "step": 2667
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 4.4917473793029785,
+      "learning_rate": 5.383222907408094e-06,
+      "loss": 0.1266,
+      "step": 2668
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 6.799139976501465,
+      "learning_rate": 5.380431090255475e-06,
+      "loss": 0.0372,
+      "step": 2669
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 8.629815101623535,
+      "learning_rate": 5.377639153800229e-06,
+      "loss": 0.1131,
+      "step": 2670
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 2.0219626426696777,
+      "learning_rate": 5.374847098917899e-06,
+      "loss": 0.0179,
+      "step": 2671
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 6.67774772644043,
+      "learning_rate": 5.3720549264840725e-06,
+      "loss": 0.0435,
+      "step": 2672
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 7.634645938873291,
+      "learning_rate": 5.36926263737437e-06,
+      "loss": 0.0741,
+      "step": 2673
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 7.9138288497924805,
+      "learning_rate": 5.36647023246445e-06,
+      "loss": 0.0766,
+      "step": 2674
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 1.8411310911178589,
+      "learning_rate": 5.363677712630004e-06,
+      "loss": 0.0133,
+      "step": 2675
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 15.702991485595703,
+      "learning_rate": 5.360885078746764e-06,
+      "loss": 0.151,
+      "step": 2676
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 5.965345859527588,
+      "learning_rate": 5.358092331690497e-06,
+      "loss": 0.0286,
+      "step": 2677
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 6.246129512786865,
+      "learning_rate": 5.355299472337001e-06,
+      "loss": 0.022,
+      "step": 2678
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 12.065853118896484,
+      "learning_rate": 5.352506501562114e-06,
+      "loss": 0.1208,
+      "step": 2679
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 2.620753288269043,
+      "learning_rate": 5.34971342024171e-06,
+      "loss": 0.0189,
+      "step": 2680
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 5.675437927246094,
+      "learning_rate": 5.346920229251692e-06,
+      "loss": 0.0253,
+      "step": 2681
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 12.870821952819824,
+      "learning_rate": 5.344126929468001e-06,
+      "loss": 0.1329,
+      "step": 2682
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 14.453897476196289,
+      "learning_rate": 5.341333521766614e-06,
+      "loss": 0.1916,
+      "step": 2683
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 12.161906242370605,
+      "learning_rate": 5.338540007023538e-06,
+      "loss": 0.0354,
+      "step": 2684
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 9.858351707458496,
+      "learning_rate": 5.335746386114814e-06,
+      "loss": 0.1384,
+      "step": 2685
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 8.193168640136719,
+      "learning_rate": 5.33295265991652e-06,
+      "loss": 0.0482,
+      "step": 2686
+    },
+    {
+      "epoch": 4.78,
+      "grad_norm": 4.574099063873291,
+      "learning_rate": 5.3301588293047615e-06,
+      "loss": 0.024,
+      "step": 2687
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 9.764312744140625,
+      "learning_rate": 5.327364895155682e-06,
+      "loss": 0.097,
+      "step": 2688
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 8.077651977539062,
+      "learning_rate": 5.3245708583454545e-06,
+      "loss": 0.0861,
+      "step": 2689
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 2.504058837890625,
+      "learning_rate": 5.321776719750283e-06,
+      "loss": 0.0103,
+      "step": 2690
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 11.202048301696777,
+      "learning_rate": 5.318982480246407e-06,
+      "loss": 0.0565,
+      "step": 2691
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 5.679230690002441,
+      "learning_rate": 5.316188140710097e-06,
+      "loss": 0.0621,
+      "step": 2692
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 2.778287887573242,
+      "learning_rate": 5.3133937020176505e-06,
+      "loss": 0.0251,
+      "step": 2693
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 5.3225836753845215,
+      "learning_rate": 5.3105991650454025e-06,
+      "loss": 0.0277,
+      "step": 2694
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 8.463663101196289,
+      "learning_rate": 5.3078045306697154e-06,
+      "loss": 0.0492,
+      "step": 2695
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 10.663856506347656,
+      "learning_rate": 5.3050097997669795e-06,
+      "loss": 0.2149,
+      "step": 2696
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 4.482873439788818,
+      "learning_rate": 5.302214973213621e-06,
+      "loss": 0.0194,
+      "step": 2697
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 5.888377666473389,
+      "learning_rate": 5.299420051886093e-06,
+      "loss": 0.1121,
+      "step": 2698
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 14.177448272705078,
+      "learning_rate": 5.296625036660879e-06,
+      "loss": 0.0923,
+      "step": 2699
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 11.147114753723145,
+      "learning_rate": 5.29382992841449e-06,
+      "loss": 0.0708,
+      "step": 2700
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 4.332546710968018,
+      "learning_rate": 5.2910347280234675e-06,
+      "loss": 0.0217,
+      "step": 2701
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 13.859275817871094,
+      "learning_rate": 5.288239436364385e-06,
+      "loss": 0.1744,
+      "step": 2702
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 10.313837051391602,
+      "learning_rate": 5.285444054313841e-06,
+      "loss": 0.0567,
+      "step": 2703
+    },
+    {
+      "epoch": 4.81,
+      "grad_norm": 3.458085775375366,
+      "learning_rate": 5.282648582748461e-06,
+      "loss": 0.0222,
+      "step": 2704
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 1.8777897357940674,
+      "learning_rate": 5.279853022544904e-06,
+      "loss": 0.0117,
+      "step": 2705
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 9.928890228271484,
+      "learning_rate": 5.27705737457985e-06,
+      "loss": 0.1745,
+      "step": 2706
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 8.442804336547852,
+      "learning_rate": 5.274261639730013e-06,
+      "loss": 0.1289,
+      "step": 2707
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 11.894110679626465,
+      "learning_rate": 5.27146581887213e-06,
+      "loss": 0.0845,
+      "step": 2708
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 5.771883964538574,
+      "learning_rate": 5.268669912882968e-06,
+      "loss": 0.0566,
+      "step": 2709
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 8.149015426635742,
+      "learning_rate": 5.265873922639315e-06,
+      "loss": 0.0571,
+      "step": 2710
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 5.640145778656006,
+      "learning_rate": 5.2630778490179944e-06,
+      "loss": 0.0327,
+      "step": 2711
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 8.473013877868652,
+      "learning_rate": 5.260281692895847e-06,
+      "loss": 0.0879,
+      "step": 2712
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 6.626067638397217,
+      "learning_rate": 5.257485455149746e-06,
+      "loss": 0.0495,
+      "step": 2713
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 1.3715640306472778,
+      "learning_rate": 5.254689136656585e-06,
+      "loss": 0.0098,
+      "step": 2714
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 4.873822212219238,
+      "learning_rate": 5.251892738293285e-06,
+      "loss": 0.0527,
+      "step": 2715
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 6.908078670501709,
+      "learning_rate": 5.2490962609367946e-06,
+      "loss": 0.0909,
+      "step": 2716
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 15.399928092956543,
+      "learning_rate": 5.246299705464085e-06,
+      "loss": 0.1451,
+      "step": 2717
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 16.401798248291016,
+      "learning_rate": 5.24350307275215e-06,
+      "loss": 0.2271,
+      "step": 2718
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 13.5808744430542,
+      "learning_rate": 5.240706363678012e-06,
+      "loss": 0.0682,
+      "step": 2719
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 3.2704877853393555,
+      "learning_rate": 5.237909579118713e-06,
+      "loss": 0.0289,
+      "step": 2720
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 1.5447269678115845,
+      "learning_rate": 5.235112719951319e-06,
+      "loss": 0.0165,
+      "step": 2721
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 6.992668151855469,
+      "learning_rate": 5.232315787052924e-06,
+      "loss": 0.0544,
+      "step": 2722
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 6.947348117828369,
+      "learning_rate": 5.229518781300642e-06,
+      "loss": 0.1014,
+      "step": 2723
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 15.235512733459473,
+      "learning_rate": 5.226721703571607e-06,
+      "loss": 0.2196,
+      "step": 2724
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 8.97320556640625,
+      "learning_rate": 5.223924554742982e-06,
+      "loss": 0.0761,
+      "step": 2725
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 5.240776538848877,
+      "learning_rate": 5.221127335691944e-06,
+      "loss": 0.044,
+      "step": 2726
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 10.253002166748047,
+      "learning_rate": 5.2183300472957025e-06,
+      "loss": 0.0766,
+      "step": 2727
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 2.1749401092529297,
+      "learning_rate": 5.2155326904314795e-06,
+      "loss": 0.0101,
+      "step": 2728
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 9.856504440307617,
+      "learning_rate": 5.212735265976521e-06,
+      "loss": 0.1331,
+      "step": 2729
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 5.966109752655029,
+      "learning_rate": 5.209937774808098e-06,
+      "loss": 0.1126,
+      "step": 2730
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 3.245793581008911,
+      "learning_rate": 5.207140217803499e-06,
+      "loss": 0.0238,
+      "step": 2731
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 6.464186191558838,
+      "learning_rate": 5.204342595840032e-06,
+      "loss": 0.1049,
+      "step": 2732
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 14.940474510192871,
+      "learning_rate": 5.201544909795029e-06,
+      "loss": 0.0721,
+      "step": 2733
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 9.911134719848633,
+      "learning_rate": 5.19874716054584e-06,
+      "loss": 0.0841,
+      "step": 2734
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 6.523481845855713,
+      "learning_rate": 5.195949348969833e-06,
+      "loss": 0.0471,
+      "step": 2735
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 12.622416496276855,
+      "learning_rate": 5.193151475944403e-06,
+      "loss": 0.2357,
+      "step": 2736
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 6.429666519165039,
+      "learning_rate": 5.190353542346951e-06,
+      "loss": 0.0738,
+      "step": 2737
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 8.847264289855957,
+      "learning_rate": 5.18755554905491e-06,
+      "loss": 0.2387,
+      "step": 2738
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 12.472099304199219,
+      "learning_rate": 5.184757496945726e-06,
+      "loss": 0.0807,
+      "step": 2739
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 5.8747406005859375,
+      "learning_rate": 5.181959386896862e-06,
+      "loss": 0.0811,
+      "step": 2740
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 14.258538246154785,
+      "learning_rate": 5.179161219785805e-06,
+      "loss": 0.1428,
+      "step": 2741
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 2.3904967308044434,
+      "learning_rate": 5.176362996490052e-06,
+      "loss": 0.0147,
+      "step": 2742
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 2.730851411819458,
+      "learning_rate": 5.173564717887123e-06,
+      "loss": 0.023,
+      "step": 2743
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 5.645975589752197,
+      "learning_rate": 5.170766384854555e-06,
+      "loss": 0.0757,
+      "step": 2744
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 8.96431827545166,
+      "learning_rate": 5.167967998269902e-06,
+      "loss": 0.1441,
+      "step": 2745
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 8.41584300994873,
+      "learning_rate": 5.16516955901073e-06,
+      "loss": 0.0364,
+      "step": 2746
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 10.911765098571777,
+      "learning_rate": 5.162371067954629e-06,
+      "loss": 0.0622,
+      "step": 2747
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 8.0665922164917,
+      "learning_rate": 5.1595725259792005e-06,
+      "loss": 0.0551,
+      "step": 2748
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 14.112918853759766,
+      "learning_rate": 5.156773933962065e-06,
+      "loss": 0.19,
+      "step": 2749
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 5.135739326477051,
+      "learning_rate": 5.153975292780852e-06,
+      "loss": 0.0553,
+      "step": 2750
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 6.7532525062561035,
+      "learning_rate": 5.151176603313216e-06,
+      "loss": 0.0509,
+      "step": 2751
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 12.819918632507324,
+      "learning_rate": 5.14837786643682e-06,
+      "loss": 0.1019,
+      "step": 2752
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 8.494109153747559,
+      "learning_rate": 5.145579083029344e-06,
+      "loss": 0.2726,
+      "step": 2753
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 3.105408191680908,
+      "learning_rate": 5.142780253968481e-06,
+      "loss": 0.0215,
+      "step": 2754
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 4.559239387512207,
+      "learning_rate": 5.139981380131943e-06,
+      "loss": 0.0243,
+      "step": 2755
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 7.062232971191406,
+      "learning_rate": 5.13718246239745e-06,
+      "loss": 0.0465,
+      "step": 2756
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 0.8230646848678589,
+      "learning_rate": 5.134383501642738e-06,
+      "loss": 0.0083,
+      "step": 2757
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 20.786720275878906,
+      "learning_rate": 5.131584498745559e-06,
+      "loss": 0.1604,
+      "step": 2758
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 17.771560668945312,
+      "learning_rate": 5.128785454583676e-06,
+      "loss": 0.0296,
+      "step": 2759
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 5.696237087249756,
+      "learning_rate": 5.125986370034862e-06,
+      "loss": 0.0366,
+      "step": 2760
+    },
+    {
+      "epoch": 4.91,
+      "grad_norm": 8.164920806884766,
+      "learning_rate": 5.123187245976912e-06,
+      "loss": 0.1546,
+      "step": 2761
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 9.50357723236084,
+      "learning_rate": 5.1203880832876196e-06,
+      "loss": 0.0389,
+      "step": 2762
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 9.390918731689453,
+      "learning_rate": 5.117588882844804e-06,
+      "loss": 0.0506,
+      "step": 2763
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 12.810199737548828,
+      "learning_rate": 5.114789645526287e-06,
+      "loss": 0.1152,
+      "step": 2764
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 1.3772250413894653,
+      "learning_rate": 5.111990372209906e-06,
+      "loss": 0.0112,
+      "step": 2765
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 0.9400896430015564,
+      "learning_rate": 5.109191063773511e-06,
+      "loss": 0.0102,
+      "step": 2766
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 1.7021467685699463,
+      "learning_rate": 5.106391721094959e-06,
+      "loss": 0.0146,
+      "step": 2767
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 8.357072830200195,
+      "learning_rate": 5.10359234505212e-06,
+      "loss": 0.1671,
+      "step": 2768
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 2.2806241512298584,
+      "learning_rate": 5.100792936522874e-06,
+      "loss": 0.0154,
+      "step": 2769
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 8.753829956054688,
+      "learning_rate": 5.097993496385112e-06,
+      "loss": 0.047,
+      "step": 2770
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 9.171425819396973,
+      "learning_rate": 5.095194025516733e-06,
+      "loss": 0.0776,
+      "step": 2771
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 6.075747013092041,
+      "learning_rate": 5.09239452479565e-06,
+      "loss": 0.0241,
+      "step": 2772
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 12.471604347229004,
+      "learning_rate": 5.089594995099781e-06,
+      "loss": 0.1894,
+      "step": 2773
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 11.119457244873047,
+      "learning_rate": 5.0867954373070516e-06,
+      "loss": 0.0586,
+      "step": 2774
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 6.315127372741699,
+      "learning_rate": 5.083995852295402e-06,
+      "loss": 0.0678,
+      "step": 2775
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 15.665904998779297,
+      "learning_rate": 5.081196240942775e-06,
+      "loss": 0.2833,
+      "step": 2776
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 1.0348546504974365,
+      "learning_rate": 5.078396604127128e-06,
+      "loss": 0.0087,
+      "step": 2777
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 2.4084620475769043,
+      "learning_rate": 5.0755969427264225e-06,
+      "loss": 0.0159,
+      "step": 2778
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 9.69312572479248,
+      "learning_rate": 5.072797257618626e-06,
+      "loss": 0.1703,
+      "step": 2779
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 7.593048572540283,
+      "learning_rate": 5.069997549681718e-06,
+      "loss": 0.0213,
+      "step": 2780
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 3.0674936771392822,
+      "learning_rate": 5.067197819793681e-06,
+      "loss": 0.0173,
+      "step": 2781
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 13.582511901855469,
+      "learning_rate": 5.064398068832506e-06,
+      "loss": 0.2305,
+      "step": 2782
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 9.983345985412598,
+      "learning_rate": 5.061598297676192e-06,
+      "loss": 0.128,
+      "step": 2783
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 17.17118263244629,
+      "learning_rate": 5.0587985072027445e-06,
+      "loss": 0.1621,
+      "step": 2784
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 5.015740871429443,
+      "learning_rate": 5.0559986982901695e-06,
+      "loss": 0.0282,
+      "step": 2785
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 14.40271282196045,
+      "learning_rate": 5.053198871816487e-06,
+      "loss": 0.0765,
+      "step": 2786
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 10.400226593017578,
+      "learning_rate": 5.050399028659716e-06,
+      "loss": 0.0548,
+      "step": 2787
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 7.40225887298584,
+      "learning_rate": 5.0475991696978844e-06,
+      "loss": 0.0439,
+      "step": 2788
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 6.870412349700928,
+      "learning_rate": 5.044799295809023e-06,
+      "loss": 0.0275,
+      "step": 2789
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 14.774957656860352,
+      "learning_rate": 5.041999407871168e-06,
+      "loss": 0.1064,
+      "step": 2790
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 4.733208656311035,
+      "learning_rate": 5.0391995067623615e-06,
+      "loss": 0.0277,
+      "step": 2791
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 5.212532997131348,
+      "learning_rate": 5.036399593360647e-06,
+      "loss": 0.086,
+      "step": 2792
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 4.455263137817383,
+      "learning_rate": 5.0335996685440734e-06,
+      "loss": 0.0273,
+      "step": 2793
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 22.480396270751953,
+      "learning_rate": 5.030799733190694e-06,
+      "loss": 0.1929,
+      "step": 2794
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 8.301018714904785,
+      "learning_rate": 5.0279997881785635e-06,
+      "loss": 0.0895,
+      "step": 2795
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 9.432550430297852,
+      "learning_rate": 5.02519983438574e-06,
+      "loss": 0.0699,
+      "step": 2796
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 10.095921516418457,
+      "learning_rate": 5.022399872690287e-06,
+      "loss": 0.1126,
+      "step": 2797
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 9.012378692626953,
+      "learning_rate": 5.019599903970266e-06,
+      "loss": 0.0313,
+      "step": 2798
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 8.083922386169434,
+      "learning_rate": 5.016799929103745e-06,
+      "loss": 0.07,
+      "step": 2799
+    },
+    {
+      "epoch": 4.98,
+      "grad_norm": 8.131073951721191,
+      "learning_rate": 5.01399994896879e-06,
+      "loss": 0.0334,
+      "step": 2800
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 8.166223526000977,
+      "learning_rate": 5.0111999644434716e-06,
+      "loss": 0.065,
+      "step": 2801
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 6.536017894744873,
+      "learning_rate": 5.008399976405861e-06,
+      "loss": 0.0361,
+      "step": 2802
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 11.405527114868164,
+      "learning_rate": 5.0055999857340305e-06,
+      "loss": 0.0674,
+      "step": 2803
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 6.004866600036621,
+      "learning_rate": 5.0027999933060526e-06,
+      "loss": 0.0336,
+      "step": 2804
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 5.752556800842285,
+      "learning_rate": 5e-06,
+      "loss": 0.1072,
+      "step": 2805
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 7.376157283782959,
+      "learning_rate": 4.997200006693949e-06,
+      "loss": 0.0963,
+      "step": 2806
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 5.855232238769531,
+      "learning_rate": 4.99440001426597e-06,
+      "loss": 0.0773,
+      "step": 2807
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 10.316385269165039,
+      "learning_rate": 4.9916000235941415e-06,
+      "loss": 0.0461,
+      "step": 2808
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.8478605388272583,
+      "eval_loss": 0.6093880534172058,
+      "eval_runtime": 11.4982,
+      "eval_samples_per_second": 164.634,
+      "eval_steps_per_second": 20.612,
+      "step": 2808
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 5.046910762786865,
+      "learning_rate": 4.98880003555653e-06,
+      "loss": 0.0335,
+      "step": 2809
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 3.638169765472412,
+      "learning_rate": 4.986000051031212e-06,
+      "loss": 0.02,
+      "step": 2810
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 5.096895694732666,
+      "learning_rate": 4.983200070896257e-06,
+      "loss": 0.0277,
+      "step": 2811
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 6.07396125793457,
+      "learning_rate": 4.980400096029735e-06,
+      "loss": 0.0155,
+      "step": 2812
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 3.8821420669555664,
+      "learning_rate": 4.977600127309714e-06,
+      "loss": 0.0255,
+      "step": 2813
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 1.5839569568634033,
+      "learning_rate": 4.974800165614261e-06,
+      "loss": 0.0177,
+      "step": 2814
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 11.879055976867676,
+      "learning_rate": 4.972000211821438e-06,
+      "loss": 0.1054,
+      "step": 2815
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 8.067184448242188,
+      "learning_rate": 4.9692002668093075e-06,
+      "loss": 0.0685,
+      "step": 2816
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 4.063624382019043,
+      "learning_rate": 4.966400331455927e-06,
+      "loss": 0.0237,
+      "step": 2817
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 1.5427782535552979,
+      "learning_rate": 4.963600406639354e-06,
+      "loss": 0.0104,
+      "step": 2818
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 3.956857919692993,
+      "learning_rate": 4.960800493237641e-06,
+      "loss": 0.0237,
+      "step": 2819
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 6.984810829162598,
+      "learning_rate": 4.958000592128834e-06,
+      "loss": 0.0324,
+      "step": 2820
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 5.051743507385254,
+      "learning_rate": 4.955200704190979e-06,
+      "loss": 0.0327,
+      "step": 2821
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 0.7572007775306702,
+      "learning_rate": 4.952400830302117e-06,
+      "loss": 0.0092,
+      "step": 2822
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 21.561058044433594,
+      "learning_rate": 4.9496009713402856e-06,
+      "loss": 0.0612,
+      "step": 2823
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 5.18229866027832,
+      "learning_rate": 4.946801128183514e-06,
+      "loss": 0.0253,
+      "step": 2824
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 0.8317416310310364,
+      "learning_rate": 4.944001301709832e-06,
+      "loss": 0.0082,
+      "step": 2825
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 1.8185746669769287,
+      "learning_rate": 4.941201492797257e-06,
+      "loss": 0.0135,
+      "step": 2826
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 5.126213073730469,
+      "learning_rate": 4.9384017023238085e-06,
+      "loss": 0.0274,
+      "step": 2827
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 11.109391212463379,
+      "learning_rate": 4.935601931167494e-06,
+      "loss": 0.107,
+      "step": 2828
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 2.228433847427368,
+      "learning_rate": 4.932802180206321e-06,
+      "loss": 0.0153,
+      "step": 2829
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 2.3907721042633057,
+      "learning_rate": 4.930002450318282e-06,
+      "loss": 0.0143,
+      "step": 2830
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 1.930088996887207,
+      "learning_rate": 4.927202742381376e-06,
+      "loss": 0.0132,
+      "step": 2831
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 1.5097506046295166,
+      "learning_rate": 4.92440305727358e-06,
+      "loss": 0.0102,
+      "step": 2832
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 8.362695693969727,
+      "learning_rate": 4.921603395872873e-06,
+      "loss": 0.0648,
+      "step": 2833
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 1.019399881362915,
+      "learning_rate": 4.918803759057226e-06,
+      "loss": 0.0087,
+      "step": 2834
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 4.341545581817627,
+      "learning_rate": 4.9160041477046e-06,
+      "loss": 0.0116,
+      "step": 2835
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 6.950345516204834,
+      "learning_rate": 4.913204562692952e-06,
+      "loss": 0.0257,
+      "step": 2836
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 4.180719375610352,
+      "learning_rate": 4.910405004900222e-06,
+      "loss": 0.0111,
+      "step": 2837
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 20.29711151123047,
+      "learning_rate": 4.907605475204352e-06,
+      "loss": 0.0596,
+      "step": 2838
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 1.1610355377197266,
+      "learning_rate": 4.904805974483267e-06,
+      "loss": 0.0097,
+      "step": 2839
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 0.5406525731086731,
+      "learning_rate": 4.9020065036148885e-06,
+      "loss": 0.0064,
+      "step": 2840
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 0.45107054710388184,
+      "learning_rate": 4.899207063477126e-06,
+      "loss": 0.0081,
+      "step": 2841
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 7.769826412200928,
+      "learning_rate": 4.896407654947883e-06,
+      "loss": 0.0564,
+      "step": 2842
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 3.181713104248047,
+      "learning_rate": 4.8936082789050435e-06,
+      "loss": 0.0178,
+      "step": 2843
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 13.849215507507324,
+      "learning_rate": 4.890808936226491e-06,
+      "loss": 0.1497,
+      "step": 2844
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 5.563639163970947,
+      "learning_rate": 4.888009627790095e-06,
+      "loss": 0.1135,
+      "step": 2845
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 17.732934951782227,
+      "learning_rate": 4.885210354473714e-06,
+      "loss": 0.1885,
+      "step": 2846
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 4.892922401428223,
+      "learning_rate": 4.882411117155199e-06,
+      "loss": 0.0193,
+      "step": 2847
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 8.718681335449219,
+      "learning_rate": 4.879611916712382e-06,
+      "loss": 0.0172,
+      "step": 2848
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 10.502619743347168,
+      "learning_rate": 4.876812754023092e-06,
+      "loss": 0.2273,
+      "step": 2849
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 1.877105951309204,
+      "learning_rate": 4.874013629965138e-06,
+      "loss": 0.0119,
+      "step": 2850
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 10.657464027404785,
+      "learning_rate": 4.871214545416326e-06,
+      "loss": 0.0777,
+      "step": 2851
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 13.938342094421387,
+      "learning_rate": 4.868415501254441e-06,
+      "loss": 0.1008,
+      "step": 2852
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 1.3590036630630493,
+      "learning_rate": 4.865616498357263e-06,
+      "loss": 0.0082,
+      "step": 2853
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 22.8161678314209,
+      "learning_rate": 4.862817537602552e-06,
+      "loss": 0.0712,
+      "step": 2854
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 4.318600654602051,
+      "learning_rate": 4.860018619868058e-06,
+      "loss": 0.0231,
+      "step": 2855
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 6.258166313171387,
+      "learning_rate": 4.85721974603152e-06,
+      "loss": 0.0358,
+      "step": 2856
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 7.913333892822266,
+      "learning_rate": 4.854420916970657e-06,
+      "loss": 0.1154,
+      "step": 2857
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 14.685039520263672,
+      "learning_rate": 4.851622133563183e-06,
+      "loss": 0.0666,
+      "step": 2858
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 9.767109870910645,
+      "learning_rate": 4.848823396686785e-06,
+      "loss": 0.0218,
+      "step": 2859
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 3.9029414653778076,
+      "learning_rate": 4.846024707219149e-06,
+      "loss": 0.0194,
+      "step": 2860
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 0.7807303667068481,
+      "learning_rate": 4.843226066037937e-06,
+      "loss": 0.0077,
+      "step": 2861
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 22.420757293701172,
+      "learning_rate": 4.8404274740208e-06,
+      "loss": 0.0972,
+      "step": 2862
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 4.401296615600586,
+      "learning_rate": 4.837628932045371e-06,
+      "loss": 0.0186,
+      "step": 2863
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 9.882148742675781,
+      "learning_rate": 4.834830440989272e-06,
+      "loss": 0.0198,
+      "step": 2864
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 17.181625366210938,
+      "learning_rate": 4.8320320017301e-06,
+      "loss": 0.079,
+      "step": 2865
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 2.00813364982605,
+      "learning_rate": 4.8292336151454465e-06,
+      "loss": 0.0119,
+      "step": 2866
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 0.5745384693145752,
+      "learning_rate": 4.826435282112878e-06,
+      "loss": 0.0077,
+      "step": 2867
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 6.458466053009033,
+      "learning_rate": 4.82363700350995e-06,
+      "loss": 0.0891,
+      "step": 2868
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 2.7553234100341797,
+      "learning_rate": 4.820838780214199e-06,
+      "loss": 0.0106,
+      "step": 2869
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 23.68199348449707,
+      "learning_rate": 4.818040613103139e-06,
+      "loss": 0.1096,
+      "step": 2870
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 12.00549030303955,
+      "learning_rate": 4.815242503054277e-06,
+      "loss": 0.0333,
+      "step": 2871
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 1.6793766021728516,
+      "learning_rate": 4.812444450945091e-06,
+      "loss": 0.0107,
+      "step": 2872
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 1.3825937509536743,
+      "learning_rate": 4.809646457653051e-06,
+      "loss": 0.0111,
+      "step": 2873
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 11.49808406829834,
+      "learning_rate": 4.8068485240556e-06,
+      "loss": 0.0121,
+      "step": 2874
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 10.940629959106445,
+      "learning_rate": 4.804050651030168e-06,
+      "loss": 0.0599,
+      "step": 2875
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 7.200607776641846,
+      "learning_rate": 4.801252839454161e-06,
+      "loss": 0.0357,
+      "step": 2876
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 2.7786855697631836,
+      "learning_rate": 4.798455090204972e-06,
+      "loss": 0.0184,
+      "step": 2877
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 6.984340190887451,
+      "learning_rate": 4.795657404159969e-06,
+      "loss": 0.0517,
+      "step": 2878
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 1.919773817062378,
+      "learning_rate": 4.792859782196503e-06,
+      "loss": 0.0099,
+      "step": 2879
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 5.009191989898682,
+      "learning_rate": 4.790062225191902e-06,
+      "loss": 0.0194,
+      "step": 2880
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 7.391280651092529,
+      "learning_rate": 4.787264734023481e-06,
+      "loss": 0.1331,
+      "step": 2881
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 1.7528316974639893,
+      "learning_rate": 4.784467309568524e-06,
+      "loss": 0.0085,
+      "step": 2882
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 1.2343730926513672,
+      "learning_rate": 4.781669952704299e-06,
+      "loss": 0.0088,
+      "step": 2883
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 0.2193189114332199,
+      "learning_rate": 4.778872664308057e-06,
+      "loss": 0.0046,
+      "step": 2884
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 0.6194412112236023,
+      "learning_rate": 4.77607544525702e-06,
+      "loss": 0.0071,
+      "step": 2885
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 9.382741928100586,
+      "learning_rate": 4.773278296428395e-06,
+      "loss": 0.0255,
+      "step": 2886
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 2.3163859844207764,
+      "learning_rate": 4.77048121869936e-06,
+      "loss": 0.0153,
+      "step": 2887
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 4.6898884773254395,
+      "learning_rate": 4.767684212947077e-06,
+      "loss": 0.0158,
+      "step": 2888
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 2.7231438159942627,
+      "learning_rate": 4.764887280048681e-06,
+      "loss": 0.0154,
+      "step": 2889
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 3.5279743671417236,
+      "learning_rate": 4.762090420881289e-06,
+      "loss": 0.0209,
+      "step": 2890
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 4.8647637367248535,
+      "learning_rate": 4.759293636321989e-06,
+      "loss": 0.0123,
+      "step": 2891
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 1.8748981952667236,
+      "learning_rate": 4.756496927247851e-06,
+      "loss": 0.0135,
+      "step": 2892
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.7683350443840027,
+      "learning_rate": 4.753700294535916e-06,
+      "loss": 0.0067,
+      "step": 2893
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 7.221288681030273,
+      "learning_rate": 4.750903739063206e-06,
+      "loss": 0.0572,
+      "step": 2894
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 3.297537088394165,
+      "learning_rate": 4.748107261706716e-06,
+      "loss": 0.0234,
+      "step": 2895
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 1.643818974494934,
+      "learning_rate": 4.745310863343417e-06,
+      "loss": 0.0087,
+      "step": 2896
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 6.469111919403076,
+      "learning_rate": 4.742514544850258e-06,
+      "loss": 0.1255,
+      "step": 2897
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 2.2265846729278564,
+      "learning_rate": 4.739718307104155e-06,
+      "loss": 0.0098,
+      "step": 2898
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 6.750699996948242,
+      "learning_rate": 4.736922150982008e-06,
+      "loss": 0.0296,
+      "step": 2899
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 2.472545862197876,
+      "learning_rate": 4.734126077360685e-06,
+      "loss": 0.0156,
+      "step": 2900
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 3.2475786209106445,
+      "learning_rate": 4.731330087117034e-06,
+      "loss": 0.0146,
+      "step": 2901
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 1.8263052701950073,
+      "learning_rate": 4.72853418112787e-06,
+      "loss": 0.0114,
+      "step": 2902
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 2.7932496070861816,
+      "learning_rate": 4.725738360269989e-06,
+      "loss": 0.0206,
+      "step": 2903
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 7.901010036468506,
+      "learning_rate": 4.7229426254201504e-06,
+      "loss": 0.1015,
+      "step": 2904
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 10.85585880279541,
+      "learning_rate": 4.720146977455098e-06,
+      "loss": 0.0659,
+      "step": 2905
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 3.7799477577209473,
+      "learning_rate": 4.71735141725154e-06,
+      "loss": 0.0151,
+      "step": 2906
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 11.781479835510254,
+      "learning_rate": 4.71455594568616e-06,
+      "loss": 0.0562,
+      "step": 2907
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 5.620331287384033,
+      "learning_rate": 4.711760563635617e-06,
+      "loss": 0.0842,
+      "step": 2908
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 10.326717376708984,
+      "learning_rate": 4.708965271976533e-06,
+      "loss": 0.06,
+      "step": 2909
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 1.262880563735962,
+      "learning_rate": 4.706170071585513e-06,
+      "loss": 0.0076,
+      "step": 2910
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 29.089683532714844,
+      "learning_rate": 4.703374963339123e-06,
+      "loss": 0.0656,
+      "step": 2911
+    },
+    {
+      "epoch": 5.18,
+      "grad_norm": 3.3051445484161377,
+      "learning_rate": 4.700579948113908e-06,
+      "loss": 0.0203,
+      "step": 2912
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 5.932816028594971,
+      "learning_rate": 4.697785026786379e-06,
+      "loss": 0.0144,
+      "step": 2913
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 3.2065038681030273,
+      "learning_rate": 4.694990200233022e-06,
+      "loss": 0.0153,
+      "step": 2914
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 11.653398513793945,
+      "learning_rate": 4.692195469330286e-06,
+      "loss": 0.0668,
+      "step": 2915
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 1.1626932621002197,
+      "learning_rate": 4.689400834954598e-06,
+      "loss": 0.0057,
+      "step": 2916
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 7.356791973114014,
+      "learning_rate": 4.6866062979823494e-06,
+      "loss": 0.1288,
+      "step": 2917
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 3.9244933128356934,
+      "learning_rate": 4.6838118592899045e-06,
+      "loss": 0.0208,
+      "step": 2918
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 14.007989883422852,
+      "learning_rate": 4.681017519753595e-06,
+      "loss": 0.0682,
+      "step": 2919
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 15.476340293884277,
+      "learning_rate": 4.678223280249718e-06,
+      "loss": 0.0349,
+      "step": 2920
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 12.090557098388672,
+      "learning_rate": 4.675429141654548e-06,
+      "loss": 0.0737,
+      "step": 2921
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 11.13563346862793,
+      "learning_rate": 4.672635104844319e-06,
+      "loss": 0.0667,
+      "step": 2922
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 7.828865051269531,
+      "learning_rate": 4.669841170695239e-06,
+      "loss": 0.0378,
+      "step": 2923
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 2.238959550857544,
+      "learning_rate": 4.667047340083481e-06,
+      "loss": 0.0114,
+      "step": 2924
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 5.894798755645752,
+      "learning_rate": 4.664253613885187e-06,
+      "loss": 0.0897,
+      "step": 2925
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 4.557458877563477,
+      "learning_rate": 4.661459992976463e-06,
+      "loss": 0.017,
+      "step": 2926
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 5.18538761138916,
+      "learning_rate": 4.658666478233388e-06,
+      "loss": 0.0252,
+      "step": 2927
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 2.478604316711426,
+      "learning_rate": 4.655873070531999e-06,
+      "loss": 0.0119,
+      "step": 2928
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 1.2854171991348267,
+      "learning_rate": 4.653079770748309e-06,
+      "loss": 0.0081,
+      "step": 2929
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 0.3432590663433075,
+      "learning_rate": 4.650286579758291e-06,
+      "loss": 0.0046,
+      "step": 2930
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 1.3417695760726929,
+      "learning_rate": 4.647493498437887e-06,
+      "loss": 0.0074,
+      "step": 2931
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 21.46624183654785,
+      "learning_rate": 4.644700527663002e-06,
+      "loss": 0.029,
+      "step": 2932
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 5.116425514221191,
+      "learning_rate": 4.641907668309506e-06,
+      "loss": 0.0401,
+      "step": 2933
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 3.0705060958862305,
+      "learning_rate": 4.639114921253238e-06,
+      "loss": 0.011,
+      "step": 2934
+    },
+    {
+      "epoch": 5.22,
+      "grad_norm": 9.141608238220215,
+      "learning_rate": 4.636322287369997e-06,
+      "loss": 0.1044,
+      "step": 2935
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 0.44009220600128174,
+      "learning_rate": 4.6335297675355525e-06,
+      "loss": 0.0053,
+      "step": 2936
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 11.009571075439453,
+      "learning_rate": 4.630737362625631e-06,
+      "loss": 0.0397,
+      "step": 2937
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 6.7480292320251465,
+      "learning_rate": 4.627945073515928e-06,
+      "loss": 0.0259,
+      "step": 2938
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 5.355017185211182,
+      "learning_rate": 4.625152901082101e-06,
+      "loss": 0.0202,
+      "step": 2939
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 1.574902057647705,
+      "learning_rate": 4.622360846199772e-06,
+      "loss": 0.0083,
+      "step": 2940
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 7.912002086639404,
+      "learning_rate": 4.619568909744524e-06,
+      "loss": 0.0513,
+      "step": 2941
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 1.6618866920471191,
+      "learning_rate": 4.616777092591907e-06,
+      "loss": 0.0089,
+      "step": 2942
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 9.620646476745605,
+      "learning_rate": 4.613985395617427e-06,
+      "loss": 0.0925,
+      "step": 2943
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 0.49985671043395996,
+      "learning_rate": 4.611193819696557e-06,
+      "loss": 0.0055,
+      "step": 2944
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 4.609509468078613,
+      "learning_rate": 4.608402365704734e-06,
+      "loss": 0.0137,
+      "step": 2945
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 6.1092143058776855,
+      "learning_rate": 4.605611034517348e-06,
+      "loss": 0.0442,
+      "step": 2946
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 10.232038497924805,
+      "learning_rate": 4.602819827009763e-06,
+      "loss": 0.0483,
+      "step": 2947
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 10.360949516296387,
+      "learning_rate": 4.6000287440572925e-06,
+      "loss": 0.0458,
+      "step": 2948
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 0.3105195462703705,
+      "learning_rate": 4.597237786535219e-06,
+      "loss": 0.0048,
+      "step": 2949
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 30.41156768798828,
+      "learning_rate": 4.594446955318781e-06,
+      "loss": 0.0559,
+      "step": 2950
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 5.6721367835998535,
+      "learning_rate": 4.5916562512831794e-06,
+      "loss": 0.019,
+      "step": 2951
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 0.9055770039558411,
+      "learning_rate": 4.588865675303575e-06,
+      "loss": 0.0084,
+      "step": 2952
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 13.359508514404297,
+      "learning_rate": 4.586075228255091e-06,
+      "loss": 0.0261,
+      "step": 2953
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 4.268441677093506,
+      "learning_rate": 4.583284911012803e-06,
+      "loss": 0.018,
+      "step": 2954
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 2.3447859287261963,
+      "learning_rate": 4.580494724451752e-06,
+      "loss": 0.0154,
+      "step": 2955
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 0.7731708884239197,
+      "learning_rate": 4.577704669446939e-06,
+      "loss": 0.0064,
+      "step": 2956
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 7.194380760192871,
+      "learning_rate": 4.5749147468733155e-06,
+      "loss": 0.0272,
+      "step": 2957
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 1.9442142248153687,
+      "learning_rate": 4.572124957605803e-06,
+      "loss": 0.0051,
+      "step": 2958
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 12.710718154907227,
+      "learning_rate": 4.569335302519271e-06,
+      "loss": 0.0171,
+      "step": 2959
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 2.487877368927002,
+      "learning_rate": 4.566545782488554e-06,
+      "loss": 0.0061,
+      "step": 2960
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 7.652891635894775,
+      "learning_rate": 4.563756398388439e-06,
+      "loss": 0.0231,
+      "step": 2961
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 8.442100524902344,
+      "learning_rate": 4.560967151093675e-06,
+      "loss": 0.0661,
+      "step": 2962
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 9.832866668701172,
+      "learning_rate": 4.558178041478964e-06,
+      "loss": 0.1648,
+      "step": 2963
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 6.281636714935303,
+      "learning_rate": 4.5553890704189704e-06,
+      "loss": 0.0281,
+      "step": 2964
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 3.533468723297119,
+      "learning_rate": 4.552600238788306e-06,
+      "loss": 0.0098,
+      "step": 2965
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 5.150492191314697,
+      "learning_rate": 4.549811547461548e-06,
+      "loss": 0.1028,
+      "step": 2966
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 19.28959846496582,
+      "learning_rate": 4.547022997313224e-06,
+      "loss": 0.0351,
+      "step": 2967
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 5.966936111450195,
+      "learning_rate": 4.54423458921782e-06,
+      "loss": 0.0388,
+      "step": 2968
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 6.619085788726807,
+      "learning_rate": 4.541446324049779e-06,
+      "loss": 0.0863,
+      "step": 2969
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 0.9254005551338196,
+      "learning_rate": 4.53865820268349e-06,
+      "loss": 0.0065,
+      "step": 2970
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 16.13799285888672,
+      "learning_rate": 4.535870225993312e-06,
+      "loss": 0.1069,
+      "step": 2971
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 4.677116870880127,
+      "learning_rate": 4.533082394853544e-06,
+      "loss": 0.0934,
+      "step": 2972
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 7.554133415222168,
+      "learning_rate": 4.53029471013845e-06,
+      "loss": 0.0357,
+      "step": 2973
+    },
+    {
+      "epoch": 5.29,
+      "grad_norm": 1.606514573097229,
+      "learning_rate": 4.527507172722241e-06,
+      "loss": 0.0119,
+      "step": 2974
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 4.885801792144775,
+      "learning_rate": 4.524719783479088e-06,
+      "loss": 0.014,
+      "step": 2975
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 4.343081474304199,
+      "learning_rate": 4.521932543283107e-06,
+      "loss": 0.0432,
+      "step": 2976
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 9.279769897460938,
+      "learning_rate": 4.519145453008378e-06,
+      "loss": 0.0113,
+      "step": 2977
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 6.086661338806152,
+      "learning_rate": 4.516358513528923e-06,
+      "loss": 0.028,
+      "step": 2978
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 0.6768794655799866,
+      "learning_rate": 4.513571725718727e-06,
+      "loss": 0.0065,
+      "step": 2979
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 7.197216987609863,
+      "learning_rate": 4.510785090451719e-06,
+      "loss": 0.038,
+      "step": 2980
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 3.5045995712280273,
+      "learning_rate": 4.507998608601787e-06,
+      "loss": 0.0139,
+      "step": 2981
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 6.454619884490967,
+      "learning_rate": 4.5052122810427655e-06,
+      "loss": 0.0244,
+      "step": 2982
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 0.8454774022102356,
+      "learning_rate": 4.502426108648441e-06,
+      "loss": 0.0066,
+      "step": 2983
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 2.858727216720581,
+      "learning_rate": 4.499640092292558e-06,
+      "loss": 0.0115,
+      "step": 2984
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 10.26382827758789,
+      "learning_rate": 4.4968542328488e-06,
+      "loss": 0.0313,
+      "step": 2985
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 7.934657573699951,
+      "learning_rate": 4.4940685311908155e-06,
+      "loss": 0.0139,
+      "step": 2986
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 3.515362501144409,
+      "learning_rate": 4.491282988192192e-06,
+      "loss": 0.0121,
+      "step": 2987
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 29.577497482299805,
+      "learning_rate": 4.4884976047264736e-06,
+      "loss": 0.0668,
+      "step": 2988
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 4.061966419219971,
+      "learning_rate": 4.485712381667152e-06,
+      "loss": 0.0827,
+      "step": 2989
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 7.68365478515625,
+      "learning_rate": 4.482927319887669e-06,
+      "loss": 0.0469,
+      "step": 2990
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 1.9097245931625366,
+      "learning_rate": 4.480142420261415e-06,
+      "loss": 0.0121,
+      "step": 2991
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 4.898718357086182,
+      "learning_rate": 4.477357683661734e-06,
+      "loss": 0.0359,
+      "step": 2992
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 8.903642654418945,
+      "learning_rate": 4.47457311096191e-06,
+      "loss": 0.0552,
+      "step": 2993
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 29.25267219543457,
+      "learning_rate": 4.471788703035184e-06,
+      "loss": 0.0263,
+      "step": 2994
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 4.304629802703857,
+      "learning_rate": 4.469004460754743e-06,
+      "loss": 0.0193,
+      "step": 2995
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 0.42354318499565125,
+      "learning_rate": 4.466220384993718e-06,
+      "loss": 0.0067,
+      "step": 2996
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 1.306233525276184,
+      "learning_rate": 4.463436476625195e-06,
+      "loss": 0.0067,
+      "step": 2997
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 13.254376411437988,
+      "learning_rate": 4.4606527365222e-06,
+      "loss": 0.0115,
+      "step": 2998
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 0.9308993816375732,
+      "learning_rate": 4.457869165557712e-06,
+      "loss": 0.0074,
+      "step": 2999
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 7.601993560791016,
+      "learning_rate": 4.455085764604653e-06,
+      "loss": 0.0182,
+      "step": 3000
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 6.207484245300293,
+      "learning_rate": 4.452302534535896e-06,
+      "loss": 0.0928,
+      "step": 3001
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 0.7907255291938782,
+      "learning_rate": 4.449519476224255e-06,
+      "loss": 0.0061,
+      "step": 3002
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 7.393033027648926,
+      "learning_rate": 4.446736590542497e-06,
+      "loss": 0.0221,
+      "step": 3003
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 1.6112476587295532,
+      "learning_rate": 4.443953878363325e-06,
+      "loss": 0.0121,
+      "step": 3004
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 19.294822692871094,
+      "learning_rate": 4.441171340559399e-06,
+      "loss": 0.0376,
+      "step": 3005
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 4.080855369567871,
+      "learning_rate": 4.4383889780033155e-06,
+      "loss": 0.0239,
+      "step": 3006
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 0.514409065246582,
+      "learning_rate": 4.43560679156762e-06,
+      "loss": 0.0054,
+      "step": 3007
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 13.382828712463379,
+      "learning_rate": 4.432824782124804e-06,
+      "loss": 0.0301,
+      "step": 3008
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.9480242729187012,
+      "learning_rate": 4.430042950547298e-06,
+      "loss": 0.0066,
+      "step": 3009
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 4.6385040283203125,
+      "learning_rate": 4.427261297707482e-06,
+      "loss": 0.0175,
+      "step": 3010
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.4864596128463745,
+      "learning_rate": 4.4244798244776785e-06,
+      "loss": 0.0058,
+      "step": 3011
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 0.7241849899291992,
+      "learning_rate": 4.4216985317301545e-06,
+      "loss": 0.0071,
+      "step": 3012
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 4.744707107543945,
+      "learning_rate": 4.418917420337115e-06,
+      "loss": 0.0252,
+      "step": 3013
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 38.527984619140625,
+      "learning_rate": 4.41613649117072e-06,
+      "loss": 0.1863,
+      "step": 3014
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 7.461682319641113,
+      "learning_rate": 4.413355745103057e-06,
+      "loss": 0.0207,
+      "step": 3015
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 8.728968620300293,
+      "learning_rate": 4.410575183006168e-06,
+      "loss": 0.0288,
+      "step": 3016
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 10.876944541931152,
+      "learning_rate": 4.4077948057520315e-06,
+      "loss": 0.1015,
+      "step": 3017
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 2.547455310821533,
+      "learning_rate": 4.4050146142125714e-06,
+      "loss": 0.0225,
+      "step": 3018
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 3.195225954055786,
+      "learning_rate": 4.402234609259652e-06,
+      "loss": 0.0118,
+      "step": 3019
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 7.228540897369385,
+      "learning_rate": 4.399454791765076e-06,
+      "loss": 0.0173,
+      "step": 3020
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 22.580270767211914,
+      "learning_rate": 4.396675162600591e-06,
+      "loss": 0.0166,
+      "step": 3021
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 14.562980651855469,
+      "learning_rate": 4.393895722637885e-06,
+      "loss": 0.0944,
+      "step": 3022
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 0.5571821928024292,
+      "learning_rate": 4.39111647274859e-06,
+      "loss": 0.0054,
+      "step": 3023
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 0.2470635622739792,
+      "learning_rate": 4.3883374138042655e-06,
+      "loss": 0.0048,
+      "step": 3024
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 5.594621658325195,
+      "learning_rate": 4.3855585466764305e-06,
+      "loss": 0.0564,
+      "step": 3025
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 1.8275127410888672,
+      "learning_rate": 4.382779872236527e-06,
+      "loss": 0.0123,
+      "step": 3026
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 19.060928344726562,
+      "learning_rate": 4.380001391355946e-06,
+      "loss": 0.1254,
+      "step": 3027
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 2.35212779045105,
+      "learning_rate": 4.377223104906013e-06,
+      "loss": 0.094,
+      "step": 3028
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 6.824979305267334,
+      "learning_rate": 4.374445013757998e-06,
+      "loss": 0.0209,
+      "step": 3029
+    },
+    {
+      "epoch": 5.39,
+      "grad_norm": 1.945109486579895,
+      "learning_rate": 4.371667118783101e-06,
+      "loss": 0.0076,
+      "step": 3030
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 9.746783256530762,
+      "learning_rate": 4.3688894208524725e-06,
+      "loss": 0.1011,
+      "step": 3031
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 9.363571166992188,
+      "learning_rate": 4.366111920837189e-06,
+      "loss": 0.0581,
+      "step": 3032
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 2.7941741943359375,
+      "learning_rate": 4.363334619608272e-06,
+      "loss": 0.0091,
+      "step": 3033
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 13.223176002502441,
+      "learning_rate": 4.36055751803668e-06,
+      "loss": 0.0325,
+      "step": 3034
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 7.201384544372559,
+      "learning_rate": 4.357780616993305e-06,
+      "loss": 0.0613,
+      "step": 3035
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 8.571467399597168,
+      "learning_rate": 4.355003917348985e-06,
+      "loss": 0.0373,
+      "step": 3036
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 20.103792190551758,
+      "learning_rate": 4.352227419974483e-06,
+      "loss": 0.1503,
+      "step": 3037
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 1.0826451778411865,
+      "learning_rate": 4.349451125740507e-06,
+      "loss": 0.0066,
+      "step": 3038
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 0.2576778829097748,
+      "learning_rate": 4.346675035517698e-06,
+      "loss": 0.0042,
+      "step": 3039
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 1.7643768787384033,
+      "learning_rate": 4.343899150176635e-06,
+      "loss": 0.0087,
+      "step": 3040
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 4.956659317016602,
+      "learning_rate": 4.34112347058783e-06,
+      "loss": 0.0168,
+      "step": 3041
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 2.179227352142334,
+      "learning_rate": 4.338347997621735e-06,
+      "loss": 0.0114,
+      "step": 3042
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 5.4347381591796875,
+      "learning_rate": 4.33557273214873e-06,
+      "loss": 0.0195,
+      "step": 3043
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 11.137736320495605,
+      "learning_rate": 4.3327976750391376e-06,
+      "loss": 0.0389,
+      "step": 3044
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 7.524304389953613,
+      "learning_rate": 4.3300228271632105e-06,
+      "loss": 0.0531,
+      "step": 3045
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 1.6475909948349,
+      "learning_rate": 4.3272481893911345e-06,
+      "loss": 0.0085,
+      "step": 3046
+    },
+    {
+      "epoch": 5.42,
+      "grad_norm": 1.2443203926086426,
+      "learning_rate": 4.324473762593037e-06,
+      "loss": 0.0087,
+      "step": 3047
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 5.003077030181885,
+      "learning_rate": 4.321699547638968e-06,
+      "loss": 0.0157,
+      "step": 3048
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 4.975095748901367,
+      "learning_rate": 4.318925545398923e-06,
+      "loss": 0.0335,
+      "step": 3049
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 13.602434158325195,
+      "learning_rate": 4.316151756742821e-06,
+      "loss": 0.0577,
+      "step": 3050
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 1.5983210802078247,
+      "learning_rate": 4.313378182540521e-06,
+      "loss": 0.0078,
+      "step": 3051
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 2.0449740886688232,
+      "learning_rate": 4.310604823661808e-06,
+      "loss": 0.0116,
+      "step": 3052
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 0.27072277665138245,
+      "learning_rate": 4.307831680976409e-06,
+      "loss": 0.0031,
+      "step": 3053
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 1.4098490476608276,
+      "learning_rate": 4.30505875535397e-06,
+      "loss": 0.0088,
+      "step": 3054
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 6.435232639312744,
+      "learning_rate": 4.302286047664083e-06,
+      "loss": 0.0541,
+      "step": 3055
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 8.448822975158691,
+      "learning_rate": 4.299513558776261e-06,
+      "loss": 0.0362,
+      "step": 3056
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 0.5037113428115845,
+      "learning_rate": 4.296741289559955e-06,
+      "loss": 0.0045,
+      "step": 3057
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 3.304753541946411,
+      "learning_rate": 4.293969240884545e-06,
+      "loss": 0.0211,
+      "step": 3058
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 6.899008274078369,
+      "learning_rate": 4.2911974136193364e-06,
+      "loss": 0.0207,
+      "step": 3059
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 0.43501126766204834,
+      "learning_rate": 4.2884258086335755e-06,
+      "loss": 0.0051,
+      "step": 3060
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 3.236902952194214,
+      "learning_rate": 4.28565442679643e-06,
+      "loss": 0.0095,
+      "step": 3061
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 5.151862621307373,
+      "learning_rate": 4.282883268977005e-06,
+      "loss": 0.0247,
+      "step": 3062
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 14.261865615844727,
+      "learning_rate": 4.280112336044326e-06,
+      "loss": 0.2213,
+      "step": 3063
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 5.435333728790283,
+      "learning_rate": 4.2773416288673595e-06,
+      "loss": 0.0291,
+      "step": 3064
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 1.3885453939437866,
+      "learning_rate": 4.274571148314991e-06,
+      "loss": 0.0108,
+      "step": 3065
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 10.406073570251465,
+      "learning_rate": 4.271800895256041e-06,
+      "loss": 0.0522,
+      "step": 3066
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 0.41064420342445374,
+      "learning_rate": 4.269030870559256e-06,
+      "loss": 0.005,
+      "step": 3067
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 7.533734321594238,
+      "learning_rate": 4.266261075093312e-06,
+      "loss": 0.0359,
+      "step": 3068
+    },
+    {
+      "epoch": 5.46,
+      "grad_norm": 3.569594383239746,
+      "learning_rate": 4.263491509726812e-06,
+      "loss": 0.0137,
+      "step": 3069
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 21.590696334838867,
+      "learning_rate": 4.26072217532829e-06,
+      "loss": 0.13,
+      "step": 3070
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 13.952478408813477,
+      "learning_rate": 4.257953072766203e-06,
+      "loss": 0.1932,
+      "step": 3071
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 9.996973991394043,
+      "learning_rate": 4.255184202908937e-06,
+      "loss": 0.0359,
+      "step": 3072
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 2.616981029510498,
+      "learning_rate": 4.252415566624809e-06,
+      "loss": 0.0133,
+      "step": 3073
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 11.211518287658691,
+      "learning_rate": 4.249647164782052e-06,
+      "loss": 0.1107,
+      "step": 3074
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 2.6415328979492188,
+      "learning_rate": 4.2468789982488415e-06,
+      "loss": 0.0177,
+      "step": 3075
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 0.9873169660568237,
+      "learning_rate": 4.244111067893264e-06,
+      "loss": 0.0077,
+      "step": 3076
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 2.77599835395813,
+      "learning_rate": 4.241343374583343e-06,
+      "loss": 0.0077,
+      "step": 3077
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 4.522502422332764,
+      "learning_rate": 4.238575919187019e-06,
+      "loss": 0.0149,
+      "step": 3078
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 13.320879936218262,
+      "learning_rate": 4.235808702572164e-06,
+      "loss": 0.0733,
+      "step": 3079
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 3.6741559505462646,
+      "learning_rate": 4.233041725606573e-06,
+      "loss": 0.0088,
+      "step": 3080
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 1.0996060371398926,
+      "learning_rate": 4.230274989157967e-06,
+      "loss": 0.0073,
+      "step": 3081
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 4.639238357543945,
+      "learning_rate": 4.227508494093988e-06,
+      "loss": 0.03,
+      "step": 3082
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 6.663539409637451,
+      "learning_rate": 4.224742241282206e-06,
+      "loss": 0.0459,
+      "step": 3083
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 1.308127522468567,
+      "learning_rate": 4.2219762315901146e-06,
+      "loss": 0.0094,
+      "step": 3084
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 6.144577503204346,
+      "learning_rate": 4.219210465885127e-06,
+      "loss": 0.0188,
+      "step": 3085
+    },
+    {
+      "epoch": 5.49,
+      "grad_norm": 7.55708646774292,
+      "learning_rate": 4.216444945034589e-06,
+      "loss": 0.1638,
+      "step": 3086
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 1.3362748622894287,
+      "learning_rate": 4.213679669905758e-06,
+      "loss": 0.009,
+      "step": 3087
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 4.557662487030029,
+      "learning_rate": 4.210914641365824e-06,
+      "loss": 0.0078,
+      "step": 3088
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 9.635598182678223,
+      "learning_rate": 4.208149860281894e-06,
+      "loss": 0.1004,
+      "step": 3089
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 15.055778503417969,
+      "learning_rate": 4.205385327521002e-06,
+      "loss": 0.0507,
+      "step": 3090
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 1.0495803356170654,
+      "learning_rate": 4.202621043950096e-06,
+      "loss": 0.0076,
+      "step": 3091
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 0.8096566200256348,
+      "learning_rate": 4.1998570104360576e-06,
+      "loss": 0.0043,
+      "step": 3092
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 16.117124557495117,
+      "learning_rate": 4.197093227845679e-06,
+      "loss": 0.0585,
+      "step": 3093
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 4.371682643890381,
+      "learning_rate": 4.194329697045681e-06,
+      "loss": 0.0145,
+      "step": 3094
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 3.3942480087280273,
+      "learning_rate": 4.191566418902701e-06,
+      "loss": 0.0926,
+      "step": 3095
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 7.581375598907471,
+      "learning_rate": 4.1888033942832985e-06,
+      "loss": 0.0104,
+      "step": 3096
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 0.8671513199806213,
+      "learning_rate": 4.186040624053958e-06,
+      "loss": 0.0076,
+      "step": 3097
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 7.040656566619873,
+      "learning_rate": 4.183278109081075e-06,
+      "loss": 0.0688,
+      "step": 3098
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 12.789340019226074,
+      "learning_rate": 4.180515850230974e-06,
+      "loss": 0.0394,
+      "step": 3099
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 1.891412615776062,
+      "learning_rate": 4.177753848369892e-06,
+      "loss": 0.0154,
+      "step": 3100
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 3.723687171936035,
+      "learning_rate": 4.174992104363993e-06,
+      "loss": 0.0174,
+      "step": 3101
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 0.21531502902507782,
+      "learning_rate": 4.17223061907935e-06,
+      "loss": 0.0035,
+      "step": 3102
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 7.024179935455322,
+      "learning_rate": 4.169469393381968e-06,
+      "loss": 0.0226,
+      "step": 3103
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 1.5251938104629517,
+      "learning_rate": 4.1667084281377554e-06,
+      "loss": 0.0108,
+      "step": 3104
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 1.9741802215576172,
+      "learning_rate": 4.163947724212553e-06,
+      "loss": 0.0079,
+      "step": 3105
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 15.232251167297363,
+      "learning_rate": 4.16118728247211e-06,
+      "loss": 0.1181,
+      "step": 3106
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 4.226377010345459,
+      "learning_rate": 4.158427103782099e-06,
+      "loss": 0.027,
+      "step": 3107
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 1.7914578914642334,
+      "learning_rate": 4.155667189008108e-06,
+      "loss": 0.0082,
+      "step": 3108
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 11.6802396774292,
+      "learning_rate": 4.15290753901564e-06,
+      "loss": 0.033,
+      "step": 3109
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 10.51911449432373,
+      "learning_rate": 4.1501481546701185e-06,
+      "loss": 0.0633,
+      "step": 3110
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 6.4429802894592285,
+      "learning_rate": 4.147389036836881e-06,
+      "loss": 0.0144,
+      "step": 3111
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 5.510497570037842,
+      "learning_rate": 4.144630186381187e-06,
+      "loss": 0.0278,
+      "step": 3112
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 6.589241027832031,
+      "learning_rate": 4.141871604168201e-06,
+      "loss": 0.1051,
+      "step": 3113
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 3.638979434967041,
+      "learning_rate": 4.139113291063019e-06,
+      "loss": 0.0137,
+      "step": 3114
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 0.46482717990875244,
+      "learning_rate": 4.136355247930636e-06,
+      "loss": 0.0052,
+      "step": 3115
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 1.6225818395614624,
+      "learning_rate": 4.133597475635975e-06,
+      "loss": 0.0077,
+      "step": 3116
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 2.3030502796173096,
+      "learning_rate": 4.130839975043866e-06,
+      "loss": 0.0145,
+      "step": 3117
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 3.662914514541626,
+      "learning_rate": 4.128082747019061e-06,
+      "loss": 0.0154,
+      "step": 3118
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 4.000531196594238,
+      "learning_rate": 4.125325792426218e-06,
+      "loss": 0.0215,
+      "step": 3119
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 0.701994001865387,
+      "learning_rate": 4.12256911212992e-06,
+      "loss": 0.0052,
+      "step": 3120
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 4.184095859527588,
+      "learning_rate": 4.119812706994654e-06,
+      "loss": 0.0225,
+      "step": 3121
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 0.35647228360176086,
+      "learning_rate": 4.117056577884822e-06,
+      "loss": 0.0038,
+      "step": 3122
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 0.6790317296981812,
+      "learning_rate": 4.114300725664749e-06,
+      "loss": 0.0049,
+      "step": 3123
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 0.9610785841941833,
+      "learning_rate": 4.111545151198657e-06,
+      "loss": 0.0046,
+      "step": 3124
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 9.039873123168945,
+      "learning_rate": 4.108789855350699e-06,
+      "loss": 0.1026,
+      "step": 3125
+    },
+    {
+      "epoch": 5.56,
+      "grad_norm": 2.0333786010742188,
+      "learning_rate": 4.106034838984926e-06,
+      "loss": 0.012,
+      "step": 3126
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 11.582330703735352,
+      "learning_rate": 4.10328010296531e-06,
+      "loss": 0.0453,
+      "step": 3127
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 12.281033515930176,
+      "learning_rate": 4.100525648155731e-06,
+      "loss": 0.0571,
+      "step": 3128
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 10.511531829833984,
+      "learning_rate": 4.0977714754199834e-06,
+      "loss": 0.0395,
+      "step": 3129
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 1.0815774202346802,
+      "learning_rate": 4.095017585621767e-06,
+      "loss": 0.0076,
+      "step": 3130
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 2.2818872928619385,
+      "learning_rate": 4.092263979624705e-06,
+      "loss": 0.0118,
+      "step": 3131
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 3.5232996940612793,
+      "learning_rate": 4.089510658292318e-06,
+      "loss": 0.0872,
+      "step": 3132
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 2.3113527297973633,
+      "learning_rate": 4.086757622488046e-06,
+      "loss": 0.0087,
+      "step": 3133
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 21.18972396850586,
+      "learning_rate": 4.084004873075238e-06,
+      "loss": 0.1222,
+      "step": 3134
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 9.379136085510254,
+      "learning_rate": 4.081252410917148e-06,
+      "loss": 0.0233,
+      "step": 3135
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 1.2989639043807983,
+      "learning_rate": 4.078500236876951e-06,
+      "loss": 0.0066,
+      "step": 3136
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 5.594966411590576,
+      "learning_rate": 4.075748351817717e-06,
+      "loss": 0.023,
+      "step": 3137
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 0.4272444546222687,
+      "learning_rate": 4.072996756602438e-06,
+      "loss": 0.0053,
+      "step": 3138
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 10.132466316223145,
+      "learning_rate": 4.0702454520940085e-06,
+      "loss": 0.0553,
+      "step": 3139
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 6.897375583648682,
+      "learning_rate": 4.067494439155236e-06,
+      "loss": 0.0322,
+      "step": 3140
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 3.279681921005249,
+      "learning_rate": 4.064743718648828e-06,
+      "loss": 0.1439,
+      "step": 3141
+    },
+    {
+      "epoch": 5.59,
+      "grad_norm": 4.131958484649658,
+      "learning_rate": 4.061993291437415e-06,
+      "loss": 0.1506,
+      "step": 3142
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 7.7318220138549805,
+      "learning_rate": 4.059243158383519e-06,
+      "loss": 0.0293,
+      "step": 3143
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 7.522507190704346,
+      "learning_rate": 4.056493320349583e-06,
+      "loss": 0.0168,
+      "step": 3144
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.9558770656585693,
+      "learning_rate": 4.053743778197951e-06,
+      "loss": 0.0075,
+      "step": 3145
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 17.022613525390625,
+      "learning_rate": 4.050994532790871e-06,
+      "loss": 0.0809,
+      "step": 3146
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.25400209426879883,
+      "learning_rate": 4.048245584990509e-06,
+      "loss": 0.004,
+      "step": 3147
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 2.7277350425720215,
+      "learning_rate": 4.0454969356589265e-06,
+      "loss": 0.0111,
+      "step": 3148
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 4.057364463806152,
+      "learning_rate": 4.042748585658099e-06,
+      "loss": 0.0254,
+      "step": 3149
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 3.999810218811035,
+      "learning_rate": 4.0400005358499e-06,
+      "loss": 0.0106,
+      "step": 3150
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 0.6298880577087402,
+      "learning_rate": 4.037252787096121e-06,
+      "loss": 0.0062,
+      "step": 3151
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 10.421594619750977,
+      "learning_rate": 4.034505340258444e-06,
+      "loss": 0.1353,
+      "step": 3152
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 6.203615188598633,
+      "learning_rate": 4.031758196198471e-06,
+      "loss": 0.0844,
+      "step": 3153
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 6.753244400024414,
+      "learning_rate": 4.029011355777698e-06,
+      "loss": 0.0183,
+      "step": 3154
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 13.250808715820312,
+      "learning_rate": 4.026264819857533e-06,
+      "loss": 0.0424,
+      "step": 3155
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 0.2509465217590332,
+      "learning_rate": 4.023518589299283e-06,
+      "loss": 0.004,
+      "step": 3156
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 0.3963233530521393,
+      "learning_rate": 4.020772664964166e-06,
+      "loss": 0.0037,
+      "step": 3157
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 3.466156482696533,
+      "learning_rate": 4.018027047713296e-06,
+      "loss": 0.0125,
+      "step": 3158
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 10.74256420135498,
+      "learning_rate": 4.015281738407694e-06,
+      "loss": 0.0653,
+      "step": 3159
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 0.6151906251907349,
+      "learning_rate": 4.012536737908288e-06,
+      "loss": 0.0038,
+      "step": 3160
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 11.528644561767578,
+      "learning_rate": 4.009792047075903e-06,
+      "loss": 0.1395,
+      "step": 3161
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 11.325761795043945,
+      "learning_rate": 4.007047666771274e-06,
+      "loss": 0.0232,
+      "step": 3162
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 5.415079116821289,
+      "learning_rate": 4.004303597855029e-06,
+      "loss": 0.021,
+      "step": 3163
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 4.048696041107178,
+      "learning_rate": 4.001559841187711e-06,
+      "loss": 0.0132,
+      "step": 3164
+    },
+    {
+      "epoch": 5.63,
+      "grad_norm": 3.589747667312622,
+      "learning_rate": 3.998816397629752e-06,
+      "loss": 0.0152,
+      "step": 3165
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 10.95584487915039,
+      "learning_rate": 3.996073268041494e-06,
+      "loss": 0.023,
+      "step": 3166
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 24.226438522338867,
+      "learning_rate": 3.993330453283178e-06,
+      "loss": 0.0624,
+      "step": 3167
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 6.432476997375488,
+      "learning_rate": 3.99058795421495e-06,
+      "loss": 0.0932,
+      "step": 3168
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 12.196669578552246,
+      "learning_rate": 3.9878457716968475e-06,
+      "loss": 0.1043,
+      "step": 3169
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 3.11474871635437,
+      "learning_rate": 3.985103906588821e-06,
+      "loss": 0.0137,
+      "step": 3170
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 11.783096313476562,
+      "learning_rate": 3.9823623597507115e-06,
+      "loss": 0.0393,
+      "step": 3171
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 11.002062797546387,
+      "learning_rate": 3.979621132042264e-06,
+      "loss": 0.0241,
+      "step": 3172
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 22.669658660888672,
+      "learning_rate": 3.976880224323126e-06,
+      "loss": 0.1204,
+      "step": 3173
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 0.6093739867210388,
+      "learning_rate": 3.9741396374528385e-06,
+      "loss": 0.0053,
+      "step": 3174
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 6.095942974090576,
+      "learning_rate": 3.971399372290851e-06,
+      "loss": 0.0318,
+      "step": 3175
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 2.985624074935913,
+      "learning_rate": 3.968659429696501e-06,
+      "loss": 0.0095,
+      "step": 3176
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 7.241685390472412,
+      "learning_rate": 3.965919810529034e-06,
+      "loss": 0.0111,
+      "step": 3177
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 14.395182609558105,
+      "learning_rate": 3.963180515647587e-06,
+      "loss": 0.0993,
+      "step": 3178
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 1.9134804010391235,
+      "learning_rate": 3.960441545911205e-06,
+      "loss": 0.007,
+      "step": 3179
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 1.0754880905151367,
+      "learning_rate": 3.957702902178816e-06,
+      "loss": 0.0075,
+      "step": 3180
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 16.6334285736084,
+      "learning_rate": 3.9549645853092644e-06,
+      "loss": 0.0237,
+      "step": 3181
+    },
+    {
+      "epoch": 5.66,
+      "grad_norm": 3.7156412601470947,
+      "learning_rate": 3.952226596161275e-06,
+      "loss": 0.0119,
+      "step": 3182
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 5.4813551902771,
+      "learning_rate": 3.949488935593481e-06,
+      "loss": 0.0183,
+      "step": 3183
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 1.6109819412231445,
+      "learning_rate": 3.946751604464408e-06,
+      "loss": 0.0094,
+      "step": 3184
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 10.058215141296387,
+      "learning_rate": 3.944014603632476e-06,
+      "loss": 0.0379,
+      "step": 3185
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 0.8709433674812317,
+      "learning_rate": 3.941277933956009e-06,
+      "loss": 0.0062,
+      "step": 3186
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 4.163586139678955,
+      "learning_rate": 3.9385415962932184e-06,
+      "loss": 0.0165,
+      "step": 3187
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 19.757957458496094,
+      "learning_rate": 3.9358055915022195e-06,
+      "loss": 0.2221,
+      "step": 3188
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 11.940670013427734,
+      "learning_rate": 3.933069920441016e-06,
+      "loss": 0.2114,
+      "step": 3189
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 27.949064254760742,
+      "learning_rate": 3.930334583967514e-06,
+      "loss": 0.0482,
+      "step": 3190
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 4.719934463500977,
+      "learning_rate": 3.927599582939507e-06,
+      "loss": 0.0333,
+      "step": 3191
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 1.8781254291534424,
+      "learning_rate": 3.92486491821469e-06,
+      "loss": 0.0065,
+      "step": 3192
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 6.13599157333374,
+      "learning_rate": 3.922130590650649e-06,
+      "loss": 0.0451,
+      "step": 3193
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 0.8611292839050293,
+      "learning_rate": 3.919396601104864e-06,
+      "loss": 0.006,
+      "step": 3194
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 1.2676491737365723,
+      "learning_rate": 3.916662950434714e-06,
+      "loss": 0.0084,
+      "step": 3195
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 0.3905896842479706,
+      "learning_rate": 3.913929639497462e-06,
+      "loss": 0.0049,
+      "step": 3196
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 0.680129885673523,
+      "learning_rate": 3.911196669150275e-06,
+      "loss": 0.0071,
+      "step": 3197
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 1.5073635578155518,
+      "learning_rate": 3.908464040250205e-06,
+      "loss": 0.0068,
+      "step": 3198
+    },
+    {
+      "epoch": 5.69,
+      "grad_norm": 6.638657093048096,
+      "learning_rate": 3.905731753654204e-06,
+      "loss": 0.0386,
+      "step": 3199
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 3.0773699283599854,
+      "learning_rate": 3.902999810219109e-06,
+      "loss": 0.0121,
+      "step": 3200
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 4.284664630889893,
+      "learning_rate": 3.9002682108016585e-06,
+      "loss": 0.0518,
+      "step": 3201
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 20.06857681274414,
+      "learning_rate": 3.8975369562584714e-06,
+      "loss": 0.0654,
+      "step": 3202
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 11.702192306518555,
+      "learning_rate": 3.894806047446072e-06,
+      "loss": 0.0648,
+      "step": 3203
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 6.13560676574707,
+      "learning_rate": 3.892075485220865e-06,
+      "loss": 0.0173,
+      "step": 3204
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 1.1685351133346558,
+      "learning_rate": 3.889345270439152e-06,
+      "loss": 0.0073,
+      "step": 3205
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 28.330968856811523,
+      "learning_rate": 3.886615403957124e-06,
+      "loss": 0.1141,
+      "step": 3206
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 0.8932600021362305,
+      "learning_rate": 3.883885886630863e-06,
+      "loss": 0.006,
+      "step": 3207
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 3.7747061252593994,
+      "learning_rate": 3.8811567193163435e-06,
+      "loss": 0.018,
+      "step": 3208
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 1.6706373691558838,
+      "learning_rate": 3.878427902869424e-06,
+      "loss": 0.0098,
+      "step": 3209
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 17.291275024414062,
+      "learning_rate": 3.875699438145862e-06,
+      "loss": 0.1975,
+      "step": 3210
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 7.880677223205566,
+      "learning_rate": 3.872971326001297e-06,
+      "loss": 0.1253,
+      "step": 3211
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 12.836236000061035,
+      "learning_rate": 3.870243567291263e-06,
+      "loss": 0.1313,
+      "step": 3212
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 7.391240119934082,
+      "learning_rate": 3.867516162871177e-06,
+      "loss": 0.0462,
+      "step": 3213
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 4.499868869781494,
+      "learning_rate": 3.8647891135963565e-06,
+      "loss": 0.014,
+      "step": 3214
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 0.9906122088432312,
+      "learning_rate": 3.862062420321993e-06,
+      "loss": 0.0067,
+      "step": 3215
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 15.029112815856934,
+      "learning_rate": 3.859336083903177e-06,
+      "loss": 0.0836,
+      "step": 3216
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 1.3181432485580444,
+      "learning_rate": 3.856610105194882e-06,
+      "loss": 0.0079,
+      "step": 3217
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 13.37888240814209,
+      "learning_rate": 3.853884485051974e-06,
+      "loss": 0.0467,
+      "step": 3218
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 0.5053179860115051,
+      "learning_rate": 3.851159224329198e-06,
+      "loss": 0.0045,
+      "step": 3219
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 6.548157691955566,
+      "learning_rate": 3.8484343238811976e-06,
+      "loss": 0.027,
+      "step": 3220
+    },
+    {
+      "epoch": 5.73,
+      "grad_norm": 6.2808756828308105,
+      "learning_rate": 3.845709784562493e-06,
+      "loss": 0.0117,
+      "step": 3221
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 1.022453784942627,
+      "learning_rate": 3.842985607227497e-06,
+      "loss": 0.0078,
+      "step": 3222
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 15.117263793945312,
+      "learning_rate": 3.840261792730511e-06,
+      "loss": 0.0714,
+      "step": 3223
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 20.592336654663086,
+      "learning_rate": 3.837538341925713e-06,
+      "loss": 0.0359,
+      "step": 3224
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 10.69933032989502,
+      "learning_rate": 3.834815255667179e-06,
+      "loss": 0.2598,
+      "step": 3225
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 9.357650756835938,
+      "learning_rate": 3.832092534808861e-06,
+      "loss": 0.0677,
+      "step": 3226
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 9.830259323120117,
+      "learning_rate": 3.8293701802046025e-06,
+      "loss": 0.1644,
+      "step": 3227
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 9.49650764465332,
+      "learning_rate": 3.826648192708127e-06,
+      "loss": 0.0841,
+      "step": 3228
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 8.590238571166992,
+      "learning_rate": 3.823926573173051e-06,
+      "loss": 0.0335,
+      "step": 3229
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 9.233003616333008,
+      "learning_rate": 3.821205322452863e-06,
+      "loss": 0.0404,
+      "step": 3230
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 5.256519794464111,
+      "learning_rate": 3.818484441400951e-06,
+      "loss": 0.0124,
+      "step": 3231
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 2.6738791465759277,
+      "learning_rate": 3.815763930870574e-06,
+      "loss": 0.0174,
+      "step": 3232
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 5.61312198638916,
+      "learning_rate": 3.813043791714882e-06,
+      "loss": 0.017,
+      "step": 3233
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 2.8102102279663086,
+      "learning_rate": 3.8103240247869077e-06,
+      "loss": 0.0113,
+      "step": 3234
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 5.5402655601501465,
+      "learning_rate": 3.8076046309395627e-06,
+      "loss": 0.0146,
+      "step": 3235
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 10.580025672912598,
+      "learning_rate": 3.8048856110256475e-06,
+      "loss": 0.1915,
+      "step": 3236
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 4.756521224975586,
+      "learning_rate": 3.802166965897841e-06,
+      "loss": 0.0115,
+      "step": 3237
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 9.86764907836914,
+      "learning_rate": 3.7994486964087087e-06,
+      "loss": 0.0631,
+      "step": 3238
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 7.6868743896484375,
+      "learning_rate": 3.7967308034106933e-06,
+      "loss": 0.0278,
+      "step": 3239
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 37.86629867553711,
+      "learning_rate": 3.794013287756125e-06,
+      "loss": 0.1272,
+      "step": 3240
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 3.699517011642456,
+      "learning_rate": 3.791296150297209e-06,
+      "loss": 0.0196,
+      "step": 3241
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 0.45253846049308777,
+      "learning_rate": 3.7885793918860404e-06,
+      "loss": 0.0049,
+      "step": 3242
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 8.679898262023926,
+      "learning_rate": 3.785863013374585e-06,
+      "loss": 0.0357,
+      "step": 3243
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 21.255889892578125,
+      "learning_rate": 3.7831470156146992e-06,
+      "loss": 0.2025,
+      "step": 3244
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 8.794122695922852,
+      "learning_rate": 3.7804313994581143e-06,
+      "loss": 0.1213,
+      "step": 3245
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 12.479826927185059,
+      "learning_rate": 3.7777161657564452e-06,
+      "loss": 0.0605,
+      "step": 3246
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 8.304293632507324,
+      "learning_rate": 3.775001315361183e-06,
+      "loss": 0.025,
+      "step": 3247
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 0.33791854977607727,
+      "learning_rate": 3.7722868491237e-06,
+      "loss": 0.0036,
+      "step": 3248
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 3.7195181846618652,
+      "learning_rate": 3.7695727678952527e-06,
+      "loss": 0.0161,
+      "step": 3249
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 7.297631740570068,
+      "learning_rate": 3.766859072526969e-06,
+      "loss": 0.0736,
+      "step": 3250
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 7.411331653594971,
+      "learning_rate": 3.7641457638698632e-06,
+      "loss": 0.018,
+      "step": 3251
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 4.079013824462891,
+      "learning_rate": 3.7614328427748204e-06,
+      "loss": 0.0162,
+      "step": 3252
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 11.752937316894531,
+      "learning_rate": 3.758720310092614e-06,
+      "loss": 0.0463,
+      "step": 3253
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 0.555255115032196,
+      "learning_rate": 3.756008166673886e-06,
+      "loss": 0.005,
+      "step": 3254
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 1.1109877824783325,
+      "learning_rate": 3.7532964133691634e-06,
+      "loss": 0.0061,
+      "step": 3255
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 0.8529955148696899,
+      "learning_rate": 3.7505850510288455e-06,
+      "loss": 0.0063,
+      "step": 3256
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 8.886727333068848,
+      "learning_rate": 3.7478740805032156e-06,
+      "loss": 0.0354,
+      "step": 3257
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 9.054100036621094,
+      "learning_rate": 3.745163502642426e-06,
+      "loss": 0.0937,
+      "step": 3258
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 1.5699316263198853,
+      "learning_rate": 3.7424533182965094e-06,
+      "loss": 0.0092,
+      "step": 3259
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 10.259729385375977,
+      "learning_rate": 3.7397435283153795e-06,
+      "loss": 0.097,
+      "step": 3260
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 5.402533531188965,
+      "learning_rate": 3.7370341335488193e-06,
+      "loss": 0.0153,
+      "step": 3261
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 2.8133490085601807,
+      "learning_rate": 3.7343251348464937e-06,
+      "loss": 0.0107,
+      "step": 3262
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 5.641959190368652,
+      "learning_rate": 3.731616533057937e-06,
+      "loss": 0.0228,
+      "step": 3263
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 1.1667739152908325,
+      "learning_rate": 3.7289083290325668e-06,
+      "loss": 0.0075,
+      "step": 3264
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 4.648532390594482,
+      "learning_rate": 3.726200523619668e-06,
+      "loss": 0.0178,
+      "step": 3265
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 28.618602752685547,
+      "learning_rate": 3.7234931176684076e-06,
+      "loss": 0.1068,
+      "step": 3266
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 1.0919698476791382,
+      "learning_rate": 3.720786112027822e-06,
+      "loss": 0.0056,
+      "step": 3267
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 0.7564353346824646,
+      "learning_rate": 3.7180795075468266e-06,
+      "loss": 0.0061,
+      "step": 3268
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 1.9278502464294434,
+      "learning_rate": 3.715373305074205e-06,
+      "loss": 0.0069,
+      "step": 3269
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 0.9137459397315979,
+      "learning_rate": 3.712667505458622e-06,
+      "loss": 0.0064,
+      "step": 3270
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 1.2438137531280518,
+      "learning_rate": 3.7099621095486107e-06,
+      "loss": 0.0079,
+      "step": 3271
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 1.436510443687439,
+      "learning_rate": 3.7072571181925778e-06,
+      "loss": 0.0077,
+      "step": 3272
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 0.7877283692359924,
+      "learning_rate": 3.7045525322388076e-06,
+      "loss": 0.0056,
+      "step": 3273
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 2.679215669631958,
+      "learning_rate": 3.701848352535451e-06,
+      "loss": 0.0142,
+      "step": 3274
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 10.482553482055664,
+      "learning_rate": 3.6991445799305376e-06,
+      "loss": 0.0994,
+      "step": 3275
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 7.972808361053467,
+      "learning_rate": 3.696441215271964e-06,
+      "loss": 0.1037,
+      "step": 3276
+    },
+    {
+      "epoch": 5.83,
+      "grad_norm": 9.933003425598145,
+      "learning_rate": 3.6937382594075045e-06,
+      "loss": 0.023,
+      "step": 3277
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 12.992663383483887,
+      "learning_rate": 3.6910357131847986e-06,
+      "loss": 0.1255,
+      "step": 3278
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 6.983480453491211,
+      "learning_rate": 3.6883335774513644e-06,
+      "loss": 0.0437,
+      "step": 3279
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 13.951624870300293,
+      "learning_rate": 3.685631853054583e-06,
+      "loss": 0.1823,
+      "step": 3280
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 7.585327625274658,
+      "learning_rate": 3.682930540841717e-06,
+      "loss": 0.1218,
+      "step": 3281
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 8.526784896850586,
+      "learning_rate": 3.6802296416598893e-06,
+      "loss": 0.0158,
+      "step": 3282
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 1.7102826833724976,
+      "learning_rate": 3.6775291563561e-06,
+      "loss": 0.008,
+      "step": 3283
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 2.8827550411224365,
+      "learning_rate": 3.6748290857772183e-06,
+      "loss": 0.0087,
+      "step": 3284
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 1.6113271713256836,
+      "learning_rate": 3.6721294307699786e-06,
+      "loss": 0.0057,
+      "step": 3285
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 0.8327566981315613,
+      "learning_rate": 3.6694301921809917e-06,
+      "loss": 0.006,
+      "step": 3286
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 0.208288311958313,
+      "learning_rate": 3.6667313708567337e-06,
+      "loss": 0.0034,
+      "step": 3287
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 7.669879913330078,
+      "learning_rate": 3.664032967643552e-06,
+      "loss": 0.0461,
+      "step": 3288
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 2.14546275138855,
+      "learning_rate": 3.6613349833876607e-06,
+      "loss": 0.0985,
+      "step": 3289
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 2.5684449672698975,
+      "learning_rate": 3.658637418935146e-06,
+      "loss": 0.0156,
+      "step": 3290
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 3.2200441360473633,
+      "learning_rate": 3.655940275131955e-06,
+      "loss": 0.0166,
+      "step": 3291
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 4.327977657318115,
+      "learning_rate": 3.6532435528239153e-06,
+      "loss": 0.0163,
+      "step": 3292
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 6.53366756439209,
+      "learning_rate": 3.650547252856708e-06,
+      "loss": 0.0373,
+      "step": 3293
+    },
+    {
+      "epoch": 5.86,
+      "grad_norm": 3.1286063194274902,
+      "learning_rate": 3.6478513760758938e-06,
+      "loss": 0.0108,
+      "step": 3294
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 3.601699113845825,
+      "learning_rate": 3.645155923326893e-06,
+      "loss": 0.1475,
+      "step": 3295
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 8.553617477416992,
+      "learning_rate": 3.6424608954549977e-06,
+      "loss": 0.0324,
+      "step": 3296
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 1.333288550376892,
+      "learning_rate": 3.639766293305363e-06,
+      "loss": 0.0059,
+      "step": 3297
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 0.8069380521774292,
+      "learning_rate": 3.637072117723012e-06,
+      "loss": 0.0046,
+      "step": 3298
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 0.9580917358398438,
+      "learning_rate": 3.6343783695528343e-06,
+      "loss": 0.0049,
+      "step": 3299
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 0.8078251481056213,
+      "learning_rate": 3.6316850496395863e-06,
+      "loss": 0.0054,
+      "step": 3300
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 7.415097236633301,
+      "learning_rate": 3.6289921588278896e-06,
+      "loss": 0.0272,
+      "step": 3301
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 9.078073501586914,
+      "learning_rate": 3.6262996979622286e-06,
+      "loss": 0.1443,
+      "step": 3302
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 11.67236042022705,
+      "learning_rate": 3.623607667886957e-06,
+      "loss": 0.0257,
+      "step": 3303
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 0.40362921357154846,
+      "learning_rate": 3.62091606944629e-06,
+      "loss": 0.0031,
+      "step": 3304
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 10.868256568908691,
+      "learning_rate": 3.61822490348431e-06,
+      "loss": 0.0456,
+      "step": 3305
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 11.86292552947998,
+      "learning_rate": 3.6155341708449615e-06,
+      "loss": 0.0526,
+      "step": 3306
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 3.739980697631836,
+      "learning_rate": 3.612843872372057e-06,
+      "loss": 0.0098,
+      "step": 3307
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 1.0641306638717651,
+      "learning_rate": 3.6101540089092655e-06,
+      "loss": 0.0064,
+      "step": 3308
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 22.345069885253906,
+      "learning_rate": 3.607464581300131e-06,
+      "loss": 0.1096,
+      "step": 3309
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 1.2245696783065796,
+      "learning_rate": 3.6047755903880478e-06,
+      "loss": 0.0059,
+      "step": 3310
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 4.380024433135986,
+      "learning_rate": 3.602087037016281e-06,
+      "loss": 0.0094,
+      "step": 3311
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 14.796142578125,
+      "learning_rate": 3.59939892202796e-06,
+      "loss": 0.1734,
+      "step": 3312
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 11.782308578491211,
+      "learning_rate": 3.5967112462660693e-06,
+      "loss": 0.0436,
+      "step": 3313
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 9.295053482055664,
+      "learning_rate": 3.5940240105734624e-06,
+      "loss": 0.1153,
+      "step": 3314
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 1.603499412536621,
+      "learning_rate": 3.5913372157928515e-06,
+      "loss": 0.0066,
+      "step": 3315
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 1.402740478515625,
+      "learning_rate": 3.5886508627668137e-06,
+      "loss": 0.0065,
+      "step": 3316
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 5.5805888175964355,
+      "learning_rate": 3.5859649523377827e-06,
+      "loss": 0.0168,
+      "step": 3317
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 5.7353315353393555,
+      "learning_rate": 3.583279485348059e-06,
+      "loss": 0.1481,
+      "step": 3318
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 1.08493173122406,
+      "learning_rate": 3.580594462639796e-06,
+      "loss": 0.0046,
+      "step": 3319
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 12.422683715820312,
+      "learning_rate": 3.577909885055019e-06,
+      "loss": 0.1218,
+      "step": 3320
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 6.536487102508545,
+      "learning_rate": 3.575225753435604e-06,
+      "loss": 0.0414,
+      "step": 3321
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 1.4267656803131104,
+      "learning_rate": 3.5725420686232903e-06,
+      "loss": 0.0046,
+      "step": 3322
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 4.949660301208496,
+      "learning_rate": 3.5698588314596803e-06,
+      "loss": 0.0409,
+      "step": 3323
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 9.60941219329834,
+      "learning_rate": 3.56717604278623e-06,
+      "loss": 0.0311,
+      "step": 3324
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 6.353866100311279,
+      "learning_rate": 3.56449370344426e-06,
+      "loss": 0.0145,
+      "step": 3325
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 6.283907890319824,
+      "learning_rate": 3.561811814274947e-06,
+      "loss": 0.0808,
+      "step": 3326
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 11.975065231323242,
+      "learning_rate": 3.5591303761193287e-06,
+      "loss": 0.033,
+      "step": 3327
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 15.531221389770508,
+      "learning_rate": 3.5564493898182976e-06,
+      "loss": 0.1168,
+      "step": 3328
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 14.0263671875,
+      "learning_rate": 3.5537688562126115e-06,
+      "loss": 0.1086,
+      "step": 3329
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 6.333539009094238,
+      "learning_rate": 3.5510887761428764e-06,
+      "loss": 0.021,
+      "step": 3330
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 11.94481086730957,
+      "learning_rate": 3.5484091504495667e-06,
+      "loss": 0.1655,
+      "step": 3331
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 6.357314586639404,
+      "learning_rate": 3.5457299799730047e-06,
+      "loss": 0.0819,
+      "step": 3332
+    },
+    {
+      "epoch": 5.93,
+      "grad_norm": 0.4023033380508423,
+      "learning_rate": 3.5430512655533774e-06,
+      "loss": 0.0048,
+      "step": 3333
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 3.723346471786499,
+      "learning_rate": 3.540373008030725e-06,
+      "loss": 0.0194,
+      "step": 3334
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 12.900154113769531,
+      "learning_rate": 3.5376952082449425e-06,
+      "loss": 0.0813,
+      "step": 3335
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 2.1727397441864014,
+      "learning_rate": 3.5350178670357873e-06,
+      "loss": 0.0128,
+      "step": 3336
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 9.685309410095215,
+      "learning_rate": 3.5323409852428667e-06,
+      "loss": 0.1419,
+      "step": 3337
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 11.391621589660645,
+      "learning_rate": 3.52966456370565e-06,
+      "loss": 0.0474,
+      "step": 3338
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 0.5889759063720703,
+      "learning_rate": 3.526988603263457e-06,
+      "loss": 0.0041,
+      "step": 3339
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 1.5406478643417358,
+      "learning_rate": 3.524313104755468e-06,
+      "loss": 0.0062,
+      "step": 3340
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 1.318064570426941,
+      "learning_rate": 3.5216380690207098e-06,
+      "loss": 0.0061,
+      "step": 3341
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 5.284785270690918,
+      "learning_rate": 3.518963496898074e-06,
+      "loss": 0.0219,
+      "step": 3342
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 7.006436347961426,
+      "learning_rate": 3.5162893892263006e-06,
+      "loss": 0.0631,
+      "step": 3343
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 4.556286811828613,
+      "learning_rate": 3.513615746843987e-06,
+      "loss": 0.009,
+      "step": 3344
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 7.543275833129883,
+      "learning_rate": 3.510942570589583e-06,
+      "loss": 0.07,
+      "step": 3345
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 1.4481792449951172,
+      "learning_rate": 3.508269861301393e-06,
+      "loss": 0.0075,
+      "step": 3346
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 0.29023146629333496,
+      "learning_rate": 3.5055976198175747e-06,
+      "loss": 0.0035,
+      "step": 3347
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 10.710103034973145,
+      "learning_rate": 3.502925846976137e-06,
+      "loss": 0.0515,
+      "step": 3348
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 1.3541955947875977,
+      "learning_rate": 3.5002545436149478e-06,
+      "loss": 0.0061,
+      "step": 3349
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 0.9578431844711304,
+      "learning_rate": 3.4975837105717203e-06,
+      "loss": 0.0063,
+      "step": 3350
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 1.8527026176452637,
+      "learning_rate": 3.494913348684027e-06,
+      "loss": 0.007,
+      "step": 3351
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 4.511735439300537,
+      "learning_rate": 3.492243458789286e-06,
+      "loss": 0.0434,
+      "step": 3352
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 38.76997756958008,
+      "learning_rate": 3.4895740417247737e-06,
+      "loss": 0.1412,
+      "step": 3353
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 7.172122001647949,
+      "learning_rate": 3.486905098327613e-06,
+      "loss": 0.0257,
+      "step": 3354
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 5.521944046020508,
+      "learning_rate": 3.484236629434783e-06,
+      "loss": 0.0459,
+      "step": 3355
+    },
+    {
+      "epoch": 5.97,
+      "grad_norm": 4.459950923919678,
+      "learning_rate": 3.481568635883109e-06,
+      "loss": 0.0168,
+      "step": 3356
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 16.559486389160156,
+      "learning_rate": 3.4789011185092746e-06,
+      "loss": 0.1327,
+      "step": 3357
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 10.520103454589844,
+      "learning_rate": 3.4762340781498016e-06,
+      "loss": 0.0348,
+      "step": 3358
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 8.427156448364258,
+      "learning_rate": 3.4735675156410787e-06,
+      "loss": 0.0148,
+      "step": 3359
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 1.6250112056732178,
+      "learning_rate": 3.4709014318193298e-06,
+      "loss": 0.0698,
+      "step": 3360
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 6.689555644989014,
+      "learning_rate": 3.4682358275206353e-06,
+      "loss": 0.0282,
+      "step": 3361
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 12.303206443786621,
+      "learning_rate": 3.4655707035809287e-06,
+      "loss": 0.0685,
+      "step": 3362
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 4.018043041229248,
+      "learning_rate": 3.4629060608359845e-06,
+      "loss": 0.0233,
+      "step": 3363
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 3.9291880130767822,
+      "learning_rate": 3.460241900121434e-06,
+      "loss": 0.0141,
+      "step": 3364
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 0.36454322934150696,
+      "learning_rate": 3.4575782222727507e-06,
+      "loss": 0.0039,
+      "step": 3365
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 3.365579128265381,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.1061,
+      "step": 3366
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 3.9354300498962402,
+      "learning_rate": 3.452252318514144e-06,
+      "loss": 0.0313,
+      "step": 3367
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.6507599949836731,
+      "learning_rate": 3.4495900942744165e-06,
+      "loss": 0.0057,
+      "step": 3368
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 2.7286109924316406,
+      "learning_rate": 3.446928356240946e-06,
+      "loss": 0.0332,
+      "step": 3369
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 10.784978866577148,
+      "learning_rate": 3.4442671052484545e-06,
+      "loss": 0.0363,
+      "step": 3370
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.8642366613840465,
+      "eval_loss": 0.6670230031013489,
+      "eval_runtime": 11.4418,
+      "eval_samples_per_second": 165.446,
+      "eval_steps_per_second": 20.714,
+      "step": 3370
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5610,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 2.3939730804781875e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}