{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 3415,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0014654698662758747,
      "grad_norm": 3.7154194217547345,
      "learning_rate": 0.0,
      "loss": 1.1894,
      "num_tokens": 3198322.0,
      "step": 1
    },
    {
      "epoch": 0.0029309397325517493,
      "grad_norm": 3.8159096754478,
      "learning_rate": 3.883495145631068e-07,
      "loss": 1.1959,
      "num_tokens": 6267889.0,
      "step": 2
    },
    {
      "epoch": 0.004396409598827624,
      "grad_norm": 3.831445269987945,
      "learning_rate": 7.766990291262136e-07,
      "loss": 1.237,
      "num_tokens": 9183528.0,
      "step": 3
    },
    {
      "epoch": 0.005861879465103499,
      "grad_norm": 3.756428669522261,
      "learning_rate": 1.1650485436893206e-06,
      "loss": 1.2097,
      "num_tokens": 12005997.0,
      "step": 4
    },
    {
      "epoch": 0.007327349331379373,
      "grad_norm": 3.7049281074937817,
      "learning_rate": 1.5533980582524272e-06,
      "loss": 1.2168,
      "num_tokens": 14867401.0,
      "step": 5
    },
    {
      "epoch": 0.008792819197655248,
      "grad_norm": 3.2899890256573694,
      "learning_rate": 1.941747572815534e-06,
      "loss": 1.2075,
      "num_tokens": 17926912.0,
      "step": 6
    },
    {
      "epoch": 0.010258289063931123,
      "grad_norm": 3.0589726023363477,
      "learning_rate": 2.330097087378641e-06,
      "loss": 1.183,
      "num_tokens": 20972782.0,
      "step": 7
    },
    {
      "epoch": 0.011723758930206997,
      "grad_norm": 2.290087708190892,
      "learning_rate": 2.718446601941748e-06,
      "loss": 1.1912,
      "num_tokens": 23831942.0,
      "step": 8
    },
    {
      "epoch": 0.013189228796482872,
      "grad_norm": 2.205407335694658,
      "learning_rate": 3.1067961165048544e-06,
      "loss": 1.1797,
      "num_tokens": 26808849.0,
      "step": 9
    },
    {
      "epoch": 0.014654698662758746,
      "grad_norm": 2.1597756695335093,
      "learning_rate": 3.4951456310679615e-06,
      "loss": 1.1746,
      "num_tokens": 29800658.0,
      "step": 10
    },
    {
      "epoch": 0.016120168529034622,
      "grad_norm": 3.2808602516524483,
      "learning_rate": 3.883495145631068e-06,
      "loss": 1.1718,
      "num_tokens": 32709194.0,
      "step": 11
    },
    {
      "epoch": 0.017585638395310497,
      "grad_norm": 3.6239058889427964,
      "learning_rate": 4.271844660194175e-06,
      "loss": 1.1736,
      "num_tokens": 35774224.0,
      "step": 12
    },
    {
      "epoch": 0.01905110826158637,
      "grad_norm": 3.2224743488745724,
      "learning_rate": 4.660194174757282e-06,
      "loss": 1.1688,
      "num_tokens": 38705681.0,
      "step": 13
    },
    {
      "epoch": 0.020516578127862246,
      "grad_norm": 2.188551653370355,
      "learning_rate": 5.048543689320389e-06,
      "loss": 1.1255,
      "num_tokens": 41755116.0,
      "step": 14
    },
    {
      "epoch": 0.02198204799413812,
      "grad_norm": 2.455305133554404,
      "learning_rate": 5.436893203883496e-06,
      "loss": 1.1036,
      "num_tokens": 44857316.0,
      "step": 15
    },
    {
      "epoch": 0.023447517860413995,
      "grad_norm": 2.6182275949629936,
      "learning_rate": 5.825242718446602e-06,
      "loss": 1.1218,
      "num_tokens": 47585617.0,
      "step": 16
    },
    {
      "epoch": 0.02491298772668987,
      "grad_norm": 6.20879928353618,
      "learning_rate": 6.213592233009709e-06,
      "loss": 1.0892,
      "num_tokens": 50672319.0,
      "step": 17
    },
    {
      "epoch": 0.026378457592965743,
      "grad_norm": 2.209383359381836,
      "learning_rate": 6.601941747572816e-06,
      "loss": 1.0662,
      "num_tokens": 53726652.0,
      "step": 18
    },
    {
      "epoch": 0.027843927459241618,
      "grad_norm": 2.2063531579858724,
      "learning_rate": 6.990291262135923e-06,
      "loss": 1.0671,
      "num_tokens": 56942695.0,
      "step": 19
    },
    {
      "epoch": 0.029309397325517492,
      "grad_norm": 1.6889738500376674,
      "learning_rate": 7.37864077669903e-06,
      "loss": 1.0496,
      "num_tokens": 60025080.0,
      "step": 20
    },
    {
      "epoch": 0.03077486719179337,
      "grad_norm": 5.050365716432631,
      "learning_rate": 7.766990291262136e-06,
      "loss": 1.0,
      "num_tokens": 63190690.0,
      "step": 21
    },
    {
      "epoch": 0.032240337058069245,
      "grad_norm": 1.3669283816660662,
      "learning_rate": 8.155339805825243e-06,
      "loss": 0.9853,
      "num_tokens": 66159787.0,
      "step": 22
    },
    {
      "epoch": 0.03370580692434512,
      "grad_norm": 1.2365578960470651,
      "learning_rate": 8.54368932038835e-06,
      "loss": 0.9868,
      "num_tokens": 69372254.0,
      "step": 23
    },
    {
      "epoch": 0.035171276790620994,
      "grad_norm": 1.004333292885707,
      "learning_rate": 8.932038834951458e-06,
      "loss": 0.9841,
      "num_tokens": 72158598.0,
      "step": 24
    },
    {
      "epoch": 0.03663674665689687,
      "grad_norm": 0.8552509734385357,
      "learning_rate": 9.320388349514565e-06,
      "loss": 0.9954,
      "num_tokens": 75199277.0,
      "step": 25
    },
    {
      "epoch": 0.03810221652317274,
      "grad_norm": 0.8435970179653405,
      "learning_rate": 9.708737864077671e-06,
      "loss": 0.9613,
      "num_tokens": 78505230.0,
      "step": 26
    },
    {
      "epoch": 0.03956768638944862,
      "grad_norm": 0.7965616227829095,
      "learning_rate": 1.0097087378640778e-05,
      "loss": 0.9647,
      "num_tokens": 81528695.0,
      "step": 27
    },
    {
      "epoch": 0.04103315625572449,
      "grad_norm": 0.6986712691976343,
      "learning_rate": 1.0485436893203885e-05,
      "loss": 0.9534,
      "num_tokens": 84629623.0,
      "step": 28
    },
    {
      "epoch": 0.042498626122000366,
      "grad_norm": 0.6499330969978084,
      "learning_rate": 1.0873786407766991e-05,
      "loss": 0.9424,
      "num_tokens": 87822492.0,
      "step": 29
    },
    {
      "epoch": 0.04396409598827624,
      "grad_norm": 0.5704142147162633,
      "learning_rate": 1.1262135922330098e-05,
      "loss": 0.9647,
      "num_tokens": 91028813.0,
      "step": 30
    },
    {
      "epoch": 0.045429565854552115,
      "grad_norm": 0.5687018137715497,
      "learning_rate": 1.1650485436893204e-05,
      "loss": 0.9564,
      "num_tokens": 94210012.0,
      "step": 31
    },
    {
      "epoch": 0.04689503572082799,
      "grad_norm": 0.5057684258625181,
      "learning_rate": 1.2038834951456311e-05,
      "loss": 0.9535,
      "num_tokens": 97431648.0,
      "step": 32
    },
    {
      "epoch": 0.048360505587103864,
      "grad_norm": 0.5368187773607731,
      "learning_rate": 1.2427184466019418e-05,
      "loss": 0.9419,
      "num_tokens": 100240083.0,
      "step": 33
    },
    {
      "epoch": 0.04982597545337974,
      "grad_norm": 0.521600938283564,
      "learning_rate": 1.2815533980582526e-05,
      "loss": 0.9342,
      "num_tokens": 103326466.0,
      "step": 34
    },
    {
      "epoch": 0.05129144531965561,
      "grad_norm": 0.48855050271178957,
      "learning_rate": 1.3203883495145633e-05,
      "loss": 0.9254,
      "num_tokens": 106214597.0,
      "step": 35
    },
    {
      "epoch": 0.05275691518593149,
      "grad_norm": 0.455737524836322,
      "learning_rate": 1.359223300970874e-05,
      "loss": 0.9206,
      "num_tokens": 109354828.0,
      "step": 36
    },
    {
      "epoch": 0.05422238505220736,
      "grad_norm": 0.508354078655406,
      "learning_rate": 1.3980582524271846e-05,
      "loss": 0.926,
      "num_tokens": 112421290.0,
      "step": 37
    },
    {
      "epoch": 0.055687854918483236,
      "grad_norm": 0.4786981791275268,
      "learning_rate": 1.4368932038834953e-05,
      "loss": 0.9123,
      "num_tokens": 115343538.0,
      "step": 38
    },
    {
      "epoch": 0.05715332478475911,
      "grad_norm": 0.46294861923447456,
      "learning_rate": 1.475728155339806e-05,
      "loss": 0.9113,
      "num_tokens": 118367356.0,
      "step": 39
    },
    {
      "epoch": 0.058618794651034985,
      "grad_norm": 0.5295701920957653,
      "learning_rate": 1.5145631067961166e-05,
      "loss": 0.9226,
      "num_tokens": 121358192.0,
      "step": 40
    },
    {
      "epoch": 0.060084264517310866,
      "grad_norm": 0.49306899051866787,
      "learning_rate": 1.5533980582524273e-05,
      "loss": 0.9091,
      "num_tokens": 124534846.0,
      "step": 41
    },
    {
      "epoch": 0.06154973438358674,
      "grad_norm": 0.46334592033574157,
      "learning_rate": 1.592233009708738e-05,
      "loss": 0.911,
      "num_tokens": 127636604.0,
      "step": 42
    },
    {
      "epoch": 0.06301520424986261,
      "grad_norm": 0.5062020274630149,
      "learning_rate": 1.6310679611650486e-05,
      "loss": 0.9267,
      "num_tokens": 130520629.0,
      "step": 43
    },
    {
      "epoch": 0.06448067411613849,
      "grad_norm": 0.44534995182779014,
      "learning_rate": 1.6699029126213594e-05,
      "loss": 0.9073,
      "num_tokens": 133517640.0,
      "step": 44
    },
    {
      "epoch": 0.06594614398241436,
      "grad_norm": 0.4864222301298436,
      "learning_rate": 1.70873786407767e-05,
      "loss": 0.915,
      "num_tokens": 136872123.0,
      "step": 45
    },
    {
      "epoch": 0.06741161384869024,
      "grad_norm": 0.43845702373502604,
      "learning_rate": 1.7475728155339808e-05,
      "loss": 0.9046,
      "num_tokens": 139899803.0,
      "step": 46
    },
    {
      "epoch": 0.0688770837149661,
      "grad_norm": 0.48413509568112467,
      "learning_rate": 1.7864077669902916e-05,
      "loss": 0.9004,
      "num_tokens": 143029695.0,
      "step": 47
    },
    {
      "epoch": 0.07034255358124199,
      "grad_norm": 0.4979005902336758,
      "learning_rate": 1.825242718446602e-05,
      "loss": 0.9081,
      "num_tokens": 146136067.0,
      "step": 48
    },
    {
      "epoch": 0.07180802344751785,
      "grad_norm": 0.48474843137917467,
      "learning_rate": 1.864077669902913e-05,
      "loss": 0.8975,
      "num_tokens": 149134152.0,
      "step": 49
    },
    {
      "epoch": 0.07327349331379374,
      "grad_norm": 0.5520629070878562,
      "learning_rate": 1.9029126213592234e-05,
      "loss": 0.9135,
      "num_tokens": 152171548.0,
      "step": 50
    },
    {
      "epoch": 0.0747389631800696,
      "grad_norm": 0.41162920146592064,
      "learning_rate": 1.9417475728155343e-05,
      "loss": 0.8686,
      "num_tokens": 155385344.0,
      "step": 51
    },
    {
      "epoch": 0.07620443304634548,
      "grad_norm": 0.46665397621944266,
      "learning_rate": 1.9805825242718447e-05,
      "loss": 0.8828,
      "num_tokens": 158641919.0,
      "step": 52
    },
    {
      "epoch": 0.07766990291262135,
      "grad_norm": 0.49616593118930535,
      "learning_rate": 2.0194174757281556e-05,
      "loss": 0.8942,
      "num_tokens": 161685973.0,
      "step": 53
    },
    {
      "epoch": 0.07913537277889723,
      "grad_norm": 0.4689210013321571,
      "learning_rate": 2.058252427184466e-05,
      "loss": 0.8968,
      "num_tokens": 164656830.0,
      "step": 54
    },
    {
      "epoch": 0.08060084264517312,
      "grad_norm": 0.5546915841150304,
      "learning_rate": 2.097087378640777e-05,
      "loss": 0.8843,
      "num_tokens": 167703016.0,
      "step": 55
    },
    {
      "epoch": 0.08206631251144898,
      "grad_norm": 0.5253100741527619,
      "learning_rate": 2.1359223300970874e-05,
      "loss": 0.8954,
      "num_tokens": 170862180.0,
      "step": 56
    },
    {
      "epoch": 0.08353178237772486,
      "grad_norm": 0.4871450938003483,
      "learning_rate": 2.1747572815533982e-05,
      "loss": 0.8791,
      "num_tokens": 173771441.0,
      "step": 57
    },
    {
      "epoch": 0.08499725224400073,
      "grad_norm": 0.7115493810638506,
      "learning_rate": 2.2135922330097087e-05,
      "loss": 0.9003,
      "num_tokens": 176699325.0,
      "step": 58
    },
    {
      "epoch": 0.08646272211027661,
      "grad_norm": 0.5788787720979568,
      "learning_rate": 2.2524271844660196e-05,
      "loss": 0.8965,
      "num_tokens": 179878886.0,
      "step": 59
    },
    {
      "epoch": 0.08792819197655248,
      "grad_norm": 0.7218463417280757,
      "learning_rate": 2.29126213592233e-05,
      "loss": 0.872,
      "num_tokens": 182854320.0,
      "step": 60
    },
    {
      "epoch": 0.08939366184282836,
      "grad_norm": 0.6405516461924528,
      "learning_rate": 2.330097087378641e-05,
      "loss": 0.8908,
      "num_tokens": 185743356.0,
      "step": 61
    },
    {
      "epoch": 0.09085913170910423,
      "grad_norm": 0.7642503850701262,
      "learning_rate": 2.3689320388349514e-05,
      "loss": 0.8984,
      "num_tokens": 188709848.0,
      "step": 62
    },
    {
      "epoch": 0.09232460157538011,
      "grad_norm": 0.5282646333755459,
      "learning_rate": 2.4077669902912622e-05,
      "loss": 0.895,
      "num_tokens": 192129742.0,
      "step": 63
    },
    {
      "epoch": 0.09379007144165598,
      "grad_norm": 0.7703478916276936,
      "learning_rate": 2.4466019417475727e-05,
      "loss": 0.8567,
      "num_tokens": 195114755.0,
      "step": 64
    },
    {
      "epoch": 0.09525554130793186,
      "grad_norm": 0.5769241566237074,
      "learning_rate": 2.4854368932038836e-05,
      "loss": 0.8861,
      "num_tokens": 198140939.0,
      "step": 65
    },
    {
      "epoch": 0.09672101117420773,
      "grad_norm": 0.7287654485975635,
      "learning_rate": 2.5242718446601947e-05,
      "loss": 0.8627,
      "num_tokens": 201019131.0,
      "step": 66
    },
    {
      "epoch": 0.09818648104048361,
      "grad_norm": 0.6390322091922136,
      "learning_rate": 2.5631067961165052e-05,
      "loss": 0.8757,
      "num_tokens": 204203222.0,
      "step": 67
    },
    {
      "epoch": 0.09965195090675948,
      "grad_norm": 0.6642161473425475,
      "learning_rate": 2.601941747572816e-05,
      "loss": 0.8699,
      "num_tokens": 207301837.0,
      "step": 68
    },
    {
      "epoch": 0.10111742077303536,
      "grad_norm": 0.6093151136946741,
      "learning_rate": 2.6407766990291266e-05,
      "loss": 0.8684,
      "num_tokens": 210327220.0,
      "step": 69
    },
    {
      "epoch": 0.10258289063931122,
      "grad_norm": 0.6789884618682116,
      "learning_rate": 2.6796116504854374e-05,
      "loss": 0.8644,
      "num_tokens": 213155631.0,
      "step": 70
    },
    {
      "epoch": 0.1040483605055871,
      "grad_norm": 0.5795597272375321,
      "learning_rate": 2.718446601941748e-05,
      "loss": 0.8469,
      "num_tokens": 216233990.0,
      "step": 71
    },
    {
      "epoch": 0.10551383037186297,
      "grad_norm": 0.6394234581377922,
      "learning_rate": 2.7572815533980587e-05,
      "loss": 0.876,
      "num_tokens": 219210949.0,
      "step": 72
    },
    {
      "epoch": 0.10697930023813886,
      "grad_norm": 0.6103946781659355,
      "learning_rate": 2.7961165048543692e-05,
      "loss": 0.8733,
      "num_tokens": 222342989.0,
      "step": 73
    },
    {
      "epoch": 0.10844477010441472,
      "grad_norm": 0.8026490815081038,
      "learning_rate": 2.83495145631068e-05,
      "loss": 0.8632,
      "num_tokens": 225683535.0,
      "step": 74
    },
    {
      "epoch": 0.1099102399706906,
      "grad_norm": 0.6498093598726404,
      "learning_rate": 2.8737864077669905e-05,
      "loss": 0.8549,
      "num_tokens": 228993739.0,
      "step": 75
    },
    {
      "epoch": 0.11137570983696647,
      "grad_norm": 0.6796573878122779,
      "learning_rate": 2.9126213592233014e-05,
      "loss": 0.8701,
      "num_tokens": 232307768.0,
      "step": 76
    },
    {
      "epoch": 0.11284117970324235,
      "grad_norm": 0.5991007871834552,
      "learning_rate": 2.951456310679612e-05,
      "loss": 0.8778,
      "num_tokens": 235284000.0,
      "step": 77
    },
    {
      "epoch": 0.11430664956951822,
      "grad_norm": 0.6225913604825317,
      "learning_rate": 2.9902912621359227e-05,
      "loss": 0.8496,
      "num_tokens": 238369671.0,
      "step": 78
    },
    {
      "epoch": 0.1157721194357941,
      "grad_norm": 0.805492420861617,
      "learning_rate": 3.0291262135922332e-05,
      "loss": 0.8519,
      "num_tokens": 241594371.0,
      "step": 79
    },
    {
      "epoch": 0.11723758930206997,
      "grad_norm": 0.7420095626227796,
      "learning_rate": 3.067961165048544e-05,
      "loss": 0.8547,
      "num_tokens": 244622780.0,
      "step": 80
    },
    {
      "epoch": 0.11870305916834585,
      "grad_norm": 0.6037490765727215,
      "learning_rate": 3.1067961165048545e-05,
      "loss": 0.8685,
      "num_tokens": 247756320.0,
      "step": 81
    },
    {
      "epoch": 0.12016852903462173,
      "grad_norm": 0.7097892538296358,
      "learning_rate": 3.1456310679611654e-05,
      "loss": 0.8499,
      "num_tokens": 250888113.0,
      "step": 82
    },
    {
      "epoch": 0.1216339989008976,
      "grad_norm": 0.6423028146339036,
      "learning_rate": 3.184466019417476e-05,
      "loss": 0.8774,
      "num_tokens": 253982723.0,
      "step": 83
    },
    {
      "epoch": 0.12309946876717348,
      "grad_norm": 0.7476981768912153,
      "learning_rate": 3.2233009708737864e-05,
      "loss": 0.8595,
      "num_tokens": 257284764.0,
      "step": 84
    },
    {
      "epoch": 0.12456493863344935,
      "grad_norm": 0.6000248431150975,
      "learning_rate": 3.262135922330097e-05,
      "loss": 0.8564,
      "num_tokens": 260479493.0,
      "step": 85
    },
    {
      "epoch": 0.12603040849972522,
      "grad_norm": 0.7338133703090208,
      "learning_rate": 3.300970873786408e-05,
      "loss": 0.8363,
      "num_tokens": 263554740.0,
      "step": 86
    },
    {
      "epoch": 0.1274958783660011,
      "grad_norm": 0.7838147196534231,
      "learning_rate": 3.339805825242719e-05,
      "loss": 0.8619,
      "num_tokens": 266523921.0,
      "step": 87
    },
    {
      "epoch": 0.12896134823227698,
      "grad_norm": 0.5395144411175702,
      "learning_rate": 3.378640776699029e-05,
      "loss": 0.8535,
      "num_tokens": 269868631.0,
      "step": 88
    },
    {
      "epoch": 0.13042681809855286,
      "grad_norm": 0.9256661786843077,
      "learning_rate": 3.41747572815534e-05,
      "loss": 0.8367,
      "num_tokens": 273242182.0,
      "step": 89
    },
    {
      "epoch": 0.1318922879648287,
      "grad_norm": 0.6320711229649394,
      "learning_rate": 3.456310679611651e-05,
      "loss": 0.8676,
      "num_tokens": 276420774.0,
      "step": 90
    },
    {
      "epoch": 0.1333577578311046,
      "grad_norm": 0.9111610484555086,
      "learning_rate": 3.4951456310679615e-05,
      "loss": 0.8525,
      "num_tokens": 279322789.0,
      "step": 91
    },
    {
      "epoch": 0.13482322769738048,
      "grad_norm": 0.6630400558870672,
      "learning_rate": 3.5339805825242724e-05,
      "loss": 0.8459,
      "num_tokens": 282594946.0,
      "step": 92
    },
    {
      "epoch": 0.13628869756365636,
      "grad_norm": 0.7872835328404213,
      "learning_rate": 3.572815533980583e-05,
      "loss": 0.8696,
      "num_tokens": 285739114.0,
      "step": 93
    },
    {
      "epoch": 0.1377541674299322,
      "grad_norm": 0.7757734012108687,
      "learning_rate": 3.6116504854368933e-05,
      "loss": 0.8404,
      "num_tokens": 288842961.0,
      "step": 94
    },
    {
      "epoch": 0.1392196372962081,
      "grad_norm": 0.8172131583040233,
      "learning_rate": 3.650485436893204e-05,
      "loss": 0.8349,
      "num_tokens": 291857499.0,
      "step": 95
    },
    {
      "epoch": 0.14068510716248397,
      "grad_norm": 0.7505284006119732,
      "learning_rate": 3.689320388349515e-05,
      "loss": 0.8506,
      "num_tokens": 294786027.0,
      "step": 96
    },
    {
      "epoch": 0.14215057702875986,
      "grad_norm": 0.8544005803634707,
      "learning_rate": 3.728155339805826e-05,
      "loss": 0.8381,
      "num_tokens": 297533328.0,
      "step": 97
    },
    {
      "epoch": 0.1436160468950357,
      "grad_norm": 0.5925127226227879,
      "learning_rate": 3.766990291262136e-05,
      "loss": 0.8218,
      "num_tokens": 300605580.0,
      "step": 98
    },
    {
      "epoch": 0.1450815167613116,
      "grad_norm": 0.9521999060037152,
      "learning_rate": 3.805825242718447e-05,
      "loss": 0.8471,
      "num_tokens": 303698254.0,
      "step": 99
    },
    {
      "epoch": 0.14654698662758747,
      "grad_norm": 0.7010064402363387,
      "learning_rate": 3.844660194174758e-05,
      "loss": 0.8273,
      "num_tokens": 306988972.0,
      "step": 100
    },
    {
      "epoch": 0.14801245649386335,
      "grad_norm": 0.9654068200560665,
      "learning_rate": 3.8834951456310685e-05,
      "loss": 0.8429,
      "num_tokens": 309998481.0,
      "step": 101
    },
    {
      "epoch": 0.1494779263601392,
      "grad_norm": 0.7428113583562915,
      "learning_rate": 3.9223300970873787e-05,
      "loss": 0.8606,
      "num_tokens": 312761538.0,
      "step": 102
    },
    {
      "epoch": 0.1509433962264151,
      "grad_norm": 1.1685023275237025,
      "learning_rate": 3.9611650485436895e-05,
      "loss": 0.8526,
      "num_tokens": 316066460.0,
      "step": 103
    },
    {
      "epoch": 0.15240886609269097,
      "grad_norm": 0.8975147633009299,
      "learning_rate": 4e-05,
      "loss": 0.8401,
      "num_tokens": 319285989.0,
      "step": 104
    },
    {
      "epoch": 0.15387433595896685,
      "grad_norm": 0.9617782907220275,
      "learning_rate": 3.999999190230219e-05,
      "loss": 0.8521,
      "num_tokens": 322459463.0,
      "step": 105
    },
    {
      "epoch": 0.1553398058252427,
      "grad_norm": 0.9014787361849639,
      "learning_rate": 3.9999967609216026e-05,
      "loss": 0.8682,
      "num_tokens": 325398105.0,
      "step": 106
    },
    {
      "epoch": 0.1568052756915186,
      "grad_norm": 0.8448968557840734,
      "learning_rate": 3.9999927120763375e-05,
      "loss": 0.8515,
      "num_tokens": 328354575.0,
      "step": 107
    },
    {
      "epoch": 0.15827074555779447,
      "grad_norm": 0.7675935741857141,
      "learning_rate": 3.9999870436980666e-05,
      "loss": 0.849,
      "num_tokens": 331380106.0,
      "step": 108
    },
    {
      "epoch": 0.15973621542407035,
      "grad_norm": 0.7819100282863429,
      "learning_rate": 3.99997975579189e-05,
      "loss": 0.8492,
      "num_tokens": 334337989.0,
      "step": 109
    },
    {
      "epoch": 0.16120168529034623,
      "grad_norm": 0.6316748610620255,
      "learning_rate": 3.999970848364365e-05,
      "loss": 0.8335,
      "num_tokens": 337316918.0,
      "step": 110
    },
    {
      "epoch": 0.16266715515662208,
      "grad_norm": 0.9880671696016186,
      "learning_rate": 3.999960321423505e-05,
      "loss": 0.8565,
      "num_tokens": 340569368.0,
      "step": 111
    },
    {
      "epoch": 0.16413262502289797,
      "grad_norm": 0.8459771138972552,
      "learning_rate": 3.9999481749787836e-05,
      "loss": 0.8351,
      "num_tokens": 343785795.0,
      "step": 112
    },
    {
      "epoch": 0.16559809488917385,
      "grad_norm": 0.7228713159313074,
      "learning_rate": 3.9999344090411276e-05,
      "loss": 0.8432,
      "num_tokens": 347009560.0,
      "step": 113
    },
    {
      "epoch": 0.16706356475544973,
      "grad_norm": 0.6915081540386461,
      "learning_rate": 3.9999190236229236e-05,
      "loss": 0.8427,
      "num_tokens": 350077813.0,
      "step": 114
    },
    {
      "epoch": 0.16852903462172558,
      "grad_norm": 0.7399809312221232,
      "learning_rate": 3.999902018738014e-05,
      "loss": 0.854,
      "num_tokens": 353365170.0,
      "step": 115
    },
    {
      "epoch": 0.16999450448800146,
      "grad_norm": 0.6204926157232024,
      "learning_rate": 3.9998833944017004e-05,
      "loss": 0.8672,
      "num_tokens": 356249147.0,
      "step": 116
    },
    {
      "epoch": 0.17145997435427734,
      "grad_norm": 0.8167529790814838,
      "learning_rate": 3.9998631506307376e-05,
      "loss": 0.8396,
      "num_tokens": 359194019.0,
      "step": 117
    },
    {
      "epoch": 0.17292544422055323,
      "grad_norm": 0.7448155552014681,
      "learning_rate": 3.9998412874433426e-05,
      "loss": 0.8432,
      "num_tokens": 362220812.0,
      "step": 118
    },
    {
      "epoch": 0.17439091408682908,
      "grad_norm": 0.7704376811076306,
      "learning_rate": 3.999817804859184e-05,
      "loss": 0.8335,
      "num_tokens": 365054187.0,
      "step": 119
    },
    {
      "epoch": 0.17585638395310496,
      "grad_norm": 0.8236095964572204,
      "learning_rate": 3.999792702899391e-05,
      "loss": 0.8401,
      "num_tokens": 368095469.0,
      "step": 120
    },
    {
      "epoch": 0.17732185381938084,
      "grad_norm": 0.7242411939290744,
      "learning_rate": 3.999765981586551e-05,
      "loss": 0.8421,
      "num_tokens": 371179280.0,
      "step": 121
    },
    {
      "epoch": 0.17878732368565672,
      "grad_norm": 0.7602144061016769,
      "learning_rate": 3.9997376409447025e-05,
      "loss": 0.82,
      "num_tokens": 374510666.0,
      "step": 122
    },
    {
      "epoch": 0.18025279355193258,
      "grad_norm": 0.6611171707745039,
      "learning_rate": 3.9997076809993475e-05,
      "loss": 0.818,
      "num_tokens": 377732134.0,
      "step": 123
    },
    {
      "epoch": 0.18171826341820846,
      "grad_norm": 0.8899181285907293,
      "learning_rate": 3.999676101777443e-05,
      "loss": 0.8437,
      "num_tokens": 380914452.0,
      "step": 124
    },
    {
      "epoch": 0.18318373328448434,
      "grad_norm": 0.7437575189790447,
      "learning_rate": 3.999642903307399e-05,
      "loss": 0.8385,
      "num_tokens": 383855630.0,
      "step": 125
    },
    {
      "epoch": 0.18464920315076022,
      "grad_norm": 0.79914125957133,
      "learning_rate": 3.999608085619088e-05,
      "loss": 0.8438,
      "num_tokens": 386829036.0,
      "step": 126
    },
    {
      "epoch": 0.18611467301703608,
      "grad_norm": 0.7133956104576291,
      "learning_rate": 3.999571648743837e-05,
      "loss": 0.8319,
      "num_tokens": 389956141.0,
      "step": 127
    },
    {
      "epoch": 0.18758014288331196,
      "grad_norm": 0.7472988405317954,
      "learning_rate": 3.999533592714429e-05,
      "loss": 0.8323,
      "num_tokens": 392958264.0,
      "step": 128
    },
    {
      "epoch": 0.18904561274958784,
      "grad_norm": 0.5783827236951032,
      "learning_rate": 3.999493917565105e-05,
      "loss": 0.8223,
      "num_tokens": 396253470.0,
      "step": 129
    },
    {
      "epoch": 0.19051108261586372,
      "grad_norm": 0.677932958212559,
      "learning_rate": 3.999452623331563e-05,
      "loss": 0.8381,
      "num_tokens": 399257893.0,
      "step": 130
    },
    {
      "epoch": 0.19197655248213957,
      "grad_norm": 0.6926833614130089,
      "learning_rate": 3.999409710050957e-05,
      "loss": 0.8576,
      "num_tokens": 402276427.0,
      "step": 131
    },
    {
      "epoch": 0.19344202234841545,
      "grad_norm": 0.6312795952253336,
      "learning_rate": 3.999365177761897e-05,
      "loss": 0.8461,
      "num_tokens": 405193016.0,
      "step": 132
    },
    {
      "epoch": 0.19490749221469134,
      "grad_norm": 1.0920656953343262,
      "learning_rate": 3.999319026504452e-05,
      "loss": 0.8329,
      "num_tokens": 408193549.0,
      "step": 133
    },
    {
      "epoch": 0.19637296208096722,
      "grad_norm": 0.5899652588744977,
      "learning_rate": 3.999271256320145e-05,
      "loss": 0.8472,
      "num_tokens": 411359687.0,
      "step": 134
    },
    {
      "epoch": 0.1978384319472431,
      "grad_norm": 1.0499494065913662,
      "learning_rate": 3.9992218672519585e-05,
      "loss": 0.8424,
      "num_tokens": 414484226.0,
      "step": 135
    },
    {
      "epoch": 0.19930390181351895,
      "grad_norm": 0.7735587808705342,
      "learning_rate": 3.99917085934433e-05,
      "loss": 0.8192,
      "num_tokens": 417469898.0,
      "step": 136
    },
    {
      "epoch": 0.20076937167979483,
      "grad_norm": 0.7893087239695173,
      "learning_rate": 3.999118232643152e-05,
      "loss": 0.8353,
      "num_tokens": 420518881.0,
      "step": 137
    },
    {
      "epoch": 0.20223484154607071,
      "grad_norm": 0.8844659631297803,
      "learning_rate": 3.9990639871957755e-05,
      "loss": 0.8505,
      "num_tokens": 423606313.0,
      "step": 138
    },
    {
      "epoch": 0.2037003114123466,
      "grad_norm": 0.7422524640015375,
      "learning_rate": 3.9990081230510086e-05,
      "loss": 0.8201,
      "num_tokens": 426791799.0,
      "step": 139
    },
    {
      "epoch": 0.20516578127862245,
      "grad_norm": 0.7895678346896589,
      "learning_rate": 3.998950640259113e-05,
      "loss": 0.8532,
      "num_tokens": 429715234.0,
      "step": 140
    },
    {
      "epoch": 0.20663125114489833,
      "grad_norm": 0.7162901158544891,
      "learning_rate": 3.998891538871811e-05,
      "loss": 0.8367,
      "num_tokens": 432885607.0,
      "step": 141
    },
    {
      "epoch": 0.2080967210111742,
      "grad_norm": 1.0051799893599505,
      "learning_rate": 3.9988308189422764e-05,
      "loss": 0.8187,
      "num_tokens": 436138685.0,
      "step": 142
    },
    {
      "epoch": 0.2095621908774501,
      "grad_norm": 0.6031350174028013,
      "learning_rate": 3.9987684805251434e-05,
      "loss": 0.8293,
      "num_tokens": 439190104.0,
      "step": 143
    },
    {
      "epoch": 0.21102766074372595,
      "grad_norm": 1.08526344989608,
      "learning_rate": 3.998704523676499e-05,
      "loss": 0.8226,
      "num_tokens": 442137425.0,
      "step": 144
    },
    {
      "epoch": 0.21249313061000183,
      "grad_norm": 0.6482151176096894,
      "learning_rate": 3.9986389484538886e-05,
      "loss": 0.8287,
      "num_tokens": 445161524.0,
      "step": 145
    },
    {
      "epoch": 0.2139586004762777,
      "grad_norm": 0.860680740857215,
      "learning_rate": 3.998571754916314e-05,
      "loss": 0.8153,
      "num_tokens": 448239655.0,
      "step": 146
    },
    {
      "epoch": 0.2154240703425536,
      "grad_norm": 0.7415041111321264,
      "learning_rate": 3.9985029431242305e-05,
      "loss": 0.8497,
      "num_tokens": 451132218.0,
      "step": 147
    },
    {
      "epoch": 0.21688954020882945,
      "grad_norm": 0.8029762433268514,
      "learning_rate": 3.998432513139553e-05,
      "loss": 0.8384,
      "num_tokens": 454285865.0,
      "step": 148
    },
    {
      "epoch": 0.21835501007510533,
      "grad_norm": 0.6349460139245732,
      "learning_rate": 3.998360465025649e-05,
      "loss": 0.8313,
      "num_tokens": 457309502.0,
      "step": 149
    },
    {
      "epoch": 0.2198204799413812,
      "grad_norm": 0.8030936589895405,
      "learning_rate": 3.9982867988473446e-05,
      "loss": 0.8156,
      "num_tokens": 460241737.0,
      "step": 150
    },
    {
      "epoch": 0.2212859498076571,
      "grad_norm": 0.5663180192493148,
      "learning_rate": 3.998211514670919e-05,
      "loss": 0.8299,
      "num_tokens": 463436310.0,
      "step": 151
    },
    {
      "epoch": 0.22275141967393294,
      "grad_norm": 0.819436376636797,
      "learning_rate": 3.99813461256411e-05,
      "loss": 0.8193,
      "num_tokens": 466381548.0,
      "step": 152
    },
    {
      "epoch": 0.22421688954020882,
      "grad_norm": 0.7434237957955598,
      "learning_rate": 3.998056092596109e-05,
      "loss": 0.8621,
      "num_tokens": 469350892.0,
      "step": 153
    },
    {
      "epoch": 0.2256823594064847,
      "grad_norm": 0.7452172603634878,
      "learning_rate": 3.997975954837564e-05,
      "loss": 0.8249,
      "num_tokens": 472530190.0,
      "step": 154
    },
    {
      "epoch": 0.2271478292727606,
      "grad_norm": 0.7123150549669498,
      "learning_rate": 3.99789419936058e-05,
      "loss": 0.8407,
      "num_tokens": 475507490.0,
      "step": 155
    },
    {
      "epoch": 0.22861329913903644,
      "grad_norm": 0.7865493328792948,
      "learning_rate": 3.9978108262387135e-05,
      "loss": 0.8165,
      "num_tokens": 478670275.0,
      "step": 156
    },
    {
      "epoch": 0.23007876900531232,
      "grad_norm": 0.6101056307376534,
      "learning_rate": 3.997725835546981e-05,
      "loss": 0.827,
      "num_tokens": 481730944.0,
      "step": 157
    },
    {
      "epoch": 0.2315442388715882,
      "grad_norm": 0.8565396599611025,
      "learning_rate": 3.9976392273618514e-05,
      "loss": 0.8385,
      "num_tokens": 484600898.0,
      "step": 158
    },
    {
      "epoch": 0.23300970873786409,
      "grad_norm": 0.7336100996232997,
      "learning_rate": 3.99755100176125e-05,
      "loss": 0.8263,
      "num_tokens": 487778777.0,
      "step": 159
    },
    {
      "epoch": 0.23447517860413994,
      "grad_norm": 0.6937429258421548,
      "learning_rate": 3.9974611588245576e-05,
      "loss": 0.827,
      "num_tokens": 490635037.0,
      "step": 160
    },
    {
      "epoch": 0.23594064847041582,
      "grad_norm": 0.6441213653525039,
      "learning_rate": 3.99736969863261e-05,
      "loss": 0.8332,
      "num_tokens": 493613420.0,
      "step": 161
    },
    {
      "epoch": 0.2374061183366917,
      "grad_norm": 0.7216488874387245,
      "learning_rate": 3.997276621267697e-05,
      "loss": 0.8252,
      "num_tokens": 496857348.0,
      "step": 162
    },
    {
      "epoch": 0.23887158820296758,
      "grad_norm": 0.8074819977336892,
      "learning_rate": 3.9971819268135646e-05,
      "loss": 0.8374,
      "num_tokens": 499788261.0,
      "step": 163
    },
    {
      "epoch": 0.24033705806924346,
      "grad_norm": 0.6077381796651184,
      "learning_rate": 3.997085615355416e-05,
      "loss": 0.8164,
      "num_tokens": 503170980.0,
      "step": 164
    },
    {
      "epoch": 0.24180252793551932,
      "grad_norm": 0.7944494723827582,
      "learning_rate": 3.996987686979903e-05,
      "loss": 0.8154,
      "num_tokens": 506409642.0,
      "step": 165
    },
    {
      "epoch": 0.2432679978017952,
      "grad_norm": 0.6690511398862766,
      "learning_rate": 3.996888141775139e-05,
      "loss": 0.8171,
      "num_tokens": 509537723.0,
      "step": 166
    },
    {
      "epoch": 0.24473346766807108,
      "grad_norm": 0.8344368839873392,
      "learning_rate": 3.996786979830687e-05,
      "loss": 0.8265,
      "num_tokens": 512407522.0,
      "step": 167
    },
    {
      "epoch": 0.24619893753434696,
      "grad_norm": 0.6592804149080247,
      "learning_rate": 3.996684201237569e-05,
      "loss": 0.8545,
      "num_tokens": 515316936.0,
      "step": 168
    },
    {
      "epoch": 0.24766440740062282,
      "grad_norm": 0.6466670112342553,
      "learning_rate": 3.996579806088259e-05,
      "loss": 0.8315,
      "num_tokens": 518454253.0,
      "step": 169
    },
    {
      "epoch": 0.2491298772668987,
      "grad_norm": 0.803154966017149,
      "learning_rate": 3.996473794476685e-05,
      "loss": 0.8316,
      "num_tokens": 521488424.0,
      "step": 170
    },
    {
      "epoch": 0.25059534713317455,
      "grad_norm": 0.6940984560980675,
      "learning_rate": 3.996366166498231e-05,
      "loss": 0.8201,
      "num_tokens": 524663901.0,
      "step": 171
    },
    {
      "epoch": 0.25206081699945043,
      "grad_norm": 0.6805309224566704,
      "learning_rate": 3.996256922249734e-05,
      "loss": 0.8312,
      "num_tokens": 527436321.0,
      "step": 172
    },
    {
      "epoch": 0.2535262868657263,
      "grad_norm": 0.6749512403454886,
      "learning_rate": 3.996146061829487e-05,
      "loss": 0.8161,
      "num_tokens": 530418450.0,
      "step": 173
    },
    {
      "epoch": 0.2549917567320022,
      "grad_norm": 0.7730068368828539,
      "learning_rate": 3.9960335853372345e-05,
      "loss": 0.821,
      "num_tokens": 533701141.0,
      "step": 174
    },
    {
      "epoch": 0.2564572265982781,
      "grad_norm": 0.7203670802264126,
      "learning_rate": 3.995919492874178e-05,
      "loss": 0.8104,
      "num_tokens": 536860257.0,
      "step": 175
    },
    {
      "epoch": 0.25792269646455396,
      "grad_norm": 0.7984702395294277,
      "learning_rate": 3.995803784542971e-05,
      "loss": 0.8465,
      "num_tokens": 539725159.0,
      "step": 176
    },
    {
      "epoch": 0.25938816633082984,
      "grad_norm": 0.5362726653124847,
      "learning_rate": 3.9956864604477214e-05,
      "loss": 0.8267,
      "num_tokens": 542644241.0,
      "step": 177
    },
    {
      "epoch": 0.2608536361971057,
      "grad_norm": 0.6835100341938316,
      "learning_rate": 3.9955675206939906e-05,
      "loss": 0.8317,
      "num_tokens": 545588641.0,
      "step": 178
    },
    {
      "epoch": 0.26231910606338155,
      "grad_norm": 0.8074313766534024,
      "learning_rate": 3.995446965388794e-05,
      "loss": 0.8174,
      "num_tokens": 548795621.0,
      "step": 179
    },
    {
      "epoch": 0.2637845759296574,
      "grad_norm": 0.5507924702900725,
      "learning_rate": 3.995324794640601e-05,
      "loss": 0.8387,
      "num_tokens": 551569263.0,
      "step": 180
    },
    {
      "epoch": 0.2652500457959333,
      "grad_norm": 0.6814789822425229,
      "learning_rate": 3.995201008559334e-05,
      "loss": 0.836,
      "num_tokens": 554671550.0,
      "step": 181
    },
    {
      "epoch": 0.2667155156622092,
      "grad_norm": 0.811174854448483,
      "learning_rate": 3.995075607256368e-05,
      "loss": 0.8184,
      "num_tokens": 557598005.0,
      "step": 182
    },
    {
      "epoch": 0.26818098552848507,
      "grad_norm": 0.5709985096651433,
      "learning_rate": 3.994948590844533e-05,
      "loss": 0.7927,
      "num_tokens": 560902990.0,
      "step": 183
    },
    {
      "epoch": 0.26964645539476095,
      "grad_norm": 0.9184572254839238,
      "learning_rate": 3.994819959438111e-05,
      "loss": 0.8215,
      "num_tokens": 563930630.0,
      "step": 184
    },
    {
      "epoch": 0.27111192526103683,
      "grad_norm": 0.6528820775338945,
      "learning_rate": 3.9946897131528374e-05,
      "loss": 0.8105,
      "num_tokens": 567208936.0,
      "step": 185
    },
    {
      "epoch": 0.2725773951273127,
      "grad_norm": 0.7296617750225377,
      "learning_rate": 3.9945578521059e-05,
      "loss": 0.812,
      "num_tokens": 570240635.0,
      "step": 186
    },
    {
      "epoch": 0.27404286499358854,
      "grad_norm": 0.7285613438531743,
      "learning_rate": 3.994424376415941e-05,
      "loss": 0.82,
      "num_tokens": 573184515.0,
      "step": 187
    },
    {
      "epoch": 0.2755083348598644,
      "grad_norm": 0.6901016626567648,
      "learning_rate": 3.994289286203053e-05,
      "loss": 0.811,
      "num_tokens": 576074626.0,
      "step": 188
    },
    {
      "epoch": 0.2769738047261403,
      "grad_norm": 0.5757190070173106,
      "learning_rate": 3.9941525815887836e-05,
      "loss": 0.824,
      "num_tokens": 579444824.0,
      "step": 189
    },
    {
      "epoch": 0.2784392745924162,
      "grad_norm": 0.8966354598014828,
      "learning_rate": 3.994014262696133e-05,
      "loss": 0.817,
      "num_tokens": 582565685.0,
      "step": 190
    },
    {
      "epoch": 0.27990474445869207,
      "grad_norm": 0.6254396798953378,
      "learning_rate": 3.99387432964955e-05,
      "loss": 0.8342,
      "num_tokens": 585677800.0,
      "step": 191
    },
    {
      "epoch": 0.28137021432496795,
      "grad_norm": 0.7372678089826386,
      "learning_rate": 3.993732782574942e-05,
      "loss": 0.7932,
      "num_tokens": 588847079.0,
      "step": 192
    },
    {
      "epoch": 0.28283568419124383,
      "grad_norm": 0.7299211035819455,
      "learning_rate": 3.993589621599662e-05,
      "loss": 0.7952,
      "num_tokens": 591843647.0,
      "step": 193
    },
    {
      "epoch": 0.2843011540575197,
      "grad_norm": 0.7200112059284413,
      "learning_rate": 3.99344484685252e-05,
      "loss": 0.8195,
      "num_tokens": 594992889.0,
      "step": 194
    },
    {
      "epoch": 0.2857666239237956,
      "grad_norm": 0.8431571261967636,
      "learning_rate": 3.993298458463776e-05,
      "loss": 0.8351,
      "num_tokens": 598058589.0,
      "step": 195
    },
    {
      "epoch": 0.2872320937900714,
      "grad_norm": 0.5891969513358577,
      "learning_rate": 3.993150456565143e-05,
      "loss": 0.8348,
      "num_tokens": 601074188.0,
      "step": 196
    },
    {
      "epoch": 0.2886975636563473,
      "grad_norm": 0.7584934823993437,
      "learning_rate": 3.993000841289782e-05,
      "loss": 0.8292,
      "num_tokens": 604082070.0,
      "step": 197
    },
    {
      "epoch": 0.2901630335226232,
      "grad_norm": 0.6632460309991204,
      "learning_rate": 3.9928496127723114e-05,
      "loss": 0.8002,
      "num_tokens": 607498039.0,
      "step": 198
    },
    {
      "epoch": 0.29162850338889906,
      "grad_norm": 0.7422033104844754,
      "learning_rate": 3.9926967711487966e-05,
      "loss": 0.8176,
      "num_tokens": 610437913.0,
      "step": 199
    },
    {
      "epoch": 0.29309397325517494,
      "grad_norm": 0.5154286101153708,
      "learning_rate": 3.9925423165567564e-05,
      "loss": 0.8074,
      "num_tokens": 613524270.0,
      "step": 200
    },
    {
      "epoch": 0.2945594431214508,
      "grad_norm": 0.7019171226982792,
      "learning_rate": 3.992386249135161e-05,
      "loss": 0.8238,
      "num_tokens": 616438860.0,
      "step": 201
    },
    {
      "epoch": 0.2960249129877267,
      "grad_norm": 0.7459400715093033,
      "learning_rate": 3.99222856902443e-05,
      "loss": 0.8105,
      "num_tokens": 619976734.0,
      "step": 202
    },
    {
      "epoch": 0.2974903828540026,
      "grad_norm": 0.8354419436277302,
      "learning_rate": 3.992069276366436e-05,
      "loss": 0.8177,
      "num_tokens": 623066903.0,
      "step": 203
    },
    {
      "epoch": 0.2989558527202784,
      "grad_norm": 0.6196032193238038,
      "learning_rate": 3.991908371304502e-05,
      "loss": 0.81,
      "num_tokens": 626141402.0,
      "step": 204
    },
    {
      "epoch": 0.3004213225865543,
      "grad_norm": 0.7017892859541583,
      "learning_rate": 3.9917458539834e-05,
      "loss": 0.8213,
      "num_tokens": 629307909.0,
      "step": 205
    },
    {
      "epoch": 0.3018867924528302,
      "grad_norm": 0.6712223525214189,
      "learning_rate": 3.9915817245493555e-05,
      "loss": 0.8089,
      "num_tokens": 632370796.0,
      "step": 206
    },
    {
      "epoch": 0.30335226231910606,
      "grad_norm": 0.6752590187047599,
      "learning_rate": 3.991415983150042e-05,
      "loss": 0.8492,
      "num_tokens": 635295397.0,
      "step": 207
    },
    {
      "epoch": 0.30481773218538194,
      "grad_norm": 0.8611888424687264,
      "learning_rate": 3.991248629934585e-05,
      "loss": 0.8288,
      "num_tokens": 638322212.0,
      "step": 208
    },
    {
      "epoch": 0.3062832020516578,
      "grad_norm": 0.5131364545213245,
      "learning_rate": 3.9910796650535594e-05,
      "loss": 0.8076,
      "num_tokens": 641478294.0,
      "step": 209
    },
    {
      "epoch": 0.3077486719179337,
      "grad_norm": 0.7907027963530079,
      "learning_rate": 3.99090908865899e-05,
      "loss": 0.8292,
      "num_tokens": 644684314.0,
      "step": 210
    },
    {
      "epoch": 0.3092141417842096,
      "grad_norm": 0.6935867625715167,
      "learning_rate": 3.9907369009043525e-05,
      "loss": 0.8186,
      "num_tokens": 647703251.0,
      "step": 211
    },
    {
      "epoch": 0.3106796116504854,
      "grad_norm": 0.6529157200526053,
      "learning_rate": 3.9905631019445706e-05,
      "loss": 0.8106,
      "num_tokens": 650647298.0,
      "step": 212
    },
    {
      "epoch": 0.3121450815167613,
      "grad_norm": 0.7831886799826963,
      "learning_rate": 3.9903876919360207e-05,
      "loss": 0.8209,
      "num_tokens": 653422516.0,
      "step": 213
    },
    {
      "epoch": 0.3136105513830372,
      "grad_norm": 0.6510852093617102,
      "learning_rate": 3.990210671036527e-05,
      "loss": 0.8177,
      "num_tokens": 656293002.0,
      "step": 214
    },
    {
      "epoch": 0.31507602124931305,
      "grad_norm": 0.6425154337484317,
      "learning_rate": 3.99003203940536e-05,
      "loss": 0.8283,
      "num_tokens": 659414419.0,
      "step": 215
    },
    {
      "epoch": 0.31654149111558894,
      "grad_norm": 0.837178388828837,
      "learning_rate": 3.9898517972032456e-05,
      "loss": 0.8187,
      "num_tokens": 662466418.0,
      "step": 216
    },
    {
      "epoch": 0.3180069609818648,
      "grad_norm": 0.7419767469506745,
      "learning_rate": 3.989669944592355e-05,
      "loss": 0.8152,
      "num_tokens": 665332711.0,
      "step": 217
    },
    {
      "epoch": 0.3194724308481407,
      "grad_norm": 0.6899215374772198,
      "learning_rate": 3.989486481736308e-05,
      "loss": 0.8053,
      "num_tokens": 668191041.0,
      "step": 218
    },
    {
      "epoch": 0.3209379007144166,
      "grad_norm": 0.9102526563098543,
      "learning_rate": 3.9893014088001754e-05,
      "loss": 0.8002,
      "num_tokens": 671019312.0,
      "step": 219
    },
    {
      "epoch": 0.32240337058069246,
      "grad_norm": 0.7618911735458993,
      "learning_rate": 3.989114725950475e-05,
      "loss": 0.7975,
      "num_tokens": 674135552.0,
      "step": 220
    },
    {
      "epoch": 0.3238688404469683,
      "grad_norm": 0.8828199745356196,
      "learning_rate": 3.988926433355174e-05,
      "loss": 0.8024,
      "num_tokens": 677512820.0,
      "step": 221
    },
    {
      "epoch": 0.32533431031324417,
      "grad_norm": 0.5871783833268484,
      "learning_rate": 3.9887365311836865e-05,
      "loss": 0.8033,
      "num_tokens": 680449923.0,
      "step": 222
    },
    {
      "epoch": 0.32679978017952005,
      "grad_norm": 1.3220464396030602,
      "learning_rate": 3.9885450196068774e-05,
      "loss": 0.8136,
      "num_tokens": 683398145.0,
      "step": 223
    },
    {
      "epoch": 0.32826525004579593,
      "grad_norm": 0.8448212854927531,
      "learning_rate": 3.9883518987970564e-05,
      "loss": 0.8169,
      "num_tokens": 686301937.0,
      "step": 224
    },
    {
      "epoch": 0.3297307199120718,
      "grad_norm": 1.337087513449658,
      "learning_rate": 3.988157168927984e-05,
      "loss": 0.82,
      "num_tokens": 689608883.0,
      "step": 225
    },
    {
      "epoch": 0.3311961897783477,
      "grad_norm": 1.0891499449411013,
      "learning_rate": 3.9879608301748676e-05,
      "loss": 0.8196,
      "num_tokens": 692335164.0,
      "step": 226
    },
    {
      "epoch": 0.3326616596446236,
      "grad_norm": 1.1160541404724815,
      "learning_rate": 3.987762882714361e-05,
      "loss": 0.8112,
      "num_tokens": 695609166.0,
      "step": 227
    },
    {
      "epoch": 0.33412712951089946,
      "grad_norm": 1.0211109905827904,
      "learning_rate": 3.987563326724566e-05,
      "loss": 0.8064,
      "num_tokens": 698716359.0,
      "step": 228
    },
    {
      "epoch": 0.3355925993771753,
      "grad_norm": 0.8574404998176582,
      "learning_rate": 3.987362162385033e-05,
      "loss": 0.8264,
      "num_tokens": 701775841.0,
      "step": 229
    },
    {
      "epoch": 0.33705806924345116,
      "grad_norm": 0.964153884020382,
      "learning_rate": 3.987159389876758e-05,
      "loss": 0.8316,
      "num_tokens": 704772315.0,
      "step": 230
    },
    {
      "epoch": 0.33852353910972705,
      "grad_norm": 0.6003963708381315,
      "learning_rate": 3.986955009382185e-05,
      "loss": 0.8013,
      "num_tokens": 707787887.0,
      "step": 231
    },
    {
      "epoch": 0.3399890089760029,
      "grad_norm": 0.9262843585184378,
      "learning_rate": 3.9867490210852023e-05,
      "loss": 0.828,
      "num_tokens": 710898819.0,
      "step": 232
    },
    {
      "epoch": 0.3414544788422788,
      "grad_norm": 0.5636810797185664,
      "learning_rate": 3.9865414251711484e-05,
      "loss": 0.805,
      "num_tokens": 714001836.0,
      "step": 233
    },
    {
      "epoch": 0.3429199487085547,
      "grad_norm": 0.9796068494628346,
      "learning_rate": 3.986332221826806e-05,
      "loss": 0.8104,
      "num_tokens": 716937618.0,
      "step": 234
    },
    {
      "epoch": 0.34438541857483057,
      "grad_norm": 0.8212024547318494,
      "learning_rate": 3.986121411240404e-05,
      "loss": 0.8119,
      "num_tokens": 720053014.0,
      "step": 235
    },
    {
      "epoch": 0.34585088844110645,
      "grad_norm": 0.7990815267851159,
      "learning_rate": 3.9859089936016194e-05,
      "loss": 0.8088,
      "num_tokens": 723012456.0,
      "step": 236
    },
    {
      "epoch": 0.3473163583073823,
      "grad_norm": 0.932304241352169,
      "learning_rate": 3.985694969101573e-05,
      "loss": 0.8095,
      "num_tokens": 725810540.0,
      "step": 237
    },
    {
      "epoch": 0.34878182817365816,
      "grad_norm": 0.6575298005513099,
      "learning_rate": 3.9854793379328314e-05,
      "loss": 0.8253,
      "num_tokens": 728859351.0,
      "step": 238
    },
    {
      "epoch": 0.35024729803993404,
      "grad_norm": 0.9884944468854905,
      "learning_rate": 3.9852621002894084e-05,
      "loss": 0.8187,
      "num_tokens": 731985755.0,
      "step": 239
    },
    {
      "epoch": 0.3517127679062099,
      "grad_norm": 0.6499706972996803,
      "learning_rate": 3.9850432563667624e-05,
      "loss": 0.7985,
      "num_tokens": 734818087.0,
      "step": 240
    },
    {
      "epoch": 0.3531782377724858,
      "grad_norm": 0.9432189990448124,
      "learning_rate": 3.984822806361797e-05,
      "loss": 0.8061,
      "num_tokens": 737882416.0,
      "step": 241
    },
    {
      "epoch": 0.3546437076387617,
      "grad_norm": 0.7416121449625138,
      "learning_rate": 3.9846007504728593e-05,
      "loss": 0.7946,
      "num_tokens": 740872334.0,
      "step": 242
    },
    {
      "epoch": 0.35610917750503757,
      "grad_norm": 0.7958157545430461,
      "learning_rate": 3.9843770888997444e-05,
      "loss": 0.7975,
      "num_tokens": 744025224.0,
      "step": 243
    },
    {
      "epoch": 0.35757464737131345,
      "grad_norm": 0.8168175520110875,
      "learning_rate": 3.98415182184369e-05,
      "loss": 0.825,
      "num_tokens": 747083529.0,
      "step": 244
    },
    {
      "epoch": 0.35904011723758933,
      "grad_norm": 0.6333502231811122,
      "learning_rate": 3.983924949507379e-05,
      "loss": 0.8089,
      "num_tokens": 749927489.0,
      "step": 245
    },
    {
      "epoch": 0.36050558710386515,
      "grad_norm": 0.7167453216406626,
      "learning_rate": 3.9836964720949376e-05,
      "loss": 0.7852,
      "num_tokens": 753003946.0,
      "step": 246
    },
    {
      "epoch": 0.36197105697014104,
      "grad_norm": 0.6586488885722229,
      "learning_rate": 3.983466389811938e-05,
      "loss": 0.8022,
      "num_tokens": 756206664.0,
      "step": 247
    },
    {
      "epoch": 0.3634365268364169,
      "grad_norm": 0.6739528099479933,
      "learning_rate": 3.9832347028653956e-05,
      "loss": 0.7856,
      "num_tokens": 759305905.0,
      "step": 248
    },
    {
      "epoch": 0.3649019967026928,
      "grad_norm": 0.756683568899693,
      "learning_rate": 3.9830014114637685e-05,
      "loss": 0.7916,
      "num_tokens": 762317675.0,
      "step": 249
    },
    {
      "epoch": 0.3663674665689687,
      "grad_norm": 0.5972547793831667,
      "learning_rate": 3.98276651581696e-05,
      "loss": 0.8033,
      "num_tokens": 765233847.0,
      "step": 250
    },
    {
      "epoch": 0.36783293643524456,
      "grad_norm": 0.7968804539912702,
      "learning_rate": 3.982530016136316e-05,
      "loss": 0.8171,
      "num_tokens": 768110450.0,
      "step": 251
    },
    {
      "epoch": 0.36929840630152044,
      "grad_norm": 0.6402463575133113,
      "learning_rate": 3.982291912634625e-05,
      "loss": 0.8081,
      "num_tokens": 770993870.0,
      "step": 252
    },
    {
      "epoch": 0.3707638761677963,
      "grad_norm": 0.5854244804235733,
      "learning_rate": 3.9820522055261205e-05,
      "loss": 0.8046,
      "num_tokens": 773870954.0,
      "step": 253
    },
    {
      "epoch": 0.37222934603407215,
      "grad_norm": 0.8352382465839165,
      "learning_rate": 3.981810895026476e-05,
      "loss": 0.8152,
      "num_tokens": 776774139.0,
      "step": 254
    },
    {
      "epoch": 0.37369481590034803,
      "grad_norm": 0.5477078216784473,
      "learning_rate": 3.9815679813528107e-05,
      "loss": 0.7987,
      "num_tokens": 779627344.0,
      "step": 255
    },
    {
      "epoch": 0.3751602857666239,
      "grad_norm": 0.9948671635332358,
      "learning_rate": 3.9813234647236835e-05,
      "loss": 0.7874,
      "num_tokens": 782724891.0,
      "step": 256
    },
    {
      "epoch": 0.3766257556328998,
      "grad_norm": 0.7177257458743718,
      "learning_rate": 3.981077345359098e-05,
      "loss": 0.7984,
      "num_tokens": 785864127.0,
      "step": 257
    },
    {
      "epoch": 0.3780912254991757,
      "grad_norm": 1.0248541853552995,
      "learning_rate": 3.980829623480498e-05,
      "loss": 0.8065,
      "num_tokens": 788860443.0,
      "step": 258
    },
    {
      "epoch": 0.37955669536545156,
      "grad_norm": 0.8269486779942563,
      "learning_rate": 3.98058029931077e-05,
      "loss": 0.7892,
      "num_tokens": 791965513.0,
      "step": 259
    },
    {
      "epoch": 0.38102216523172744,
      "grad_norm": 0.8210289323787733,
      "learning_rate": 3.980329373074242e-05,
      "loss": 0.7909,
      "num_tokens": 795188967.0,
      "step": 260
    },
    {
      "epoch": 0.3824876350980033,
      "grad_norm": 0.7413417161587943,
      "learning_rate": 3.980076844996683e-05,
      "loss": 0.8017,
      "num_tokens": 798214311.0,
      "step": 261
    },
    {
      "epoch": 0.38395310496427915,
      "grad_norm": 0.6897198702390833,
      "learning_rate": 3.979822715305305e-05,
      "loss": 0.8075,
      "num_tokens": 801079842.0,
      "step": 262
    },
    {
      "epoch": 0.385418574830555,
      "grad_norm": 0.6951560105884733,
      "learning_rate": 3.9795669842287575e-05,
      "loss": 0.8122,
      "num_tokens": 804219719.0,
      "step": 263
    },
    {
      "epoch": 0.3868840446968309,
      "grad_norm": 0.6195012813677117,
      "learning_rate": 3.9793096519971354e-05,
      "loss": 0.8068,
      "num_tokens": 807280123.0,
      "step": 264
    },
    {
      "epoch": 0.3883495145631068,
      "grad_norm": 0.7587752207285182,
      "learning_rate": 3.97905071884197e-05,
      "loss": 0.8067,
      "num_tokens": 810041493.0,
      "step": 265
    },
    {
      "epoch": 0.38981498442938267,
      "grad_norm": 0.6222903416283405,
      "learning_rate": 3.9787901849962355e-05,
      "loss": 0.8065,
      "num_tokens": 812919486.0,
      "step": 266
    },
    {
      "epoch": 0.39128045429565855,
      "grad_norm": 0.6674186345466703,
      "learning_rate": 3.9785280506943465e-05,
      "loss": 0.7986,
      "num_tokens": 816094940.0,
      "step": 267
    },
    {
      "epoch": 0.39274592416193443,
      "grad_norm": 0.7153698861223848,
      "learning_rate": 3.978264316172156e-05,
      "loss": 0.8086,
      "num_tokens": 819154728.0,
      "step": 268
    },
    {
      "epoch": 0.3942113940282103,
      "grad_norm": 0.5121286382129685,
      "learning_rate": 3.977998981666957e-05,
      "loss": 0.8062,
      "num_tokens": 822342162.0,
      "step": 269
    },
    {
      "epoch": 0.3956768638944862,
      "grad_norm": 0.8154009432621302,
      "learning_rate": 3.977732047417485e-05,
      "loss": 0.8087,
      "num_tokens": 825271625.0,
      "step": 270
    },
    {
      "epoch": 0.397142333760762,
      "grad_norm": 0.6427132566769365,
      "learning_rate": 3.9774635136639106e-05,
      "loss": 0.7945,
      "num_tokens": 828216400.0,
      "step": 271
    },
    {
      "epoch": 0.3986078036270379,
      "grad_norm": 0.7439761634674241,
      "learning_rate": 3.9771933806478455e-05,
      "loss": 0.8059,
      "num_tokens": 831222721.0,
      "step": 272
    },
    {
      "epoch": 0.4000732734933138,
      "grad_norm": 0.7090772876336,
      "learning_rate": 3.976921648612341e-05,
      "loss": 0.7848,
      "num_tokens": 834382978.0,
      "step": 273
    },
    {
      "epoch": 0.40153874335958967,
      "grad_norm": 0.6604548780890596,
      "learning_rate": 3.976648317801886e-05,
      "loss": 0.7996,
      "num_tokens": 837453495.0,
      "step": 274
    },
    {
      "epoch": 0.40300421322586555,
      "grad_norm": 0.6988608505240668,
      "learning_rate": 3.976373388462409e-05,
      "loss": 0.8125,
      "num_tokens": 840495565.0,
      "step": 275
    },
    {
      "epoch": 0.40446968309214143,
      "grad_norm": 0.5309357857778855,
      "learning_rate": 3.9760968608412755e-05,
      "loss": 0.8153,
      "num_tokens": 843584708.0,
      "step": 276
    },
    {
      "epoch": 0.4059351529584173,
      "grad_norm": 0.7055946980845533,
      "learning_rate": 3.9758187351872895e-05,
      "loss": 0.8005,
      "num_tokens": 846486455.0,
      "step": 277
    },
    {
      "epoch": 0.4074006228246932,
      "grad_norm": 0.6784624675781239,
      "learning_rate": 3.9755390117506924e-05,
      "loss": 0.7907,
      "num_tokens": 849756701.0,
      "step": 278
    },
    {
      "epoch": 0.408866092690969,
      "grad_norm": 0.5924931852271991,
      "learning_rate": 3.975257690783166e-05,
      "loss": 0.7945,
      "num_tokens": 852723313.0,
      "step": 279
    },
    {
      "epoch": 0.4103315625572449,
      "grad_norm": 0.7632430943044748,
      "learning_rate": 3.9749747725378245e-05,
      "loss": 0.7974,
      "num_tokens": 855700614.0,
      "step": 280
    },
    {
      "epoch": 0.4117970324235208,
      "grad_norm": 0.5546325040716105,
      "learning_rate": 3.974690257269224e-05,
      "loss": 0.7984,
      "num_tokens": 858771918.0,
      "step": 281
    },
    {
      "epoch": 0.41326250228979666,
      "grad_norm": 0.8281035270624759,
      "learning_rate": 3.974404145233354e-05,
      "loss": 0.8052,
      "num_tokens": 861559560.0,
      "step": 282
    },
    {
      "epoch": 0.41472797215607254,
      "grad_norm": 0.570178531770508,
      "learning_rate": 3.974116436687643e-05,
      "loss": 0.7844,
      "num_tokens": 864562982.0,
      "step": 283
    },
    {
      "epoch": 0.4161934420223484,
      "grad_norm": 0.6470628944736802,
      "learning_rate": 3.9738271318909544e-05,
      "loss": 0.7975,
      "num_tokens": 867609919.0,
      "step": 284
    },
    {
      "epoch": 0.4176589118886243,
      "grad_norm": 0.7056418954685535,
      "learning_rate": 3.9735362311035894e-05,
      "loss": 0.8221,
      "num_tokens": 870665592.0,
      "step": 285
    },
    {
      "epoch": 0.4191243817549002,
      "grad_norm": 0.7723744780339872,
      "learning_rate": 3.973243734587284e-05,
      "loss": 0.7755,
      "num_tokens": 873694884.0,
      "step": 286
    },
    {
      "epoch": 0.420589851621176,
      "grad_norm": 0.6891367616734481,
      "learning_rate": 3.97294964260521e-05,
      "loss": 0.796,
      "num_tokens": 876730774.0,
      "step": 287
    },
    {
      "epoch": 0.4220553214874519,
      "grad_norm": 0.5892571573919205,
      "learning_rate": 3.972653955421975e-05,
      "loss": 0.8076,
      "num_tokens": 879715258.0,
      "step": 288
    },
    {
      "epoch": 0.4235207913537278,
      "grad_norm": 0.8101736963219275,
      "learning_rate": 3.972356673303622e-05,
      "loss": 0.8176,
      "num_tokens": 882831143.0,
      "step": 289
    },
    {
      "epoch": 0.42498626122000366,
      "grad_norm": 0.6682034633143411,
      "learning_rate": 3.972057796517628e-05,
      "loss": 0.8112,
      "num_tokens": 886199642.0,
      "step": 290
    },
    {
      "epoch": 0.42645173108627954,
      "grad_norm": 0.6904065303589332,
      "learning_rate": 3.9717573253329076e-05,
      "loss": 0.813,
      "num_tokens": 889091703.0,
      "step": 291
    },
    {
      "epoch": 0.4279172009525554,
      "grad_norm": 0.676075982662874,
      "learning_rate": 3.9714552600198056e-05,
      "loss": 0.8026,
      "num_tokens": 892457899.0,
      "step": 292
    },
    {
      "epoch": 0.4293826708188313,
      "grad_norm": 0.6386175341949273,
      "learning_rate": 3.971151600850105e-05,
      "loss": 0.7951,
      "num_tokens": 895354175.0,
      "step": 293
    },
    {
      "epoch": 0.4308481406851072,
      "grad_norm": 0.8040523475084294,
      "learning_rate": 3.9708463480970224e-05,
      "loss": 0.7881,
      "num_tokens": 898455626.0,
      "step": 294
    },
    {
      "epoch": 0.432313610551383,
      "grad_norm": 0.6037494861758412,
      "learning_rate": 3.970539502035205e-05,
      "loss": 0.8072,
      "num_tokens": 901682272.0,
      "step": 295
    },
    {
      "epoch": 0.4337790804176589,
      "grad_norm": 0.8445458036330848,
      "learning_rate": 3.970231062940736e-05,
      "loss": 0.8455,
      "num_tokens": 904622788.0,
      "step": 296
    },
    {
      "epoch": 0.4352445502839348,
      "grad_norm": 0.6079029180527582,
      "learning_rate": 3.969921031091133e-05,
      "loss": 0.7967,
      "num_tokens": 907679701.0,
      "step": 297
    },
    {
      "epoch": 0.43671002015021065,
      "grad_norm": 0.8638414962718668,
      "learning_rate": 3.969609406765345e-05,
      "loss": 0.7715,
      "num_tokens": 910720194.0,
      "step": 298
    },
    {
      "epoch": 0.43817549001648654,
      "grad_norm": 0.5879101715228864,
      "learning_rate": 3.9692961902437536e-05,
      "loss": 0.7968,
      "num_tokens": 913883318.0,
      "step": 299
    },
    {
      "epoch": 0.4396409598827624,
      "grad_norm": 1.1033230276301016,
      "learning_rate": 3.968981381808174e-05,
      "loss": 0.8025,
      "num_tokens": 916930317.0,
      "step": 300
    },
    {
      "epoch": 0.4411064297490383,
      "grad_norm": 0.7204023357354239,
      "learning_rate": 3.968664981741852e-05,
      "loss": 0.803,
      "num_tokens": 919860522.0,
      "step": 301
    },
    {
      "epoch": 0.4425718996153142,
      "grad_norm": 1.1140147070786401,
      "learning_rate": 3.968346990329469e-05,
      "loss": 0.7883,
      "num_tokens": 923091105.0,
      "step": 302
    },
    {
      "epoch": 0.44403736948159006,
      "grad_norm": 0.931350919822376,
      "learning_rate": 3.968027407857134e-05,
      "loss": 0.794,
      "num_tokens": 926028964.0,
      "step": 303
    },
    {
      "epoch": 0.4455028393478659,
      "grad_norm": 0.9984343179523539,
      "learning_rate": 3.967706234612391e-05,
      "loss": 0.7777,
      "num_tokens": 929348858.0,
      "step": 304
    },
    {
      "epoch": 0.44696830921414177,
      "grad_norm": 0.8807420148602723,
      "learning_rate": 3.967383470884213e-05,
      "loss": 0.8031,
      "num_tokens": 932222204.0,
      "step": 305
    },
    {
      "epoch": 0.44843377908041765,
      "grad_norm": 0.8750755663477989,
      "learning_rate": 3.967059116963004e-05,
      "loss": 0.8033,
      "num_tokens": 935364724.0,
      "step": 306
    },
    {
      "epoch": 0.44989924894669353,
      "grad_norm": 0.8628230525343615,
      "learning_rate": 3.9667331731406015e-05,
      "loss": 0.8133,
      "num_tokens": 938525625.0,
      "step": 307
    },
    {
      "epoch": 0.4513647188129694,
      "grad_norm": 0.7845540994685867,
      "learning_rate": 3.966405639710269e-05,
      "loss": 0.8156,
      "num_tokens": 941740823.0,
      "step": 308
    },
    {
      "epoch": 0.4528301886792453,
      "grad_norm": 0.8150532723202197,
      "learning_rate": 3.966076516966705e-05,
      "loss": 0.7963,
      "num_tokens": 944914711.0,
      "step": 309
    },
    {
      "epoch": 0.4542956585455212,
      "grad_norm": 0.6892761616796849,
      "learning_rate": 3.965745805206035e-05,
      "loss": 0.7926,
      "num_tokens": 948008105.0,
      "step": 310
    },
    {
      "epoch": 0.45576112841179706,
      "grad_norm": 0.7309392964250384,
      "learning_rate": 3.965413504725815e-05,
      "loss": 0.7914,
      "num_tokens": 951233502.0,
      "step": 311
    },
    {
      "epoch": 0.4572265982780729,
      "grad_norm": 0.7253252405163354,
      "learning_rate": 3.9650796158250304e-05,
      "loss": 0.7882,
      "num_tokens": 954103566.0,
      "step": 312
    },
    {
      "epoch": 0.45869206814434876,
      "grad_norm": 0.66944422228442,
      "learning_rate": 3.964744138804096e-05,
      "loss": 0.7825,
      "num_tokens": 957109159.0,
      "step": 313
    },
    {
      "epoch": 0.46015753801062464,
      "grad_norm": 0.569515057687289,
      "learning_rate": 3.964407073964854e-05,
      "loss": 0.7941,
      "num_tokens": 960136750.0,
      "step": 314
    },
    {
      "epoch": 0.4616230078769005,
      "grad_norm": 0.6938206879198919,
      "learning_rate": 3.964068421610579e-05,
      "loss": 0.8057,
      "num_tokens": 963127053.0,
      "step": 315
    },
    {
      "epoch": 0.4630884777431764,
      "grad_norm": 0.6512216286182886,
      "learning_rate": 3.9637281820459696e-05,
      "loss": 0.794,
      "num_tokens": 966385018.0,
      "step": 316
    },
    {
      "epoch": 0.4645539476094523,
      "grad_norm": 0.5956523162117932,
      "learning_rate": 3.963386355577155e-05,
      "loss": 0.7958,
      "num_tokens": 969223008.0,
      "step": 317
    },
    {
      "epoch": 0.46601941747572817,
      "grad_norm": 0.6460599725708294,
      "learning_rate": 3.9630429425116924e-05,
      "loss": 0.7853,
      "num_tokens": 972338208.0,
      "step": 318
    },
    {
      "epoch": 0.46748488734200405,
      "grad_norm": 0.6323153278934952,
      "learning_rate": 3.962697943158564e-05,
      "loss": 0.7834,
      "num_tokens": 975501698.0,
      "step": 319
    },
    {
      "epoch": 0.4689503572082799,
      "grad_norm": 0.6910925634491862,
      "learning_rate": 3.9623513578281814e-05,
      "loss": 0.7978,
      "num_tokens": 978725425.0,
      "step": 320
    },
    {
      "epoch": 0.47041582707455576,
      "grad_norm": 0.5623156008534718,
      "learning_rate": 3.9620031868323834e-05,
      "loss": 0.7916,
      "num_tokens": 982024542.0,
      "step": 321
    },
    {
      "epoch": 0.47188129694083164,
      "grad_norm": 0.5869379770798241,
      "learning_rate": 3.9616534304844355e-05,
      "loss": 0.7983,
      "num_tokens": 985153492.0,
      "step": 322
    },
    {
      "epoch": 0.4733467668071075,
      "grad_norm": 0.645479405742131,
      "learning_rate": 3.961302089099027e-05,
      "loss": 0.7844,
      "num_tokens": 988100099.0,
      "step": 323
    },
    {
      "epoch": 0.4748122366733834,
      "grad_norm": 0.6116473211266792,
      "learning_rate": 3.960949162992276e-05,
      "loss": 0.7909,
      "num_tokens": 991146505.0,
      "step": 324
    },
    {
      "epoch": 0.4762777065396593,
      "grad_norm": 0.753635641137463,
      "learning_rate": 3.960594652481727e-05,
      "loss": 0.7947,
      "num_tokens": 994303276.0,
      "step": 325
    },
    {
      "epoch": 0.47774317640593517,
      "grad_norm": 0.5061771930164656,
      "learning_rate": 3.960238557886347e-05,
      "loss": 0.7685,
      "num_tokens": 997379414.0,
      "step": 326
    },
    {
      "epoch": 0.47920864627221105,
      "grad_norm": 0.6105881092651655,
      "learning_rate": 3.959880879526531e-05,
      "loss": 0.7825,
      "num_tokens": 1000390205.0,
      "step": 327
    },
    {
      "epoch": 0.48067411613848693,
      "grad_norm": 0.5438321751943127,
      "learning_rate": 3.959521617724098e-05,
      "loss": 0.7822,
      "num_tokens": 1003438133.0,
      "step": 328
    },
    {
      "epoch": 0.48213958600476275,
      "grad_norm": 0.8333424366739319,
      "learning_rate": 3.9591607728022915e-05,
      "loss": 0.7938,
      "num_tokens": 1006560380.0,
      "step": 329
    },
    {
      "epoch": 0.48360505587103864,
      "grad_norm": 0.5779230897698857,
      "learning_rate": 3.95879834508578e-05,
      "loss": 0.7972,
      "num_tokens": 1009754494.0,
      "step": 330
    },
    {
      "epoch": 0.4850705257373145,
      "grad_norm": 0.6645147682746726,
      "learning_rate": 3.9584343349006554e-05,
      "loss": 0.7991,
      "num_tokens": 1012777513.0,
      "step": 331
    },
    {
      "epoch": 0.4865359956035904,
      "grad_norm": 0.6547200183154098,
      "learning_rate": 3.9580687425744336e-05,
      "loss": 0.7958,
      "num_tokens": 1015900012.0,
      "step": 332
    },
    {
      "epoch": 0.4880014654698663,
      "grad_norm": 0.725519819194977,
      "learning_rate": 3.957701568436054e-05,
      "loss": 0.7794,
      "num_tokens": 1018865976.0,
      "step": 333
    },
    {
      "epoch": 0.48946693533614216,
      "grad_norm": 0.5428382191286174,
      "learning_rate": 3.95733281281588e-05,
      "loss": 0.7831,
      "num_tokens": 1021912685.0,
      "step": 334
    },
    {
      "epoch": 0.49093240520241804,
      "grad_norm": 0.5866903093772561,
      "learning_rate": 3.956962476045698e-05,
      "loss": 0.7921,
      "num_tokens": 1024858000.0,
      "step": 335
    },
    {
      "epoch": 0.4923978750686939,
      "grad_norm": 0.7383254938022301,
      "learning_rate": 3.956590558458714e-05,
      "loss": 0.7817,
      "num_tokens": 1028067724.0,
      "step": 336
    },
    {
      "epoch": 0.49386334493496975,
      "grad_norm": 0.7228307517265506,
      "learning_rate": 3.956217060389561e-05,
      "loss": 0.809,
      "num_tokens": 1030949134.0,
      "step": 337
    },
    {
      "epoch": 0.49532881480124563,
      "grad_norm": 0.5758838905550404,
      "learning_rate": 3.955841982174292e-05,
      "loss": 0.8036,
      "num_tokens": 1034039321.0,
      "step": 338
    },
    {
      "epoch": 0.4967942846675215,
      "grad_norm": 0.6632232135488345,
      "learning_rate": 3.9554653241503785e-05,
      "loss": 0.7936,
      "num_tokens": 1037238595.0,
      "step": 339
    },
    {
      "epoch": 0.4982597545337974,
      "grad_norm": 0.6954498018354363,
      "learning_rate": 3.9550870866567186e-05,
      "loss": 0.7831,
      "num_tokens": 1040511123.0,
      "step": 340
    },
    {
      "epoch": 0.4997252244000733,
      "grad_norm": 0.6315183800034614,
      "learning_rate": 3.95470727003363e-05,
      "loss": 0.8094,
      "num_tokens": 1043604871.0,
      "step": 341
    },
    {
      "epoch": 0.5011906942663491,
      "grad_norm": 0.5996472827135996,
      "learning_rate": 3.954325874622848e-05,
      "loss": 0.7767,
      "num_tokens": 1046621373.0,
      "step": 342
    },
    {
      "epoch": 0.502656164132625,
      "grad_norm": 0.5474040962151834,
      "learning_rate": 3.953942900767533e-05,
      "loss": 0.8032,
      "num_tokens": 1049599585.0,
      "step": 343
    },
    {
      "epoch": 0.5041216339989009,
      "grad_norm": 0.5543290180624266,
      "learning_rate": 3.953558348812263e-05,
      "loss": 0.8225,
      "num_tokens": 1052567894.0,
      "step": 344
    },
    {
      "epoch": 0.5055871038651768,
      "grad_norm": 0.7581832677374636,
      "learning_rate": 3.953172219103036e-05,
      "loss": 0.7849,
      "num_tokens": 1055667078.0,
      "step": 345
    },
    {
      "epoch": 0.5070525737314526,
      "grad_norm": 0.5984245809336709,
      "learning_rate": 3.952784511987269e-05,
      "loss": 0.802,
      "num_tokens": 1058805802.0,
      "step": 346
    },
    {
      "epoch": 0.5085180435977286,
      "grad_norm": 0.5481121604127609,
      "learning_rate": 3.952395227813802e-05,
      "loss": 0.8092,
      "num_tokens": 1061690228.0,
      "step": 347
    },
    {
      "epoch": 0.5099835134640044,
      "grad_norm": 0.6257314219554806,
      "learning_rate": 3.9520043669328896e-05,
      "loss": 0.7891,
      "num_tokens": 1064713785.0,
      "step": 348
    },
    {
      "epoch": 0.5114489833302802,
      "grad_norm": 0.5963159753362327,
      "learning_rate": 3.951611929696206e-05,
      "loss": 0.8011,
      "num_tokens": 1067670764.0,
      "step": 349
    },
    {
      "epoch": 0.5129144531965562,
      "grad_norm": 0.6794277560003524,
      "learning_rate": 3.9512179164568446e-05,
      "loss": 0.7741,
      "num_tokens": 1070805078.0,
      "step": 350
    },
    {
      "epoch": 0.514379923062832,
      "grad_norm": 0.6359463570012431,
      "learning_rate": 3.950822327569318e-05,
      "loss": 0.7971,
      "num_tokens": 1073917951.0,
      "step": 351
    },
    {
      "epoch": 0.5158453929291079,
      "grad_norm": 0.6999400118469834,
      "learning_rate": 3.950425163389553e-05,
      "loss": 0.8135,
      "num_tokens": 1077041988.0,
      "step": 352
    },
    {
      "epoch": 0.5173108627953837,
      "grad_norm": 0.6467988992899758,
      "learning_rate": 3.950026424274897e-05,
      "loss": 0.8002,
      "num_tokens": 1080239292.0,
      "step": 353
    },
    {
      "epoch": 0.5187763326616597,
      "grad_norm": 0.4342948331930541,
      "learning_rate": 3.949626110584112e-05,
      "loss": 0.8039,
      "num_tokens": 1083360690.0,
      "step": 354
    },
    {
      "epoch": 0.5202418025279355,
      "grad_norm": 0.6917638123972273,
      "learning_rate": 3.9492242226773796e-05,
      "loss": 0.7702,
      "num_tokens": 1086576694.0,
      "step": 355
    },
    {
      "epoch": 0.5217072723942114,
      "grad_norm": 0.686982527796773,
      "learning_rate": 3.9488207609162946e-05,
      "loss": 0.7715,
      "num_tokens": 1089946956.0,
      "step": 356
    },
    {
      "epoch": 0.5231727422604873,
      "grad_norm": 0.5805771358591818,
      "learning_rate": 3.948415725663871e-05,
      "loss": 0.7941,
      "num_tokens": 1093097656.0,
      "step": 357
    },
    {
      "epoch": 0.5246382121267631,
      "grad_norm": 0.6947326498781151,
      "learning_rate": 3.948009117284536e-05,
      "loss": 0.7844,
      "num_tokens": 1095943615.0,
      "step": 358
    },
    {
      "epoch": 0.526103681993039,
      "grad_norm": 0.6461136086064133,
      "learning_rate": 3.9476009361441316e-05,
      "loss": 0.7932,
      "num_tokens": 1099018381.0,
      "step": 359
    },
    {
      "epoch": 0.5275691518593149,
      "grad_norm": 0.4945645826622396,
      "learning_rate": 3.947191182609919e-05,
      "loss": 0.7752,
      "num_tokens": 1102099212.0,
      "step": 360
    },
    {
      "epoch": 0.5290346217255908,
      "grad_norm": 0.9569932332248041,
      "learning_rate": 3.946779857050571e-05,
      "loss": 0.7959,
      "num_tokens": 1105067286.0,
      "step": 361
    },
    {
      "epoch": 0.5305000915918666,
      "grad_norm": 0.5557588276975498,
      "learning_rate": 3.946366959836174e-05,
      "loss": 0.7781,
      "num_tokens": 1107966322.0,
      "step": 362
    },
    {
      "epoch": 0.5319655614581426,
      "grad_norm": 0.773491420016794,
      "learning_rate": 3.9459524913382315e-05,
      "loss": 0.7671,
      "num_tokens": 1111109458.0,
      "step": 363
    },
    {
      "epoch": 0.5334310313244184,
      "grad_norm": 0.6832784929303869,
      "learning_rate": 3.945536451929657e-05,
      "loss": 0.782,
      "num_tokens": 1114146576.0,
      "step": 364
    },
    {
      "epoch": 0.5348965011906943,
      "grad_norm": 0.6103917935833321,
      "learning_rate": 3.9451188419847825e-05,
      "loss": 0.8011,
      "num_tokens": 1117297618.0,
      "step": 365
    },
    {
      "epoch": 0.5363619710569701,
      "grad_norm": 0.754558711716202,
      "learning_rate": 3.9446996618793475e-05,
      "loss": 0.8023,
      "num_tokens": 1120224816.0,
      "step": 366
    },
    {
      "epoch": 0.537827440923246,
      "grad_norm": 0.5819548538974069,
      "learning_rate": 3.944278911990509e-05,
      "loss": 0.7984,
      "num_tokens": 1123350152.0,
      "step": 367
    },
    {
      "epoch": 0.5392929107895219,
      "grad_norm": 0.7092411227189372,
      "learning_rate": 3.943856592696832e-05,
      "loss": 0.771,
      "num_tokens": 1126551904.0,
      "step": 368
    },
    {
      "epoch": 0.5407583806557977,
      "grad_norm": 0.584353224153154,
      "learning_rate": 3.943432704378297e-05,
      "loss": 0.7857,
      "num_tokens": 1129793804.0,
      "step": 369
    },
    {
      "epoch": 0.5422238505220737,
      "grad_norm": 0.6509116472948473,
      "learning_rate": 3.9430072474162956e-05,
      "loss": 0.8032,
      "num_tokens": 1132646681.0,
      "step": 370
    },
    {
      "epoch": 0.5436893203883495,
      "grad_norm": 0.658935817704477,
      "learning_rate": 3.9425802221936297e-05,
      "loss": 0.7915,
      "num_tokens": 1135669439.0,
      "step": 371
    },
    {
      "epoch": 0.5451547902546254,
      "grad_norm": 0.6326213942256899,
      "learning_rate": 3.942151629094513e-05,
      "loss": 0.8007,
      "num_tokens": 1138920242.0,
      "step": 372
    },
    {
      "epoch": 0.5466202601209013,
      "grad_norm": 0.5359650740188716,
      "learning_rate": 3.941721468504569e-05,
      "loss": 0.7814,
      "num_tokens": 1142055088.0,
      "step": 373
    },
    {
      "epoch": 0.5480857299871771,
      "grad_norm": 0.627386838556622,
      "learning_rate": 3.941289740810833e-05,
      "loss": 0.7864,
      "num_tokens": 1145189774.0,
      "step": 374
    },
    {
      "epoch": 0.549551199853453,
      "grad_norm": 0.597519572284839,
      "learning_rate": 3.940856446401749e-05,
      "loss": 0.7735,
      "num_tokens": 1147993618.0,
      "step": 375
    },
    {
      "epoch": 0.5510166697197288,
      "grad_norm": 0.7502438584293525,
      "learning_rate": 3.9404215856671714e-05,
      "loss": 0.793,
      "num_tokens": 1151068236.0,
      "step": 376
    },
    {
      "epoch": 0.5524821395860048,
      "grad_norm": 0.5907966795842524,
      "learning_rate": 3.9399851589983636e-05,
      "loss": 0.7858,
      "num_tokens": 1154095768.0,
      "step": 377
    },
    {
      "epoch": 0.5539476094522806,
      "grad_norm": 0.6486592223973918,
      "learning_rate": 3.939547166787997e-05,
      "loss": 0.7732,
      "num_tokens": 1157114624.0,
      "step": 378
    },
    {
      "epoch": 0.5554130793185565,
      "grad_norm": 0.5086127371666345,
      "learning_rate": 3.9391076094301544e-05,
      "loss": 0.7734,
      "num_tokens": 1160391302.0,
      "step": 379
    },
    {
      "epoch": 0.5568785491848324,
      "grad_norm": 0.7111022612740963,
      "learning_rate": 3.938666487320323e-05,
      "loss": 0.789,
      "num_tokens": 1163468792.0,
      "step": 380
    },
    {
      "epoch": 0.5583440190511083,
      "grad_norm": 0.6089368963135149,
      "learning_rate": 3.938223800855402e-05,
      "loss": 0.7846,
      "num_tokens": 1166565708.0,
      "step": 381
    },
    {
      "epoch": 0.5598094889173841,
      "grad_norm": 0.525040717736477,
      "learning_rate": 3.937779550433694e-05,
      "loss": 0.7641,
      "num_tokens": 1170031638.0,
      "step": 382
    },
    {
      "epoch": 0.56127495878366,
      "grad_norm": 0.6863441114088674,
      "learning_rate": 3.9373337364549115e-05,
      "loss": 0.7636,
      "num_tokens": 1173067839.0,
      "step": 383
    },
    {
      "epoch": 0.5627404286499359,
      "grad_norm": 0.5270062193185172,
      "learning_rate": 3.936886359320174e-05,
      "loss": 0.7802,
      "num_tokens": 1176040768.0,
      "step": 384
    },
    {
      "epoch": 0.5642058985162117,
      "grad_norm": 0.6740752203015982,
      "learning_rate": 3.9364374194320045e-05,
      "loss": 0.7902,
      "num_tokens": 1178909285.0,
      "step": 385
    },
    {
      "epoch": 0.5656713683824877,
      "grad_norm": 0.5466221291761898,
      "learning_rate": 3.9359869171943356e-05,
      "loss": 0.7744,
      "num_tokens": 1181817393.0,
      "step": 386
    },
    {
      "epoch": 0.5671368382487635,
      "grad_norm": 0.6160945556506481,
      "learning_rate": 3.935534853012504e-05,
      "loss": 0.7791,
      "num_tokens": 1184603430.0,
      "step": 387
    },
    {
      "epoch": 0.5686023081150394,
      "grad_norm": 0.6409266957731405,
      "learning_rate": 3.9350812272932504e-05,
      "loss": 0.7794,
      "num_tokens": 1187698443.0,
      "step": 388
    },
    {
      "epoch": 0.5700677779813152,
      "grad_norm": 0.6141934112720102,
      "learning_rate": 3.934626040444724e-05,
      "loss": 0.7885,
      "num_tokens": 1190909470.0,
      "step": 389
    },
    {
      "epoch": 0.5715332478475912,
      "grad_norm": 0.5316224934856943,
      "learning_rate": 3.934169292876475e-05,
      "loss": 0.7822,
      "num_tokens": 1193782360.0,
      "step": 390
    },
    {
      "epoch": 0.572998717713867,
      "grad_norm": 0.6855633623383222,
      "learning_rate": 3.9337109849994606e-05,
      "loss": 0.7819,
      "num_tokens": 1196698697.0,
      "step": 391
    },
    {
      "epoch": 0.5744641875801428,
      "grad_norm": 0.5043655130170671,
      "learning_rate": 3.933251117226039e-05,
      "loss": 0.7861,
      "num_tokens": 1199962102.0,
      "step": 392
    },
    {
      "epoch": 0.5759296574464188,
      "grad_norm": 0.7055760117776626,
      "learning_rate": 3.932789689969975e-05,
      "loss": 0.7836,
      "num_tokens": 1202749896.0,
      "step": 393
    },
    {
      "epoch": 0.5773951273126946,
      "grad_norm": 0.6052393958224018,
      "learning_rate": 3.932326703646435e-05,
      "loss": 0.783,
      "num_tokens": 1205844591.0,
      "step": 394
    },
    {
      "epoch": 0.5788605971789705,
      "grad_norm": 0.6709060336300964,
      "learning_rate": 3.9318621586719865e-05,
      "loss": 0.7868,
      "num_tokens": 1208875270.0,
      "step": 395
    },
    {
      "epoch": 0.5803260670452464,
      "grad_norm": 0.5504767987809124,
      "learning_rate": 3.931396055464603e-05,
      "loss": 0.7945,
      "num_tokens": 1211788947.0,
      "step": 396
    },
    {
      "epoch": 0.5817915369115223,
      "grad_norm": 0.6845434695696377,
      "learning_rate": 3.930928394443658e-05,
      "loss": 0.7871,
      "num_tokens": 1214838846.0,
      "step": 397
    },
    {
      "epoch": 0.5832570067777981,
      "grad_norm": 0.5786687718557336,
      "learning_rate": 3.9304591760299266e-05,
      "loss": 0.7889,
      "num_tokens": 1217622512.0,
      "step": 398
    },
    {
      "epoch": 0.584722476644074,
      "grad_norm": 0.6684765415797735,
      "learning_rate": 3.929988400645585e-05,
      "loss": 0.7703,
      "num_tokens": 1220437989.0,
      "step": 399
    },
    {
      "epoch": 0.5861879465103499,
      "grad_norm": 0.6565781443921442,
      "learning_rate": 3.929516068714211e-05,
      "loss": 0.7759,
      "num_tokens": 1223454041.0,
      "step": 400
    },
    {
      "epoch": 0.5876534163766257,
      "grad_norm": 0.5449202898670066,
      "learning_rate": 3.929042180660782e-05,
      "loss": 0.787,
      "num_tokens": 1226435353.0,
      "step": 401
    },
    {
      "epoch": 0.5891188862429017,
      "grad_norm": 0.6875512812145042,
      "learning_rate": 3.9285667369116764e-05,
      "loss": 0.7888,
      "num_tokens": 1229405928.0,
      "step": 402
    },
    {
      "epoch": 0.5905843561091775,
      "grad_norm": 0.4735567416076701,
      "learning_rate": 3.928089737894672e-05,
      "loss": 0.775,
      "num_tokens": 1232720332.0,
      "step": 403
    },
    {
      "epoch": 0.5920498259754534,
      "grad_norm": 0.735516944930375,
      "learning_rate": 3.9276111840389456e-05,
      "loss": 0.7752,
      "num_tokens": 1235824664.0,
      "step": 404
    },
    {
      "epoch": 0.5935152958417292,
      "grad_norm": 0.6645257659658121,
      "learning_rate": 3.927131075775074e-05,
      "loss": 0.805,
      "num_tokens": 1238529918.0,
      "step": 405
    },
    {
      "epoch": 0.5949807657080052,
      "grad_norm": 0.5638827920306675,
      "learning_rate": 3.9266494135350306e-05,
      "loss": 0.784,
      "num_tokens": 1241546540.0,
      "step": 406
    },
    {
      "epoch": 0.596446235574281,
      "grad_norm": 0.6112222243437431,
      "learning_rate": 3.9261661977521894e-05,
      "loss": 0.781,
      "num_tokens": 1244682810.0,
      "step": 407
    },
    {
      "epoch": 0.5979117054405568,
      "grad_norm": 0.6838035001907536,
      "learning_rate": 3.92568142886132e-05,
      "loss": 0.7665,
      "num_tokens": 1247620565.0,
      "step": 408
    },
    {
      "epoch": 0.5993771753068328,
      "grad_norm": 0.5170317822843736,
      "learning_rate": 3.925195107298592e-05,
      "loss": 0.7842,
      "num_tokens": 1250627433.0,
      "step": 409
    },
    {
      "epoch": 0.6008426451731086,
      "grad_norm": 0.6347012338565816,
      "learning_rate": 3.924707233501568e-05,
      "loss": 0.7931,
      "num_tokens": 1253693480.0,
      "step": 410
    },
    {
      "epoch": 0.6023081150393845,
      "grad_norm": 0.6736550040624822,
      "learning_rate": 3.9242178079092116e-05,
      "loss": 0.7742,
      "num_tokens": 1256774656.0,
      "step": 411
    },
    {
      "epoch": 0.6037735849056604,
      "grad_norm": 0.593308429622272,
      "learning_rate": 3.9237268309618796e-05,
      "loss": 0.7807,
      "num_tokens": 1259897203.0,
      "step": 412
    },
    {
      "epoch": 0.6052390547719363,
      "grad_norm": 0.6138921785316901,
      "learning_rate": 3.923234303101326e-05,
      "loss": 0.781,
      "num_tokens": 1263043492.0,
      "step": 413
    },
    {
      "epoch": 0.6067045246382121,
      "grad_norm": 0.5987593300522766,
      "learning_rate": 3.922740224770701e-05,
      "loss": 0.7732,
      "num_tokens": 1266154425.0,
      "step": 414
    },
    {
      "epoch": 0.608169994504488,
      "grad_norm": 0.5212218843700888,
      "learning_rate": 3.9222445964145466e-05,
      "loss": 0.7817,
      "num_tokens": 1269169061.0,
      "step": 415
    },
    {
      "epoch": 0.6096354643707639,
      "grad_norm": 0.5665301085024289,
      "learning_rate": 3.921747418478803e-05,
      "loss": 0.7866,
      "num_tokens": 1272417238.0,
      "step": 416
    },
    {
      "epoch": 0.6111009342370397,
      "grad_norm": 0.5137543210301353,
      "learning_rate": 3.921248691410802e-05,
      "loss": 0.776,
      "num_tokens": 1275474812.0,
      "step": 417
    },
    {
      "epoch": 0.6125664041033156,
      "grad_norm": 0.5527216864080352,
      "learning_rate": 3.920748415659272e-05,
      "loss": 0.7871,
      "num_tokens": 1278861034.0,
      "step": 418
    },
    {
      "epoch": 0.6140318739695915,
      "grad_norm": 0.6134234420326042,
      "learning_rate": 3.920246591674332e-05,
      "loss": 0.7845,
      "num_tokens": 1281795150.0,
      "step": 419
    },
    {
      "epoch": 0.6154973438358674,
      "grad_norm": 0.7442124436399065,
      "learning_rate": 3.9197432199074955e-05,
      "loss": 0.7883,
      "num_tokens": 1284921095.0,
      "step": 420
    },
    {
      "epoch": 0.6169628137021432,
      "grad_norm": 0.5380476473910203,
      "learning_rate": 3.9192383008116684e-05,
      "loss": 0.7809,
      "num_tokens": 1287873067.0,
      "step": 421
    },
    {
      "epoch": 0.6184282835684192,
      "grad_norm": 0.6606212619924012,
      "learning_rate": 3.9187318348411484e-05,
      "loss": 0.7732,
      "num_tokens": 1290760966.0,
      "step": 422
    },
    {
      "epoch": 0.619893753434695,
      "grad_norm": 0.592132378648681,
      "learning_rate": 3.9182238224516264e-05,
      "loss": 0.766,
      "num_tokens": 1293817398.0,
      "step": 423
    },
    {
      "epoch": 0.6213592233009708,
      "grad_norm": 0.8815843889587126,
      "learning_rate": 3.917714264100183e-05,
      "loss": 0.7811,
      "num_tokens": 1296831296.0,
      "step": 424
    },
    {
      "epoch": 0.6228246931672468,
      "grad_norm": 0.47232065460420103,
      "learning_rate": 3.9172031602452884e-05,
      "loss": 0.7817,
      "num_tokens": 1299973604.0,
      "step": 425
    },
    {
      "epoch": 0.6242901630335226,
      "grad_norm": 1.115093151183982,
      "learning_rate": 3.9166905113468086e-05,
      "loss": 0.7751,
      "num_tokens": 1303233366.0,
      "step": 426
    },
    {
      "epoch": 0.6257556328997985,
      "grad_norm": 0.7426681872407431,
      "learning_rate": 3.916176317865995e-05,
      "loss": 0.7888,
      "num_tokens": 1306379980.0,
      "step": 427
    },
    {
      "epoch": 0.6272211027660743,
      "grad_norm": 1.0384039770021642,
      "learning_rate": 3.9156605802654896e-05,
      "loss": 0.7799,
      "num_tokens": 1309404593.0,
      "step": 428
    },
    {
      "epoch": 0.6286865726323503,
      "grad_norm": 0.7670076455276811,
      "learning_rate": 3.9151432990093255e-05,
      "loss": 0.774,
      "num_tokens": 1312596883.0,
      "step": 429
    },
    {
      "epoch": 0.6301520424986261,
      "grad_norm": 0.9763673607777046,
      "learning_rate": 3.914624474562923e-05,
      "loss": 0.7653,
      "num_tokens": 1315917367.0,
      "step": 430
    },
    {
      "epoch": 0.631617512364902,
      "grad_norm": 0.7540824912315313,
      "learning_rate": 3.914104107393091e-05,
      "loss": 0.7566,
      "num_tokens": 1318837842.0,
      "step": 431
    },
    {
      "epoch": 0.6330829822311779,
      "grad_norm": 0.953332107944831,
      "learning_rate": 3.913582197968027e-05,
      "loss": 0.7644,
      "num_tokens": 1322279369.0,
      "step": 432
    },
    {
      "epoch": 0.6345484520974537,
      "grad_norm": 0.8161246900306962,
      "learning_rate": 3.9130587467573156e-05,
      "loss": 0.7815,
      "num_tokens": 1325342829.0,
      "step": 433
    },
    {
      "epoch": 0.6360139219637296,
      "grad_norm": 0.7848078108515719,
      "learning_rate": 3.9125337542319306e-05,
      "loss": 0.7904,
      "num_tokens": 1328280112.0,
      "step": 434
    },
    {
      "epoch": 0.6374793918300055,
      "grad_norm": 0.7229667163099274,
      "learning_rate": 3.9120072208642296e-05,
      "loss": 0.7868,
      "num_tokens": 1331197587.0,
      "step": 435
    },
    {
      "epoch": 0.6389448616962814,
      "grad_norm": 0.8101404125823519,
      "learning_rate": 3.911479147127958e-05,
      "loss": 0.7858,
      "num_tokens": 1334364662.0,
      "step": 436
    },
    {
      "epoch": 0.6404103315625572,
      "grad_norm": 0.621806279389199,
      "learning_rate": 3.9109495334982484e-05,
      "loss": 0.789,
      "num_tokens": 1337342389.0,
      "step": 437
    },
    {
      "epoch": 0.6418758014288332,
      "grad_norm": 0.7762252985115091,
      "learning_rate": 3.910418380451615e-05,
      "loss": 0.7683,
      "num_tokens": 1340543873.0,
      "step": 438
    },
    {
      "epoch": 0.643341271295109,
      "grad_norm": 0.6468013970310809,
      "learning_rate": 3.909885688465962e-05,
      "loss": 0.7954,
      "num_tokens": 1343548675.0,
      "step": 439
    },
    {
      "epoch": 0.6448067411613849,
      "grad_norm": 0.8112026541833866,
      "learning_rate": 3.909351458020575e-05,
      "loss": 0.7896,
      "num_tokens": 1346577580.0,
      "step": 440
    },
    {
      "epoch": 0.6462722110276607,
      "grad_norm": 0.7295288410932435,
      "learning_rate": 3.908815689596125e-05,
      "loss": 0.7651,
      "num_tokens": 1349560639.0,
      "step": 441
    },
    {
      "epoch": 0.6477376808939366,
      "grad_norm": 0.6982691689196974,
      "learning_rate": 3.9082783836746665e-05,
      "loss": 0.788,
      "num_tokens": 1352759086.0,
      "step": 442
    },
    {
      "epoch": 0.6492031507602125,
      "grad_norm": 0.8147723140051017,
      "learning_rate": 3.9077395407396365e-05,
      "loss": 0.7754,
      "num_tokens": 1355817913.0,
      "step": 443
    },
    {
      "epoch": 0.6506686206264883,
      "grad_norm": 0.6167307174187188,
      "learning_rate": 3.907199161275858e-05,
      "loss": 0.7621,
      "num_tokens": 1358828142.0,
      "step": 444
    },
    {
      "epoch": 0.6521340904927643,
      "grad_norm": 0.729215913207261,
      "learning_rate": 3.906657245769532e-05,
      "loss": 0.7753,
      "num_tokens": 1361889206.0,
      "step": 445
    },
    {
      "epoch": 0.6535995603590401,
      "grad_norm": 0.6522801125783623,
      "learning_rate": 3.9061137947082445e-05,
      "loss": 0.7879,
      "num_tokens": 1365138469.0,
      "step": 446
    },
    {
      "epoch": 0.655065030225316,
      "grad_norm": 0.7112828083812891,
      "learning_rate": 3.9055688085809635e-05,
      "loss": 0.7735,
      "num_tokens": 1368277984.0,
      "step": 447
    },
    {
      "epoch": 0.6565305000915919,
      "grad_norm": 0.5650826012003861,
      "learning_rate": 3.905022287878036e-05,
      "loss": 0.7747,
      "num_tokens": 1371434001.0,
      "step": 448
    },
    {
      "epoch": 0.6579959699578677,
      "grad_norm": 0.6451291680871525,
      "learning_rate": 3.9044742330911904e-05,
      "loss": 0.7715,
      "num_tokens": 1374392968.0,
      "step": 449
    },
    {
      "epoch": 0.6594614398241436,
      "grad_norm": 0.5762246494944382,
      "learning_rate": 3.9039246447135374e-05,
      "loss": 0.79,
      "num_tokens": 1377171680.0,
      "step": 450
    },
    {
      "epoch": 0.6609269096904195,
      "grad_norm": 0.6953658230451008,
      "learning_rate": 3.903373523239565e-05,
      "loss": 0.7936,
      "num_tokens": 1380071522.0,
      "step": 451
    },
    {
      "epoch": 0.6623923795566954,
      "grad_norm": 0.5468782132147643,
      "learning_rate": 3.902820869165141e-05,
      "loss": 0.7618,
      "num_tokens": 1383012986.0,
      "step": 452
    },
    {
      "epoch": 0.6638578494229712,
      "grad_norm": 0.8415941788046309,
      "learning_rate": 3.902266682987514e-05,
      "loss": 0.7966,
      "num_tokens": 1385910595.0,
      "step": 453
    },
    {
      "epoch": 0.6653233192892472,
      "grad_norm": 0.6087189391945046,
      "learning_rate": 3.9017109652053085e-05,
      "loss": 0.757,
      "num_tokens": 1389136825.0,
      "step": 454
    },
    {
      "epoch": 0.666788789155523,
      "grad_norm": 0.8174995843209238,
      "learning_rate": 3.9011537163185294e-05,
      "loss": 0.7779,
      "num_tokens": 1392384785.0,
      "step": 455
    },
    {
      "epoch": 0.6682542590217989,
      "grad_norm": 0.6976440952330858,
      "learning_rate": 3.9005949368285575e-05,
      "loss": 0.8005,
      "num_tokens": 1395371553.0,
      "step": 456
    },
    {
      "epoch": 0.6697197288880747,
      "grad_norm": 0.661072210064578,
      "learning_rate": 3.9000346272381516e-05,
      "loss": 0.7781,
      "num_tokens": 1398462633.0,
      "step": 457
    },
    {
      "epoch": 0.6711851987543506,
      "grad_norm": 0.6777063662822085,
      "learning_rate": 3.899472788051448e-05,
      "loss": 0.7852,
      "num_tokens": 1401596380.0,
      "step": 458
    },
    {
      "epoch": 0.6726506686206265,
      "grad_norm": 0.7089751388455204,
      "learning_rate": 3.898909419773956e-05,
      "loss": 0.7788,
      "num_tokens": 1404918933.0,
      "step": 459
    },
    {
      "epoch": 0.6741161384869023,
      "grad_norm": 0.580366557626494,
      "learning_rate": 3.898344522912565e-05,
      "loss": 0.7719,
      "num_tokens": 1407976306.0,
      "step": 460
    },
    {
      "epoch": 0.6755816083531783,
      "grad_norm": 0.7152891481812426,
      "learning_rate": 3.897778097975537e-05,
      "loss": 0.7836,
      "num_tokens": 1411177240.0,
      "step": 461
    },
    {
      "epoch": 0.6770470782194541,
      "grad_norm": 0.534737628124812,
      "learning_rate": 3.897210145472509e-05,
      "loss": 0.7817,
      "num_tokens": 1414218311.0,
      "step": 462
    },
    {
      "epoch": 0.67851254808573,
      "grad_norm": 0.6488712988579897,
      "learning_rate": 3.896640665914494e-05,
      "loss": 0.7755,
      "num_tokens": 1417136629.0,
      "step": 463
    },
    {
      "epoch": 0.6799780179520059,
      "grad_norm": 0.48440960634838226,
      "learning_rate": 3.896069659813878e-05,
      "loss": 0.7725,
      "num_tokens": 1420178133.0,
      "step": 464
    },
    {
      "epoch": 0.6814434878182818,
      "grad_norm": 0.7834086421903245,
      "learning_rate": 3.8954971276844184e-05,
      "loss": 0.7938,
      "num_tokens": 1423121554.0,
      "step": 465
    },
    {
      "epoch": 0.6829089576845576,
      "grad_norm": 0.599529069777397,
      "learning_rate": 3.89492307004125e-05,
      "loss": 0.7888,
      "num_tokens": 1426180217.0,
      "step": 466
    },
    {
      "epoch": 0.6843744275508334,
      "grad_norm": 0.6866984516490249,
      "learning_rate": 3.894347487400876e-05,
      "loss": 0.7763,
      "num_tokens": 1429299039.0,
      "step": 467
    },
    {
      "epoch": 0.6858398974171094,
      "grad_norm": 0.7533196608962499,
      "learning_rate": 3.893770380281175e-05,
      "loss": 0.7718,
      "num_tokens": 1432360004.0,
      "step": 468
    },
    {
      "epoch": 0.6873053672833852,
      "grad_norm": 0.5443359379346597,
      "learning_rate": 3.893191749201395e-05,
      "loss": 0.7578,
      "num_tokens": 1435359898.0,
      "step": 469
    },
    {
      "epoch": 0.6887708371496611,
      "grad_norm": 0.853703474916966,
      "learning_rate": 3.892611594682156e-05,
      "loss": 0.7773,
      "num_tokens": 1438209499.0,
      "step": 470
    },
    {
      "epoch": 0.690236307015937,
      "grad_norm": 0.6040732201550099,
      "learning_rate": 3.892029917245449e-05,
      "loss": 0.7915,
      "num_tokens": 1441072799.0,
      "step": 471
    },
    {
      "epoch": 0.6917017768822129,
      "grad_norm": 0.7784781059909279,
      "learning_rate": 3.891446717414635e-05,
      "loss": 0.7753,
      "num_tokens": 1444188254.0,
      "step": 472
    },
    {
      "epoch": 0.6931672467484887,
      "grad_norm": 0.634458856385789,
      "learning_rate": 3.8908619957144446e-05,
      "loss": 0.7839,
      "num_tokens": 1447221260.0,
      "step": 473
    },
    {
      "epoch": 0.6946327166147646,
      "grad_norm": 0.764673041289811,
      "learning_rate": 3.890275752670978e-05,
      "loss": 0.788,
      "num_tokens": 1450300624.0,
      "step": 474
    },
    {
      "epoch": 0.6960981864810405,
      "grad_norm": 0.6606186030664977,
      "learning_rate": 3.8896879888117026e-05,
      "loss": 0.7798,
      "num_tokens": 1453288079.0,
      "step": 475
    },
    {
      "epoch": 0.6975636563473163,
      "grad_norm": 0.6083189697546619,
      "learning_rate": 3.889098704665457e-05,
      "loss": 0.7699,
      "num_tokens": 1456315233.0,
      "step": 476
    },
    {
      "epoch": 0.6990291262135923,
      "grad_norm": 0.7083192251172743,
      "learning_rate": 3.8885079007624456e-05,
      "loss": 0.7735,
      "num_tokens": 1459527291.0,
      "step": 477
    },
    {
      "epoch": 0.7004945960798681,
      "grad_norm": 0.5412000913569363,
      "learning_rate": 3.8879155776342405e-05,
      "loss": 0.7885,
      "num_tokens": 1462729527.0,
      "step": 478
    },
    {
      "epoch": 0.701960065946144,
      "grad_norm": 0.9034052043802873,
      "learning_rate": 3.8873217358137814e-05,
      "loss": 0.7653,
      "num_tokens": 1465896073.0,
      "step": 479
    },
    {
      "epoch": 0.7034255358124198,
      "grad_norm": 0.6303286250366069,
      "learning_rate": 3.8867263758353746e-05,
      "loss": 0.7532,
      "num_tokens": 1469030392.0,
      "step": 480
    },
    {
      "epoch": 0.7048910056786958,
      "grad_norm": 0.7924394172209809,
      "learning_rate": 3.8861294982346905e-05,
      "loss": 0.7538,
      "num_tokens": 1472130503.0,
      "step": 481
    },
    {
      "epoch": 0.7063564755449716,
      "grad_norm": 0.602683271414693,
      "learning_rate": 3.885531103548768e-05,
      "loss": 0.768,
      "num_tokens": 1475017493.0,
      "step": 482
    },
    {
      "epoch": 0.7078219454112474,
      "grad_norm": 0.7463579993927153,
      "learning_rate": 3.884931192316007e-05,
      "loss": 0.7825,
      "num_tokens": 1478011774.0,
      "step": 483
    },
    {
      "epoch": 0.7092874152775234,
      "grad_norm": 0.5652187937745072,
      "learning_rate": 3.884329765076176e-05,
      "loss": 0.7591,
      "num_tokens": 1481019740.0,
      "step": 484
    },
    {
      "epoch": 0.7107528851437992,
      "grad_norm": 0.7411279672493576,
      "learning_rate": 3.8837268223704045e-05,
      "loss": 0.7722,
      "num_tokens": 1483932004.0,
      "step": 485
    },
    {
      "epoch": 0.7122183550100751,
      "grad_norm": 0.5838860770566946,
      "learning_rate": 3.883122364741188e-05,
      "loss": 0.7861,
      "num_tokens": 1486898777.0,
      "step": 486
    },
    {
      "epoch": 0.713683824876351,
      "grad_norm": 0.8460749934271568,
      "learning_rate": 3.8825163927323825e-05,
      "loss": 0.7775,
      "num_tokens": 1490122731.0,
      "step": 487
    },
    {
      "epoch": 0.7151492947426269,
      "grad_norm": 0.6077038252936527,
      "learning_rate": 3.881908906889208e-05,
      "loss": 0.7933,
      "num_tokens": 1493182135.0,
      "step": 488
    },
    {
      "epoch": 0.7166147646089027,
      "grad_norm": 0.7473797197512128,
      "learning_rate": 3.881299907758247e-05,
      "loss": 0.7685,
      "num_tokens": 1496024863.0,
      "step": 489
    },
    {
      "epoch": 0.7180802344751787,
      "grad_norm": 0.6096257828107025,
      "learning_rate": 3.880689395887443e-05,
      "loss": 0.7585,
      "num_tokens": 1499232607.0,
      "step": 490
    },
    {
      "epoch": 0.7195457043414545,
      "grad_norm": 0.7702386432638039,
      "learning_rate": 3.8800773718261e-05,
      "loss": 0.7686,
      "num_tokens": 1502277307.0,
      "step": 491
    },
    {
      "epoch": 0.7210111742077303,
      "grad_norm": 0.6071741645971063,
      "learning_rate": 3.879463836124882e-05,
      "loss": 0.8017,
      "num_tokens": 1505081067.0,
      "step": 492
    },
    {
      "epoch": 0.7224766440740062,
      "grad_norm": 0.833216656808556,
      "learning_rate": 3.878848789335817e-05,
      "loss": 0.7591,
      "num_tokens": 1508130314.0,
      "step": 493
    },
    {
      "epoch": 0.7239421139402821,
      "grad_norm": 0.7333919896146535,
      "learning_rate": 3.878232232012287e-05,
      "loss": 0.7818,
      "num_tokens": 1511194457.0,
      "step": 494
    },
    {
      "epoch": 0.725407583806558,
      "grad_norm": 0.7226789755837114,
      "learning_rate": 3.8776141647090375e-05,
      "loss": 0.7846,
      "num_tokens": 1514382240.0,
      "step": 495
    },
    {
      "epoch": 0.7268730536728338,
      "grad_norm": 0.7569312313476318,
      "learning_rate": 3.87699458798217e-05,
      "loss": 0.8025,
      "num_tokens": 1517380000.0,
      "step": 496
    },
    {
      "epoch": 0.7283385235391098,
      "grad_norm": 0.528979561273253,
      "learning_rate": 3.8763735023891464e-05,
      "loss": 0.7741,
      "num_tokens": 1520176959.0,
      "step": 497
    },
    {
      "epoch": 0.7298039934053856,
      "grad_norm": 0.8142373663026293,
      "learning_rate": 3.875750908488784e-05,
      "loss": 0.7628,
      "num_tokens": 1523340234.0,
      "step": 498
    },
    {
      "epoch": 0.7312694632716614,
      "grad_norm": 0.5947926582686632,
      "learning_rate": 3.875126806841258e-05,
      "loss": 0.7675,
      "num_tokens": 1526236003.0,
      "step": 499
    },
    {
      "epoch": 0.7327349331379374,
      "grad_norm": 0.7104407265737775,
      "learning_rate": 3.8745011980081e-05,
      "loss": 0.7724,
      "num_tokens": 1529336302.0,
      "step": 500
    },
    {
      "epoch": 0.7342004030042132,
      "grad_norm": 0.6697752916424362,
      "learning_rate": 3.873874082552199e-05,
      "loss": 0.7758,
      "num_tokens": 1532401849.0,
      "step": 501
    },
    {
      "epoch": 0.7356658728704891,
      "grad_norm": 0.7068147486297662,
      "learning_rate": 3.873245461037797e-05,
      "loss": 0.7964,
      "num_tokens": 1535510117.0,
      "step": 502
    },
    {
      "epoch": 0.737131342736765,
      "grad_norm": 0.6843236221096626,
      "learning_rate": 3.8726153340304926e-05,
      "loss": 0.7732,
      "num_tokens": 1538520247.0,
      "step": 503
    },
    {
      "epoch": 0.7385968126030409,
      "grad_norm": 0.6172541260094148,
      "learning_rate": 3.871983702097241e-05,
      "loss": 0.7662,
      "num_tokens": 1541583964.0,
      "step": 504
    },
    {
      "epoch": 0.7400622824693167,
      "grad_norm": 0.6765277604121307,
      "learning_rate": 3.8713505658063476e-05,
      "loss": 0.771,
      "num_tokens": 1544463505.0,
      "step": 505
    },
    {
      "epoch": 0.7415277523355926,
      "grad_norm": 0.5230724744281858,
      "learning_rate": 3.8707159257274734e-05,
      "loss": 0.7809,
      "num_tokens": 1547557309.0,
      "step": 506
    },
    {
      "epoch": 0.7429932222018685,
      "grad_norm": 0.6284111259896931,
      "learning_rate": 3.870079782431632e-05,
      "loss": 0.7791,
      "num_tokens": 1550458461.0,
      "step": 507
    },
    {
      "epoch": 0.7444586920681443,
      "grad_norm": 0.6094597467251823,
      "learning_rate": 3.869442136491191e-05,
      "loss": 0.7733,
      "num_tokens": 1553466783.0,
      "step": 508
    },
    {
      "epoch": 0.7459241619344202,
      "grad_norm": 0.6263105032387869,
      "learning_rate": 3.868802988479866e-05,
      "loss": 0.7565,
      "num_tokens": 1557008221.0,
      "step": 509
    },
    {
      "epoch": 0.7473896318006961,
      "grad_norm": 0.5884792669598419,
      "learning_rate": 3.8681623389727295e-05,
      "loss": 0.7713,
      "num_tokens": 1560029652.0,
      "step": 510
    },
    {
      "epoch": 0.748855101666972,
      "grad_norm": 0.5006096746404923,
      "learning_rate": 3.8675201885462e-05,
      "loss": 0.766,
      "num_tokens": 1563004556.0,
      "step": 511
    },
    {
      "epoch": 0.7503205715332478,
      "grad_norm": 0.7167339651060948,
      "learning_rate": 3.866876537778049e-05,
      "loss": 0.7824,
      "num_tokens": 1566005000.0,
      "step": 512
    },
    {
      "epoch": 0.7517860413995238,
      "grad_norm": 0.4953478599593932,
      "learning_rate": 3.8662313872473995e-05,
      "loss": 0.7686,
      "num_tokens": 1569059505.0,
      "step": 513
    },
    {
      "epoch": 0.7532515112657996,
      "grad_norm": 0.5031118002515562,
      "learning_rate": 3.8655847375347206e-05,
      "loss": 0.7757,
      "num_tokens": 1572217300.0,
      "step": 514
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 0.5426351733471624,
      "learning_rate": 3.864936589221831e-05,
      "loss": 0.7659,
      "num_tokens": 1575210357.0,
      "step": 515
    },
    {
      "epoch": 0.7561824509983514,
      "grad_norm": 0.6525469950913664,
      "learning_rate": 3.864286942891898e-05,
      "loss": 0.7663,
      "num_tokens": 1578123445.0,
      "step": 516
    },
    {
      "epoch": 0.7576479208646272,
      "grad_norm": 0.630243241204321,
      "learning_rate": 3.86363579912944e-05,
      "loss": 0.7809,
      "num_tokens": 1581090645.0,
      "step": 517
    },
    {
      "epoch": 0.7591133907309031,
      "grad_norm": 0.49214684456991553,
      "learning_rate": 3.862983158520316e-05,
      "loss": 0.7747,
      "num_tokens": 1584330857.0,
      "step": 518
    },
    {
      "epoch": 0.7605788605971789,
      "grad_norm": 0.5863879783395399,
      "learning_rate": 3.862329021651739e-05,
      "loss": 0.768,
      "num_tokens": 1587377799.0,
      "step": 519
    },
    {
      "epoch": 0.7620443304634549,
      "grad_norm": 0.6522482556156715,
      "learning_rate": 3.861673389112262e-05,
      "loss": 0.7742,
      "num_tokens": 1590573823.0,
      "step": 520
    },
    {
      "epoch": 0.7635098003297307,
      "grad_norm": 0.41971785510632237,
      "learning_rate": 3.8610162614917894e-05,
      "loss": 0.7707,
      "num_tokens": 1593652876.0,
      "step": 521
    },
    {
      "epoch": 0.7649752701960066,
      "grad_norm": 0.8835081541898636,
      "learning_rate": 3.860357639381566e-05,
      "loss": 0.7801,
      "num_tokens": 1596867208.0,
      "step": 522
    },
    {
      "epoch": 0.7664407400622825,
      "grad_norm": 0.5152010993737639,
      "learning_rate": 3.859697523374183e-05,
      "loss": 0.7809,
      "num_tokens": 1599825066.0,
      "step": 523
    },
    {
      "epoch": 0.7679062099285583,
      "grad_norm": 0.8714571453596084,
      "learning_rate": 3.859035914063577e-05,
      "loss": 0.7873,
      "num_tokens": 1602974774.0,
      "step": 524
    },
    {
      "epoch": 0.7693716797948342,
      "grad_norm": 0.5784098116916205,
      "learning_rate": 3.858372812045028e-05,
      "loss": 0.7771,
      "num_tokens": 1606035021.0,
      "step": 525
    },
    {
      "epoch": 0.77083714966111,
      "grad_norm": 0.8381661227962233,
      "learning_rate": 3.857708217915156e-05,
      "loss": 0.7522,
      "num_tokens": 1609266943.0,
      "step": 526
    },
    {
      "epoch": 0.772302619527386,
      "grad_norm": 0.5982072885783272,
      "learning_rate": 3.857042132271926e-05,
      "loss": 0.7529,
      "num_tokens": 1612830214.0,
      "step": 527
    },
    {
      "epoch": 0.7737680893936618,
      "grad_norm": 0.7758919332290647,
      "learning_rate": 3.8563745557146466e-05,
      "loss": 0.7943,
      "num_tokens": 1615457501.0,
      "step": 528
    },
    {
      "epoch": 0.7752335592599378,
      "grad_norm": 0.5661228425225252,
      "learning_rate": 3.8557054888439636e-05,
      "loss": 0.772,
      "num_tokens": 1618288030.0,
      "step": 529
    },
    {
      "epoch": 0.7766990291262136,
      "grad_norm": 0.7197739030423916,
      "learning_rate": 3.8550349322618685e-05,
      "loss": 0.7735,
      "num_tokens": 1621146668.0,
      "step": 530
    },
    {
      "epoch": 0.7781644989924895,
      "grad_norm": 0.660519962074317,
      "learning_rate": 3.8543628865716886e-05,
      "loss": 0.8066,
      "num_tokens": 1624154648.0,
      "step": 531
    },
    {
      "epoch": 0.7796299688587653,
      "grad_norm": 0.8140158300335816,
      "learning_rate": 3.8536893523780944e-05,
      "loss": 0.7739,
      "num_tokens": 1627411378.0,
      "step": 532
    },
    {
      "epoch": 0.7810954387250412,
      "grad_norm": 0.6517068414951401,
      "learning_rate": 3.853014330287093e-05,
      "loss": 0.7677,
      "num_tokens": 1630447325.0,
      "step": 533
    },
    {
      "epoch": 0.7825609085913171,
      "grad_norm": 0.7886290194470019,
      "learning_rate": 3.852337820906033e-05,
      "loss": 0.78,
      "num_tokens": 1633236728.0,
      "step": 534
    },
    {
      "epoch": 0.7840263784575929,
      "grad_norm": 0.6662349152469276,
      "learning_rate": 3.8516598248436e-05,
      "loss": 0.7866,
      "num_tokens": 1636053550.0,
      "step": 535
    },
    {
      "epoch": 0.7854918483238689,
      "grad_norm": 0.6597644024068506,
      "learning_rate": 3.850980342709816e-05,
      "loss": 0.7674,
      "num_tokens": 1639185387.0,
      "step": 536
    },
    {
      "epoch": 0.7869573181901447,
      "grad_norm": 0.6628934212702776,
      "learning_rate": 3.8502993751160406e-05,
      "loss": 0.7733,
      "num_tokens": 1642324692.0,
      "step": 537
    },
    {
      "epoch": 0.7884227880564206,
      "grad_norm": 0.5531629243895193,
      "learning_rate": 3.8496169226749725e-05,
      "loss": 0.7857,
      "num_tokens": 1645256217.0,
      "step": 538
    },
    {
      "epoch": 0.7898882579226965,
      "grad_norm": 0.7072817019038333,
      "learning_rate": 3.8489329860006426e-05,
      "loss": 0.7546,
      "num_tokens": 1648541265.0,
      "step": 539
    },
    {
      "epoch": 0.7913537277889724,
      "grad_norm": 0.5655990813132724,
      "learning_rate": 3.848247565708419e-05,
      "loss": 0.7524,
      "num_tokens": 1651817564.0,
      "step": 540
    },
    {
      "epoch": 0.7928191976552482,
      "grad_norm": 0.518083058589993,
      "learning_rate": 3.8475606624150055e-05,
      "loss": 0.7666,
      "num_tokens": 1654699908.0,
      "step": 541
    },
    {
      "epoch": 0.794284667521524,
      "grad_norm": 0.6114597863287824,
      "learning_rate": 3.8468722767384386e-05,
      "loss": 0.7505,
      "num_tokens": 1657639430.0,
      "step": 542
    },
    {
      "epoch": 0.7957501373878,
      "grad_norm": 0.5686664510550941,
      "learning_rate": 3.84618240929809e-05,
      "loss": 0.781,
      "num_tokens": 1660594851.0,
      "step": 543
    },
    {
      "epoch": 0.7972156072540758,
      "grad_norm": 0.6263953588269895,
      "learning_rate": 3.8454910607146634e-05,
      "loss": 0.7592,
      "num_tokens": 1663687166.0,
      "step": 544
    },
    {
      "epoch": 0.7986810771203517,
      "grad_norm": 0.593559663144848,
      "learning_rate": 3.844798231610196e-05,
      "loss": 0.7649,
      "num_tokens": 1666804213.0,
      "step": 545
    },
    {
      "epoch": 0.8001465469866276,
      "grad_norm": 0.6393386034179077,
      "learning_rate": 3.844103922608057e-05,
      "loss": 0.7995,
      "num_tokens": 1669948659.0,
      "step": 546
    },
    {
      "epoch": 0.8016120168529035,
      "grad_norm": 0.5159120808173705,
      "learning_rate": 3.843408134332946e-05,
      "loss": 0.7532,
      "num_tokens": 1672947803.0,
      "step": 547
    },
    {
      "epoch": 0.8030774867191793,
      "grad_norm": 0.859704414753804,
      "learning_rate": 3.842710867410895e-05,
      "loss": 0.7712,
      "num_tokens": 1675898262.0,
      "step": 548
    },
    {
      "epoch": 0.8045429565854552,
      "grad_norm": 0.5255441946988123,
      "learning_rate": 3.842012122469266e-05,
      "loss": 0.7569,
      "num_tokens": 1678870562.0,
      "step": 549
    },
    {
      "epoch": 0.8060084264517311,
      "grad_norm": 0.6058251553544091,
      "learning_rate": 3.841311900136751e-05,
      "loss": 0.7463,
      "num_tokens": 1681961893.0,
      "step": 550
    },
    {
      "epoch": 0.8074738963180069,
      "grad_norm": 0.7387923965593198,
      "learning_rate": 3.84061020104337e-05,
      "loss": 0.7774,
      "num_tokens": 1684915961.0,
      "step": 551
    },
    {
      "epoch": 0.8089393661842829,
      "grad_norm": 0.508863830599595,
      "learning_rate": 3.839907025820474e-05,
      "loss": 0.7832,
      "num_tokens": 1687823301.0,
      "step": 552
    },
    {
      "epoch": 0.8104048360505587,
      "grad_norm": 0.5867690563669935,
      "learning_rate": 3.839202375100739e-05,
      "loss": 0.7934,
      "num_tokens": 1690689894.0,
      "step": 553
    },
    {
      "epoch": 0.8118703059168346,
      "grad_norm": 0.6860803174324955,
      "learning_rate": 3.8384962495181724e-05,
      "loss": 0.7835,
      "num_tokens": 1693640395.0,
      "step": 554
    },
    {
      "epoch": 0.8133357757831104,
      "grad_norm": 0.5034383439781328,
      "learning_rate": 3.8377886497081054e-05,
      "loss": 0.7659,
      "num_tokens": 1696721998.0,
      "step": 555
    },
    {
      "epoch": 0.8148012456493864,
      "grad_norm": 0.5723866443275422,
      "learning_rate": 3.8370795763071974e-05,
      "loss": 0.7715,
      "num_tokens": 1699940440.0,
      "step": 556
    },
    {
      "epoch": 0.8162667155156622,
      "grad_norm": 0.7362265331009978,
      "learning_rate": 3.836369029953432e-05,
      "loss": 0.7753,
      "num_tokens": 1703301089.0,
      "step": 557
    },
    {
      "epoch": 0.817732185381938,
      "grad_norm": 0.5840870466351289,
      "learning_rate": 3.835657011286121e-05,
      "loss": 0.7538,
      "num_tokens": 1706454921.0,
      "step": 558
    },
    {
      "epoch": 0.819197655248214,
      "grad_norm": 0.5685976903083878,
      "learning_rate": 3.834943520945897e-05,
      "loss": 0.7741,
      "num_tokens": 1709625437.0,
      "step": 559
    },
    {
      "epoch": 0.8206631251144898,
      "grad_norm": 0.8711292180046611,
      "learning_rate": 3.83422855957472e-05,
      "loss": 0.7546,
      "num_tokens": 1712752714.0,
      "step": 560
    },
    {
      "epoch": 0.8221285949807657,
      "grad_norm": 0.513888958021349,
      "learning_rate": 3.833512127815873e-05,
      "loss": 0.7712,
      "num_tokens": 1715832151.0,
      "step": 561
    },
    {
      "epoch": 0.8235940648470416,
      "grad_norm": 1.123361662531118,
      "learning_rate": 3.83279422631396e-05,
      "loss": 0.75,
      "num_tokens": 1718946506.0,
      "step": 562
    },
    {
      "epoch": 0.8250595347133175,
      "grad_norm": 0.9046599041199087,
      "learning_rate": 3.832074855714909e-05,
      "loss": 0.789,
      "num_tokens": 1722009749.0,
      "step": 563
    },
    {
      "epoch": 0.8265250045795933,
      "grad_norm": 1.0516863082542982,
      "learning_rate": 3.83135401666597e-05,
      "loss": 0.7683,
      "num_tokens": 1724900146.0,
      "step": 564
    },
    {
      "epoch": 0.8279904744458693,
      "grad_norm": 0.9850461210213872,
      "learning_rate": 3.8306317098157135e-05,
      "loss": 0.763,
      "num_tokens": 1728198817.0,
      "step": 565
    },
    {
      "epoch": 0.8294559443121451,
      "grad_norm": 0.848996674112302,
      "learning_rate": 3.829907935814031e-05,
      "loss": 0.7533,
      "num_tokens": 1731133358.0,
      "step": 566
    },
    {
      "epoch": 0.8309214141784209,
      "grad_norm": 0.7782184320669764,
      "learning_rate": 3.829182695312134e-05,
      "loss": 0.7608,
      "num_tokens": 1734403770.0,
      "step": 567
    },
    {
      "epoch": 0.8323868840446969,
      "grad_norm": 0.7828890158174929,
      "learning_rate": 3.828455988962553e-05,
      "loss": 0.7819,
      "num_tokens": 1737385413.0,
      "step": 568
    },
    {
      "epoch": 0.8338523539109727,
      "grad_norm": 0.7079152085683248,
      "learning_rate": 3.827727817419138e-05,
      "loss": 0.7766,
      "num_tokens": 1740556447.0,
      "step": 569
    },
    {
      "epoch": 0.8353178237772486,
      "grad_norm": 0.7639860590735518,
      "learning_rate": 3.8269981813370576e-05,
      "loss": 0.7727,
      "num_tokens": 1743547906.0,
      "step": 570
    },
    {
      "epoch": 0.8367832936435244,
      "grad_norm": 0.6579434434211288,
      "learning_rate": 3.8262670813727964e-05,
      "loss": 0.7725,
      "num_tokens": 1746704671.0,
      "step": 571
    },
    {
      "epoch": 0.8382487635098004,
      "grad_norm": 0.8593002193880989,
      "learning_rate": 3.825534518184159e-05,
      "loss": 0.7614,
      "num_tokens": 1749743059.0,
      "step": 572
    },
    {
      "epoch": 0.8397142333760762,
      "grad_norm": 0.6201447637005403,
      "learning_rate": 3.824800492430264e-05,
      "loss": 0.7417,
      "num_tokens": 1753037690.0,
      "step": 573
    },
    {
      "epoch": 0.841179703242352,
      "grad_norm": 0.7348326542706543,
      "learning_rate": 3.824065004771547e-05,
      "loss": 0.7621,
      "num_tokens": 1755987574.0,
      "step": 574
    },
    {
      "epoch": 0.842645173108628,
      "grad_norm": 0.5753543623117091,
      "learning_rate": 3.823328055869759e-05,
      "loss": 0.7687,
      "num_tokens": 1759155412.0,
      "step": 575
    },
    {
      "epoch": 0.8441106429749038,
      "grad_norm": 0.867551702155227,
      "learning_rate": 3.8225896463879645e-05,
      "loss": 0.7711,
      "num_tokens": 1762221945.0,
      "step": 576
    },
    {
      "epoch": 0.8455761128411797,
      "grad_norm": 0.6886997315319308,
      "learning_rate": 3.821849776990544e-05,
      "loss": 0.7593,
      "num_tokens": 1765184272.0,
      "step": 577
    },
    {
      "epoch": 0.8470415827074556,
      "grad_norm": 0.7583845133299721,
      "learning_rate": 3.8211084483431905e-05,
      "loss": 0.7552,
      "num_tokens": 1768176944.0,
      "step": 578
    },
    {
      "epoch": 0.8485070525737315,
      "grad_norm": 0.6706691330864654,
      "learning_rate": 3.820365661112911e-05,
      "loss": 0.7614,
      "num_tokens": 1771268843.0,
      "step": 579
    },
    {
      "epoch": 0.8499725224400073,
      "grad_norm": 0.8022246493600097,
      "learning_rate": 3.819621415968022e-05,
      "loss": 0.7765,
      "num_tokens": 1774237204.0,
      "step": 580
    },
    {
      "epoch": 0.8514379923062833,
      "grad_norm": 0.686071693150319,
      "learning_rate": 3.8188757135781555e-05,
      "loss": 0.7601,
      "num_tokens": 1777192729.0,
      "step": 581
    },
    {
      "epoch": 0.8529034621725591,
      "grad_norm": 0.7164221100816927,
      "learning_rate": 3.8181285546142516e-05,
      "loss": 0.7696,
      "num_tokens": 1780591122.0,
      "step": 582
    },
    {
      "epoch": 0.8543689320388349,
      "grad_norm": 0.6258251071096054,
      "learning_rate": 3.817379939748564e-05,
      "loss": 0.7793,
      "num_tokens": 1783720426.0,
      "step": 583
    },
    {
      "epoch": 0.8558344019051108,
      "grad_norm": 0.6613147576084321,
      "learning_rate": 3.8166298696546535e-05,
      "loss": 0.7492,
      "num_tokens": 1786456272.0,
      "step": 584
    },
    {
      "epoch": 0.8572998717713867,
      "grad_norm": 0.5617988990411139,
      "learning_rate": 3.815878345007391e-05,
      "loss": 0.7699,
      "num_tokens": 1789577159.0,
      "step": 585
    },
    {
      "epoch": 0.8587653416376626,
      "grad_norm": 0.7354277155930304,
      "learning_rate": 3.815125366482957e-05,
      "loss": 0.7583,
      "num_tokens": 1792698276.0,
      "step": 586
    },
    {
      "epoch": 0.8602308115039384,
      "grad_norm": 0.6010858642506439,
      "learning_rate": 3.8143709347588396e-05,
      "loss": 0.755,
      "num_tokens": 1795728576.0,
      "step": 587
    },
    {
      "epoch": 0.8616962813702144,
      "grad_norm": 0.6919995060659836,
      "learning_rate": 3.8136150505138336e-05,
      "loss": 0.786,
      "num_tokens": 1798751032.0,
      "step": 588
    },
    {
      "epoch": 0.8631617512364902,
      "grad_norm": 0.6231253593139175,
      "learning_rate": 3.8128577144280424e-05,
      "loss": 0.7697,
      "num_tokens": 1801685360.0,
      "step": 589
    },
    {
      "epoch": 0.864627221102766,
      "grad_norm": 0.5251147821580963,
      "learning_rate": 3.812098927182874e-05,
      "loss": 0.7834,
      "num_tokens": 1804558735.0,
      "step": 590
    },
    {
      "epoch": 0.866092690969042,
      "grad_norm": 0.5855076540161251,
      "learning_rate": 3.811338689461044e-05,
      "loss": 0.7675,
      "num_tokens": 1807850994.0,
      "step": 591
    },
    {
      "epoch": 0.8675581608353178,
      "grad_norm": 0.6121148974997432,
      "learning_rate": 3.8105770019465706e-05,
      "loss": 0.7601,
      "num_tokens": 1810969593.0,
      "step": 592
    },
    {
      "epoch": 0.8690236307015937,
      "grad_norm": 0.6031889914215367,
      "learning_rate": 3.8098138653247785e-05,
      "loss": 0.7798,
      "num_tokens": 1814052780.0,
      "step": 593
    },
    {
      "epoch": 0.8704891005678695,
      "grad_norm": 0.558087348577981,
      "learning_rate": 3.8090492802822946e-05,
      "loss": 0.7539,
      "num_tokens": 1817391223.0,
      "step": 594
    },
    {
      "epoch": 0.8719545704341455,
      "grad_norm": 0.5549099055701825,
      "learning_rate": 3.8082832475070505e-05,
      "loss": 0.7763,
      "num_tokens": 1820417584.0,
      "step": 595
    },
    {
      "epoch": 0.8734200403004213,
      "grad_norm": 0.6208111128014616,
      "learning_rate": 3.80751576768828e-05,
      "loss": 0.785,
      "num_tokens": 1823515514.0,
      "step": 596
    },
    {
      "epoch": 0.8748855101666972,
      "grad_norm": 0.4976925243765284,
      "learning_rate": 3.806746841516518e-05,
      "loss": 0.7591,
      "num_tokens": 1826899894.0,
      "step": 597
    },
    {
      "epoch": 0.8763509800329731,
      "grad_norm": 0.7321468338160054,
      "learning_rate": 3.805976469683602e-05,
      "loss": 0.7824,
      "num_tokens": 1829798308.0,
      "step": 598
    },
    {
      "epoch": 0.8778164498992489,
      "grad_norm": 0.47468106625610196,
      "learning_rate": 3.805204652882669e-05,
      "loss": 0.77,
      "num_tokens": 1832599869.0,
      "step": 599
    },
    {
      "epoch": 0.8792819197655248,
      "grad_norm": 0.589599283303487,
      "learning_rate": 3.804431391808157e-05,
      "loss": 0.7773,
      "num_tokens": 1835854114.0,
      "step": 600
    },
    {
      "epoch": 0.8807473896318007,
      "grad_norm": 0.5939112988176937,
      "learning_rate": 3.8036566871558035e-05,
      "loss": 0.7534,
      "num_tokens": 1839016372.0,
      "step": 601
    },
    {
      "epoch": 0.8822128594980766,
      "grad_norm": 0.7221522422621756,
      "learning_rate": 3.802880539622644e-05,
      "loss": 0.7578,
      "num_tokens": 1841980827.0,
      "step": 602
    },
    {
      "epoch": 0.8836783293643524,
      "grad_norm": 0.5087678196844709,
      "learning_rate": 3.802102949907012e-05,
      "loss": 0.7538,
      "num_tokens": 1845008150.0,
      "step": 603
    },
    {
      "epoch": 0.8851437992306284,
      "grad_norm": 0.7152874057852808,
      "learning_rate": 3.801323918708541e-05,
      "loss": 0.7638,
      "num_tokens": 1847736990.0,
      "step": 604
    },
    {
      "epoch": 0.8866092690969042,
      "grad_norm": 0.5427245440311039,
      "learning_rate": 3.800543446728159e-05,
      "loss": 0.7684,
      "num_tokens": 1850914429.0,
      "step": 605
    },
    {
      "epoch": 0.8880747389631801,
      "grad_norm": 0.7468173045710687,
      "learning_rate": 3.799761534668091e-05,
      "loss": 0.7644,
      "num_tokens": 1853981556.0,
      "step": 606
    },
    {
      "epoch": 0.889540208829456,
      "grad_norm": 0.6290630816339953,
      "learning_rate": 3.798978183231858e-05,
      "loss": 0.7542,
      "num_tokens": 1856959268.0,
      "step": 607
    },
    {
      "epoch": 0.8910056786957318,
      "grad_norm": 0.5729155712992845,
      "learning_rate": 3.7981933931242754e-05,
      "loss": 0.7735,
      "num_tokens": 1859919603.0,
      "step": 608
    },
    {
      "epoch": 0.8924711485620077,
      "grad_norm": 0.6710013158198507,
      "learning_rate": 3.797407165051455e-05,
      "loss": 0.7634,
      "num_tokens": 1863028848.0,
      "step": 609
    },
    {
      "epoch": 0.8939366184282835,
      "grad_norm": 0.654982737002341,
      "learning_rate": 3.796619499720799e-05,
      "loss": 0.7688,
      "num_tokens": 1866030242.0,
      "step": 610
    },
    {
      "epoch": 0.8954020882945595,
      "grad_norm": 0.486423032708723,
      "learning_rate": 3.795830397841007e-05,
      "loss": 0.761,
      "num_tokens": 1869043166.0,
      "step": 611
    },
    {
      "epoch": 0.8968675581608353,
      "grad_norm": 0.638014776803668,
      "learning_rate": 3.795039860122066e-05,
      "loss": 0.7688,
      "num_tokens": 1872129976.0,
      "step": 612
    },
    {
      "epoch": 0.8983330280271112,
      "grad_norm": 0.7837014284517078,
      "learning_rate": 3.7942478872752606e-05,
      "loss": 0.7694,
      "num_tokens": 1875247846.0,
      "step": 613
    },
    {
      "epoch": 0.8997984978933871,
      "grad_norm": 0.4713128952747065,
      "learning_rate": 3.7934544800131616e-05,
      "loss": 0.7573,
      "num_tokens": 1878495853.0,
      "step": 614
    },
    {
      "epoch": 0.9012639677596629,
      "grad_norm": 0.6479293136023899,
      "learning_rate": 3.792659639049634e-05,
      "loss": 0.7692,
      "num_tokens": 1881595328.0,
      "step": 615
    },
    {
      "epoch": 0.9027294376259388,
      "grad_norm": 0.5824132333654548,
      "learning_rate": 3.79186336509983e-05,
      "loss": 0.7742,
      "num_tokens": 1884535469.0,
      "step": 616
    },
    {
      "epoch": 0.9041949074922147,
      "grad_norm": 0.6436379597988445,
      "learning_rate": 3.7910656588801945e-05,
      "loss": 0.7686,
      "num_tokens": 1887531076.0,
      "step": 617
    },
    {
      "epoch": 0.9056603773584906,
      "grad_norm": 0.5656678989177333,
      "learning_rate": 3.7902665211084564e-05,
      "loss": 0.7419,
      "num_tokens": 1890365326.0,
      "step": 618
    },
    {
      "epoch": 0.9071258472247664,
      "grad_norm": 0.5674126326743792,
      "learning_rate": 3.789465952503638e-05,
      "loss": 0.7436,
      "num_tokens": 1893640789.0,
      "step": 619
    },
    {
      "epoch": 0.9085913170910423,
      "grad_norm": 0.7276819727120908,
      "learning_rate": 3.7886639537860444e-05,
      "loss": 0.7546,
      "num_tokens": 1896828926.0,
      "step": 620
    },
    {
      "epoch": 0.9100567869573182,
      "grad_norm": 0.4572270931522491,
      "learning_rate": 3.78786052567727e-05,
      "loss": 0.7852,
      "num_tokens": 1899616146.0,
      "step": 621
    },
    {
      "epoch": 0.9115222568235941,
      "grad_norm": 0.6660650488574197,
      "learning_rate": 3.7870556689001945e-05,
      "loss": 0.7516,
      "num_tokens": 1902694705.0,
      "step": 622
    },
    {
      "epoch": 0.9129877266898699,
      "grad_norm": 0.6345765146342996,
      "learning_rate": 3.7862493841789826e-05,
      "loss": 0.7715,
      "num_tokens": 1905706613.0,
      "step": 623
    },
    {
      "epoch": 0.9144531965561458,
      "grad_norm": 0.4783344256116178,
      "learning_rate": 3.785441672239085e-05,
      "loss": 0.7779,
      "num_tokens": 1908570121.0,
      "step": 624
    },
    {
      "epoch": 0.9159186664224217,
      "grad_norm": 0.686450956164066,
      "learning_rate": 3.784632533807235e-05,
      "loss": 0.7729,
      "num_tokens": 1911550249.0,
      "step": 625
    },
    {
      "epoch": 0.9173841362886975,
      "grad_norm": 0.6899407427286569,
      "learning_rate": 3.783821969611451e-05,
      "loss": 0.7472,
      "num_tokens": 1914636649.0,
      "step": 626
    },
    {
      "epoch": 0.9188496061549735,
      "grad_norm": 0.4934670114726233,
      "learning_rate": 3.783009980381034e-05,
      "loss": 0.7747,
      "num_tokens": 1917739554.0,
      "step": 627
    },
    {
      "epoch": 0.9203150760212493,
      "grad_norm": 0.8639122391165722,
      "learning_rate": 3.782196566846565e-05,
      "loss": 0.7715,
      "num_tokens": 1920751438.0,
      "step": 628
    },
    {
      "epoch": 0.9217805458875252,
      "grad_norm": 0.5585272785901677,
      "learning_rate": 3.7813817297399096e-05,
      "loss": 0.7525,
      "num_tokens": 1924023122.0,
      "step": 629
    },
    {
      "epoch": 0.923246015753801,
      "grad_norm": 0.9341746802404657,
      "learning_rate": 3.780565469794212e-05,
      "loss": 0.7738,
      "num_tokens": 1927229317.0,
      "step": 630
    },
    {
      "epoch": 0.924711485620077,
      "grad_norm": 0.6941226597369331,
      "learning_rate": 3.779747787743897e-05,
      "loss": 0.7849,
      "num_tokens": 1930194495.0,
      "step": 631
    },
    {
      "epoch": 0.9261769554863528,
      "grad_norm": 0.8787799800712817,
      "learning_rate": 3.77892868432467e-05,
      "loss": 0.7719,
      "num_tokens": 1933226447.0,
      "step": 632
    },
    {
      "epoch": 0.9276424253526286,
      "grad_norm": 0.6634970779532314,
      "learning_rate": 3.7781081602735145e-05,
      "loss": 0.759,
      "num_tokens": 1936221228.0,
      "step": 633
    },
    {
      "epoch": 0.9291078952189046,
      "grad_norm": 0.887136527526139,
      "learning_rate": 3.777286216328692e-05,
      "loss": 0.781,
      "num_tokens": 1939211467.0,
      "step": 634
    },
    {
      "epoch": 0.9305733650851804,
      "grad_norm": 0.6418875797886834,
      "learning_rate": 3.776462853229741e-05,
      "loss": 0.7721,
      "num_tokens": 1942214597.0,
      "step": 635
    },
    {
      "epoch": 0.9320388349514563,
      "grad_norm": 0.7994318117480244,
      "learning_rate": 3.7756380717174796e-05,
      "loss": 0.7591,
      "num_tokens": 1945156758.0,
      "step": 636
    },
    {
      "epoch": 0.9335043048177322,
      "grad_norm": 0.685468606703183,
      "learning_rate": 3.774811872533998e-05,
      "loss": 0.7615,
      "num_tokens": 1948231699.0,
      "step": 637
    },
    {
      "epoch": 0.9349697746840081,
      "grad_norm": 0.7091267136673098,
      "learning_rate": 3.7739842564226667e-05,
      "loss": 0.756,
      "num_tokens": 1951320148.0,
      "step": 638
    },
    {
      "epoch": 0.9364352445502839,
      "grad_norm": 0.6395586406376436,
      "learning_rate": 3.773155224128126e-05,
      "loss": 0.7481,
      "num_tokens": 1954425861.0,
      "step": 639
    },
    {
      "epoch": 0.9379007144165598,
      "grad_norm": 0.6514377068482021,
      "learning_rate": 3.772324776396294e-05,
      "loss": 0.765,
      "num_tokens": 1957327430.0,
      "step": 640
    },
    {
      "epoch": 0.9393661842828357,
      "grad_norm": 0.604915929784694,
      "learning_rate": 3.771492913974362e-05,
      "loss": 0.766,
      "num_tokens": 1960281867.0,
      "step": 641
    },
    {
      "epoch": 0.9408316541491115,
      "grad_norm": 0.5561202514864331,
      "learning_rate": 3.7706596376107916e-05,
      "loss": 0.7326,
      "num_tokens": 1963542347.0,
      "step": 642
    },
    {
      "epoch": 0.9422971240153875,
      "grad_norm": 0.6524460318326223,
      "learning_rate": 3.769824948055319e-05,
      "loss": 0.7667,
      "num_tokens": 1966572145.0,
      "step": 643
    },
    {
      "epoch": 0.9437625938816633,
      "grad_norm": 0.5037735999069346,
      "learning_rate": 3.768988846058953e-05,
      "loss": 0.7592,
      "num_tokens": 1969679554.0,
      "step": 644
    },
    {
      "epoch": 0.9452280637479392,
      "grad_norm": 0.5786731912448202,
      "learning_rate": 3.768151332373969e-05,
      "loss": 0.7491,
      "num_tokens": 1972803651.0,
      "step": 645
    },
    {
      "epoch": 0.946693533614215,
      "grad_norm": 0.523808608927916,
      "learning_rate": 3.767312407753917e-05,
      "loss": 0.7738,
      "num_tokens": 1975809709.0,
      "step": 646
    },
    {
      "epoch": 0.948159003480491,
      "grad_norm": 0.5995120358534207,
      "learning_rate": 3.766472072953613e-05,
      "loss": 0.7642,
      "num_tokens": 1978728863.0,
      "step": 647
    },
    {
      "epoch": 0.9496244733467668,
      "grad_norm": 0.4845529389777227,
      "learning_rate": 3.765630328729145e-05,
      "loss": 0.7471,
      "num_tokens": 1981859746.0,
      "step": 648
    },
    {
      "epoch": 0.9510899432130426,
      "grad_norm": 0.7277835379583072,
      "learning_rate": 3.7647871758378654e-05,
      "loss": 0.7462,
      "num_tokens": 1985036527.0,
      "step": 649
    },
    {
      "epoch": 0.9525554130793186,
      "grad_norm": 0.5203820949638468,
      "learning_rate": 3.7639426150383986e-05,
      "loss": 0.7756,
      "num_tokens": 1987994905.0,
      "step": 650
    },
    {
      "epoch": 0.9540208829455944,
      "grad_norm": 0.8262214620143367,
      "learning_rate": 3.7630966470906307e-05,
      "loss": 0.7588,
      "num_tokens": 1990952582.0,
      "step": 651
    },
    {
      "epoch": 0.9554863528118703,
      "grad_norm": 0.6958722537962205,
      "learning_rate": 3.762249272755719e-05,
      "loss": 0.7767,
      "num_tokens": 1994021014.0,
      "step": 652
    },
    {
      "epoch": 0.9569518226781462,
      "grad_norm": 0.6926954380394293,
      "learning_rate": 3.761400492796082e-05,
      "loss": 0.7654,
      "num_tokens": 1996988852.0,
      "step": 653
    },
    {
      "epoch": 0.9584172925444221,
      "grad_norm": 0.6978551806189677,
      "learning_rate": 3.760550307975405e-05,
      "loss": 0.7748,
      "num_tokens": 2000054813.0,
      "step": 654
    },
    {
      "epoch": 0.9598827624106979,
      "grad_norm": 0.605957194001,
      "learning_rate": 3.7596987190586374e-05,
      "loss": 0.7509,
      "num_tokens": 2003042990.0,
      "step": 655
    },
    {
      "epoch": 0.9613482322769739,
      "grad_norm": 0.6368133440966786,
      "learning_rate": 3.75884572681199e-05,
      "loss": 0.7508,
      "num_tokens": 2006319675.0,
      "step": 656
    },
    {
      "epoch": 0.9628137021432497,
      "grad_norm": 0.476910320318197,
      "learning_rate": 3.757991332002939e-05,
      "loss": 0.7661,
      "num_tokens": 2009422886.0,
      "step": 657
    },
    {
      "epoch": 0.9642791720095255,
      "grad_norm": 0.7325847239666787,
      "learning_rate": 3.75713553540022e-05,
      "loss": 0.7522,
      "num_tokens": 2012454764.0,
      "step": 658
    },
    {
      "epoch": 0.9657446418758014,
      "grad_norm": 0.5828842999264109,
      "learning_rate": 3.756278337773832e-05,
      "loss": 0.7772,
      "num_tokens": 2015465754.0,
      "step": 659
    },
    {
      "epoch": 0.9672101117420773,
      "grad_norm": 0.6543938148531748,
      "learning_rate": 3.7554197398950335e-05,
      "loss": 0.7525,
      "num_tokens": 2018379848.0,
      "step": 660
    },
    {
      "epoch": 0.9686755816083532,
      "grad_norm": 0.6074608961465975,
      "learning_rate": 3.7545597425363426e-05,
      "loss": 0.753,
      "num_tokens": 2021448637.0,
      "step": 661
    },
    {
      "epoch": 0.970141051474629,
      "grad_norm": 0.5277739122604715,
      "learning_rate": 3.7536983464715374e-05,
      "loss": 0.7794,
      "num_tokens": 2024365660.0,
      "step": 662
    },
    {
      "epoch": 0.971606521340905,
      "grad_norm": 0.6490356458883493,
      "learning_rate": 3.752835552475653e-05,
      "loss": 0.7609,
      "num_tokens": 2027404131.0,
      "step": 663
    },
    {
      "epoch": 0.9730719912071808,
      "grad_norm": 0.48663515563816123,
      "learning_rate": 3.751971361324985e-05,
      "loss": 0.7531,
      "num_tokens": 2030472960.0,
      "step": 664
    },
    {
      "epoch": 0.9745374610734566,
      "grad_norm": 0.8271509981819511,
      "learning_rate": 3.7511057737970824e-05,
      "loss": 0.7635,
      "num_tokens": 2033648017.0,
      "step": 665
    },
    {
      "epoch": 0.9760029309397326,
      "grad_norm": 0.6448244407421377,
      "learning_rate": 3.7502387906707536e-05,
      "loss": 0.785,
      "num_tokens": 2036775729.0,
      "step": 666
    },
    {
      "epoch": 0.9774684008060084,
      "grad_norm": 0.7868198580042994,
      "learning_rate": 3.7493704127260616e-05,
      "loss": 0.768,
      "num_tokens": 2039845178.0,
      "step": 667
    },
    {
      "epoch": 0.9789338706722843,
      "grad_norm": 0.6423425164607255,
      "learning_rate": 3.748500640744325e-05,
      "loss": 0.7589,
      "num_tokens": 2042913140.0,
      "step": 668
    },
    {
      "epoch": 0.9803993405385601,
      "grad_norm": 0.7312382148283119,
      "learning_rate": 3.747629475508115e-05,
      "loss": 0.7511,
      "num_tokens": 2045883851.0,
      "step": 669
    },
    {
      "epoch": 0.9818648104048361,
      "grad_norm": 0.5684524585243846,
      "learning_rate": 3.7467569178012575e-05,
      "loss": 0.7571,
      "num_tokens": 2048883833.0,
      "step": 670
    },
    {
      "epoch": 0.9833302802711119,
      "grad_norm": 0.7301807075780937,
      "learning_rate": 3.745882968408832e-05,
      "loss": 0.7452,
      "num_tokens": 2051893376.0,
      "step": 671
    },
    {
      "epoch": 0.9847957501373878,
      "grad_norm": 0.5149366367790847,
      "learning_rate": 3.7450076281171696e-05,
      "loss": 0.7633,
      "num_tokens": 2055080057.0,
      "step": 672
    },
    {
      "epoch": 0.9862612200036637,
      "grad_norm": 0.7592593842889864,
      "learning_rate": 3.7441308977138514e-05,
      "loss": 0.7622,
      "num_tokens": 2058308501.0,
      "step": 673
    },
    {
      "epoch": 0.9877266898699395,
      "grad_norm": 0.5840995350881396,
      "learning_rate": 3.743252777987712e-05,
      "loss": 0.7657,
      "num_tokens": 2061474649.0,
      "step": 674
    },
    {
      "epoch": 0.9891921597362154,
      "grad_norm": 0.7238127166623062,
      "learning_rate": 3.742373269728833e-05,
      "loss": 0.7641,
      "num_tokens": 2064280704.0,
      "step": 675
    },
    {
      "epoch": 0.9906576296024913,
      "grad_norm": 0.5656155902952759,
      "learning_rate": 3.741492373728548e-05,
      "loss": 0.7631,
      "num_tokens": 2067358410.0,
      "step": 676
    },
    {
      "epoch": 0.9921230994687672,
      "grad_norm": 0.8681921379583608,
      "learning_rate": 3.7406100907794375e-05,
      "loss": 0.7655,
      "num_tokens": 2070338357.0,
      "step": 677
    },
    {
      "epoch": 0.993588569335043,
      "grad_norm": 0.6811209766453731,
      "learning_rate": 3.739726421675331e-05,
      "loss": 0.7822,
      "num_tokens": 2073323968.0,
      "step": 678
    },
    {
      "epoch": 0.995054039201319,
      "grad_norm": 0.6764551214662876,
      "learning_rate": 3.738841367211304e-05,
      "loss": 0.7586,
      "num_tokens": 2076382759.0,
      "step": 679
    },
    {
      "epoch": 0.9965195090675948,
      "grad_norm": 0.7250122547172594,
      "learning_rate": 3.73795492818368e-05,
      "loss": 0.7664,
      "num_tokens": 2079519161.0,
      "step": 680
    },
    {
      "epoch": 0.9979849789338707,
      "grad_norm": 0.547601968793431,
      "learning_rate": 3.737067105390026e-05,
      "loss": 0.7553,
      "num_tokens": 2082558483.0,
      "step": 681
    },
    {
      "epoch": 0.9994504488001466,
      "grad_norm": 0.6722580383412239,
      "learning_rate": 3.736177899629156e-05,
      "loss": 0.7608,
      "num_tokens": 2085711886.0,
      "step": 682
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6722580383412239,
      "learning_rate": 3.735287311701129e-05,
      "loss": 0.7411,
      "num_tokens": 2086354946.0,
      "step": 683
    },
    {
      "epoch": 1.001465469866276,
      "grad_norm": 0.965701733730156,
      "learning_rate": 3.734395342407245e-05,
      "loss": 0.7464,
      "num_tokens": 2089424116.0,
      "step": 684
    },
    {
      "epoch": 1.0029309397325517,
      "grad_norm": 0.5133810627523421,
      "learning_rate": 3.733501992550048e-05,
      "loss": 0.7469,
      "num_tokens": 2092504087.0,
      "step": 685
    },
    {
      "epoch": 1.0043964095988276,
      "grad_norm": 0.6846944303065458,
      "learning_rate": 3.732607262933325e-05,
      "loss": 0.7501,
      "num_tokens": 2095695436.0,
      "step": 686
    },
    {
      "epoch": 1.0058618794651035,
      "grad_norm": 0.5492008160627736,
      "learning_rate": 3.7317111543621035e-05,
      "loss": 0.7623,
      "num_tokens": 2098818040.0,
      "step": 687
    },
    {
      "epoch": 1.0073273493313795,
      "grad_norm": 0.608048644578349,
      "learning_rate": 3.730813667642652e-05,
      "loss": 0.7297,
      "num_tokens": 2102081547.0,
      "step": 688
    },
    {
      "epoch": 1.0087928191976552,
      "grad_norm": 0.42408006198989767,
      "learning_rate": 3.729914803582479e-05,
      "loss": 0.7375,
      "num_tokens": 2105193885.0,
      "step": 689
    },
    {
      "epoch": 1.0102582890639311,
      "grad_norm": 0.5928279425557839,
      "learning_rate": 3.729014562990333e-05,
      "loss": 0.7562,
      "num_tokens": 2108095244.0,
      "step": 690
    },
    {
      "epoch": 1.011723758930207,
      "grad_norm": 0.6051913181331622,
      "learning_rate": 3.7281129466761995e-05,
      "loss": 0.7603,
      "num_tokens": 2111400323.0,
      "step": 691
    },
    {
      "epoch": 1.0131892287964828,
      "grad_norm": 0.6295552731757489,
      "learning_rate": 3.727209955451302e-05,
      "loss": 0.7417,
      "num_tokens": 2114436479.0,
      "step": 692
    },
    {
      "epoch": 1.0146546986627587,
      "grad_norm": 0.5359591982659879,
      "learning_rate": 3.7263055901281026e-05,
      "loss": 0.7568,
      "num_tokens": 2117679032.0,
      "step": 693
    },
    {
      "epoch": 1.0161201685290346,
      "grad_norm": 0.6670312571939421,
      "learning_rate": 3.7253998515202986e-05,
      "loss": 0.7493,
      "num_tokens": 2120717761.0,
      "step": 694
    },
    {
      "epoch": 1.0175856383953106,
      "grad_norm": 0.5887974320624524,
      "learning_rate": 3.724492740442822e-05,
      "loss": 0.732,
      "num_tokens": 2123668167.0,
      "step": 695
    },
    {
      "epoch": 1.0190511082615863,
      "grad_norm": 0.6428688747718982,
      "learning_rate": 3.723584257711842e-05,
      "loss": 0.7689,
      "num_tokens": 2126540267.0,
      "step": 696
    },
    {
      "epoch": 1.0205165781278622,
      "grad_norm": 0.5164418224075241,
      "learning_rate": 3.7226744041447607e-05,
      "loss": 0.7413,
      "num_tokens": 2129476421.0,
      "step": 697
    },
    {
      "epoch": 1.0219820479941382,
      "grad_norm": 0.6054376335161614,
      "learning_rate": 3.7217631805602125e-05,
      "loss": 0.734,
      "num_tokens": 2132390995.0,
      "step": 698
    },
    {
      "epoch": 1.023447517860414,
      "grad_norm": 0.5744791864215427,
      "learning_rate": 3.720850587778066e-05,
      "loss": 0.7429,
      "num_tokens": 2135359608.0,
      "step": 699
    },
    {
      "epoch": 1.0249129877266898,
      "grad_norm": 0.5649861377147157,
      "learning_rate": 3.719936626619422e-05,
      "loss": 0.7432,
      "num_tokens": 2138501286.0,
      "step": 700
    },
    {
      "epoch": 1.0263784575929658,
      "grad_norm": 0.5091744191362831,
      "learning_rate": 3.7190212979066107e-05,
      "loss": 0.7512,
      "num_tokens": 2141374314.0,
      "step": 701
    },
    {
      "epoch": 1.0278439274592417,
      "grad_norm": 0.5725742428395841,
      "learning_rate": 3.718104602463194e-05,
      "loss": 0.7407,
      "num_tokens": 2144285726.0,
      "step": 702
    },
    {
      "epoch": 1.0293093973255174,
      "grad_norm": 0.5304200237730711,
      "learning_rate": 3.717186541113964e-05,
      "loss": 0.7415,
      "num_tokens": 2147618227.0,
      "step": 703
    },
    {
      "epoch": 1.0307748671917933,
      "grad_norm": 0.5377789103815364,
      "learning_rate": 3.7162671146849414e-05,
      "loss": 0.7708,
      "num_tokens": 2150711331.0,
      "step": 704
    },
    {
      "epoch": 1.0322403370580693,
      "grad_norm": 0.5107631684950187,
      "learning_rate": 3.715346324003373e-05,
      "loss": 0.7681,
      "num_tokens": 2153694895.0,
      "step": 705
    },
    {
      "epoch": 1.0337058069243452,
      "grad_norm": 0.7161134476985407,
      "learning_rate": 3.714424169897737e-05,
      "loss": 0.7456,
      "num_tokens": 2157005666.0,
      "step": 706
    },
    {
      "epoch": 1.035171276790621,
      "grad_norm": 0.5206373746222844,
      "learning_rate": 3.713500653197734e-05,
      "loss": 0.7355,
      "num_tokens": 2160039775.0,
      "step": 707
    },
    {
      "epoch": 1.0366367466568969,
      "grad_norm": 0.5404919957857687,
      "learning_rate": 3.712575774734294e-05,
      "loss": 0.7595,
      "num_tokens": 2163129477.0,
      "step": 708
    },
    {
      "epoch": 1.0381022165231728,
      "grad_norm": 0.6472966435467952,
      "learning_rate": 3.7116495353395714e-05,
      "loss": 0.7479,
      "num_tokens": 2166037896.0,
      "step": 709
    },
    {
      "epoch": 1.0395676863894485,
      "grad_norm": 0.5231631163973381,
      "learning_rate": 3.710721935846944e-05,
      "loss": 0.7889,
      "num_tokens": 2169029002.0,
      "step": 710
    },
    {
      "epoch": 1.0410331562557245,
      "grad_norm": 0.530060468539575,
      "learning_rate": 3.7097929770910146e-05,
      "loss": 0.7482,
      "num_tokens": 2171870391.0,
      "step": 711
    },
    {
      "epoch": 1.0424986261220004,
      "grad_norm": 0.5622453631172125,
      "learning_rate": 3.708862659907608e-05,
      "loss": 0.7573,
      "num_tokens": 2175024201.0,
      "step": 712
    },
    {
      "epoch": 1.0439640959882763,
      "grad_norm": 0.5963062854922916,
      "learning_rate": 3.707930985133772e-05,
      "loss": 0.7462,
      "num_tokens": 2178280799.0,
      "step": 713
    },
    {
      "epoch": 1.045429565854552,
      "grad_norm": 0.5313545131659448,
      "learning_rate": 3.706997953607776e-05,
      "loss": 0.7648,
      "num_tokens": 2181266981.0,
      "step": 714
    },
    {
      "epoch": 1.046895035720828,
      "grad_norm": 0.5466249742925424,
      "learning_rate": 3.7060635661691084e-05,
      "loss": 0.7343,
      "num_tokens": 2184188084.0,
      "step": 715
    },
    {
      "epoch": 1.048360505587104,
      "grad_norm": 0.5663932311432857,
      "learning_rate": 3.70512782365848e-05,
      "loss": 0.7536,
      "num_tokens": 2187430737.0,
      "step": 716
    },
    {
      "epoch": 1.0498259754533796,
      "grad_norm": 0.4838039672619207,
      "learning_rate": 3.7041907269178195e-05,
      "loss": 0.7572,
      "num_tokens": 2190128313.0,
      "step": 717
    },
    {
      "epoch": 1.0512914453196556,
      "grad_norm": 0.5631770530991895,
      "learning_rate": 3.703252276790273e-05,
      "loss": 0.7513,
      "num_tokens": 2193187300.0,
      "step": 718
    },
    {
      "epoch": 1.0527569151859315,
      "grad_norm": 0.509897390962656,
      "learning_rate": 3.702312474120208e-05,
      "loss": 0.7551,
      "num_tokens": 2196209970.0,
      "step": 719
    },
    {
      "epoch": 1.0542223850522074,
      "grad_norm": 0.6061100376432039,
      "learning_rate": 3.701371319753205e-05,
      "loss": 0.7389,
      "num_tokens": 2199075223.0,
      "step": 720
    },
    {
      "epoch": 1.0556878549184832,
      "grad_norm": 0.49065247253595917,
      "learning_rate": 3.700428814536062e-05,
      "loss": 0.7411,
      "num_tokens": 2202166743.0,
      "step": 721
    },
    {
      "epoch": 1.057153324784759,
      "grad_norm": 0.6047032526105094,
      "learning_rate": 3.699484959316793e-05,
      "loss": 0.7482,
      "num_tokens": 2205269004.0,
      "step": 722
    },
    {
      "epoch": 1.058618794651035,
      "grad_norm": 0.5069813223229449,
      "learning_rate": 3.698539754944626e-05,
      "loss": 0.7453,
      "num_tokens": 2208177599.0,
      "step": 723
    },
    {
      "epoch": 1.060084264517311,
      "grad_norm": 0.5198781724310066,
      "learning_rate": 3.697593202270004e-05,
      "loss": 0.7732,
      "num_tokens": 2211161130.0,
      "step": 724
    },
    {
      "epoch": 1.0615497343835867,
      "grad_norm": 0.5297285255553046,
      "learning_rate": 3.696645302144582e-05,
      "loss": 0.7703,
      "num_tokens": 2214106845.0,
      "step": 725
    },
    {
      "epoch": 1.0630152042498626,
      "grad_norm": 0.5810591529235009,
      "learning_rate": 3.6956960554212264e-05,
      "loss": 0.74,
      "num_tokens": 2217036287.0,
      "step": 726
    },
    {
      "epoch": 1.0644806741161386,
      "grad_norm": 0.5253126415054268,
      "learning_rate": 3.694745462954018e-05,
      "loss": 0.7714,
      "num_tokens": 2219849225.0,
      "step": 727
    },
    {
      "epoch": 1.0659461439824143,
      "grad_norm": 0.5787768130087091,
      "learning_rate": 3.693793525598246e-05,
      "loss": 0.7594,
      "num_tokens": 2222920367.0,
      "step": 728
    },
    {
      "epoch": 1.0674116138486902,
      "grad_norm": 0.5951506746902782,
      "learning_rate": 3.6928402442104106e-05,
      "loss": 0.7481,
      "num_tokens": 2225714824.0,
      "step": 729
    },
    {
      "epoch": 1.0688770837149661,
      "grad_norm": 0.620649507315363,
      "learning_rate": 3.6918856196482204e-05,
      "loss": 0.7417,
      "num_tokens": 2228896692.0,
      "step": 730
    },
    {
      "epoch": 1.070342553581242,
      "grad_norm": 0.4214075647142451,
      "learning_rate": 3.6909296527705956e-05,
      "loss": 0.7537,
      "num_tokens": 2231821181.0,
      "step": 731
    },
    {
      "epoch": 1.0718080234475178,
      "grad_norm": 0.7835099958453671,
      "learning_rate": 3.68997234443766e-05,
      "loss": 0.746,
      "num_tokens": 2234900557.0,
      "step": 732
    },
    {
      "epoch": 1.0732734933137937,
      "grad_norm": 0.5186298880331794,
      "learning_rate": 3.6890136955107466e-05,
      "loss": 0.7435,
      "num_tokens": 2238144344.0,
      "step": 733
    },
    {
      "epoch": 1.0747389631800697,
      "grad_norm": 0.6085105303091098,
      "learning_rate": 3.6880537068523944e-05,
      "loss": 0.7423,
      "num_tokens": 2241146810.0,
      "step": 734
    },
    {
      "epoch": 1.0762044330463454,
      "grad_norm": 0.5636060930723763,
      "learning_rate": 3.6870923793263475e-05,
      "loss": 0.7577,
      "num_tokens": 2244224387.0,
      "step": 735
    },
    {
      "epoch": 1.0776699029126213,
      "grad_norm": 0.6067940184516105,
      "learning_rate": 3.686129713797555e-05,
      "loss": 0.7479,
      "num_tokens": 2247524768.0,
      "step": 736
    },
    {
      "epoch": 1.0791353727788973,
      "grad_norm": 0.5148018865782866,
      "learning_rate": 3.685165711132169e-05,
      "loss": 0.7427,
      "num_tokens": 2250688717.0,
      "step": 737
    },
    {
      "epoch": 1.0806008426451732,
      "grad_norm": 0.4802690919252581,
      "learning_rate": 3.684200372197547e-05,
      "loss": 0.7359,
      "num_tokens": 2253700064.0,
      "step": 738
    },
    {
      "epoch": 1.082066312511449,
      "grad_norm": 0.5275861366733628,
      "learning_rate": 3.683233697862245e-05,
      "loss": 0.7472,
      "num_tokens": 2256735602.0,
      "step": 739
    },
    {
      "epoch": 1.0835317823777249,
      "grad_norm": 0.6110310461543157,
      "learning_rate": 3.6822656889960233e-05,
      "loss": 0.7333,
      "num_tokens": 2259924265.0,
      "step": 740
    },
    {
      "epoch": 1.0849972522440008,
      "grad_norm": 0.5372957147826037,
      "learning_rate": 3.681296346469842e-05,
      "loss": 0.7551,
      "num_tokens": 2262916078.0,
      "step": 741
    },
    {
      "epoch": 1.0864627221102765,
      "grad_norm": 0.6659513036118253,
      "learning_rate": 3.680325671155863e-05,
      "loss": 0.7625,
      "num_tokens": 2265813741.0,
      "step": 742
    },
    {
      "epoch": 1.0879281919765524,
      "grad_norm": 0.4416955811201939,
      "learning_rate": 3.679353663927445e-05,
      "loss": 0.7484,
      "num_tokens": 2268793250.0,
      "step": 743
    },
    {
      "epoch": 1.0893936618428284,
      "grad_norm": 0.7027031517387361,
      "learning_rate": 3.678380325659145e-05,
      "loss": 0.7333,
      "num_tokens": 2271828634.0,
      "step": 744
    },
    {
      "epoch": 1.0908591317091043,
      "grad_norm": 0.5281969383089093,
      "learning_rate": 3.677405657226718e-05,
      "loss": 0.7676,
      "num_tokens": 2274614334.0,
      "step": 745
    },
    {
      "epoch": 1.09232460157538,
      "grad_norm": 0.5968650308781397,
      "learning_rate": 3.6764296595071185e-05,
      "loss": 0.7293,
      "num_tokens": 2277613326.0,
      "step": 746
    },
    {
      "epoch": 1.093790071441656,
      "grad_norm": 0.5649618616317241,
      "learning_rate": 3.6754523333784925e-05,
      "loss": 0.7567,
      "num_tokens": 2280784695.0,
      "step": 747
    },
    {
      "epoch": 1.095255541307932,
      "grad_norm": 0.5942567444664041,
      "learning_rate": 3.6744736797201856e-05,
      "loss": 0.7434,
      "num_tokens": 2283843683.0,
      "step": 748
    },
    {
      "epoch": 1.0967210111742078,
      "grad_norm": 0.5741612199460744,
      "learning_rate": 3.673493699412734e-05,
      "loss": 0.7318,
      "num_tokens": 2286716216.0,
      "step": 749
    },
    {
      "epoch": 1.0981864810404836,
      "grad_norm": 0.42116528502576295,
      "learning_rate": 3.672512393337871e-05,
      "loss": 0.7243,
      "num_tokens": 2289952952.0,
      "step": 750
    },
    {
      "epoch": 1.0996519509067595,
      "grad_norm": 0.6044648116673002,
      "learning_rate": 3.671529762378519e-05,
      "loss": 0.7452,
      "num_tokens": 2293060517.0,
      "step": 751
    },
    {
      "epoch": 1.1011174207730354,
      "grad_norm": 0.46348240332111157,
      "learning_rate": 3.670545807418796e-05,
      "loss": 0.7551,
      "num_tokens": 2296232202.0,
      "step": 752
    },
    {
      "epoch": 1.1025828906393111,
      "grad_norm": 0.5180548128420432,
      "learning_rate": 3.669560529344009e-05,
      "loss": 0.7579,
      "num_tokens": 2299313476.0,
      "step": 753
    },
    {
      "epoch": 1.104048360505587,
      "grad_norm": 0.5645230320912202,
      "learning_rate": 3.668573929040657e-05,
      "loss": 0.7316,
      "num_tokens": 2302211722.0,
      "step": 754
    },
    {
      "epoch": 1.105513830371863,
      "grad_norm": 0.6507893652662134,
      "learning_rate": 3.6675860073964275e-05,
      "loss": 0.7444,
      "num_tokens": 2305146236.0,
      "step": 755
    },
    {
      "epoch": 1.106979300238139,
      "grad_norm": 0.4689356093792895,
      "learning_rate": 3.666596765300197e-05,
      "loss": 0.7595,
      "num_tokens": 2308004712.0,
      "step": 756
    },
    {
      "epoch": 1.1084447701044147,
      "grad_norm": 0.5578643684173523,
      "learning_rate": 3.665606203642031e-05,
      "loss": 0.7456,
      "num_tokens": 2311076276.0,
      "step": 757
    },
    {
      "epoch": 1.1099102399706906,
      "grad_norm": 0.4923974582451954,
      "learning_rate": 3.664614323313181e-05,
      "loss": 0.7844,
      "num_tokens": 2314222276.0,
      "step": 758
    },
    {
      "epoch": 1.1113757098369665,
      "grad_norm": 0.5791975983862808,
      "learning_rate": 3.663621125206086e-05,
      "loss": 0.7539,
      "num_tokens": 2317277231.0,
      "step": 759
    },
    {
      "epoch": 1.1128411797032423,
      "grad_norm": 0.4948512630493004,
      "learning_rate": 3.66262661021437e-05,
      "loss": 0.7542,
      "num_tokens": 2320363787.0,
      "step": 760
    },
    {
      "epoch": 1.1143066495695182,
      "grad_norm": 0.6162555073830158,
      "learning_rate": 3.661630779232842e-05,
      "loss": 0.7462,
      "num_tokens": 2323562029.0,
      "step": 761
    },
    {
      "epoch": 1.1157721194357941,
      "grad_norm": 0.44356108900421054,
      "learning_rate": 3.660633633157495e-05,
      "loss": 0.7486,
      "num_tokens": 2326568575.0,
      "step": 762
    },
    {
      "epoch": 1.11723758930207,
      "grad_norm": 0.6717598253891943,
      "learning_rate": 3.6596351728855066e-05,
      "loss": 0.7553,
      "num_tokens": 2329403181.0,
      "step": 763
    },
    {
      "epoch": 1.1187030591683458,
      "grad_norm": 0.47370568301325094,
      "learning_rate": 3.658635399315233e-05,
      "loss": 0.7463,
      "num_tokens": 2332456267.0,
      "step": 764
    },
    {
      "epoch": 1.1201685290346217,
      "grad_norm": 0.6573236534532894,
      "learning_rate": 3.657634313346217e-05,
      "loss": 0.7609,
      "num_tokens": 2335289875.0,
      "step": 765
    },
    {
      "epoch": 1.1216339989008977,
      "grad_norm": 0.4608203279343698,
      "learning_rate": 3.6566319158791804e-05,
      "loss": 0.7717,
      "num_tokens": 2338029313.0,
      "step": 766
    },
    {
      "epoch": 1.1230994687671734,
      "grad_norm": 0.7499015540789802,
      "learning_rate": 3.655628207816023e-05,
      "loss": 0.7501,
      "num_tokens": 2341081528.0,
      "step": 767
    },
    {
      "epoch": 1.1245649386334493,
      "grad_norm": 0.5997630736891446,
      "learning_rate": 3.654623190059825e-05,
      "loss": 0.7707,
      "num_tokens": 2344094852.0,
      "step": 768
    },
    {
      "epoch": 1.1260304084997252,
      "grad_norm": 0.5725684757570828,
      "learning_rate": 3.6536168635148465e-05,
      "loss": 0.7354,
      "num_tokens": 2347198773.0,
      "step": 769
    },
    {
      "epoch": 1.1274958783660012,
      "grad_norm": 0.5888403125235339,
      "learning_rate": 3.652609229086523e-05,
      "loss": 0.7341,
      "num_tokens": 2350374813.0,
      "step": 770
    },
    {
      "epoch": 1.128961348232277,
      "grad_norm": 0.4836280198136197,
      "learning_rate": 3.6516002876814686e-05,
      "loss": 0.7431,
      "num_tokens": 2353533069.0,
      "step": 771
    },
    {
      "epoch": 1.1304268180985528,
      "grad_norm": 0.6400021899227083,
      "learning_rate": 3.650590040207473e-05,
      "loss": 0.7481,
      "num_tokens": 2356573904.0,
      "step": 772
    },
    {
      "epoch": 1.1318922879648288,
      "grad_norm": 0.4904287484823315,
      "learning_rate": 3.649578487573498e-05,
      "loss": 0.7548,
      "num_tokens": 2359819497.0,
      "step": 773
    },
    {
      "epoch": 1.1333577578311047,
      "grad_norm": 0.6436736840519235,
      "learning_rate": 3.6485656306896844e-05,
      "loss": 0.7445,
      "num_tokens": 2362852518.0,
      "step": 774
    },
    {
      "epoch": 1.1348232276973804,
      "grad_norm": 0.49343356331319965,
      "learning_rate": 3.647551470467343e-05,
      "loss": 0.7459,
      "num_tokens": 2366050595.0,
      "step": 775
    },
    {
      "epoch": 1.1362886975636564,
      "grad_norm": 0.574131251029063,
      "learning_rate": 3.6465360078189614e-05,
      "loss": 0.7397,
      "num_tokens": 2369039591.0,
      "step": 776
    },
    {
      "epoch": 1.1377541674299323,
      "grad_norm": 0.4761682632867504,
      "learning_rate": 3.645519243658193e-05,
      "loss": 0.742,
      "num_tokens": 2371936524.0,
      "step": 777
    },
    {
      "epoch": 1.139219637296208,
      "grad_norm": 0.5447938453075636,
      "learning_rate": 3.644501178899866e-05,
      "loss": 0.753,
      "num_tokens": 2374869651.0,
      "step": 778
    },
    {
      "epoch": 1.140685107162484,
      "grad_norm": 0.594985234757248,
      "learning_rate": 3.6434818144599794e-05,
      "loss": 0.7584,
      "num_tokens": 2377847693.0,
      "step": 779
    },
    {
      "epoch": 1.1421505770287599,
      "grad_norm": 0.5165556519729052,
      "learning_rate": 3.6424611512557e-05,
      "loss": 0.7413,
      "num_tokens": 2380837636.0,
      "step": 780
    },
    {
      "epoch": 1.1436160468950356,
      "grad_norm": 0.5484273098668319,
      "learning_rate": 3.641439190205363e-05,
      "loss": 0.7574,
      "num_tokens": 2384024622.0,
      "step": 781
    },
    {
      "epoch": 1.1450815167613115,
      "grad_norm": 0.49476176262851795,
      "learning_rate": 3.640415932228473e-05,
      "loss": 0.7462,
      "num_tokens": 2387127375.0,
      "step": 782
    },
    {
      "epoch": 1.1465469866275875,
      "grad_norm": 0.5673789447145892,
      "learning_rate": 3.639391378245699e-05,
      "loss": 0.7363,
      "num_tokens": 2390040686.0,
      "step": 783
    },
    {
      "epoch": 1.1480124564938634,
      "grad_norm": 0.4463761824945204,
      "learning_rate": 3.638365529178879e-05,
      "loss": 0.7225,
      "num_tokens": 2393069109.0,
      "step": 784
    },
    {
      "epoch": 1.1494779263601391,
      "grad_norm": 0.603203773557243,
      "learning_rate": 3.6373383859510135e-05,
      "loss": 0.7442,
      "num_tokens": 2396239053.0,
      "step": 785
    },
    {
      "epoch": 1.150943396226415,
      "grad_norm": 0.501386266641157,
      "learning_rate": 3.6363099494862704e-05,
      "loss": 0.7497,
      "num_tokens": 2399490843.0,
      "step": 786
    },
    {
      "epoch": 1.152408866092691,
      "grad_norm": 0.7142459542611617,
      "learning_rate": 3.635280220709977e-05,
      "loss": 0.7358,
      "num_tokens": 2402603598.0,
      "step": 787
    },
    {
      "epoch": 1.153874335958967,
      "grad_norm": 0.4884249292884803,
      "learning_rate": 3.6342492005486284e-05,
      "loss": 0.7537,
      "num_tokens": 2405554239.0,
      "step": 788
    },
    {
      "epoch": 1.1553398058252426,
      "grad_norm": 0.5448521132239552,
      "learning_rate": 3.633216889929877e-05,
      "loss": 0.7637,
      "num_tokens": 2408404215.0,
      "step": 789
    },
    {
      "epoch": 1.1568052756915186,
      "grad_norm": 0.5395111280241897,
      "learning_rate": 3.6321832897825396e-05,
      "loss": 0.7562,
      "num_tokens": 2411538139.0,
      "step": 790
    },
    {
      "epoch": 1.1582707455577945,
      "grad_norm": 0.5554376091980064,
      "learning_rate": 3.631148401036591e-05,
      "loss": 0.7623,
      "num_tokens": 2414484095.0,
      "step": 791
    },
    {
      "epoch": 1.1597362154240702,
      "grad_norm": 0.5472466773193756,
      "learning_rate": 3.630112224623167e-05,
      "loss": 0.7612,
      "num_tokens": 2417422671.0,
      "step": 792
    },
    {
      "epoch": 1.1612016852903462,
      "grad_norm": 0.47849574899464786,
      "learning_rate": 3.629074761474561e-05,
      "loss": 0.7364,
      "num_tokens": 2420428049.0,
      "step": 793
    },
    {
      "epoch": 1.1626671551566221,
      "grad_norm": 0.549409400606777,
      "learning_rate": 3.6280360125242234e-05,
      "loss": 0.7448,
      "num_tokens": 2423459433.0,
      "step": 794
    },
    {
      "epoch": 1.164132625022898,
      "grad_norm": 0.5949483510157997,
      "learning_rate": 3.626995978706765e-05,
      "loss": 0.7435,
      "num_tokens": 2426409962.0,
      "step": 795
    },
    {
      "epoch": 1.1655980948891738,
      "grad_norm": 0.43509381108765177,
      "learning_rate": 3.625954660957948e-05,
      "loss": 0.7342,
      "num_tokens": 2429570678.0,
      "step": 796
    },
    {
      "epoch": 1.1670635647554497,
      "grad_norm": 0.5173539496395994,
      "learning_rate": 3.6249120602146926e-05,
      "loss": 0.7525,
      "num_tokens": 2432651292.0,
      "step": 797
    },
    {
      "epoch": 1.1685290346217256,
      "grad_norm": 0.7109420938830574,
      "learning_rate": 3.623868177415074e-05,
      "loss": 0.7333,
      "num_tokens": 2435742723.0,
      "step": 798
    },
    {
      "epoch": 1.1699945044880016,
      "grad_norm": 0.4716566004637944,
      "learning_rate": 3.622823013498318e-05,
      "loss": 0.7341,
      "num_tokens": 2438697858.0,
      "step": 799
    },
    {
      "epoch": 1.1714599743542773,
      "grad_norm": 0.6207329486694613,
      "learning_rate": 3.621776569404806e-05,
      "loss": 0.7495,
      "num_tokens": 2441609339.0,
      "step": 800
    },
    {
      "epoch": 1.1729254442205532,
      "grad_norm": 0.5455232082217442,
      "learning_rate": 3.620728846076069e-05,
      "loss": 0.7449,
      "num_tokens": 2444460129.0,
      "step": 801
    },
    {
      "epoch": 1.1743909140868292,
      "grad_norm": 0.5715368187010545,
      "learning_rate": 3.619679844454791e-05,
      "loss": 0.7562,
      "num_tokens": 2447385702.0,
      "step": 802
    },
    {
      "epoch": 1.1758563839531049,
      "grad_norm": 0.6157891604805696,
      "learning_rate": 3.618629565484804e-05,
      "loss": 0.7425,
      "num_tokens": 2450596845.0,
      "step": 803
    },
    {
      "epoch": 1.1773218538193808,
      "grad_norm": 0.446325699112817,
      "learning_rate": 3.617578010111091e-05,
      "loss": 0.7277,
      "num_tokens": 2453748901.0,
      "step": 804
    },
    {
      "epoch": 1.1787873236856568,
      "grad_norm": 0.5441709024714138,
      "learning_rate": 3.616525179279784e-05,
      "loss": 0.7633,
      "num_tokens": 2456678338.0,
      "step": 805
    },
    {
      "epoch": 1.1802527935519325,
      "grad_norm": 0.6285552809644295,
      "learning_rate": 3.615471073938159e-05,
      "loss": 0.7302,
      "num_tokens": 2459606498.0,
      "step": 806
    },
    {
      "epoch": 1.1817182634182084,
      "grad_norm": 0.50570449381037,
      "learning_rate": 3.614415695034642e-05,
      "loss": 0.7561,
      "num_tokens": 2462718404.0,
      "step": 807
    },
    {
      "epoch": 1.1831837332844843,
      "grad_norm": 0.506231458394893,
      "learning_rate": 3.613359043518806e-05,
      "loss": 0.7354,
      "num_tokens": 2465810912.0,
      "step": 808
    },
    {
      "epoch": 1.1846492031507603,
      "grad_norm": 0.5096818741798217,
      "learning_rate": 3.612301120341365e-05,
      "loss": 0.7318,
      "num_tokens": 2468741379.0,
      "step": 809
    },
    {
      "epoch": 1.186114673017036,
      "grad_norm": 0.6600038539111857,
      "learning_rate": 3.61124192645418e-05,
      "loss": 0.7565,
      "num_tokens": 2471852662.0,
      "step": 810
    },
    {
      "epoch": 1.187580142883312,
      "grad_norm": 0.5539575573076669,
      "learning_rate": 3.610181462810254e-05,
      "loss": 0.7498,
      "num_tokens": 2475201268.0,
      "step": 811
    },
    {
      "epoch": 1.1890456127495879,
      "grad_norm": 0.5781660795578409,
      "learning_rate": 3.609119730363734e-05,
      "loss": 0.7429,
      "num_tokens": 2478183856.0,
      "step": 812
    },
    {
      "epoch": 1.1905110826158638,
      "grad_norm": 0.7656175651194363,
      "learning_rate": 3.608056730069906e-05,
      "loss": 0.7484,
      "num_tokens": 2481201221.0,
      "step": 813
    },
    {
      "epoch": 1.1919765524821395,
      "grad_norm": 0.5970484506671897,
      "learning_rate": 3.606992462885201e-05,
      "loss": 0.739,
      "num_tokens": 2484321297.0,
      "step": 814
    },
    {
      "epoch": 1.1934420223484155,
      "grad_norm": 0.6854697421273058,
      "learning_rate": 3.6059269297671844e-05,
      "loss": 0.7695,
      "num_tokens": 2487309910.0,
      "step": 815
    },
    {
      "epoch": 1.1949074922146914,
      "grad_norm": 0.7597415120672409,
      "learning_rate": 3.604860131674565e-05,
      "loss": 0.7445,
      "num_tokens": 2490232398.0,
      "step": 816
    },
    {
      "epoch": 1.196372962080967,
      "grad_norm": 0.5592664516086174,
      "learning_rate": 3.603792069567187e-05,
      "loss": 0.7512,
      "num_tokens": 2493447304.0,
      "step": 817
    },
    {
      "epoch": 1.197838431947243,
      "grad_norm": 0.8119270857697747,
      "learning_rate": 3.602722744406035e-05,
      "loss": 0.7496,
      "num_tokens": 2496716115.0,
      "step": 818
    },
    {
      "epoch": 1.199303901813519,
      "grad_norm": 0.5068081892319506,
      "learning_rate": 3.601652157153227e-05,
      "loss": 0.7572,
      "num_tokens": 2499573424.0,
      "step": 819
    },
    {
      "epoch": 1.200769371679795,
      "grad_norm": 0.8771479499729934,
      "learning_rate": 3.600580308772018e-05,
      "loss": 0.7468,
      "num_tokens": 2502631680.0,
      "step": 820
    },
    {
      "epoch": 1.2022348415460706,
      "grad_norm": 0.6003469705220338,
      "learning_rate": 3.599507200226796e-05,
      "loss": 0.7529,
      "num_tokens": 2505734677.0,
      "step": 821
    },
    {
      "epoch": 1.2037003114123466,
      "grad_norm": 0.8909743257302668,
      "learning_rate": 3.598432832483086e-05,
      "loss": 0.7424,
      "num_tokens": 2508663249.0,
      "step": 822
    },
    {
      "epoch": 1.2051657812786225,
      "grad_norm": 0.665613228337953,
      "learning_rate": 3.597357206507543e-05,
      "loss": 0.7587,
      "num_tokens": 2511523227.0,
      "step": 823
    },
    {
      "epoch": 1.2066312511448984,
      "grad_norm": 0.786596315501698,
      "learning_rate": 3.596280323267956e-05,
      "loss": 0.7335,
      "num_tokens": 2514710824.0,
      "step": 824
    },
    {
      "epoch": 1.2080967210111742,
      "grad_norm": 0.7115277495007731,
      "learning_rate": 3.595202183733244e-05,
      "loss": 0.7536,
      "num_tokens": 2517584775.0,
      "step": 825
    },
    {
      "epoch": 1.20956219087745,
      "grad_norm": 0.7014646239781099,
      "learning_rate": 3.594122788873457e-05,
      "loss": 0.7534,
      "num_tokens": 2520477375.0,
      "step": 826
    },
    {
      "epoch": 1.211027660743726,
      "grad_norm": 0.579524568878739,
      "learning_rate": 3.593042139659774e-05,
      "loss": 0.7393,
      "num_tokens": 2523649349.0,
      "step": 827
    },
    {
      "epoch": 1.2124931306100017,
      "grad_norm": 0.6768661068864853,
      "learning_rate": 3.5919602370645036e-05,
      "loss": 0.732,
      "num_tokens": 2526648143.0,
      "step": 828
    },
    {
      "epoch": 1.2139586004762777,
      "grad_norm": 0.5568210946116069,
      "learning_rate": 3.59087708206108e-05,
      "loss": 0.7336,
      "num_tokens": 2529986283.0,
      "step": 829
    },
    {
      "epoch": 1.2154240703425536,
      "grad_norm": 0.6464419441365561,
      "learning_rate": 3.589792675624067e-05,
      "loss": 0.7456,
      "num_tokens": 2533221035.0,
      "step": 830
    },
    {
      "epoch": 1.2168895402088293,
      "grad_norm": 0.6260454895826908,
      "learning_rate": 3.5887070187291516e-05,
      "loss": 0.7403,
      "num_tokens": 2536527199.0,
      "step": 831
    },
    {
      "epoch": 1.2183550100751053,
      "grad_norm": 0.49594840943509677,
      "learning_rate": 3.587620112353149e-05,
      "loss": 0.748,
      "num_tokens": 2539626744.0,
      "step": 832
    },
    {
      "epoch": 1.2198204799413812,
      "grad_norm": 0.7364413525453473,
      "learning_rate": 3.5865319574739965e-05,
      "loss": 0.7444,
      "num_tokens": 2542752358.0,
      "step": 833
    },
    {
      "epoch": 1.2212859498076571,
      "grad_norm": 0.5273930007968609,
      "learning_rate": 3.585442555070754e-05,
      "loss": 0.7242,
      "num_tokens": 2545592005.0,
      "step": 834
    },
    {
      "epoch": 1.2227514196739329,
      "grad_norm": 0.6778991032678832,
      "learning_rate": 3.584351906123606e-05,
      "loss": 0.73,
      "num_tokens": 2549012232.0,
      "step": 835
    },
    {
      "epoch": 1.2242168895402088,
      "grad_norm": 0.5183112827075003,
      "learning_rate": 3.583260011613857e-05,
      "loss": 0.752,
      "num_tokens": 2552055893.0,
      "step": 836
    },
    {
      "epoch": 1.2256823594064847,
      "grad_norm": 0.7821549164347252,
      "learning_rate": 3.582166872523933e-05,
      "loss": 0.7519,
      "num_tokens": 2555214237.0,
      "step": 837
    },
    {
      "epoch": 1.2271478292727607,
      "grad_norm": 0.6058890313797158,
      "learning_rate": 3.5810724898373794e-05,
      "loss": 0.7404,
      "num_tokens": 2558471673.0,
      "step": 838
    },
    {
      "epoch": 1.2286132991390364,
      "grad_norm": 0.8078949524200064,
      "learning_rate": 3.579976864538862e-05,
      "loss": 0.7361,
      "num_tokens": 2561663165.0,
      "step": 839
    },
    {
      "epoch": 1.2300787690053123,
      "grad_norm": 0.5594406291968924,
      "learning_rate": 3.578879997614161e-05,
      "loss": 0.7506,
      "num_tokens": 2564715693.0,
      "step": 840
    },
    {
      "epoch": 1.2315442388715883,
      "grad_norm": 0.6541570879836943,
      "learning_rate": 3.577781890050177e-05,
      "loss": 0.7299,
      "num_tokens": 2567862538.0,
      "step": 841
    },
    {
      "epoch": 1.233009708737864,
      "grad_norm": 0.6181771105707208,
      "learning_rate": 3.576682542834927e-05,
      "loss": 0.7247,
      "num_tokens": 2570985266.0,
      "step": 842
    },
    {
      "epoch": 1.23447517860414,
      "grad_norm": 0.6702058311025445,
      "learning_rate": 3.575581956957542e-05,
      "loss": 0.7581,
      "num_tokens": 2574141745.0,
      "step": 843
    },
    {
      "epoch": 1.2359406484704158,
      "grad_norm": 0.5415090090355892,
      "learning_rate": 3.5744801334082656e-05,
      "loss": 0.7366,
      "num_tokens": 2577186031.0,
      "step": 844
    },
    {
      "epoch": 1.2374061183366918,
      "grad_norm": 0.7304120868745388,
      "learning_rate": 3.57337707317846e-05,
      "loss": 0.7679,
      "num_tokens": 2580194225.0,
      "step": 845
    },
    {
      "epoch": 1.2388715882029675,
      "grad_norm": 0.5568789245984441,
      "learning_rate": 3.5722727772605954e-05,
      "loss": 0.7417,
      "num_tokens": 2583201268.0,
      "step": 846
    },
    {
      "epoch": 1.2403370580692434,
      "grad_norm": 0.6594030106680493,
      "learning_rate": 3.571167246648256e-05,
      "loss": 0.7581,
      "num_tokens": 2586260475.0,
      "step": 847
    },
    {
      "epoch": 1.2418025279355194,
      "grad_norm": 0.5883393365413924,
      "learning_rate": 3.570060482336137e-05,
      "loss": 0.7343,
      "num_tokens": 2589432034.0,
      "step": 848
    },
    {
      "epoch": 1.2432679978017953,
      "grad_norm": 0.5774334130789731,
      "learning_rate": 3.568952485320043e-05,
      "loss": 0.7388,
      "num_tokens": 2592559539.0,
      "step": 849
    },
    {
      "epoch": 1.244733467668071,
      "grad_norm": 0.5636442056489902,
      "learning_rate": 3.567843256596888e-05,
      "loss": 0.7453,
      "num_tokens": 2595456533.0,
      "step": 850
    },
    {
      "epoch": 1.246198937534347,
      "grad_norm": 0.5590164570830045,
      "learning_rate": 3.5667327971646934e-05,
      "loss": 0.7259,
      "num_tokens": 2598450360.0,
      "step": 851
    },
    {
      "epoch": 1.247664407400623,
      "grad_norm": 0.58963666002915,
      "learning_rate": 3.56562110802259e-05,
      "loss": 0.7399,
      "num_tokens": 2601810580.0,
      "step": 852
    },
    {
      "epoch": 1.2491298772668986,
      "grad_norm": 0.5532527541546471,
      "learning_rate": 3.564508190170812e-05,
      "loss": 0.7445,
      "num_tokens": 2604654098.0,
      "step": 853
    },
    {
      "epoch": 1.2505953471331746,
      "grad_norm": 0.5495565860270518,
      "learning_rate": 3.5633940446107026e-05,
      "loss": 0.7411,
      "num_tokens": 2607672619.0,
      "step": 854
    },
    {
      "epoch": 1.2520608169994505,
      "grad_norm": 0.5888307692148788,
      "learning_rate": 3.562278672344707e-05,
      "loss": 0.7543,
      "num_tokens": 2610561276.0,
      "step": 855
    },
    {
      "epoch": 1.2535262868657262,
      "grad_norm": 0.570711193868834,
      "learning_rate": 3.561162074376374e-05,
      "loss": 0.7358,
      "num_tokens": 2613497992.0,
      "step": 856
    },
    {
      "epoch": 1.2549917567320021,
      "grad_norm": 0.47406163794333306,
      "learning_rate": 3.5600442517103566e-05,
      "loss": 0.748,
      "num_tokens": 2616355657.0,
      "step": 857
    },
    {
      "epoch": 1.256457226598278,
      "grad_norm": 0.5476502333707635,
      "learning_rate": 3.5589252053524104e-05,
      "loss": 0.7371,
      "num_tokens": 2619405519.0,
      "step": 858
    },
    {
      "epoch": 1.257922696464554,
      "grad_norm": 0.48371903576850006,
      "learning_rate": 3.557804936309389e-05,
      "loss": 0.7313,
      "num_tokens": 2622608844.0,
      "step": 859
    },
    {
      "epoch": 1.25938816633083,
      "grad_norm": 0.6328645493932304,
      "learning_rate": 3.5566834455892505e-05,
      "loss": 0.751,
      "num_tokens": 2625478590.0,
      "step": 860
    },
    {
      "epoch": 1.2608536361971057,
      "grad_norm": 0.47925722201741144,
      "learning_rate": 3.555560734201047e-05,
      "loss": 0.7326,
      "num_tokens": 2628775410.0,
      "step": 861
    },
    {
      "epoch": 1.2623191060633816,
      "grad_norm": 0.5903566189409216,
      "learning_rate": 3.554436803154933e-05,
      "loss": 0.7462,
      "num_tokens": 2631955534.0,
      "step": 862
    },
    {
      "epoch": 1.2637845759296575,
      "grad_norm": 0.5574145870415723,
      "learning_rate": 3.55331165346216e-05,
      "loss": 0.7524,
      "num_tokens": 2635036510.0,
      "step": 863
    },
    {
      "epoch": 1.2652500457959333,
      "grad_norm": 0.57452401634078,
      "learning_rate": 3.552185286135072e-05,
      "loss": 0.7525,
      "num_tokens": 2638008798.0,
      "step": 864
    },
    {
      "epoch": 1.2667155156622092,
      "grad_norm": 0.4953923690178717,
      "learning_rate": 3.5510577021871145e-05,
      "loss": 0.7373,
      "num_tokens": 2640869157.0,
      "step": 865
    },
    {
      "epoch": 1.2681809855284851,
      "grad_norm": 0.4829210695497327,
      "learning_rate": 3.549928902632822e-05,
      "loss": 0.7204,
      "num_tokens": 2643766879.0,
      "step": 866
    },
    {
      "epoch": 1.2696464553947608,
      "grad_norm": 0.5408349654128393,
      "learning_rate": 3.548798888487827e-05,
      "loss": 0.7425,
      "num_tokens": 2646928103.0,
      "step": 867
    },
    {
      "epoch": 1.2711119252610368,
      "grad_norm": 0.5456776916813653,
      "learning_rate": 3.547667660768853e-05,
      "loss": 0.7375,
      "num_tokens": 2650178211.0,
      "step": 868
    },
    {
      "epoch": 1.2725773951273127,
      "grad_norm": 0.4812782524865056,
      "learning_rate": 3.5465352204937145e-05,
      "loss": 0.7466,
      "num_tokens": 2653318838.0,
      "step": 869
    },
    {
      "epoch": 1.2740428649935884,
      "grad_norm": 0.47967868325721497,
      "learning_rate": 3.54540156868132e-05,
      "loss": 0.7427,
      "num_tokens": 2656180221.0,
      "step": 870
    },
    {
      "epoch": 1.2755083348598644,
      "grad_norm": 0.6095567168366964,
      "learning_rate": 3.544266706351664e-05,
      "loss": 0.7382,
      "num_tokens": 2659195580.0,
      "step": 871
    },
    {
      "epoch": 1.2769738047261403,
      "grad_norm": 0.5163164285264217,
      "learning_rate": 3.5431306345258345e-05,
      "loss": 0.756,
      "num_tokens": 2662315281.0,
      "step": 872
    },
    {
      "epoch": 1.2784392745924162,
      "grad_norm": 0.6405609729624486,
      "learning_rate": 3.541993354226004e-05,
      "loss": 0.7477,
      "num_tokens": 2665382666.0,
      "step": 873
    },
    {
      "epoch": 1.2799047444586922,
      "grad_norm": 0.5697257241578433,
      "learning_rate": 3.540854866475433e-05,
      "loss": 0.7416,
      "num_tokens": 2668636950.0,
      "step": 874
    },
    {
      "epoch": 1.281370214324968,
      "grad_norm": 0.5449579903012629,
      "learning_rate": 3.539715172298472e-05,
      "loss": 0.758,
      "num_tokens": 2671520273.0,
      "step": 875
    },
    {
      "epoch": 1.2828356841912438,
      "grad_norm": 0.4613802454663772,
      "learning_rate": 3.5385742727205524e-05,
      "loss": 0.7383,
      "num_tokens": 2674478515.0,
      "step": 876
    },
    {
      "epoch": 1.2843011540575198,
      "grad_norm": 0.5723469994272484,
      "learning_rate": 3.5374321687681915e-05,
      "loss": 0.7431,
      "num_tokens": 2677611412.0,
      "step": 877
    },
    {
      "epoch": 1.2857666239237955,
      "grad_norm": 0.5047347499624695,
      "learning_rate": 3.5362888614689926e-05,
      "loss": 0.7466,
      "num_tokens": 2680677990.0,
      "step": 878
    },
    {
      "epoch": 1.2872320937900714,
      "grad_norm": 0.40220542916591606,
      "learning_rate": 3.5351443518516375e-05,
      "loss": 0.7338,
      "num_tokens": 2683873236.0,
      "step": 879
    },
    {
      "epoch": 1.2886975636563474,
      "grad_norm": 0.719296547305351,
      "learning_rate": 3.533998640945895e-05,
      "loss": 0.7421,
      "num_tokens": 2687040744.0,
      "step": 880
    },
    {
      "epoch": 1.290163033522623,
      "grad_norm": 0.4799917049068224,
      "learning_rate": 3.532851729782609e-05,
      "loss": 0.7463,
      "num_tokens": 2690231914.0,
      "step": 881
    },
    {
      "epoch": 1.291628503388899,
      "grad_norm": 0.649980288456009,
      "learning_rate": 3.5317036193937076e-05,
      "loss": 0.7562,
      "num_tokens": 2693452684.0,
      "step": 882
    },
    {
      "epoch": 1.293093973255175,
      "grad_norm": 0.5864776216563321,
      "learning_rate": 3.530554310812196e-05,
      "loss": 0.7281,
      "num_tokens": 2696480239.0,
      "step": 883
    },
    {
      "epoch": 1.2945594431214509,
      "grad_norm": 0.583749117807856,
      "learning_rate": 3.529403805072158e-05,
      "loss": 0.7196,
      "num_tokens": 2699452115.0,
      "step": 884
    },
    {
      "epoch": 1.2960249129877268,
      "grad_norm": 0.4439546221574545,
      "learning_rate": 3.528252103208755e-05,
      "loss": 0.7643,
      "num_tokens": 2702686208.0,
      "step": 885
    },
    {
      "epoch": 1.2974903828540025,
      "grad_norm": 0.6891188644057172,
      "learning_rate": 3.5270992062582236e-05,
      "loss": 0.7363,
      "num_tokens": 2705731990.0,
      "step": 886
    },
    {
      "epoch": 1.2989558527202785,
      "grad_norm": 0.591676696402332,
      "learning_rate": 3.525945115257876e-05,
      "loss": 0.7235,
      "num_tokens": 2708876244.0,
      "step": 887
    },
    {
      "epoch": 1.3004213225865544,
      "grad_norm": 0.4033024386014443,
      "learning_rate": 3.5247898312460994e-05,
      "loss": 0.7578,
      "num_tokens": 2711869617.0,
      "step": 888
    },
    {
      "epoch": 1.3018867924528301,
      "grad_norm": 0.5501485322714131,
      "learning_rate": 3.5236333552623536e-05,
      "loss": 0.7462,
      "num_tokens": 2714850812.0,
      "step": 889
    },
    {
      "epoch": 1.303352262319106,
      "grad_norm": 0.6634058318601604,
      "learning_rate": 3.522475688347171e-05,
      "loss": 0.7178,
      "num_tokens": 2717923189.0,
      "step": 890
    },
    {
      "epoch": 1.304817732185382,
      "grad_norm": 0.47786814637406694,
      "learning_rate": 3.5213168315421554e-05,
      "loss": 0.7275,
      "num_tokens": 2721026432.0,
      "step": 891
    },
    {
      "epoch": 1.3062832020516577,
      "grad_norm": 0.6881550321293585,
      "learning_rate": 3.5201567858899826e-05,
      "loss": 0.7455,
      "num_tokens": 2723974991.0,
      "step": 892
    },
    {
      "epoch": 1.3077486719179336,
      "grad_norm": 0.536795070157575,
      "learning_rate": 3.518995552434396e-05,
      "loss": 0.743,
      "num_tokens": 2727109890.0,
      "step": 893
    },
    {
      "epoch": 1.3092141417842096,
      "grad_norm": 0.5692289973097611,
      "learning_rate": 3.517833132220208e-05,
      "loss": 0.7436,
      "num_tokens": 2730015915.0,
      "step": 894
    },
    {
      "epoch": 1.3106796116504853,
      "grad_norm": 0.625149804943666,
      "learning_rate": 3.516669526293302e-05,
      "loss": 0.7519,
      "num_tokens": 2733262015.0,
      "step": 895
    },
    {
      "epoch": 1.3121450815167612,
      "grad_norm": 0.5464497519302005,
      "learning_rate": 3.515504735700624e-05,
      "loss": 0.7454,
      "num_tokens": 2736133138.0,
      "step": 896
    },
    {
      "epoch": 1.3136105513830372,
      "grad_norm": 0.9421700049481438,
      "learning_rate": 3.514338761490187e-05,
      "loss": 0.7372,
      "num_tokens": 2739271100.0,
      "step": 897
    },
    {
      "epoch": 1.315076021249313,
      "grad_norm": 0.6565265014193864,
      "learning_rate": 3.5131716047110704e-05,
      "loss": 0.7334,
      "num_tokens": 2742246498.0,
      "step": 898
    },
    {
      "epoch": 1.316541491115589,
      "grad_norm": 0.7901291103921996,
      "learning_rate": 3.512003266413416e-05,
      "loss": 0.725,
      "num_tokens": 2745337310.0,
      "step": 899
    },
    {
      "epoch": 1.3180069609818648,
      "grad_norm": 0.7269613575282566,
      "learning_rate": 3.51083374764843e-05,
      "loss": 0.7449,
      "num_tokens": 2748440341.0,
      "step": 900
    },
    {
      "epoch": 1.3194724308481407,
      "grad_norm": 0.7591733005729371,
      "learning_rate": 3.509663049468382e-05,
      "loss": 0.742,
      "num_tokens": 2751632761.0,
      "step": 901
    },
    {
      "epoch": 1.3209379007144166,
      "grad_norm": 0.573865126308196,
      "learning_rate": 3.508491172926597e-05,
      "loss": 0.7425,
      "num_tokens": 2754517329.0,
      "step": 902
    },
    {
      "epoch": 1.3224033705806923,
      "grad_norm": 0.8018713729767156,
      "learning_rate": 3.5073181190774665e-05,
      "loss": 0.7398,
      "num_tokens": 2757381874.0,
      "step": 903
    },
    {
      "epoch": 1.3238688404469683,
      "grad_norm": 0.513319093795368,
      "learning_rate": 3.5061438889764384e-05,
      "loss": 0.7378,
      "num_tokens": 2760492708.0,
      "step": 904
    },
    {
      "epoch": 1.3253343103132442,
      "grad_norm": 0.8144240071371247,
      "learning_rate": 3.50496848368002e-05,
      "loss": 0.7406,
      "num_tokens": 2763653827.0,
      "step": 905
    },
    {
      "epoch": 1.32679978017952,
      "grad_norm": 0.6678962157284265,
      "learning_rate": 3.503791904245774e-05,
      "loss": 0.7482,
      "num_tokens": 2766679813.0,
      "step": 906
    },
    {
      "epoch": 1.3282652500457959,
      "grad_norm": 0.7313164797008055,
      "learning_rate": 3.502614151732323e-05,
      "loss": 0.7483,
      "num_tokens": 2769515765.0,
      "step": 907
    },
    {
      "epoch": 1.3297307199120718,
      "grad_norm": 0.6108366670263453,
      "learning_rate": 3.5014352271993416e-05,
      "loss": 0.7563,
      "num_tokens": 2772513966.0,
      "step": 908
    },
    {
      "epoch": 1.3311961897783477,
      "grad_norm": 0.7368746447591218,
      "learning_rate": 3.5002551317075605e-05,
      "loss": 0.7453,
      "num_tokens": 2775430516.0,
      "step": 909
    },
    {
      "epoch": 1.3326616596446237,
      "grad_norm": 0.6535899430316167,
      "learning_rate": 3.499073866318763e-05,
      "loss": 0.7493,
      "num_tokens": 2778292388.0,
      "step": 910
    },
    {
      "epoch": 1.3341271295108994,
      "grad_norm": 0.7242669594512939,
      "learning_rate": 3.4978914320957877e-05,
      "loss": 0.7336,
      "num_tokens": 2781450279.0,
      "step": 911
    },
    {
      "epoch": 1.3355925993771753,
      "grad_norm": 0.7410910707778944,
      "learning_rate": 3.496707830102522e-05,
      "loss": 0.7348,
      "num_tokens": 2784588815.0,
      "step": 912
    },
    {
      "epoch": 1.3370580692434513,
      "grad_norm": 0.8021619705782066,
      "learning_rate": 3.495523061403904e-05,
      "loss": 0.7334,
      "num_tokens": 2787568115.0,
      "step": 913
    },
    {
      "epoch": 1.338523539109727,
      "grad_norm": 0.7166242360715578,
      "learning_rate": 3.4943371270659236e-05,
      "loss": 0.7449,
      "num_tokens": 2790342239.0,
      "step": 914
    },
    {
      "epoch": 1.339989008976003,
      "grad_norm": 0.6993206983026142,
      "learning_rate": 3.493150028155618e-05,
      "loss": 0.7402,
      "num_tokens": 2793388595.0,
      "step": 915
    },
    {
      "epoch": 1.3414544788422789,
      "grad_norm": 0.6663375919678785,
      "learning_rate": 3.491961765741073e-05,
      "loss": 0.7314,
      "num_tokens": 2796514001.0,
      "step": 916
    },
    {
      "epoch": 1.3429199487085546,
      "grad_norm": 0.6253520028667585,
      "learning_rate": 3.4907723408914196e-05,
      "loss": 0.7361,
      "num_tokens": 2799603959.0,
      "step": 917
    },
    {
      "epoch": 1.3443854185748305,
      "grad_norm": 0.5679769921239218,
      "learning_rate": 3.4895817546768365e-05,
      "loss": 0.7195,
      "num_tokens": 2802692213.0,
      "step": 918
    },
    {
      "epoch": 1.3458508884411065,
      "grad_norm": 0.6044390948647952,
      "learning_rate": 3.488390008168548e-05,
      "loss": 0.7567,
      "num_tokens": 2805650752.0,
      "step": 919
    },
    {
      "epoch": 1.3473163583073822,
      "grad_norm": 0.5294586623494583,
      "learning_rate": 3.4871971024388194e-05,
      "loss": 0.7166,
      "num_tokens": 2808966725.0,
      "step": 920
    },
    {
      "epoch": 1.348781828173658,
      "grad_norm": 0.6197481356480029,
      "learning_rate": 3.486003038560961e-05,
      "loss": 0.7558,
      "num_tokens": 2811968361.0,
      "step": 921
    },
    {
      "epoch": 1.350247298039934,
      "grad_norm": 0.4422127059512262,
      "learning_rate": 3.484807817609326e-05,
      "loss": 0.7295,
      "num_tokens": 2815190869.0,
      "step": 922
    },
    {
      "epoch": 1.35171276790621,
      "grad_norm": 0.6320052063219845,
      "learning_rate": 3.483611440659306e-05,
      "loss": 0.7287,
      "num_tokens": 2818489069.0,
      "step": 923
    },
    {
      "epoch": 1.353178237772486,
      "grad_norm": 0.4989912179242406,
      "learning_rate": 3.482413908787336e-05,
      "loss": 0.7614,
      "num_tokens": 2821507461.0,
      "step": 924
    },
    {
      "epoch": 1.3546437076387616,
      "grad_norm": 0.6948416855596531,
      "learning_rate": 3.4812152230708865e-05,
      "loss": 0.7277,
      "num_tokens": 2824764472.0,
      "step": 925
    },
    {
      "epoch": 1.3561091775050376,
      "grad_norm": 0.6054872670867794,
      "learning_rate": 3.4800153845884686e-05,
      "loss": 0.748,
      "num_tokens": 2827646531.0,
      "step": 926
    },
    {
      "epoch": 1.3575746473713135,
      "grad_norm": 0.6449690650564267,
      "learning_rate": 3.478814394419631e-05,
      "loss": 0.7273,
      "num_tokens": 2830527684.0,
      "step": 927
    },
    {
      "epoch": 1.3590401172375892,
      "grad_norm": 0.6459615081958557,
      "learning_rate": 3.477612253644956e-05,
      "loss": 0.7268,
      "num_tokens": 2833508673.0,
      "step": 928
    },
    {
      "epoch": 1.3605055871038652,
      "grad_norm": 0.6206023336750233,
      "learning_rate": 3.476408963346065e-05,
      "loss": 0.7472,
      "num_tokens": 2836495101.0,
      "step": 929
    },
    {
      "epoch": 1.361971056970141,
      "grad_norm": 0.5558160221659835,
      "learning_rate": 3.475204524605609e-05,
      "loss": 0.7375,
      "num_tokens": 2839618840.0,
      "step": 930
    },
    {
      "epoch": 1.3634365268364168,
      "grad_norm": 0.6834171666829266,
      "learning_rate": 3.473998938507276e-05,
      "loss": 0.7252,
      "num_tokens": 2842624399.0,
      "step": 931
    },
    {
      "epoch": 1.3649019967026927,
      "grad_norm": 0.535166478231198,
      "learning_rate": 3.472792206135786e-05,
      "loss": 0.7388,
      "num_tokens": 2845599578.0,
      "step": 932
    },
    {
      "epoch": 1.3663674665689687,
      "grad_norm": 0.6958789290012414,
      "learning_rate": 3.471584328576888e-05,
      "loss": 0.743,
      "num_tokens": 2849008132.0,
      "step": 933
    },
    {
      "epoch": 1.3678329364352446,
      "grad_norm": 0.5795017199080347,
      "learning_rate": 3.470375306917363e-05,
      "loss": 0.7271,
      "num_tokens": 2852241309.0,
      "step": 934
    },
    {
      "epoch": 1.3692984063015206,
      "grad_norm": 0.6430800398941781,
      "learning_rate": 3.469165142245022e-05,
      "loss": 0.7333,
      "num_tokens": 2855259819.0,
      "step": 935
    },
    {
      "epoch": 1.3707638761677963,
      "grad_norm": 0.6084882230817644,
      "learning_rate": 3.4679538356487026e-05,
      "loss": 0.734,
      "num_tokens": 2858303307.0,
      "step": 936
    },
    {
      "epoch": 1.3722293460340722,
      "grad_norm": 0.6178180948136079,
      "learning_rate": 3.4667413882182716e-05,
      "loss": 0.7472,
      "num_tokens": 2861516817.0,
      "step": 937
    },
    {
      "epoch": 1.3736948159003481,
      "grad_norm": 0.6337379342391498,
      "learning_rate": 3.465527801044622e-05,
      "loss": 0.742,
      "num_tokens": 2864492444.0,
      "step": 938
    },
    {
      "epoch": 1.3751602857666239,
      "grad_norm": 0.5602681633656569,
      "learning_rate": 3.464313075219671e-05,
      "loss": 0.738,
      "num_tokens": 2867632587.0,
      "step": 939
    },
    {
      "epoch": 1.3766257556328998,
      "grad_norm": 0.5722622089964792,
      "learning_rate": 3.463097211836361e-05,
      "loss": 0.7341,
      "num_tokens": 2870951702.0,
      "step": 940
    },
    {
      "epoch": 1.3780912254991757,
      "grad_norm": 0.5773895673524473,
      "learning_rate": 3.461880211988659e-05,
      "loss": 0.7417,
      "num_tokens": 2873623890.0,
      "step": 941
    },
    {
      "epoch": 1.3795566953654514,
      "grad_norm": 0.5525315648469733,
      "learning_rate": 3.4606620767715525e-05,
      "loss": 0.7232,
      "num_tokens": 2876713176.0,
      "step": 942
    },
    {
      "epoch": 1.3810221652317274,
      "grad_norm": 0.641720561109546,
      "learning_rate": 3.459442807281052e-05,
      "loss": 0.7448,
      "num_tokens": 2879951386.0,
      "step": 943
    },
    {
      "epoch": 1.3824876350980033,
      "grad_norm": 0.5224041688640501,
      "learning_rate": 3.458222404614188e-05,
      "loss": 0.7355,
      "num_tokens": 2883152057.0,
      "step": 944
    },
    {
      "epoch": 1.383953104964279,
      "grad_norm": 0.6819701096751417,
      "learning_rate": 3.4570008698690114e-05,
      "loss": 0.7137,
      "num_tokens": 2886164102.0,
      "step": 945
    },
    {
      "epoch": 1.385418574830555,
      "grad_norm": 0.5588874499535569,
      "learning_rate": 3.45577820414459e-05,
      "loss": 0.7346,
      "num_tokens": 2889284373.0,
      "step": 946
    },
    {
      "epoch": 1.386884044696831,
      "grad_norm": 0.5929731360099287,
      "learning_rate": 3.454554408541012e-05,
      "loss": 0.746,
      "num_tokens": 2892160330.0,
      "step": 947
    },
    {
      "epoch": 1.3883495145631068,
      "grad_norm": 0.5160411712842133,
      "learning_rate": 3.4533294841593776e-05,
      "loss": 0.7247,
      "num_tokens": 2895375745.0,
      "step": 948
    },
    {
      "epoch": 1.3898149844293828,
      "grad_norm": 0.7424513485301389,
      "learning_rate": 3.452103432101808e-05,
      "loss": 0.7565,
      "num_tokens": 2898300768.0,
      "step": 949
    },
    {
      "epoch": 1.3912804542956585,
      "grad_norm": 0.5544281865303642,
      "learning_rate": 3.450876253471434e-05,
      "loss": 0.7284,
      "num_tokens": 2901691257.0,
      "step": 950
    },
    {
      "epoch": 1.3927459241619344,
      "grad_norm": 0.6708300175017372,
      "learning_rate": 3.449647949372405e-05,
      "loss": 0.7187,
      "num_tokens": 2904669676.0,
      "step": 951
    },
    {
      "epoch": 1.3942113940282104,
      "grad_norm": 0.6765064450390005,
      "learning_rate": 3.4484185209098794e-05,
      "loss": 0.7348,
      "num_tokens": 2907777219.0,
      "step": 952
    },
    {
      "epoch": 1.395676863894486,
      "grad_norm": 0.5049804141461358,
      "learning_rate": 3.4471879691900286e-05,
      "loss": 0.7591,
      "num_tokens": 2910708073.0,
      "step": 953
    },
    {
      "epoch": 1.397142333760762,
      "grad_norm": 0.6468541829191335,
      "learning_rate": 3.445956295320033e-05,
      "loss": 0.7386,
      "num_tokens": 2913920906.0,
      "step": 954
    },
    {
      "epoch": 1.398607803627038,
      "grad_norm": 0.4467384601105551,
      "learning_rate": 3.444723500408085e-05,
      "loss": 0.746,
      "num_tokens": 2917014602.0,
      "step": 955
    },
    {
      "epoch": 1.4000732734933137,
      "grad_norm": 0.6180670226839605,
      "learning_rate": 3.443489585563385e-05,
      "loss": 0.7346,
      "num_tokens": 2919983150.0,
      "step": 956
    },
    {
      "epoch": 1.4015387433595896,
      "grad_norm": 0.5282588342729994,
      "learning_rate": 3.4422545518961414e-05,
      "loss": 0.7302,
      "num_tokens": 2922957480.0,
      "step": 957
    },
    {
      "epoch": 1.4030042132258655,
      "grad_norm": 0.7002150090693635,
      "learning_rate": 3.4410184005175664e-05,
      "loss": 0.7442,
      "num_tokens": 2926140811.0,
      "step": 958
    },
    {
      "epoch": 1.4044696830921415,
      "grad_norm": 0.47725222938137485,
      "learning_rate": 3.4397811325398814e-05,
      "loss": 0.7318,
      "num_tokens": 2929194529.0,
      "step": 959
    },
    {
      "epoch": 1.4059351529584174,
      "grad_norm": 0.747502314767629,
      "learning_rate": 3.4385427490763116e-05,
      "loss": 0.727,
      "num_tokens": 2932193635.0,
      "step": 960
    },
    {
      "epoch": 1.4074006228246931,
      "grad_norm": 0.552906821488841,
      "learning_rate": 3.4373032512410845e-05,
      "loss": 0.7563,
      "num_tokens": 2935381047.0,
      "step": 961
    },
    {
      "epoch": 1.408866092690969,
      "grad_norm": 0.8080846582881377,
      "learning_rate": 3.436062640149431e-05,
      "loss": 0.7363,
      "num_tokens": 2938530550.0,
      "step": 962
    },
    {
      "epoch": 1.410331562557245,
      "grad_norm": 0.7453059903364085,
      "learning_rate": 3.434820916917584e-05,
      "loss": 0.7401,
      "num_tokens": 2941383867.0,
      "step": 963
    },
    {
      "epoch": 1.4117970324235207,
      "grad_norm": 0.672349403884283,
      "learning_rate": 3.4335780826627764e-05,
      "loss": 0.736,
      "num_tokens": 2944459169.0,
      "step": 964
    },
    {
      "epoch": 1.4132625022897967,
      "grad_norm": 0.6891330381682266,
      "learning_rate": 3.432334138503242e-05,
      "loss": 0.7271,
      "num_tokens": 2947851010.0,
      "step": 965
    },
    {
      "epoch": 1.4147279721560726,
      "grad_norm": 0.6429726196876129,
      "learning_rate": 3.431089085558212e-05,
      "loss": 0.7572,
      "num_tokens": 2951004105.0,
      "step": 966
    },
    {
      "epoch": 1.4161934420223483,
      "grad_norm": 0.562623386368141,
      "learning_rate": 3.429842924947916e-05,
      "loss": 0.746,
      "num_tokens": 2954020786.0,
      "step": 967
    },
    {
      "epoch": 1.4176589118886243,
      "grad_norm": 0.607478971387216,
      "learning_rate": 3.428595657793578e-05,
      "loss": 0.7417,
      "num_tokens": 2957099154.0,
      "step": 968
    },
    {
      "epoch": 1.4191243817549002,
      "grad_norm": 0.5389546295318713,
      "learning_rate": 3.427347285217421e-05,
      "loss": 0.7351,
      "num_tokens": 2960108352.0,
      "step": 969
    },
    {
      "epoch": 1.420589851621176,
      "grad_norm": 0.5754693986416561,
      "learning_rate": 3.42609780834266e-05,
      "loss": 0.7389,
      "num_tokens": 2963035631.0,
      "step": 970
    },
    {
      "epoch": 1.4220553214874518,
      "grad_norm": 0.544263037545951,
      "learning_rate": 3.4248472282935065e-05,
      "loss": 0.7185,
      "num_tokens": 2966120311.0,
      "step": 971
    },
    {
      "epoch": 1.4235207913537278,
      "grad_norm": 0.46744224241911153,
      "learning_rate": 3.42359554619516e-05,
      "loss": 0.7361,
      "num_tokens": 2969538335.0,
      "step": 972
    },
    {
      "epoch": 1.4249862612200037,
      "grad_norm": 0.558341556330857,
      "learning_rate": 3.422342763173817e-05,
      "loss": 0.734,
      "num_tokens": 2972842448.0,
      "step": 973
    },
    {
      "epoch": 1.4264517310862797,
      "grad_norm": 0.5704332279149912,
      "learning_rate": 3.4210888803566586e-05,
      "loss": 0.7463,
      "num_tokens": 2975764776.0,
      "step": 974
    },
    {
      "epoch": 1.4279172009525554,
      "grad_norm": 0.5027732741836143,
      "learning_rate": 3.4198338988718606e-05,
      "loss": 0.7209,
      "num_tokens": 2978793947.0,
      "step": 975
    },
    {
      "epoch": 1.4293826708188313,
      "grad_norm": 0.4886380327133936,
      "learning_rate": 3.418577819848587e-05,
      "loss": 0.7454,
      "num_tokens": 2981910995.0,
      "step": 976
    },
    {
      "epoch": 1.4308481406851072,
      "grad_norm": 0.668011447770949,
      "learning_rate": 3.417320644416983e-05,
      "loss": 0.7591,
      "num_tokens": 2984863255.0,
      "step": 977
    },
    {
      "epoch": 1.432313610551383,
      "grad_norm": 0.4084779472147276,
      "learning_rate": 3.4160623737081886e-05,
      "loss": 0.7437,
      "num_tokens": 2987818764.0,
      "step": 978
    },
    {
      "epoch": 1.433779080417659,
      "grad_norm": 0.7710555219278353,
      "learning_rate": 3.414803008854324e-05,
      "loss": 0.7488,
      "num_tokens": 2991012288.0,
      "step": 979
    },
    {
      "epoch": 1.4352445502839348,
      "grad_norm": 0.4769981754981097,
      "learning_rate": 3.413542550988497e-05,
      "loss": 0.7411,
      "num_tokens": 2993879474.0,
      "step": 980
    },
    {
      "epoch": 1.4367100201502105,
      "grad_norm": 0.7442189314526665,
      "learning_rate": 3.412281001244796e-05,
      "loss": 0.7237,
      "num_tokens": 2996964695.0,
      "step": 981
    },
    {
      "epoch": 1.4381754900164865,
      "grad_norm": 0.5896864791581798,
      "learning_rate": 3.4110183607582924e-05,
      "loss": 0.734,
      "num_tokens": 2999946705.0,
      "step": 982
    },
    {
      "epoch": 1.4396409598827624,
      "grad_norm": 0.7182802812964383,
      "learning_rate": 3.4097546306650405e-05,
      "loss": 0.7266,
      "num_tokens": 3003064026.0,
      "step": 983
    },
    {
      "epoch": 1.4411064297490384,
      "grad_norm": 0.6154766894616646,
      "learning_rate": 3.4084898121020745e-05,
      "loss": 0.7309,
      "num_tokens": 3006311072.0,
      "step": 984
    },
    {
      "epoch": 1.4425718996153143,
      "grad_norm": 0.5629964853819998,
      "learning_rate": 3.407223906207407e-05,
      "loss": 0.7336,
      "num_tokens": 3009305560.0,
      "step": 985
    },
    {
      "epoch": 1.44403736948159,
      "grad_norm": 0.6747766488021395,
      "learning_rate": 3.40595691412003e-05,
      "loss": 0.7291,
      "num_tokens": 3012366568.0,
      "step": 986
    },
    {
      "epoch": 1.445502839347866,
      "grad_norm": 0.5401785966747638,
      "learning_rate": 3.404688836979911e-05,
      "loss": 0.748,
      "num_tokens": 3015395405.0,
      "step": 987
    },
    {
      "epoch": 1.4469683092141419,
      "grad_norm": 0.6958414132208053,
      "learning_rate": 3.403419675927997e-05,
      "loss": 0.7302,
      "num_tokens": 3018312192.0,
      "step": 988
    },
    {
      "epoch": 1.4484337790804176,
      "grad_norm": 0.7072724335636941,
      "learning_rate": 3.4021494321062075e-05,
      "loss": 0.7232,
      "num_tokens": 3021345414.0,
      "step": 989
    },
    {
      "epoch": 1.4498992489466935,
      "grad_norm": 0.619813952584958,
      "learning_rate": 3.4008781066574364e-05,
      "loss": 0.7384,
      "num_tokens": 3024209697.0,
      "step": 990
    },
    {
      "epoch": 1.4513647188129695,
      "grad_norm": 0.6326259284214336,
      "learning_rate": 3.3996057007255515e-05,
      "loss": 0.7201,
      "num_tokens": 3027275361.0,
      "step": 991
    },
    {
      "epoch": 1.4528301886792452,
      "grad_norm": 0.6490924108908067,
      "learning_rate": 3.3983322154553925e-05,
      "loss": 0.7453,
      "num_tokens": 3030208820.0,
      "step": 992
    },
    {
      "epoch": 1.4542956585455211,
      "grad_norm": 0.5454165487780195,
      "learning_rate": 3.397057651992772e-05,
      "loss": 0.7365,
      "num_tokens": 3033375472.0,
      "step": 993
    },
    {
      "epoch": 1.455761128411797,
      "grad_norm": 0.6814216328849936,
      "learning_rate": 3.3957820114844695e-05,
      "loss": 0.7269,
      "num_tokens": 3036382522.0,
      "step": 994
    },
    {
      "epoch": 1.4572265982780728,
      "grad_norm": 0.5297811835999773,
      "learning_rate": 3.3945052950782364e-05,
      "loss": 0.7474,
      "num_tokens": 3039228237.0,
      "step": 995
    },
    {
      "epoch": 1.4586920681443487,
      "grad_norm": 0.8430452099991067,
      "learning_rate": 3.39322750392279e-05,
      "loss": 0.7386,
      "num_tokens": 3042036354.0,
      "step": 996
    },
    {
      "epoch": 1.4601575380106246,
      "grad_norm": 0.618661490915788,
      "learning_rate": 3.391948639167814e-05,
      "loss": 0.7338,
      "num_tokens": 3044980209.0,
      "step": 997
    },
    {
      "epoch": 1.4616230078769006,
      "grad_norm": 0.8117242423674851,
      "learning_rate": 3.390668701963963e-05,
      "loss": 0.7414,
      "num_tokens": 3047910690.0,
      "step": 998
    },
    {
      "epoch": 1.4630884777431765,
      "grad_norm": 0.6036432914437648,
      "learning_rate": 3.389387693462849e-05,
      "loss": 0.7431,
      "num_tokens": 3051078641.0,
      "step": 999
    },
    {
      "epoch": 1.4645539476094522,
      "grad_norm": 0.7211648519234534,
      "learning_rate": 3.3881056148170555e-05,
      "loss": 0.7375,
      "num_tokens": 3054309968.0,
      "step": 1000
    },
    {
      "epoch": 1.4660194174757282,
      "grad_norm": 0.6218336784876674,
      "learning_rate": 3.386822467180124e-05,
      "loss": 0.7423,
      "num_tokens": 3057172620.0,
      "step": 1001
    },
    {
      "epoch": 1.467484887342004,
      "grad_norm": 0.5984309290919958,
      "learning_rate": 3.3855382517065596e-05,
      "loss": 0.7147,
      "num_tokens": 3060346309.0,
      "step": 1002
    },
    {
      "epoch": 1.4689503572082798,
      "grad_norm": 0.6385437456365122,
      "learning_rate": 3.384252969551827e-05,
      "loss": 0.745,
      "num_tokens": 3063391272.0,
      "step": 1003
    },
    {
      "epoch": 1.4704158270745558,
      "grad_norm": 0.6297622733277343,
      "learning_rate": 3.382966621872351e-05,
      "loss": 0.7568,
      "num_tokens": 3066181361.0,
      "step": 1004
    },
    {
      "epoch": 1.4718812969408317,
      "grad_norm": 0.5651789372439435,
      "learning_rate": 3.381679209825517e-05,
      "loss": 0.7291,
      "num_tokens": 3069156520.0,
      "step": 1005
    },
    {
      "epoch": 1.4733467668071074,
      "grad_norm": 0.5589722648461087,
      "learning_rate": 3.3803907345696646e-05,
      "loss": 0.7407,
      "num_tokens": 3072397598.0,
      "step": 1006
    },
    {
      "epoch": 1.4748122366733833,
      "grad_norm": 0.524079519668168,
      "learning_rate": 3.379101197264093e-05,
      "loss": 0.7324,
      "num_tokens": 3075555539.0,
      "step": 1007
    },
    {
      "epoch": 1.4762777065396593,
      "grad_norm": 0.6180532395165693,
      "learning_rate": 3.377810599069056e-05,
      "loss": 0.7352,
      "num_tokens": 3078543904.0,
      "step": 1008
    },
    {
      "epoch": 1.4777431764059352,
      "grad_norm": 0.5229208127610999,
      "learning_rate": 3.376518941145762e-05,
      "loss": 0.7158,
      "num_tokens": 3081761956.0,
      "step": 1009
    },
    {
      "epoch": 1.4792086462722112,
      "grad_norm": 0.6383453906067004,
      "learning_rate": 3.375226224656372e-05,
      "loss": 0.7273,
      "num_tokens": 3084811251.0,
      "step": 1010
    },
    {
      "epoch": 1.4806741161384869,
      "grad_norm": 0.49514539961983495,
      "learning_rate": 3.3739324507640015e-05,
      "loss": 0.7521,
      "num_tokens": 3087806764.0,
      "step": 1011
    },
    {
      "epoch": 1.4821395860047628,
      "grad_norm": 0.8699250735380287,
      "learning_rate": 3.372637620632715e-05,
      "loss": 0.7389,
      "num_tokens": 3090779784.0,
      "step": 1012
    },
    {
      "epoch": 1.4836050558710387,
      "grad_norm": 0.48935068934199033,
      "learning_rate": 3.371341735427528e-05,
      "loss": 0.7234,
      "num_tokens": 3094195816.0,
      "step": 1013
    },
    {
      "epoch": 1.4850705257373145,
      "grad_norm": 1.0395314552772876,
      "learning_rate": 3.370044796314407e-05,
      "loss": 0.7453,
      "num_tokens": 3097399998.0,
      "step": 1014
    },
    {
      "epoch": 1.4865359956035904,
      "grad_norm": 0.7646123752473198,
      "learning_rate": 3.3687468044602646e-05,
      "loss": 0.7234,
      "num_tokens": 3100545754.0,
      "step": 1015
    },
    {
      "epoch": 1.4880014654698663,
      "grad_norm": 0.8611666423619171,
      "learning_rate": 3.367447761032961e-05,
      "loss": 0.7499,
      "num_tokens": 3103484358.0,
      "step": 1016
    },
    {
      "epoch": 1.489466935336142,
      "grad_norm": 0.7014876343470275,
      "learning_rate": 3.366147667201304e-05,
      "loss": 0.7246,
      "num_tokens": 3106717553.0,
      "step": 1017
    },
    {
      "epoch": 1.490932405202418,
      "grad_norm": 0.8213018732023244,
      "learning_rate": 3.364846524135045e-05,
      "loss": 0.7328,
      "num_tokens": 3109883877.0,
      "step": 1018
    },
    {
      "epoch": 1.492397875068694,
      "grad_norm": 0.6751736098287298,
      "learning_rate": 3.36354433300488e-05,
      "loss": 0.7382,
      "num_tokens": 3112957536.0,
      "step": 1019
    },
    {
      "epoch": 1.4938633449349696,
      "grad_norm": 0.7140332273375319,
      "learning_rate": 3.3622410949824474e-05,
      "loss": 0.726,
      "num_tokens": 3116099381.0,
      "step": 1020
    },
    {
      "epoch": 1.4953288148012456,
      "grad_norm": 0.7338329659282037,
      "learning_rate": 3.360936811240328e-05,
      "loss": 0.7364,
      "num_tokens": 3119184190.0,
      "step": 1021
    },
    {
      "epoch": 1.4967942846675215,
      "grad_norm": 0.5879790484116247,
      "learning_rate": 3.3596314829520453e-05,
      "loss": 0.7443,
      "num_tokens": 3122460671.0,
      "step": 1022
    },
    {
      "epoch": 1.4982597545337974,
      "grad_norm": 0.6465434506831328,
      "learning_rate": 3.35832511129206e-05,
      "loss": 0.7413,
      "num_tokens": 3125486302.0,
      "step": 1023
    },
    {
      "epoch": 1.4997252244000734,
      "grad_norm": 0.6191655344923633,
      "learning_rate": 3.3570176974357714e-05,
      "loss": 0.7185,
      "num_tokens": 3128471232.0,
      "step": 1024
    },
    {
      "epoch": 1.501190694266349,
      "grad_norm": 0.5394981482070776,
      "learning_rate": 3.3557092425595176e-05,
      "loss": 0.7161,
      "num_tokens": 3131622502.0,
      "step": 1025
    },
    {
      "epoch": 1.502656164132625,
      "grad_norm": 0.5400216333786843,
      "learning_rate": 3.354399747840575e-05,
      "loss": 0.7179,
      "num_tokens": 3134599862.0,
      "step": 1026
    },
    {
      "epoch": 1.504121633998901,
      "grad_norm": 0.46173965699052566,
      "learning_rate": 3.353089214457152e-05,
      "loss": 0.721,
      "num_tokens": 3137434405.0,
      "step": 1027
    },
    {
      "epoch": 1.5055871038651767,
      "grad_norm": 0.5916615463438385,
      "learning_rate": 3.351777643588394e-05,
      "loss": 0.7391,
      "num_tokens": 3140414626.0,
      "step": 1028
    },
    {
      "epoch": 1.5070525737314526,
      "grad_norm": 0.4051633404578729,
      "learning_rate": 3.35046503641438e-05,
      "loss": 0.7105,
      "num_tokens": 3143608942.0,
      "step": 1029
    },
    {
      "epoch": 1.5085180435977286,
      "grad_norm": 0.6945349103596598,
      "learning_rate": 3.34915139411612e-05,
      "loss": 0.7415,
      "num_tokens": 3146615569.0,
      "step": 1030
    },
    {
      "epoch": 1.5099835134640043,
      "grad_norm": 0.524614205099497,
      "learning_rate": 3.347836717875556e-05,
      "loss": 0.7327,
      "num_tokens": 3149682923.0,
      "step": 1031
    },
    {
      "epoch": 1.5114489833302802,
      "grad_norm": 0.6678133863051265,
      "learning_rate": 3.346521008875561e-05,
      "loss": 0.7414,
      "num_tokens": 3152953914.0,
      "step": 1032
    },
    {
      "epoch": 1.5129144531965562,
      "grad_norm": 0.5011042569664438,
      "learning_rate": 3.345204268299936e-05,
      "loss": 0.7293,
      "num_tokens": 3155990327.0,
      "step": 1033
    },
    {
      "epoch": 1.5143799230628319,
      "grad_norm": 0.7013384833743773,
      "learning_rate": 3.343886497333411e-05,
      "loss": 0.724,
      "num_tokens": 3158890670.0,
      "step": 1034
    },
    {
      "epoch": 1.515845392929108,
      "grad_norm": 0.5873282531934446,
      "learning_rate": 3.342567697161642e-05,
      "loss": 0.7431,
      "num_tokens": 3161951741.0,
      "step": 1035
    },
    {
      "epoch": 1.5173108627953837,
      "grad_norm": 0.6791240518885524,
      "learning_rate": 3.341247868971213e-05,
      "loss": 0.733,
      "num_tokens": 3164976386.0,
      "step": 1036
    },
    {
      "epoch": 1.5187763326616597,
      "grad_norm": 0.5949906401251243,
      "learning_rate": 3.3399270139496306e-05,
      "loss": 0.7612,
      "num_tokens": 3167889904.0,
      "step": 1037
    },
    {
      "epoch": 1.5202418025279356,
      "grad_norm": 0.6706588465673512,
      "learning_rate": 3.338605133285327e-05,
      "loss": 0.7119,
      "num_tokens": 3171120853.0,
      "step": 1038
    },
    {
      "epoch": 1.5217072723942113,
      "grad_norm": 0.6429484704240662,
      "learning_rate": 3.3372822281676565e-05,
      "loss": 0.7201,
      "num_tokens": 3174232159.0,
      "step": 1039
    },
    {
      "epoch": 1.5231727422604873,
      "grad_norm": 0.5746702477684014,
      "learning_rate": 3.335958299786895e-05,
      "loss": 0.7596,
      "num_tokens": 3177062588.0,
      "step": 1040
    },
    {
      "epoch": 1.5246382121267632,
      "grad_norm": 0.545463085540361,
      "learning_rate": 3.3346333493342414e-05,
      "loss": 0.7378,
      "num_tokens": 3180182151.0,
      "step": 1041
    },
    {
      "epoch": 1.526103681993039,
      "grad_norm": 0.5929814988867435,
      "learning_rate": 3.3333073780018086e-05,
      "loss": 0.732,
      "num_tokens": 3183146742.0,
      "step": 1042
    },
    {
      "epoch": 1.5275691518593149,
      "grad_norm": 0.49123305976961607,
      "learning_rate": 3.3319803869826354e-05,
      "loss": 0.7251,
      "num_tokens": 3186262520.0,
      "step": 1043
    },
    {
      "epoch": 1.5290346217255908,
      "grad_norm": 0.6407367429991835,
      "learning_rate": 3.3306523774706715e-05,
      "loss": 0.7343,
      "num_tokens": 3189152324.0,
      "step": 1044
    },
    {
      "epoch": 1.5305000915918665,
      "grad_norm": 0.47801118859236186,
      "learning_rate": 3.329323350660787e-05,
      "loss": 0.7049,
      "num_tokens": 3192112782.0,
      "step": 1045
    },
    {
      "epoch": 1.5319655614581427,
      "grad_norm": 0.6416701796670824,
      "learning_rate": 3.327993307748766e-05,
      "loss": 0.723,
      "num_tokens": 3195045774.0,
      "step": 1046
    },
    {
      "epoch": 1.5334310313244184,
      "grad_norm": 0.5557665646133894,
      "learning_rate": 3.3266622499313066e-05,
      "loss": 0.7217,
      "num_tokens": 3198097635.0,
      "step": 1047
    },
    {
      "epoch": 1.5348965011906943,
      "grad_norm": 0.5704360442709387,
      "learning_rate": 3.3253301784060207e-05,
      "loss": 0.7299,
      "num_tokens": 3201147087.0,
      "step": 1048
    },
    {
      "epoch": 1.5363619710569703,
      "grad_norm": 0.49295754630428207,
      "learning_rate": 3.323997094371431e-05,
      "loss": 0.7339,
      "num_tokens": 3204216507.0,
      "step": 1049
    },
    {
      "epoch": 1.537827440923246,
      "grad_norm": 0.624907743916921,
      "learning_rate": 3.322662999026974e-05,
      "loss": 0.7335,
      "num_tokens": 3207563407.0,
      "step": 1050
    },
    {
      "epoch": 1.539292910789522,
      "grad_norm": 0.564770988258131,
      "learning_rate": 3.3213278935729924e-05,
      "loss": 0.7316,
      "num_tokens": 3210611152.0,
      "step": 1051
    },
    {
      "epoch": 1.5407583806557978,
      "grad_norm": 0.7179081314393227,
      "learning_rate": 3.31999177921074e-05,
      "loss": 0.7306,
      "num_tokens": 3213727576.0,
      "step": 1052
    },
    {
      "epoch": 1.5422238505220736,
      "grad_norm": 0.5848492695758731,
      "learning_rate": 3.318654657142379e-05,
      "loss": 0.7228,
      "num_tokens": 3216904552.0,
      "step": 1053
    },
    {
      "epoch": 1.5436893203883495,
      "grad_norm": 0.5371751676311499,
      "learning_rate": 3.317316528570976e-05,
      "loss": 0.7184,
      "num_tokens": 3219952403.0,
      "step": 1054
    },
    {
      "epoch": 1.5451547902546254,
      "grad_norm": 0.5781681539920612,
      "learning_rate": 3.3159773947005054e-05,
      "loss": 0.7258,
      "num_tokens": 3222983927.0,
      "step": 1055
    },
    {
      "epoch": 1.5466202601209011,
      "grad_norm": 0.4889391421591862,
      "learning_rate": 3.314637256735844e-05,
      "loss": 0.7217,
      "num_tokens": 3226075562.0,
      "step": 1056
    },
    {
      "epoch": 1.548085729987177,
      "grad_norm": 0.5275678672623512,
      "learning_rate": 3.3132961158827733e-05,
      "loss": 0.7301,
      "num_tokens": 3229168040.0,
      "step": 1057
    },
    {
      "epoch": 1.549551199853453,
      "grad_norm": 0.4633066536625952,
      "learning_rate": 3.3119539733479785e-05,
      "loss": 0.7547,
      "num_tokens": 3232171336.0,
      "step": 1058
    },
    {
      "epoch": 1.5510166697197287,
      "grad_norm": 0.5483144279440194,
      "learning_rate": 3.310610830339044e-05,
      "loss": 0.7212,
      "num_tokens": 3235061721.0,
      "step": 1059
    },
    {
      "epoch": 1.552482139586005,
      "grad_norm": 0.48880476646932947,
      "learning_rate": 3.3092666880644525e-05,
      "loss": 0.7243,
      "num_tokens": 3237903227.0,
      "step": 1060
    },
    {
      "epoch": 1.5539476094522806,
      "grad_norm": 0.46243286275003714,
      "learning_rate": 3.307921547733592e-05,
      "loss": 0.735,
      "num_tokens": 3240970329.0,
      "step": 1061
    },
    {
      "epoch": 1.5554130793185565,
      "grad_norm": 0.7132371324763109,
      "learning_rate": 3.306575410556742e-05,
      "loss": 0.723,
      "num_tokens": 3243978829.0,
      "step": 1062
    },
    {
      "epoch": 1.5568785491848325,
      "grad_norm": 0.4477501541427797,
      "learning_rate": 3.305228277745083e-05,
      "loss": 0.7218,
      "num_tokens": 3247116034.0,
      "step": 1063
    },
    {
      "epoch": 1.5583440190511082,
      "grad_norm": 0.6760630324645992,
      "learning_rate": 3.30388015051069e-05,
      "loss": 0.7285,
      "num_tokens": 3250381168.0,
      "step": 1064
    },
    {
      "epoch": 1.5598094889173841,
      "grad_norm": 0.5942818310541969,
      "learning_rate": 3.302531030066531e-05,
      "loss": 0.7117,
      "num_tokens": 3253460108.0,
      "step": 1065
    },
    {
      "epoch": 1.56127495878366,
      "grad_norm": 0.5297456474104453,
      "learning_rate": 3.301180917626471e-05,
      "loss": 0.7263,
      "num_tokens": 3256471161.0,
      "step": 1066
    },
    {
      "epoch": 1.5627404286499358,
      "grad_norm": 0.5573329355506744,
      "learning_rate": 3.299829814405265e-05,
      "loss": 0.7352,
      "num_tokens": 3259645447.0,
      "step": 1067
    },
    {
      "epoch": 1.5642058985162117,
      "grad_norm": 0.5511109591257657,
      "learning_rate": 3.298477721618561e-05,
      "loss": 0.7379,
      "num_tokens": 3262751134.0,
      "step": 1068
    },
    {
      "epoch": 1.5656713683824877,
      "grad_norm": 0.6279421627656148,
      "learning_rate": 3.297124640482896e-05,
      "loss": 0.7301,
      "num_tokens": 3265827560.0,
      "step": 1069
    },
    {
      "epoch": 1.5671368382487634,
      "grad_norm": 0.4783115147960427,
      "learning_rate": 3.295770572215697e-05,
      "loss": 0.7258,
      "num_tokens": 3268846708.0,
      "step": 1070
    },
    {
      "epoch": 1.5686023081150395,
      "grad_norm": 0.6547957915224174,
      "learning_rate": 3.2944155180352795e-05,
      "loss": 0.7408,
      "num_tokens": 3271910814.0,
      "step": 1071
    },
    {
      "epoch": 1.5700677779813152,
      "grad_norm": 0.45670962151841776,
      "learning_rate": 3.2930594791608454e-05,
      "loss": 0.7241,
      "num_tokens": 3274836099.0,
      "step": 1072
    },
    {
      "epoch": 1.5715332478475912,
      "grad_norm": 0.6451060056726136,
      "learning_rate": 3.291702456812483e-05,
      "loss": 0.7505,
      "num_tokens": 3277794487.0,
      "step": 1073
    },
    {
      "epoch": 1.5729987177138671,
      "grad_norm": 0.5679023782283982,
      "learning_rate": 3.290344452211164e-05,
      "loss": 0.7434,
      "num_tokens": 3281078531.0,
      "step": 1074
    },
    {
      "epoch": 1.5744641875801428,
      "grad_norm": 0.6597028958316802,
      "learning_rate": 3.2889854665787466e-05,
      "loss": 0.7179,
      "num_tokens": 3284324102.0,
      "step": 1075
    },
    {
      "epoch": 1.5759296574464188,
      "grad_norm": 0.56692146377955,
      "learning_rate": 3.2876255011379704e-05,
      "loss": 0.7169,
      "num_tokens": 3287622327.0,
      "step": 1076
    },
    {
      "epoch": 1.5773951273126947,
      "grad_norm": 0.6026189001325984,
      "learning_rate": 3.286264557112456e-05,
      "loss": 0.7486,
      "num_tokens": 3290657528.0,
      "step": 1077
    },
    {
      "epoch": 1.5788605971789704,
      "grad_norm": 0.5367689355581213,
      "learning_rate": 3.284902635726704e-05,
      "loss": 0.7389,
      "num_tokens": 3293587118.0,
      "step": 1078
    },
    {
      "epoch": 1.5803260670452464,
      "grad_norm": 0.47501642869900035,
      "learning_rate": 3.2835397382060966e-05,
      "loss": 0.7158,
      "num_tokens": 3296569848.0,
      "step": 1079
    },
    {
      "epoch": 1.5817915369115223,
      "grad_norm": 0.5319110225524265,
      "learning_rate": 3.282175865776892e-05,
      "loss": 0.7136,
      "num_tokens": 3299573942.0,
      "step": 1080
    },
    {
      "epoch": 1.583257006777798,
      "grad_norm": 0.5111563929543897,
      "learning_rate": 3.280811019666227e-05,
      "loss": 0.7254,
      "num_tokens": 3302835000.0,
      "step": 1081
    },
    {
      "epoch": 1.584722476644074,
      "grad_norm": 0.47152872247235283,
      "learning_rate": 3.279445201102114e-05,
      "loss": 0.7433,
      "num_tokens": 3305790323.0,
      "step": 1082
    },
    {
      "epoch": 1.5861879465103499,
      "grad_norm": 0.4350385103265398,
      "learning_rate": 3.278078411313439e-05,
      "loss": 0.7294,
      "num_tokens": 3308699330.0,
      "step": 1083
    },
    {
      "epoch": 1.5876534163766256,
      "grad_norm": 0.6991661751412634,
      "learning_rate": 3.276710651529966e-05,
      "loss": 0.7214,
      "num_tokens": 3311852068.0,
      "step": 1084
    },
    {
      "epoch": 1.5891188862429018,
      "grad_norm": 0.41310351036312054,
      "learning_rate": 3.275341922982326e-05,
      "loss": 0.7282,
      "num_tokens": 3314935902.0,
      "step": 1085
    },
    {
      "epoch": 1.5905843561091775,
      "grad_norm": 0.7182567783415769,
      "learning_rate": 3.2739722269020274e-05,
      "loss": 0.7288,
      "num_tokens": 3317773354.0,
      "step": 1086
    },
    {
      "epoch": 1.5920498259754534,
      "grad_norm": 0.5026910431209334,
      "learning_rate": 3.272601564521444e-05,
      "loss": 0.7187,
      "num_tokens": 3320777060.0,
      "step": 1087
    },
    {
      "epoch": 1.5935152958417294,
      "grad_norm": 0.7936574154842663,
      "learning_rate": 3.2712299370738216e-05,
      "loss": 0.7139,
      "num_tokens": 3323783788.0,
      "step": 1088
    },
    {
      "epoch": 1.594980765708005,
      "grad_norm": 0.5904362906427473,
      "learning_rate": 3.2698573457932756e-05,
      "loss": 0.7336,
      "num_tokens": 3327053346.0,
      "step": 1089
    },
    {
      "epoch": 1.596446235574281,
      "grad_norm": 0.6557313003915473,
      "learning_rate": 3.268483791914785e-05,
      "loss": 0.7172,
      "num_tokens": 3330003713.0,
      "step": 1090
    },
    {
      "epoch": 1.597911705440557,
      "grad_norm": 0.62142816696905,
      "learning_rate": 3.267109276674199e-05,
      "loss": 0.7163,
      "num_tokens": 3333253515.0,
      "step": 1091
    },
    {
      "epoch": 1.5993771753068327,
      "grad_norm": 0.5871516815498982,
      "learning_rate": 3.265733801308228e-05,
      "loss": 0.7301,
      "num_tokens": 3336608597.0,
      "step": 1092
    },
    {
      "epoch": 1.6008426451731086,
      "grad_norm": 0.5796492999924486,
      "learning_rate": 3.264357367054449e-05,
      "loss": 0.74,
      "num_tokens": 3339606789.0,
      "step": 1093
    },
    {
      "epoch": 1.6023081150393845,
      "grad_norm": 0.5926079023957083,
      "learning_rate": 3.2629799751512996e-05,
      "loss": 0.7332,
      "num_tokens": 3342709017.0,
      "step": 1094
    },
    {
      "epoch": 1.6037735849056602,
      "grad_norm": 0.5532692724452195,
      "learning_rate": 3.2616016268380815e-05,
      "loss": 0.7118,
      "num_tokens": 3345740170.0,
      "step": 1095
    },
    {
      "epoch": 1.6052390547719364,
      "grad_norm": 0.47389646118390866,
      "learning_rate": 3.2602223233549556e-05,
      "loss": 0.7194,
      "num_tokens": 3348752435.0,
      "step": 1096
    },
    {
      "epoch": 1.6067045246382121,
      "grad_norm": 0.49737385676964324,
      "learning_rate": 3.258842065942941e-05,
      "loss": 0.7443,
      "num_tokens": 3351883456.0,
      "step": 1097
    },
    {
      "epoch": 1.608169994504488,
      "grad_norm": 0.5505309311923081,
      "learning_rate": 3.257460855843917e-05,
      "loss": 0.7424,
      "num_tokens": 3354872328.0,
      "step": 1098
    },
    {
      "epoch": 1.609635464370764,
      "grad_norm": 0.5179267250505282,
      "learning_rate": 3.25607869430062e-05,
      "loss": 0.72,
      "num_tokens": 3357911934.0,
      "step": 1099
    },
    {
      "epoch": 1.6111009342370397,
      "grad_norm": 0.532374180336076,
      "learning_rate": 3.2546955825566404e-05,
      "loss": 0.724,
      "num_tokens": 3360937464.0,
      "step": 1100
    },
    {
      "epoch": 1.6125664041033156,
      "grad_norm": 0.4454089975890635,
      "learning_rate": 3.2533115218564265e-05,
      "loss": 0.7395,
      "num_tokens": 3363940797.0,
      "step": 1101
    },
    {
      "epoch": 1.6140318739695916,
      "grad_norm": 0.5161215969558626,
      "learning_rate": 3.2519265134452775e-05,
      "loss": 0.7724,
      "num_tokens": 3367099238.0,
      "step": 1102
    },
    {
      "epoch": 1.6154973438358673,
      "grad_norm": 0.4747798841869285,
      "learning_rate": 3.250540558569348e-05,
      "loss": 0.7602,
      "num_tokens": 3369999554.0,
      "step": 1103
    },
    {
      "epoch": 1.6169628137021432,
      "grad_norm": 0.4847535669381688,
      "learning_rate": 3.2491536584756416e-05,
      "loss": 0.7274,
      "num_tokens": 3372943501.0,
      "step": 1104
    },
    {
      "epoch": 1.6184282835684192,
      "grad_norm": 0.4115228114749518,
      "learning_rate": 3.247765814412014e-05,
      "loss": 0.7457,
      "num_tokens": 3375951689.0,
      "step": 1105
    },
    {
      "epoch": 1.6198937534346949,
      "grad_norm": 0.5336664125975602,
      "learning_rate": 3.246377027627171e-05,
      "loss": 0.7439,
      "num_tokens": 3379156676.0,
      "step": 1106
    },
    {
      "epoch": 1.6213592233009708,
      "grad_norm": 0.401389918543112,
      "learning_rate": 3.244987299370664e-05,
      "loss": 0.7501,
      "num_tokens": 3382240737.0,
      "step": 1107
    },
    {
      "epoch": 1.6228246931672468,
      "grad_norm": 0.7268465763963753,
      "learning_rate": 3.243596630892894e-05,
      "loss": 0.7243,
      "num_tokens": 3385196219.0,
      "step": 1108
    },
    {
      "epoch": 1.6242901630335225,
      "grad_norm": 0.5300005206436273,
      "learning_rate": 3.242205023445106e-05,
      "loss": 0.7368,
      "num_tokens": 3387967027.0,
      "step": 1109
    },
    {
      "epoch": 1.6257556328997986,
      "grad_norm": 0.7429931924769059,
      "learning_rate": 3.240812478279391e-05,
      "loss": 0.7206,
      "num_tokens": 3391236126.0,
      "step": 1110
    },
    {
      "epoch": 1.6272211027660743,
      "grad_norm": 0.5988607820965608,
      "learning_rate": 3.239418996648684e-05,
      "loss": 0.7397,
      "num_tokens": 3394493367.0,
      "step": 1111
    },
    {
      "epoch": 1.6286865726323503,
      "grad_norm": 0.7831629603886402,
      "learning_rate": 3.2380245798067614e-05,
      "loss": 0.7293,
      "num_tokens": 3397358120.0,
      "step": 1112
    },
    {
      "epoch": 1.6301520424986262,
      "grad_norm": 0.5698472888619368,
      "learning_rate": 3.236629229008241e-05,
      "loss": 0.7409,
      "num_tokens": 3400446904.0,
      "step": 1113
    },
    {
      "epoch": 1.631617512364902,
      "grad_norm": 0.7832338127587768,
      "learning_rate": 3.235232945508584e-05,
      "loss": 0.7277,
      "num_tokens": 3403780880.0,
      "step": 1114
    },
    {
      "epoch": 1.6330829822311779,
      "grad_norm": 0.5965913553926194,
      "learning_rate": 3.233835730564085e-05,
      "loss": 0.7287,
      "num_tokens": 3406895068.0,
      "step": 1115
    },
    {
      "epoch": 1.6345484520974538,
      "grad_norm": 0.8305549488545156,
      "learning_rate": 3.232437585431883e-05,
      "loss": 0.7454,
      "num_tokens": 3409902342.0,
      "step": 1116
    },
    {
      "epoch": 1.6360139219637295,
      "grad_norm": 0.6756638577677493,
      "learning_rate": 3.2310385113699496e-05,
      "loss": 0.7462,
      "num_tokens": 3412854131.0,
      "step": 1117
    },
    {
      "epoch": 1.6374793918300055,
      "grad_norm": 0.644462533790811,
      "learning_rate": 3.229638509637094e-05,
      "loss": 0.74,
      "num_tokens": 3415771153.0,
      "step": 1118
    },
    {
      "epoch": 1.6389448616962814,
      "grad_norm": 0.6646110036363029,
      "learning_rate": 3.228237581492959e-05,
      "loss": 0.7351,
      "num_tokens": 3418796312.0,
      "step": 1119
    },
    {
      "epoch": 1.6404103315625571,
      "grad_norm": 0.6225027121162595,
      "learning_rate": 3.226835728198023e-05,
      "loss": 0.7132,
      "num_tokens": 3421893847.0,
      "step": 1120
    },
    {
      "epoch": 1.6418758014288333,
      "grad_norm": 0.5222282192448345,
      "learning_rate": 3.225432951013593e-05,
      "loss": 0.7513,
      "num_tokens": 3424804513.0,
      "step": 1121
    },
    {
      "epoch": 1.643341271295109,
      "grad_norm": 0.6087105812141241,
      "learning_rate": 3.224029251201812e-05,
      "loss": 0.7275,
      "num_tokens": 3427716445.0,
      "step": 1122
    },
    {
      "epoch": 1.644806741161385,
      "grad_norm": 0.49733717478903516,
      "learning_rate": 3.2226246300256496e-05,
      "loss": 0.7306,
      "num_tokens": 3430705202.0,
      "step": 1123
    },
    {
      "epoch": 1.6462722110276609,
      "grad_norm": 0.6844653594208825,
      "learning_rate": 3.221219088748906e-05,
      "loss": 0.7158,
      "num_tokens": 3433730688.0,
      "step": 1124
    },
    {
      "epoch": 1.6477376808939366,
      "grad_norm": 0.54577534675777,
      "learning_rate": 3.219812628636208e-05,
      "loss": 0.7293,
      "num_tokens": 3436920229.0,
      "step": 1125
    },
    {
      "epoch": 1.6492031507602125,
      "grad_norm": 0.6712674045339279,
      "learning_rate": 3.2184052509530106e-05,
      "loss": 0.7122,
      "num_tokens": 3440037524.0,
      "step": 1126
    },
    {
      "epoch": 1.6506686206264884,
      "grad_norm": 0.5592617985393468,
      "learning_rate": 3.216996956965594e-05,
      "loss": 0.7292,
      "num_tokens": 3443366594.0,
      "step": 1127
    },
    {
      "epoch": 1.6521340904927642,
      "grad_norm": 0.6533970459944575,
      "learning_rate": 3.215587747941062e-05,
      "loss": 0.7368,
      "num_tokens": 3446537867.0,
      "step": 1128
    },
    {
      "epoch": 1.65359956035904,
      "grad_norm": 0.6034060954180704,
      "learning_rate": 3.214177625147342e-05,
      "loss": 0.7272,
      "num_tokens": 3449643899.0,
      "step": 1129
    },
    {
      "epoch": 1.655065030225316,
      "grad_norm": 0.5900912401335233,
      "learning_rate": 3.212766589853185e-05,
      "loss": 0.7237,
      "num_tokens": 3452601463.0,
      "step": 1130
    },
    {
      "epoch": 1.6565305000915918,
      "grad_norm": 0.5342510300640673,
      "learning_rate": 3.21135464332816e-05,
      "loss": 0.7171,
      "num_tokens": 3455723692.0,
      "step": 1131
    },
    {
      "epoch": 1.6579959699578677,
      "grad_norm": 0.5643493411202638,
      "learning_rate": 3.20994178684266e-05,
      "loss": 0.7256,
      "num_tokens": 3458622862.0,
      "step": 1132
    },
    {
      "epoch": 1.6594614398241436,
      "grad_norm": 0.5399316844760527,
      "learning_rate": 3.2085280216678923e-05,
      "loss": 0.7441,
      "num_tokens": 3461719654.0,
      "step": 1133
    },
    {
      "epoch": 1.6609269096904193,
      "grad_norm": 0.4921045738972708,
      "learning_rate": 3.207113349075885e-05,
      "loss": 0.716,
      "num_tokens": 3464702901.0,
      "step": 1134
    },
    {
      "epoch": 1.6623923795566955,
      "grad_norm": 0.4523643759793127,
      "learning_rate": 3.205697770339481e-05,
      "loss": 0.7316,
      "num_tokens": 3467766734.0,
      "step": 1135
    },
    {
      "epoch": 1.6638578494229712,
      "grad_norm": 0.6263062140338715,
      "learning_rate": 3.204281286732341e-05,
      "loss": 0.752,
      "num_tokens": 3470636718.0,
      "step": 1136
    },
    {
      "epoch": 1.6653233192892472,
      "grad_norm": 0.513645243389043,
      "learning_rate": 3.2028638995289345e-05,
      "loss": 0.7372,
      "num_tokens": 3473675088.0,
      "step": 1137
    },
    {
      "epoch": 1.666788789155523,
      "grad_norm": 0.588528659401566,
      "learning_rate": 3.2014456100045504e-05,
      "loss": 0.7195,
      "num_tokens": 3476699377.0,
      "step": 1138
    },
    {
      "epoch": 1.6682542590217988,
      "grad_norm": 0.5390585146923718,
      "learning_rate": 3.200026419435284e-05,
      "loss": 0.7398,
      "num_tokens": 3479592997.0,
      "step": 1139
    },
    {
      "epoch": 1.6697197288880747,
      "grad_norm": 0.5755487365152632,
      "learning_rate": 3.1986063290980457e-05,
      "loss": 0.7245,
      "num_tokens": 3482691964.0,
      "step": 1140
    },
    {
      "epoch": 1.6711851987543507,
      "grad_norm": 0.5244672541282115,
      "learning_rate": 3.1971853402705534e-05,
      "loss": 0.7271,
      "num_tokens": 3485731436.0,
      "step": 1141
    },
    {
      "epoch": 1.6726506686206264,
      "grad_norm": 0.6215608684517648,
      "learning_rate": 3.195763454231333e-05,
      "loss": 0.7117,
      "num_tokens": 3488714522.0,
      "step": 1142
    },
    {
      "epoch": 1.6741161384869023,
      "grad_norm": 0.5050276928827526,
      "learning_rate": 3.1943406722597186e-05,
      "loss": 0.7289,
      "num_tokens": 3491785471.0,
      "step": 1143
    },
    {
      "epoch": 1.6755816083531783,
      "grad_norm": 0.6954141112996651,
      "learning_rate": 3.192916995635849e-05,
      "loss": 0.7233,
      "num_tokens": 3494749318.0,
      "step": 1144
    },
    {
      "epoch": 1.677047078219454,
      "grad_norm": 0.5471953535700242,
      "learning_rate": 3.1914924256406706e-05,
      "loss": 0.7293,
      "num_tokens": 3497798166.0,
      "step": 1145
    },
    {
      "epoch": 1.6785125480857301,
      "grad_norm": 0.7340035788949208,
      "learning_rate": 3.1900669635559306e-05,
      "loss": 0.7105,
      "num_tokens": 3501072425.0,
      "step": 1146
    },
    {
      "epoch": 1.6799780179520059,
      "grad_norm": 0.6500835901284955,
      "learning_rate": 3.188640610664181e-05,
      "loss": 0.7288,
      "num_tokens": 3504045034.0,
      "step": 1147
    },
    {
      "epoch": 1.6814434878182818,
      "grad_norm": 0.5211505039482257,
      "learning_rate": 3.187213368248774e-05,
      "loss": 0.7361,
      "num_tokens": 3507246803.0,
      "step": 1148
    },
    {
      "epoch": 1.6829089576845577,
      "grad_norm": 0.7598502157506023,
      "learning_rate": 3.185785237593863e-05,
      "loss": 0.7378,
      "num_tokens": 3510298892.0,
      "step": 1149
    },
    {
      "epoch": 1.6843744275508334,
      "grad_norm": 0.4629403012605498,
      "learning_rate": 3.184356219984401e-05,
      "loss": 0.7326,
      "num_tokens": 3513347419.0,
      "step": 1150
    },
    {
      "epoch": 1.6858398974171094,
      "grad_norm": 0.7595370205256337,
      "learning_rate": 3.182926316706137e-05,
      "loss": 0.7475,
      "num_tokens": 3516145116.0,
      "step": 1151
    },
    {
      "epoch": 1.6873053672833853,
      "grad_norm": 0.6497827709852049,
      "learning_rate": 3.181495529045619e-05,
      "loss": 0.7423,
      "num_tokens": 3519224426.0,
      "step": 1152
    },
    {
      "epoch": 1.688770837149661,
      "grad_norm": 0.6208031171457175,
      "learning_rate": 3.18006385829019e-05,
      "loss": 0.7176,
      "num_tokens": 3522498714.0,
      "step": 1153
    },
    {
      "epoch": 1.690236307015937,
      "grad_norm": 0.5811303817958572,
      "learning_rate": 3.1786313057279864e-05,
      "loss": 0.7255,
      "num_tokens": 3525681709.0,
      "step": 1154
    },
    {
      "epoch": 1.691701776882213,
      "grad_norm": 0.49336714457331404,
      "learning_rate": 3.17719787264794e-05,
      "loss": 0.7217,
      "num_tokens": 3528737216.0,
      "step": 1155
    },
    {
      "epoch": 1.6931672467484886,
      "grad_norm": 0.5836098369806012,
      "learning_rate": 3.1757635603397735e-05,
      "loss": 0.7269,
      "num_tokens": 3531866415.0,
      "step": 1156
    },
    {
      "epoch": 1.6946327166147646,
      "grad_norm": 0.4697816778655563,
      "learning_rate": 3.174328370094002e-05,
      "loss": 0.7194,
      "num_tokens": 3534927930.0,
      "step": 1157
    },
    {
      "epoch": 1.6960981864810405,
      "grad_norm": 0.5838790882495166,
      "learning_rate": 3.1728923032019285e-05,
      "loss": 0.7397,
      "num_tokens": 3538129014.0,
      "step": 1158
    },
    {
      "epoch": 1.6975636563473162,
      "grad_norm": 0.5090830826740874,
      "learning_rate": 3.171455360955647e-05,
      "loss": 0.7302,
      "num_tokens": 3541246046.0,
      "step": 1159
    },
    {
      "epoch": 1.6990291262135924,
      "grad_norm": 0.6422400724318863,
      "learning_rate": 3.170017544648036e-05,
      "loss": 0.7278,
      "num_tokens": 3544645074.0,
      "step": 1160
    },
    {
      "epoch": 1.700494596079868,
      "grad_norm": 0.5383132944282785,
      "learning_rate": 3.1685788555727645e-05,
      "loss": 0.7397,
      "num_tokens": 3547839346.0,
      "step": 1161
    },
    {
      "epoch": 1.701960065946144,
      "grad_norm": 0.5343123575795636,
      "learning_rate": 3.1671392950242836e-05,
      "loss": 0.7354,
      "num_tokens": 3550811126.0,
      "step": 1162
    },
    {
      "epoch": 1.70342553581242,
      "grad_norm": 0.5795373498690242,
      "learning_rate": 3.165698864297829e-05,
      "loss": 0.7222,
      "num_tokens": 3553816084.0,
      "step": 1163
    },
    {
      "epoch": 1.7048910056786957,
      "grad_norm": 0.4559744537160204,
      "learning_rate": 3.164257564689421e-05,
      "loss": 0.7469,
      "num_tokens": 3556765666.0,
      "step": 1164
    },
    {
      "epoch": 1.7063564755449716,
      "grad_norm": 0.6471463380792962,
      "learning_rate": 3.162815397495861e-05,
      "loss": 0.7121,
      "num_tokens": 3559821001.0,
      "step": 1165
    },
    {
      "epoch": 1.7078219454112475,
      "grad_norm": 0.4850845905244907,
      "learning_rate": 3.1613723640147286e-05,
      "loss": 0.7298,
      "num_tokens": 3562873173.0,
      "step": 1166
    },
    {
      "epoch": 1.7092874152775233,
      "grad_norm": 0.6188422567781574,
      "learning_rate": 3.159928465544387e-05,
      "loss": 0.7242,
      "num_tokens": 3565849863.0,
      "step": 1167
    },
    {
      "epoch": 1.7107528851437992,
      "grad_norm": 0.4840749384988767,
      "learning_rate": 3.1584837033839736e-05,
      "loss": 0.7306,
      "num_tokens": 3568947848.0,
      "step": 1168
    },
    {
      "epoch": 1.7122183550100751,
      "grad_norm": 0.5596761016201692,
      "learning_rate": 3.1570380788334056e-05,
      "loss": 0.7105,
      "num_tokens": 3571882911.0,
      "step": 1169
    },
    {
      "epoch": 1.7136838248763508,
      "grad_norm": 0.4918055873183456,
      "learning_rate": 3.155591593193375e-05,
      "loss": 0.7288,
      "num_tokens": 3574990662.0,
      "step": 1170
    },
    {
      "epoch": 1.715149294742627,
      "grad_norm": 0.592534000360727,
      "learning_rate": 3.1541442477653504e-05,
      "loss": 0.7232,
      "num_tokens": 3577884351.0,
      "step": 1171
    },
    {
      "epoch": 1.7166147646089027,
      "grad_norm": 0.5114371719476067,
      "learning_rate": 3.15269604385157e-05,
      "loss": 0.7264,
      "num_tokens": 3580907849.0,
      "step": 1172
    },
    {
      "epoch": 1.7180802344751787,
      "grad_norm": 0.469456909143289,
      "learning_rate": 3.151246982755049e-05,
      "loss": 0.7071,
      "num_tokens": 3583984770.0,
      "step": 1173
    },
    {
      "epoch": 1.7195457043414546,
      "grad_norm": 0.4488381216659133,
      "learning_rate": 3.1497970657795704e-05,
      "loss": 0.7119,
      "num_tokens": 3587035519.0,
      "step": 1174
    },
    {
      "epoch": 1.7210111742077303,
      "grad_norm": 0.4706246660214865,
      "learning_rate": 3.1483462942296887e-05,
      "loss": 0.7301,
      "num_tokens": 3590095217.0,
      "step": 1175
    },
    {
      "epoch": 1.7224766440740062,
      "grad_norm": 0.48020071861280217,
      "learning_rate": 3.146894669410728e-05,
      "loss": 0.7354,
      "num_tokens": 3593060397.0,
      "step": 1176
    },
    {
      "epoch": 1.7239421139402822,
      "grad_norm": 0.4285191601994742,
      "learning_rate": 3.145442192628779e-05,
      "loss": 0.7279,
      "num_tokens": 3596249174.0,
      "step": 1177
    },
    {
      "epoch": 1.725407583806558,
      "grad_norm": 0.49624752039054004,
      "learning_rate": 3.143988865190699e-05,
      "loss": 0.7339,
      "num_tokens": 3599176168.0,
      "step": 1178
    },
    {
      "epoch": 1.7268730536728338,
      "grad_norm": 0.462328638319801,
      "learning_rate": 3.14253468840411e-05,
      "loss": 0.7211,
      "num_tokens": 3602206679.0,
      "step": 1179
    },
    {
      "epoch": 1.7283385235391098,
      "grad_norm": 0.41562889584227203,
      "learning_rate": 3.141079663577401e-05,
      "loss": 0.7066,
      "num_tokens": 3605320795.0,
      "step": 1180
    },
    {
      "epoch": 1.7298039934053855,
      "grad_norm": 0.527025598144562,
      "learning_rate": 3.1396237920197225e-05,
      "loss": 0.726,
      "num_tokens": 3608412928.0,
      "step": 1181
    },
    {
      "epoch": 1.7312694632716614,
      "grad_norm": 0.4446874218780101,
      "learning_rate": 3.1381670750409835e-05,
      "loss": 0.7399,
      "num_tokens": 3611443429.0,
      "step": 1182
    },
    {
      "epoch": 1.7327349331379374,
      "grad_norm": 0.5290020254165947,
      "learning_rate": 3.136709513951859e-05,
      "loss": 0.7404,
      "num_tokens": 3614404701.0,
      "step": 1183
    },
    {
      "epoch": 1.734200403004213,
      "grad_norm": 0.44132305155312357,
      "learning_rate": 3.1352511100637806e-05,
      "loss": 0.7327,
      "num_tokens": 3617466277.0,
      "step": 1184
    },
    {
      "epoch": 1.7356658728704892,
      "grad_norm": 0.4040477249423107,
      "learning_rate": 3.133791864688939e-05,
      "loss": 0.7279,
      "num_tokens": 3620521979.0,
      "step": 1185
    },
    {
      "epoch": 1.737131342736765,
      "grad_norm": 0.48742269882355593,
      "learning_rate": 3.132331779140281e-05,
      "loss": 0.7348,
      "num_tokens": 3623620004.0,
      "step": 1186
    },
    {
      "epoch": 1.7385968126030409,
      "grad_norm": 0.5296388508795381,
      "learning_rate": 3.130870854731511e-05,
      "loss": 0.7213,
      "num_tokens": 3626619270.0,
      "step": 1187
    },
    {
      "epoch": 1.7400622824693168,
      "grad_norm": 0.4144302858314123,
      "learning_rate": 3.129409092777087e-05,
      "loss": 0.7104,
      "num_tokens": 3629694777.0,
      "step": 1188
    },
    {
      "epoch": 1.7415277523355925,
      "grad_norm": 0.4799981364362493,
      "learning_rate": 3.1279464945922207e-05,
      "loss": 0.738,
      "num_tokens": 3632549218.0,
      "step": 1189
    },
    {
      "epoch": 1.7429932222018685,
      "grad_norm": 0.49866993256609593,
      "learning_rate": 3.126483061492876e-05,
      "loss": 0.7315,
      "num_tokens": 3635710550.0,
      "step": 1190
    },
    {
      "epoch": 1.7444586920681444,
      "grad_norm": 0.46700647493444075,
      "learning_rate": 3.125018794795769e-05,
      "loss": 0.737,
      "num_tokens": 3638794343.0,
      "step": 1191
    },
    {
      "epoch": 1.7459241619344201,
      "grad_norm": 0.5935741460610882,
      "learning_rate": 3.123553695818364e-05,
      "loss": 0.7356,
      "num_tokens": 3641788783.0,
      "step": 1192
    },
    {
      "epoch": 1.747389631800696,
      "grad_norm": 0.46757382601514824,
      "learning_rate": 3.1220877658788774e-05,
      "loss": 0.7267,
      "num_tokens": 3644594037.0,
      "step": 1193
    },
    {
      "epoch": 1.748855101666972,
      "grad_norm": 0.5719204535864542,
      "learning_rate": 3.120621006296269e-05,
      "loss": 0.7141,
      "num_tokens": 3647606571.0,
      "step": 1194
    },
    {
      "epoch": 1.7503205715332477,
      "grad_norm": 0.5712083735969787,
      "learning_rate": 3.119153418390249e-05,
      "loss": 0.7172,
      "num_tokens": 3650837850.0,
      "step": 1195
    },
    {
      "epoch": 1.7517860413995239,
      "grad_norm": 0.43977234853142266,
      "learning_rate": 3.1176850034812694e-05,
      "loss": 0.7488,
      "num_tokens": 3653885362.0,
      "step": 1196
    },
    {
      "epoch": 1.7532515112657996,
      "grad_norm": 0.6888557285643923,
      "learning_rate": 3.116215762890529e-05,
      "loss": 0.7265,
      "num_tokens": 3657005632.0,
      "step": 1197
    },
    {
      "epoch": 1.7547169811320755,
      "grad_norm": 0.4555981573824444,
      "learning_rate": 3.1147456979399686e-05,
      "loss": 0.7402,
      "num_tokens": 3660038114.0,
      "step": 1198
    },
    {
      "epoch": 1.7561824509983515,
      "grad_norm": 0.9133133219116264,
      "learning_rate": 3.113274809952271e-05,
      "loss": 0.7235,
      "num_tokens": 3663102093.0,
      "step": 1199
    },
    {
      "epoch": 1.7576479208646272,
      "grad_norm": 0.6001701307208664,
      "learning_rate": 3.111803100250858e-05,
      "loss": 0.7457,
      "num_tokens": 3666024379.0,
      "step": 1200
    },
    {
      "epoch": 1.7591133907309031,
      "grad_norm": 0.8691813575113173,
      "learning_rate": 3.110330570159892e-05,
      "loss": 0.7214,
      "num_tokens": 3669185912.0,
      "step": 1201
    },
    {
      "epoch": 1.760578860597179,
      "grad_norm": 0.7944104008325189,
      "learning_rate": 3.108857221004275e-05,
      "loss": 0.7133,
      "num_tokens": 3672169276.0,
      "step": 1202
    },
    {
      "epoch": 1.7620443304634548,
      "grad_norm": 0.7338254029358421,
      "learning_rate": 3.107383054109642e-05,
      "loss": 0.7189,
      "num_tokens": 3675174314.0,
      "step": 1203
    },
    {
      "epoch": 1.7635098003297307,
      "grad_norm": 0.7773729042515446,
      "learning_rate": 3.1059080708023674e-05,
      "loss": 0.7178,
      "num_tokens": 3678366344.0,
      "step": 1204
    },
    {
      "epoch": 1.7649752701960066,
      "grad_norm": 0.6663116985371427,
      "learning_rate": 3.104432272409559e-05,
      "loss": 0.7097,
      "num_tokens": 3681639036.0,
      "step": 1205
    },
    {
      "epoch": 1.7664407400622824,
      "grad_norm": 0.5910589152708715,
      "learning_rate": 3.102955660259058e-05,
      "loss": 0.7231,
      "num_tokens": 3684517526.0,
      "step": 1206
    },
    {
      "epoch": 1.7679062099285583,
      "grad_norm": 0.7038530116824047,
      "learning_rate": 3.101478235679437e-05,
      "loss": 0.7354,
      "num_tokens": 3687443110.0,
      "step": 1207
    },
    {
      "epoch": 1.7693716797948342,
      "grad_norm": 0.678595011949634,
      "learning_rate": 3.1e-05,
      "loss": 0.7257,
      "num_tokens": 3690614820.0,
      "step": 1208
    },
    {
      "epoch": 1.77083714966111,
      "grad_norm": 0.818009933965047,
      "learning_rate": 3.098520954550782e-05,
      "loss": 0.7275,
      "num_tokens": 3693621057.0,
      "step": 1209
    },
    {
      "epoch": 1.772302619527386,
      "grad_norm": 0.7179045363260441,
      "learning_rate": 3.0970411006625445e-05,
      "loss": 0.7043,
      "num_tokens": 3696534892.0,
      "step": 1210
    },
    {
      "epoch": 1.7737680893936618,
      "grad_norm": 0.6034110261983744,
      "learning_rate": 3.095560439666779e-05,
      "loss": 0.721,
      "num_tokens": 3699492714.0,
      "step": 1211
    },
    {
      "epoch": 1.7752335592599378,
      "grad_norm": 0.7017108899176389,
      "learning_rate": 3.094078972895699e-05,
      "loss": 0.7211,
      "num_tokens": 3702772210.0,
      "step": 1212
    },
    {
      "epoch": 1.7766990291262137,
      "grad_norm": 0.6500870497685952,
      "learning_rate": 3.0925967016822486e-05,
      "loss": 0.7374,
      "num_tokens": 3705916326.0,
      "step": 1213
    },
    {
      "epoch": 1.7781644989924894,
      "grad_norm": 0.5352470806940077,
      "learning_rate": 3.0911136273600906e-05,
      "loss": 0.7263,
      "num_tokens": 3708915957.0,
      "step": 1214
    },
    {
      "epoch": 1.7796299688587653,
      "grad_norm": 0.7690626070923308,
      "learning_rate": 3.089629751263614e-05,
      "loss": 0.719,
      "num_tokens": 3711971201.0,
      "step": 1215
    },
    {
      "epoch": 1.7810954387250413,
      "grad_norm": 0.5731324691113999,
      "learning_rate": 3.0881450747279255e-05,
      "loss": 0.728,
      "num_tokens": 3715236356.0,
      "step": 1216
    },
    {
      "epoch": 1.782560908591317,
      "grad_norm": 0.819380736545993,
      "learning_rate": 3.086659599088857e-05,
      "loss": 0.7065,
      "num_tokens": 3718341788.0,
      "step": 1217
    },
    {
      "epoch": 1.784026378457593,
      "grad_norm": 0.6858913601520322,
      "learning_rate": 3.085173325682955e-05,
      "loss": 0.7254,
      "num_tokens": 3721332499.0,
      "step": 1218
    },
    {
      "epoch": 1.7854918483238689,
      "grad_norm": 0.6445905221342578,
      "learning_rate": 3.0836862558474855e-05,
      "loss": 0.7249,
      "num_tokens": 3724273749.0,
      "step": 1219
    },
    {
      "epoch": 1.7869573181901446,
      "grad_norm": 0.574660481109089,
      "learning_rate": 3.0821983909204316e-05,
      "loss": 0.7195,
      "num_tokens": 3727425504.0,
      "step": 1220
    },
    {
      "epoch": 1.7884227880564207,
      "grad_norm": 0.75081816156552,
      "learning_rate": 3.08070973224049e-05,
      "loss": 0.7378,
      "num_tokens": 3730494884.0,
      "step": 1221
    },
    {
      "epoch": 1.7898882579226965,
      "grad_norm": 0.593475888457599,
      "learning_rate": 3.0792202811470735e-05,
      "loss": 0.7056,
      "num_tokens": 3733504552.0,
      "step": 1222
    },
    {
      "epoch": 1.7913537277889724,
      "grad_norm": 0.6755419208235262,
      "learning_rate": 3.077730038980307e-05,
      "loss": 0.7262,
      "num_tokens": 3736659452.0,
      "step": 1223
    },
    {
      "epoch": 1.7928191976552483,
      "grad_norm": 0.5849937750893893,
      "learning_rate": 3.0762390070810275e-05,
      "loss": 0.7366,
      "num_tokens": 3739684642.0,
      "step": 1224
    },
    {
      "epoch": 1.794284667521524,
      "grad_norm": 0.674225850230527,
      "learning_rate": 3.074747186790782e-05,
      "loss": 0.7394,
      "num_tokens": 3742597202.0,
      "step": 1225
    },
    {
      "epoch": 1.7957501373878,
      "grad_norm": 0.5711571637211587,
      "learning_rate": 3.0732545794518274e-05,
      "loss": 0.7314,
      "num_tokens": 3745617744.0,
      "step": 1226
    },
    {
      "epoch": 1.797215607254076,
      "grad_norm": 0.717277548086943,
      "learning_rate": 3.071761186407128e-05,
      "loss": 0.7375,
      "num_tokens": 3748811784.0,
      "step": 1227
    },
    {
      "epoch": 1.7986810771203516,
      "grad_norm": 0.5684042177698139,
      "learning_rate": 3.070267009000355e-05,
      "loss": 0.7162,
      "num_tokens": 3751778385.0,
      "step": 1228
    },
    {
      "epoch": 1.8001465469866276,
      "grad_norm": 0.6747659605905263,
      "learning_rate": 3.068772048575888e-05,
      "loss": 0.7134,
      "num_tokens": 3754704324.0,
      "step": 1229
    },
    {
      "epoch": 1.8016120168529035,
      "grad_norm": 0.6402452989130403,
      "learning_rate": 3.0672763064788074e-05,
      "loss": 0.7144,
      "num_tokens": 3757710461.0,
      "step": 1230
    },
    {
      "epoch": 1.8030774867191792,
      "grad_norm": 0.6290810914966188,
      "learning_rate": 3.065779784054898e-05,
      "loss": 0.7162,
      "num_tokens": 3760634495.0,
      "step": 1231
    },
    {
      "epoch": 1.8045429565854552,
      "grad_norm": 0.6439534028747723,
      "learning_rate": 3.064282482650648e-05,
      "loss": 0.7142,
      "num_tokens": 3763783677.0,
      "step": 1232
    },
    {
      "epoch": 1.806008426451731,
      "grad_norm": 0.5635779311708001,
      "learning_rate": 3.0627844036132466e-05,
      "loss": 0.7147,
      "num_tokens": 3766845635.0,
      "step": 1233
    },
    {
      "epoch": 1.8074738963180068,
      "grad_norm": 0.5929714627987843,
      "learning_rate": 3.06128554829058e-05,
      "loss": 0.7123,
      "num_tokens": 3770041904.0,
      "step": 1234
    },
    {
      "epoch": 1.808939366184283,
      "grad_norm": 0.5261616168233537,
      "learning_rate": 3.0597859180312344e-05,
      "loss": 0.7011,
      "num_tokens": 3773108762.0,
      "step": 1235
    },
    {
      "epoch": 1.8104048360505587,
      "grad_norm": 0.5032184727828983,
      "learning_rate": 3.0582855141844956e-05,
      "loss": 0.7171,
      "num_tokens": 3776214952.0,
      "step": 1236
    },
    {
      "epoch": 1.8118703059168346,
      "grad_norm": 0.6350475983716228,
      "learning_rate": 3.0567843381003416e-05,
      "loss": 0.7341,
      "num_tokens": 3779222343.0,
      "step": 1237
    },
    {
      "epoch": 1.8133357757831106,
      "grad_norm": 0.4474772270239477,
      "learning_rate": 3.055282391129446e-05,
      "loss": 0.7257,
      "num_tokens": 3782649879.0,
      "step": 1238
    },
    {
      "epoch": 1.8148012456493863,
      "grad_norm": 0.7076818118800702,
      "learning_rate": 3.0537796746231805e-05,
      "loss": 0.7336,
      "num_tokens": 3785569586.0,
      "step": 1239
    },
    {
      "epoch": 1.8162667155156622,
      "grad_norm": 0.6107843827660692,
      "learning_rate": 3.052276189933602e-05,
      "loss": 0.725,
      "num_tokens": 3788492150.0,
      "step": 1240
    },
    {
      "epoch": 1.8177321853819381,
      "grad_norm": 0.5603325806919623,
      "learning_rate": 3.050771938413463e-05,
      "loss": 0.7112,
      "num_tokens": 3791434690.0,
      "step": 1241
    },
    {
      "epoch": 1.8191976552482139,
      "grad_norm": 0.5556538930141467,
      "learning_rate": 3.049266921416205e-05,
      "loss": 0.7163,
      "num_tokens": 3794494914.0,
      "step": 1242
    },
    {
      "epoch": 1.8206631251144898,
      "grad_norm": 0.5650521624336602,
      "learning_rate": 3.0477611402959605e-05,
      "loss": 0.7281,
      "num_tokens": 3797463439.0,
      "step": 1243
    },
    {
      "epoch": 1.8221285949807657,
      "grad_norm": 0.5191335525390897,
      "learning_rate": 3.0462545964075433e-05,
      "loss": 0.7163,
      "num_tokens": 3800742071.0,
      "step": 1244
    },
    {
      "epoch": 1.8235940648470415,
      "grad_norm": 0.5726209233287751,
      "learning_rate": 3.0447472911064616e-05,
      "loss": 0.7384,
      "num_tokens": 3803769567.0,
      "step": 1245
    },
    {
      "epoch": 1.8250595347133176,
      "grad_norm": 0.4733174092666106,
      "learning_rate": 3.043239225748902e-05,
      "loss": 0.7347,
      "num_tokens": 3806715261.0,
      "step": 1246
    },
    {
      "epoch": 1.8265250045795933,
      "grad_norm": 0.6097679977025574,
      "learning_rate": 3.041730401691738e-05,
      "loss": 0.7359,
      "num_tokens": 3809710035.0,
      "step": 1247
    },
    {
      "epoch": 1.8279904744458693,
      "grad_norm": 0.4392833749665127,
      "learning_rate": 3.040220820292526e-05,
      "loss": 0.7129,
      "num_tokens": 3812762860.0,
      "step": 1248
    },
    {
      "epoch": 1.8294559443121452,
      "grad_norm": 0.642871805059476,
      "learning_rate": 3.038710482909503e-05,
      "loss": 0.7116,
      "num_tokens": 3815808336.0,
      "step": 1249
    },
    {
      "epoch": 1.830921414178421,
      "grad_norm": 0.473544473470459,
      "learning_rate": 3.0371993909015858e-05,
      "loss": 0.7303,
      "num_tokens": 3818893372.0,
      "step": 1250
    },
    {
      "epoch": 1.8323868840446969,
      "grad_norm": 0.5539277442754945,
      "learning_rate": 3.03568754562837e-05,
      "loss": 0.7301,
      "num_tokens": 3821947824.0,
      "step": 1251
    },
    {
      "epoch": 1.8338523539109728,
      "grad_norm": 0.46592363231241957,
      "learning_rate": 3.0341749484501316e-05,
      "loss": 0.7115,
      "num_tokens": 3825092070.0,
      "step": 1252
    },
    {
      "epoch": 1.8353178237772485,
      "grad_norm": 0.6924577231114566,
      "learning_rate": 3.03266160072782e-05,
      "loss": 0.7111,
      "num_tokens": 3827937601.0,
      "step": 1253
    },
    {
      "epoch": 1.8367832936435244,
      "grad_norm": 0.42918670883616356,
      "learning_rate": 3.0311475038230616e-05,
      "loss": 0.7249,
      "num_tokens": 3831020705.0,
      "step": 1254
    },
    {
      "epoch": 1.8382487635098004,
      "grad_norm": 0.660740856727847,
      "learning_rate": 3.029632659098155e-05,
      "loss": 0.7185,
      "num_tokens": 3834170742.0,
      "step": 1255
    },
    {
      "epoch": 1.839714233376076,
      "grad_norm": 0.5799651025885605,
      "learning_rate": 3.0281170679160743e-05,
      "loss": 0.725,
      "num_tokens": 3837245824.0,
      "step": 1256
    },
    {
      "epoch": 1.841179703242352,
      "grad_norm": 0.5447004998552158,
      "learning_rate": 3.0266007316404635e-05,
      "loss": 0.7119,
      "num_tokens": 3840314537.0,
      "step": 1257
    },
    {
      "epoch": 1.842645173108628,
      "grad_norm": 0.4921625001388072,
      "learning_rate": 3.025083651635636e-05,
      "loss": 0.7219,
      "num_tokens": 3843225950.0,
      "step": 1258
    },
    {
      "epoch": 1.8441106429749037,
      "grad_norm": 0.5500610059160009,
      "learning_rate": 3.0235658292665778e-05,
      "loss": 0.7111,
      "num_tokens": 3846310344.0,
      "step": 1259
    },
    {
      "epoch": 1.8455761128411798,
      "grad_norm": 0.4844317779243711,
      "learning_rate": 3.0220472658989407e-05,
      "loss": 0.7269,
      "num_tokens": 3849205778.0,
      "step": 1260
    },
    {
      "epoch": 1.8470415827074556,
      "grad_norm": 0.5577544085898276,
      "learning_rate": 3.0205279628990417e-05,
      "loss": 0.7139,
      "num_tokens": 3852222396.0,
      "step": 1261
    },
    {
      "epoch": 1.8485070525737315,
      "grad_norm": 0.4619759057969569,
      "learning_rate": 3.0190079216338656e-05,
      "loss": 0.7272,
      "num_tokens": 3855186567.0,
      "step": 1262
    },
    {
      "epoch": 1.8499725224400074,
      "grad_norm": 0.5868929021123546,
      "learning_rate": 3.0174871434710612e-05,
      "loss": 0.7234,
      "num_tokens": 3858432501.0,
      "step": 1263
    },
    {
      "epoch": 1.8514379923062831,
      "grad_norm": 0.5396731238726628,
      "learning_rate": 3.0159656297789396e-05,
      "loss": 0.7214,
      "num_tokens": 3861627148.0,
      "step": 1264
    },
    {
      "epoch": 1.852903462172559,
      "grad_norm": 0.5403695104336745,
      "learning_rate": 3.014443381926473e-05,
      "loss": 0.7019,
      "num_tokens": 3864632607.0,
      "step": 1265
    },
    {
      "epoch": 1.854368932038835,
      "grad_norm": 0.45004985065149816,
      "learning_rate": 3.0129204012832963e-05,
      "loss": 0.7052,
      "num_tokens": 3867717730.0,
      "step": 1266
    },
    {
      "epoch": 1.8558344019051107,
      "grad_norm": 0.6075585215305394,
      "learning_rate": 3.0113966892197026e-05,
      "loss": 0.7146,
      "num_tokens": 3870666202.0,
      "step": 1267
    },
    {
      "epoch": 1.8572998717713867,
      "grad_norm": 0.4328967180693791,
      "learning_rate": 3.0098722471066417e-05,
      "loss": 0.7001,
      "num_tokens": 3873591993.0,
      "step": 1268
    },
    {
      "epoch": 1.8587653416376626,
      "grad_norm": 0.539706477141303,
      "learning_rate": 3.008347076315722e-05,
      "loss": 0.7402,
      "num_tokens": 3876666875.0,
      "step": 1269
    },
    {
      "epoch": 1.8602308115039383,
      "grad_norm": 0.5367717342516909,
      "learning_rate": 3.0068211782192085e-05,
      "loss": 0.7028,
      "num_tokens": 3879833772.0,
      "step": 1270
    },
    {
      "epoch": 1.8616962813702145,
      "grad_norm": 0.4826607062342516,
      "learning_rate": 3.0052945541900175e-05,
      "loss": 0.7259,
      "num_tokens": 3882760931.0,
      "step": 1271
    },
    {
      "epoch": 1.8631617512364902,
      "grad_norm": 0.40969050374044647,
      "learning_rate": 3.00376720560172e-05,
      "loss": 0.7244,
      "num_tokens": 3885764262.0,
      "step": 1272
    },
    {
      "epoch": 1.864627221102766,
      "grad_norm": 0.691735874987434,
      "learning_rate": 3.0022391338285407e-05,
      "loss": 0.7328,
      "num_tokens": 3888743117.0,
      "step": 1273
    },
    {
      "epoch": 1.866092690969042,
      "grad_norm": 0.4928958006199086,
      "learning_rate": 3.000710340245352e-05,
      "loss": 0.7284,
      "num_tokens": 3891784760.0,
      "step": 1274
    },
    {
      "epoch": 1.8675581608353178,
      "grad_norm": 0.6032744021262607,
      "learning_rate": 2.9991808262276774e-05,
      "loss": 0.7248,
      "num_tokens": 3894970619.0,
      "step": 1275
    },
    {
      "epoch": 1.8690236307015937,
      "grad_norm": 0.6037351309961003,
      "learning_rate": 2.9976505931516878e-05,
      "loss": 0.7209,
      "num_tokens": 3897981649.0,
      "step": 1276
    },
    {
      "epoch": 1.8704891005678697,
      "grad_norm": 0.44614918124326197,
      "learning_rate": 2.9961196423942026e-05,
      "loss": 0.725,
      "num_tokens": 3901342269.0,
      "step": 1277
    },
    {
      "epoch": 1.8719545704341454,
      "grad_norm": 0.5684075035685395,
      "learning_rate": 2.9945879753326848e-05,
      "loss": 0.7283,
      "num_tokens": 3904568828.0,
      "step": 1278
    },
    {
      "epoch": 1.8734200403004213,
      "grad_norm": 0.434531047628026,
      "learning_rate": 2.9930555933452437e-05,
      "loss": 0.7067,
      "num_tokens": 3907811666.0,
      "step": 1279
    },
    {
      "epoch": 1.8748855101666972,
      "grad_norm": 0.500522785420292,
      "learning_rate": 2.9915224978106305e-05,
      "loss": 0.7126,
      "num_tokens": 3910784487.0,
      "step": 1280
    },
    {
      "epoch": 1.876350980032973,
      "grad_norm": 0.499460152151047,
      "learning_rate": 2.98998869010824e-05,
      "loss": 0.7148,
      "num_tokens": 3913881783.0,
      "step": 1281
    },
    {
      "epoch": 1.877816449899249,
      "grad_norm": 0.5777470029825018,
      "learning_rate": 2.988454171618105e-05,
      "loss": 0.7317,
      "num_tokens": 3916968803.0,
      "step": 1282
    },
    {
      "epoch": 1.8792819197655248,
      "grad_norm": 0.461422046466006,
      "learning_rate": 2.9869189437209017e-05,
      "loss": 0.7105,
      "num_tokens": 3920206115.0,
      "step": 1283
    },
    {
      "epoch": 1.8807473896318005,
      "grad_norm": 0.5384301075744651,
      "learning_rate": 2.985383007797942e-05,
      "loss": 0.7176,
      "num_tokens": 3923304339.0,
      "step": 1284
    },
    {
      "epoch": 1.8822128594980767,
      "grad_norm": 0.5054358006013112,
      "learning_rate": 2.983846365231174e-05,
      "loss": 0.7317,
      "num_tokens": 3926274071.0,
      "step": 1285
    },
    {
      "epoch": 1.8836783293643524,
      "grad_norm": 0.5318289539992627,
      "learning_rate": 2.9823090174031835e-05,
      "loss": 0.7285,
      "num_tokens": 3929194088.0,
      "step": 1286
    },
    {
      "epoch": 1.8851437992306284,
      "grad_norm": 0.4657849944244456,
      "learning_rate": 2.980770965697191e-05,
      "loss": 0.7193,
      "num_tokens": 3932347458.0,
      "step": 1287
    },
    {
      "epoch": 1.8866092690969043,
      "grad_norm": 0.4252315927345098,
      "learning_rate": 2.9792322114970495e-05,
      "loss": 0.7292,
      "num_tokens": 3935622990.0,
      "step": 1288
    },
    {
      "epoch": 1.88807473896318,
      "grad_norm": 0.5361154096964137,
      "learning_rate": 2.9776927561872434e-05,
      "loss": 0.7289,
      "num_tokens": 3938474829.0,
      "step": 1289
    },
    {
      "epoch": 1.889540208829456,
      "grad_norm": 0.46064933986285345,
      "learning_rate": 2.97615260115289e-05,
      "loss": 0.7231,
      "num_tokens": 3941721777.0,
      "step": 1290
    },
    {
      "epoch": 1.8910056786957319,
      "grad_norm": 0.44213175021100964,
      "learning_rate": 2.974611747779733e-05,
      "loss": 0.7281,
      "num_tokens": 3944819865.0,
      "step": 1291
    },
    {
      "epoch": 1.8924711485620076,
      "grad_norm": 0.5852228945647604,
      "learning_rate": 2.9730701974541478e-05,
      "loss": 0.7063,
      "num_tokens": 3948020936.0,
      "step": 1292
    },
    {
      "epoch": 1.8939366184282835,
      "grad_norm": 0.4811258644337281,
      "learning_rate": 2.9715279515631346e-05,
      "loss": 0.7156,
      "num_tokens": 3950942952.0,
      "step": 1293
    },
    {
      "epoch": 1.8954020882945595,
      "grad_norm": 0.46879300962403375,
      "learning_rate": 2.9699850114943208e-05,
      "loss": 0.7258,
      "num_tokens": 3953933033.0,
      "step": 1294
    },
    {
      "epoch": 1.8968675581608352,
      "grad_norm": 0.5570458912678902,
      "learning_rate": 2.9684413786359573e-05,
      "loss": 0.7066,
      "num_tokens": 3957025431.0,
      "step": 1295
    },
    {
      "epoch": 1.8983330280271113,
      "grad_norm": 0.4719145890572937,
      "learning_rate": 2.966897054376919e-05,
      "loss": 0.7164,
      "num_tokens": 3960114186.0,
      "step": 1296
    },
    {
      "epoch": 1.899798497893387,
      "grad_norm": 0.5369992859304673,
      "learning_rate": 2.965352040106703e-05,
      "loss": 0.7142,
      "num_tokens": 3962956785.0,
      "step": 1297
    },
    {
      "epoch": 1.9012639677596628,
      "grad_norm": 0.4683906897435623,
      "learning_rate": 2.9638063372154265e-05,
      "loss": 0.7235,
      "num_tokens": 3966035864.0,
      "step": 1298
    },
    {
      "epoch": 1.902729437625939,
      "grad_norm": 0.6021529583071882,
      "learning_rate": 2.9622599470938268e-05,
      "loss": 0.7167,
      "num_tokens": 3969163154.0,
      "step": 1299
    },
    {
      "epoch": 1.9041949074922147,
      "grad_norm": 0.46301091747249523,
      "learning_rate": 2.960712871133259e-05,
      "loss": 0.734,
      "num_tokens": 3972471784.0,
      "step": 1300
    },
    {
      "epoch": 1.9056603773584906,
      "grad_norm": 0.6292393409025439,
      "learning_rate": 2.959165110725697e-05,
      "loss": 0.7346,
      "num_tokens": 3975624326.0,
      "step": 1301
    },
    {
      "epoch": 1.9071258472247665,
      "grad_norm": 0.5152026860143738,
      "learning_rate": 2.957616667263728e-05,
      "loss": 0.7251,
      "num_tokens": 3978636213.0,
      "step": 1302
    },
    {
      "epoch": 1.9085913170910422,
      "grad_norm": 0.4502329354239532,
      "learning_rate": 2.956067542140555e-05,
      "loss": 0.7059,
      "num_tokens": 3981946784.0,
      "step": 1303
    },
    {
      "epoch": 1.9100567869573182,
      "grad_norm": 0.6162301579136784,
      "learning_rate": 2.954517736749996e-05,
      "loss": 0.7125,
      "num_tokens": 3985024580.0,
      "step": 1304
    },
    {
      "epoch": 1.9115222568235941,
      "grad_norm": 0.38837315418904206,
      "learning_rate": 2.952967252486477e-05,
      "loss": 0.7345,
      "num_tokens": 3988015338.0,
      "step": 1305
    },
    {
      "epoch": 1.9129877266898698,
      "grad_norm": 0.6197994483675844,
      "learning_rate": 2.9514160907450395e-05,
      "loss": 0.7308,
      "num_tokens": 3990995778.0,
      "step": 1306
    },
    {
      "epoch": 1.9144531965561458,
      "grad_norm": 0.483359263669142,
      "learning_rate": 2.9498642529213304e-05,
      "loss": 0.7055,
      "num_tokens": 3994217664.0,
      "step": 1307
    },
    {
      "epoch": 1.9159186664224217,
      "grad_norm": 0.5708665843435615,
      "learning_rate": 2.948311740411608e-05,
      "loss": 0.7335,
      "num_tokens": 3997315978.0,
      "step": 1308
    },
    {
      "epoch": 1.9173841362886974,
      "grad_norm": 0.5612096120423079,
      "learning_rate": 2.946758554612736e-05,
      "loss": 0.7115,
      "num_tokens": 4000347385.0,
      "step": 1309
    },
    {
      "epoch": 1.9188496061549736,
      "grad_norm": 0.47635266267691834,
      "learning_rate": 2.9452046969221845e-05,
      "loss": 0.7375,
      "num_tokens": 4003579959.0,
      "step": 1310
    },
    {
      "epoch": 1.9203150760212493,
      "grad_norm": 0.5346674676818062,
      "learning_rate": 2.9436501687380275e-05,
      "loss": 0.7078,
      "num_tokens": 4006684820.0,
      "step": 1311
    },
    {
      "epoch": 1.9217805458875252,
      "grad_norm": 0.4883809172026689,
      "learning_rate": 2.9420949714589438e-05,
      "loss": 0.7359,
      "num_tokens": 4009718818.0,
      "step": 1312
    },
    {
      "epoch": 1.9232460157538012,
      "grad_norm": 0.4297821439878945,
      "learning_rate": 2.9405391064842117e-05,
      "loss": 0.7364,
      "num_tokens": 4012738565.0,
      "step": 1313
    },
    {
      "epoch": 1.9247114856200769,
      "grad_norm": 0.5453164006040473,
      "learning_rate": 2.938982575213712e-05,
      "loss": 0.7071,
      "num_tokens": 4015733806.0,
      "step": 1314
    },
    {
      "epoch": 1.9261769554863528,
      "grad_norm": 0.49294531695059146,
      "learning_rate": 2.9374253790479264e-05,
      "loss": 0.7235,
      "num_tokens": 4018608291.0,
      "step": 1315
    },
    {
      "epoch": 1.9276424253526288,
      "grad_norm": 0.4571778837208386,
      "learning_rate": 2.9358675193879308e-05,
      "loss": 0.7289,
      "num_tokens": 4021508803.0,
      "step": 1316
    },
    {
      "epoch": 1.9291078952189045,
      "grad_norm": 0.4947590917261216,
      "learning_rate": 2.9343089976354008e-05,
      "loss": 0.7101,
      "num_tokens": 4024559584.0,
      "step": 1317
    },
    {
      "epoch": 1.9305733650851804,
      "grad_norm": 0.5507749069727216,
      "learning_rate": 2.932749815192609e-05,
      "loss": 0.6955,
      "num_tokens": 4027628747.0,
      "step": 1318
    },
    {
      "epoch": 1.9320388349514563,
      "grad_norm": 0.4632987655935729,
      "learning_rate": 2.9311899734624198e-05,
      "loss": 0.7274,
      "num_tokens": 4030826107.0,
      "step": 1319
    },
    {
      "epoch": 1.933504304817732,
      "grad_norm": 0.42026574409523243,
      "learning_rate": 2.9296294738482903e-05,
      "loss": 0.7171,
      "num_tokens": 4033947216.0,
      "step": 1320
    },
    {
      "epoch": 1.9349697746840082,
      "grad_norm": 0.5969106911659524,
      "learning_rate": 2.9280683177542733e-05,
      "loss": 0.7153,
      "num_tokens": 4036948374.0,
      "step": 1321
    },
    {
      "epoch": 1.936435244550284,
      "grad_norm": 0.3905694332410494,
      "learning_rate": 2.9265065065850086e-05,
      "loss": 0.7157,
      "num_tokens": 4039931936.0,
      "step": 1322
    },
    {
      "epoch": 1.9379007144165596,
      "grad_norm": 0.6594446442147556,
      "learning_rate": 2.9249440417457275e-05,
      "loss": 0.734,
      "num_tokens": 4042862775.0,
      "step": 1323
    },
    {
      "epoch": 1.9393661842828358,
      "grad_norm": 0.4448619573632073,
      "learning_rate": 2.9233809246422473e-05,
      "loss": 0.7223,
      "num_tokens": 4046131008.0,
      "step": 1324
    },
    {
      "epoch": 1.9408316541491115,
      "grad_norm": 0.6779458850258759,
      "learning_rate": 2.921817156680975e-05,
      "loss": 0.7184,
      "num_tokens": 4049216420.0,
      "step": 1325
    },
    {
      "epoch": 1.9422971240153875,
      "grad_norm": 0.6138260466188652,
      "learning_rate": 2.920252739268902e-05,
      "loss": 0.7032,
      "num_tokens": 4052411336.0,
      "step": 1326
    },
    {
      "epoch": 1.9437625938816634,
      "grad_norm": 0.6551602752722429,
      "learning_rate": 2.9186876738136018e-05,
      "loss": 0.7067,
      "num_tokens": 4055654355.0,
      "step": 1327
    },
    {
      "epoch": 1.945228063747939,
      "grad_norm": 0.5823608819089839,
      "learning_rate": 2.9171219617232348e-05,
      "loss": 0.7273,
      "num_tokens": 4058567839.0,
      "step": 1328
    },
    {
      "epoch": 1.946693533614215,
      "grad_norm": 0.648318666923373,
      "learning_rate": 2.915555604406541e-05,
      "loss": 0.7138,
      "num_tokens": 4061806942.0,
      "step": 1329
    },
    {
      "epoch": 1.948159003480491,
      "grad_norm": 0.557535463407278,
      "learning_rate": 2.913988603272841e-05,
      "loss": 0.7208,
      "num_tokens": 4065071761.0,
      "step": 1330
    },
    {
      "epoch": 1.9496244733467667,
      "grad_norm": 0.7194298036502226,
      "learning_rate": 2.9124209597320346e-05,
      "loss": 0.7381,
      "num_tokens": 4068299657.0,
      "step": 1331
    },
    {
      "epoch": 1.9510899432130426,
      "grad_norm": 0.5426991158983889,
      "learning_rate": 2.9108526751946006e-05,
      "loss": 0.7049,
      "num_tokens": 4071178407.0,
      "step": 1332
    },
    {
      "epoch": 1.9525554130793186,
      "grad_norm": 0.5672564952488988,
      "learning_rate": 2.9092837510715942e-05,
      "loss": 0.7543,
      "num_tokens": 4073992540.0,
      "step": 1333
    },
    {
      "epoch": 1.9540208829455943,
      "grad_norm": 0.4869103358013413,
      "learning_rate": 2.907714188774645e-05,
      "loss": 0.7175,
      "num_tokens": 4077328244.0,
      "step": 1334
    },
    {
      "epoch": 1.9554863528118704,
      "grad_norm": 0.5798219299973278,
      "learning_rate": 2.906143989715958e-05,
      "loss": 0.7508,
      "num_tokens": 4080237126.0,
      "step": 1335
    },
    {
      "epoch": 1.9569518226781462,
      "grad_norm": 0.46513049345236734,
      "learning_rate": 2.904573155308311e-05,
      "loss": 0.7069,
      "num_tokens": 4083358565.0,
      "step": 1336
    },
    {
      "epoch": 1.958417292544422,
      "grad_norm": 0.6001128590636814,
      "learning_rate": 2.9030016869650528e-05,
      "loss": 0.7336,
      "num_tokens": 4086227715.0,
      "step": 1337
    },
    {
      "epoch": 1.959882762410698,
      "grad_norm": 0.501079856171508,
      "learning_rate": 2.9014295861001023e-05,
      "loss": 0.7211,
      "num_tokens": 4089368015.0,
      "step": 1338
    },
    {
      "epoch": 1.9613482322769737,
      "grad_norm": 0.5602312410901916,
      "learning_rate": 2.899856854127949e-05,
      "loss": 0.7328,
      "num_tokens": 4092377960.0,
      "step": 1339
    },
    {
      "epoch": 1.9628137021432497,
      "grad_norm": 0.5319752150802948,
      "learning_rate": 2.8982834924636493e-05,
      "loss": 0.7221,
      "num_tokens": 4095492647.0,
      "step": 1340
    },
    {
      "epoch": 1.9642791720095256,
      "grad_norm": 0.5637811114221856,
      "learning_rate": 2.896709502522826e-05,
      "loss": 0.7255,
      "num_tokens": 4098470665.0,
      "step": 1341
    },
    {
      "epoch": 1.9657446418758013,
      "grad_norm": 0.5406982296451678,
      "learning_rate": 2.8951348857216675e-05,
      "loss": 0.7206,
      "num_tokens": 4101398879.0,
      "step": 1342
    },
    {
      "epoch": 1.9672101117420773,
      "grad_norm": 0.46364715576402576,
      "learning_rate": 2.8935596434769263e-05,
      "loss": 0.7041,
      "num_tokens": 4104234289.0,
      "step": 1343
    },
    {
      "epoch": 1.9686755816083532,
      "grad_norm": 0.5310161338306886,
      "learning_rate": 2.8919837772059173e-05,
      "loss": 0.7228,
      "num_tokens": 4107085693.0,
      "step": 1344
    },
    {
      "epoch": 1.970141051474629,
      "grad_norm": 0.530957934611928,
      "learning_rate": 2.8904072883265168e-05,
      "loss": 0.7117,
      "num_tokens": 4110242445.0,
      "step": 1345
    },
    {
      "epoch": 1.971606521340905,
      "grad_norm": 0.46089098970570264,
      "learning_rate": 2.8888301782571618e-05,
      "loss": 0.7296,
      "num_tokens": 4113351588.0,
      "step": 1346
    },
    {
      "epoch": 1.9730719912071808,
      "grad_norm": 0.4685933177800599,
      "learning_rate": 2.8872524484168482e-05,
      "loss": 0.7007,
      "num_tokens": 4116453159.0,
      "step": 1347
    },
    {
      "epoch": 1.9745374610734565,
      "grad_norm": 0.48712940482189737,
      "learning_rate": 2.885674100225128e-05,
      "loss": 0.7154,
      "num_tokens": 4119643279.0,
      "step": 1348
    },
    {
      "epoch": 1.9760029309397327,
      "grad_norm": 0.520129837952956,
      "learning_rate": 2.8840951351021133e-05,
      "loss": 0.727,
      "num_tokens": 4122618822.0,
      "step": 1349
    },
    {
      "epoch": 1.9774684008060084,
      "grad_norm": 0.39178646580988485,
      "learning_rate": 2.8825155544684665e-05,
      "loss": 0.7147,
      "num_tokens": 4125666917.0,
      "step": 1350
    },
    {
      "epoch": 1.9789338706722843,
      "grad_norm": 0.5280540580743738,
      "learning_rate": 2.8809353597454077e-05,
      "loss": 0.7011,
      "num_tokens": 4128858077.0,
      "step": 1351
    },
    {
      "epoch": 1.9803993405385603,
      "grad_norm": 0.5007308996893136,
      "learning_rate": 2.879354552354706e-05,
      "loss": 0.7139,
      "num_tokens": 4131961108.0,
      "step": 1352
    },
    {
      "epoch": 1.981864810404836,
      "grad_norm": 0.5251215565216402,
      "learning_rate": 2.8777731337186857e-05,
      "loss": 0.7508,
      "num_tokens": 4134905448.0,
      "step": 1353
    },
    {
      "epoch": 1.983330280271112,
      "grad_norm": 0.5004936037630865,
      "learning_rate": 2.876191105260218e-05,
      "loss": 0.7217,
      "num_tokens": 4137900805.0,
      "step": 1354
    },
    {
      "epoch": 1.9847957501373878,
      "grad_norm": 0.48371105608864456,
      "learning_rate": 2.874608468402724e-05,
      "loss": 0.7197,
      "num_tokens": 4140788207.0,
      "step": 1355
    },
    {
      "epoch": 1.9862612200036636,
      "grad_norm": 0.4816320746438291,
      "learning_rate": 2.8730252245701727e-05,
      "loss": 0.7147,
      "num_tokens": 4143859899.0,
      "step": 1356
    },
    {
      "epoch": 1.9877266898699395,
      "grad_norm": 0.4642541158788186,
      "learning_rate": 2.871441375187077e-05,
      "loss": 0.6947,
      "num_tokens": 4146853534.0,
      "step": 1357
    },
    {
      "epoch": 1.9891921597362154,
      "grad_norm": 0.450414638623165,
      "learning_rate": 2.8698569216784983e-05,
      "loss": 0.7215,
      "num_tokens": 4150035501.0,
      "step": 1358
    },
    {
      "epoch": 1.9906576296024912,
      "grad_norm": 0.47857178483603857,
      "learning_rate": 2.868271865470037e-05,
      "loss": 0.7286,
      "num_tokens": 4153000306.0,
      "step": 1359
    },
    {
      "epoch": 1.9921230994687673,
      "grad_norm": 0.4141431086182353,
      "learning_rate": 2.8666862079878403e-05,
      "loss": 0.7129,
      "num_tokens": 4155920615.0,
      "step": 1360
    },
    {
      "epoch": 1.993588569335043,
      "grad_norm": 0.4929319033580299,
      "learning_rate": 2.8650999506585938e-05,
      "loss": 0.7247,
      "num_tokens": 4159100804.0,
      "step": 1361
    },
    {
      "epoch": 1.995054039201319,
      "grad_norm": 0.45134386745122224,
      "learning_rate": 2.8635130949095225e-05,
      "loss": 0.6878,
      "num_tokens": 4162361276.0,
      "step": 1362
    },
    {
      "epoch": 1.996519509067595,
      "grad_norm": 0.44962153353337614,
      "learning_rate": 2.8619256421683913e-05,
      "loss": 0.742,
      "num_tokens": 4165501149.0,
      "step": 1363
    },
    {
      "epoch": 1.9979849789338706,
      "grad_norm": 0.3884680333290873,
      "learning_rate": 2.860337593863502e-05,
      "loss": 0.7382,
      "num_tokens": 4168617755.0,
      "step": 1364
    },
    {
      "epoch": 1.9994504488001466,
      "grad_norm": 0.5783997759046408,
      "learning_rate": 2.858748951423691e-05,
      "loss": 0.6998,
      "num_tokens": 4171870707.0,
      "step": 1365
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.5783997759046408,
      "learning_rate": 2.8571597162783308e-05,
      "loss": 0.7226,
      "num_tokens": 4172753218.0,
      "step": 1366
    },
    {
      "epoch": 2.0014654698662757,
      "grad_norm": 0.7150832965250836,
      "learning_rate": 2.855569889857327e-05,
      "loss": 0.6991,
      "num_tokens": 4175680256.0,
      "step": 1367
    },
    {
      "epoch": 2.002930939732552,
      "grad_norm": 0.5789439210299216,
      "learning_rate": 2.853979473591115e-05,
      "loss": 0.702,
      "num_tokens": 4178589978.0,
      "step": 1368
    },
    {
      "epoch": 2.0043964095988276,
      "grad_norm": 0.5599640352542162,
      "learning_rate": 2.852388468910663e-05,
      "loss": 0.6984,
      "num_tokens": 4181835743.0,
      "step": 1369
    },
    {
      "epoch": 2.0058618794651033,
      "grad_norm": 0.498664619477045,
      "learning_rate": 2.85079687724747e-05,
      "loss": 0.705,
      "num_tokens": 4184805022.0,
      "step": 1370
    },
    {
      "epoch": 2.0073273493313795,
      "grad_norm": 0.5672060741625693,
      "learning_rate": 2.8492047000335597e-05,
      "loss": 0.7074,
      "num_tokens": 4187914648.0,
      "step": 1371
    },
    {
      "epoch": 2.008792819197655,
      "grad_norm": 0.4219184914583131,
      "learning_rate": 2.8476119387014848e-05,
      "loss": 0.7061,
      "num_tokens": 4190953717.0,
      "step": 1372
    },
    {
      "epoch": 2.0102582890639313,
      "grad_norm": 0.5301365665518536,
      "learning_rate": 2.846018594684323e-05,
      "loss": 0.6944,
      "num_tokens": 4194002720.0,
      "step": 1373
    },
    {
      "epoch": 2.011723758930207,
      "grad_norm": 0.4685033832184126,
      "learning_rate": 2.8444246694156762e-05,
      "loss": 0.6964,
      "num_tokens": 4196862452.0,
      "step": 1374
    },
    {
      "epoch": 2.0131892287964828,
      "grad_norm": 0.4569188103619466,
      "learning_rate": 2.84283016432967e-05,
      "loss": 0.7023,
      "num_tokens": 4199912036.0,
      "step": 1375
    },
    {
      "epoch": 2.014654698662759,
      "grad_norm": 0.5211571579293084,
      "learning_rate": 2.841235080860951e-05,
      "loss": 0.6997,
      "num_tokens": 4202976305.0,
      "step": 1376
    },
    {
      "epoch": 2.0161201685290346,
      "grad_norm": 0.43930487542771857,
      "learning_rate": 2.8396394204446857e-05,
      "loss": 0.6991,
      "num_tokens": 4206371547.0,
      "step": 1377
    },
    {
      "epoch": 2.0175856383953104,
      "grad_norm": 0.48306004230362376,
      "learning_rate": 2.8380431845165604e-05,
      "loss": 0.7084,
      "num_tokens": 4209439371.0,
      "step": 1378
    },
    {
      "epoch": 2.0190511082615865,
      "grad_norm": 0.41843159337181735,
      "learning_rate": 2.8364463745127794e-05,
      "loss": 0.6951,
      "num_tokens": 4212677246.0,
      "step": 1379
    },
    {
      "epoch": 2.0205165781278622,
      "grad_norm": 0.4495191244731692,
      "learning_rate": 2.8348489918700633e-05,
      "loss": 0.7029,
      "num_tokens": 4215691506.0,
      "step": 1380
    },
    {
      "epoch": 2.021982047994138,
      "grad_norm": 0.432235280660172,
      "learning_rate": 2.833251038025648e-05,
      "loss": 0.7074,
      "num_tokens": 4218741583.0,
      "step": 1381
    },
    {
      "epoch": 2.023447517860414,
      "grad_norm": 0.4259760710863493,
      "learning_rate": 2.8316525144172828e-05,
      "loss": 0.6949,
      "num_tokens": 4221794251.0,
      "step": 1382
    },
    {
      "epoch": 2.02491298772669,
      "grad_norm": 0.4622651259480561,
      "learning_rate": 2.83005342248323e-05,
      "loss": 0.6969,
      "num_tokens": 4224925634.0,
      "step": 1383
    },
    {
      "epoch": 2.0263784575929655,
      "grad_norm": 0.43563497029818465,
      "learning_rate": 2.8284537636622636e-05,
      "loss": 0.7299,
      "num_tokens": 4227833286.0,
      "step": 1384
    },
    {
      "epoch": 2.0278439274592417,
      "grad_norm": 0.4961594987745509,
      "learning_rate": 2.826853539393668e-05,
      "loss": 0.715,
      "num_tokens": 4230774701.0,
      "step": 1385
    },
    {
      "epoch": 2.0293093973255174,
      "grad_norm": 0.41209692322352337,
      "learning_rate": 2.8252527511172337e-05,
      "loss": 0.7219,
      "num_tokens": 4234000863.0,
      "step": 1386
    },
    {
      "epoch": 2.0307748671917936,
      "grad_norm": 0.41494590263460024,
      "learning_rate": 2.8236514002732628e-05,
      "loss": 0.7125,
      "num_tokens": 4236910019.0,
      "step": 1387
    },
    {
      "epoch": 2.0322403370580693,
      "grad_norm": 0.3950802833849933,
      "learning_rate": 2.8220494883025605e-05,
      "loss": 0.7042,
      "num_tokens": 4239994217.0,
      "step": 1388
    },
    {
      "epoch": 2.033705806924345,
      "grad_norm": 0.46091084797710363,
      "learning_rate": 2.820447016646438e-05,
      "loss": 0.7096,
      "num_tokens": 4243246290.0,
      "step": 1389
    },
    {
      "epoch": 2.035171276790621,
      "grad_norm": 0.4069316813675992,
      "learning_rate": 2.8188439867467098e-05,
      "loss": 0.7101,
      "num_tokens": 4246555547.0,
      "step": 1390
    },
    {
      "epoch": 2.036636746656897,
      "grad_norm": 0.46373275034903044,
      "learning_rate": 2.8172404000456923e-05,
      "loss": 0.7271,
      "num_tokens": 4249495182.0,
      "step": 1391
    },
    {
      "epoch": 2.0381022165231726,
      "grad_norm": 0.3986559874877942,
      "learning_rate": 2.8156362579862042e-05,
      "loss": 0.696,
      "num_tokens": 4252380137.0,
      "step": 1392
    },
    {
      "epoch": 2.0395676863894487,
      "grad_norm": 0.44316063703000513,
      "learning_rate": 2.8140315620115623e-05,
      "loss": 0.7057,
      "num_tokens": 4255473789.0,
      "step": 1393
    },
    {
      "epoch": 2.0410331562557245,
      "grad_norm": 0.4111133212307321,
      "learning_rate": 2.8124263135655826e-05,
      "loss": 0.7328,
      "num_tokens": 4258313168.0,
      "step": 1394
    },
    {
      "epoch": 2.042498626122,
      "grad_norm": 0.4413632764779973,
      "learning_rate": 2.8108205140925788e-05,
      "loss": 0.6971,
      "num_tokens": 4261448179.0,
      "step": 1395
    },
    {
      "epoch": 2.0439640959882763,
      "grad_norm": 0.44624448794342847,
      "learning_rate": 2.809214165037359e-05,
      "loss": 0.7005,
      "num_tokens": 4264375389.0,
      "step": 1396
    },
    {
      "epoch": 2.045429565854552,
      "grad_norm": 0.4615703494843691,
      "learning_rate": 2.8076072678452263e-05,
      "loss": 0.7002,
      "num_tokens": 4267441589.0,
      "step": 1397
    },
    {
      "epoch": 2.046895035720828,
      "grad_norm": 0.43548063319700353,
      "learning_rate": 2.8059998239619773e-05,
      "loss": 0.7143,
      "num_tokens": 4270537801.0,
      "step": 1398
    },
    {
      "epoch": 2.048360505587104,
      "grad_norm": 0.498436171096523,
      "learning_rate": 2.8043918348339006e-05,
      "loss": 0.703,
      "num_tokens": 4273751612.0,
      "step": 1399
    },
    {
      "epoch": 2.0498259754533796,
      "grad_norm": 0.459495179106195,
      "learning_rate": 2.8027833019077753e-05,
      "loss": 0.705,
      "num_tokens": 4276672844.0,
      "step": 1400
    },
    {
      "epoch": 2.051291445319656,
      "grad_norm": 0.49853715963840534,
      "learning_rate": 2.8011742266308686e-05,
      "loss": 0.7159,
      "num_tokens": 4279608101.0,
      "step": 1401
    },
    {
      "epoch": 2.0527569151859315,
      "grad_norm": 0.41049646140061313,
      "learning_rate": 2.7995646104509377e-05,
      "loss": 0.7155,
      "num_tokens": 4282851107.0,
      "step": 1402
    },
    {
      "epoch": 2.0542223850522072,
      "grad_norm": 0.49851252493865544,
      "learning_rate": 2.7979544548162246e-05,
      "loss": 0.7092,
      "num_tokens": 4285828000.0,
      "step": 1403
    },
    {
      "epoch": 2.0556878549184834,
      "grad_norm": 0.4051887579903802,
      "learning_rate": 2.7963437611754583e-05,
      "loss": 0.7036,
      "num_tokens": 4288586316.0,
      "step": 1404
    },
    {
      "epoch": 2.057153324784759,
      "grad_norm": 0.5329319472251645,
      "learning_rate": 2.7947325309778507e-05,
      "loss": 0.7174,
      "num_tokens": 4291923561.0,
      "step": 1405
    },
    {
      "epoch": 2.058618794651035,
      "grad_norm": 0.460651114307889,
      "learning_rate": 2.7931207656730963e-05,
      "loss": 0.6846,
      "num_tokens": 4295083241.0,
      "step": 1406
    },
    {
      "epoch": 2.060084264517311,
      "grad_norm": 0.486114312359634,
      "learning_rate": 2.791508466711372e-05,
      "loss": 0.7045,
      "num_tokens": 4298237914.0,
      "step": 1407
    },
    {
      "epoch": 2.0615497343835867,
      "grad_norm": 0.4987670456764497,
      "learning_rate": 2.7898956355433352e-05,
      "loss": 0.7144,
      "num_tokens": 4301268508.0,
      "step": 1408
    },
    {
      "epoch": 2.0630152042498624,
      "grad_norm": 0.500266060603412,
      "learning_rate": 2.7882822736201204e-05,
      "loss": 0.7085,
      "num_tokens": 4304381370.0,
      "step": 1409
    },
    {
      "epoch": 2.0644806741161386,
      "grad_norm": 0.4353557074583143,
      "learning_rate": 2.7866683823933406e-05,
      "loss": 0.6974,
      "num_tokens": 4307421839.0,
      "step": 1410
    },
    {
      "epoch": 2.0659461439824143,
      "grad_norm": 0.47008725586061084,
      "learning_rate": 2.7850539633150854e-05,
      "loss": 0.6849,
      "num_tokens": 4310347288.0,
      "step": 1411
    },
    {
      "epoch": 2.0674116138486904,
      "grad_norm": 0.5069092479458464,
      "learning_rate": 2.783439017837919e-05,
      "loss": 0.7049,
      "num_tokens": 4313449301.0,
      "step": 1412
    },
    {
      "epoch": 2.068877083714966,
      "grad_norm": 0.37045911630842837,
      "learning_rate": 2.7818235474148793e-05,
      "loss": 0.7089,
      "num_tokens": 4316493785.0,
      "step": 1413
    },
    {
      "epoch": 2.070342553581242,
      "grad_norm": 0.48734419235497006,
      "learning_rate": 2.7802075534994762e-05,
      "loss": 0.7122,
      "num_tokens": 4319571389.0,
      "step": 1414
    },
    {
      "epoch": 2.071808023447518,
      "grad_norm": 0.3875173556126372,
      "learning_rate": 2.7785910375456914e-05,
      "loss": 0.7205,
      "num_tokens": 4322582684.0,
      "step": 1415
    },
    {
      "epoch": 2.0732734933137937,
      "grad_norm": 0.46711045179092997,
      "learning_rate": 2.7769740010079752e-05,
      "loss": 0.7084,
      "num_tokens": 4325662864.0,
      "step": 1416
    },
    {
      "epoch": 2.0747389631800695,
      "grad_norm": 0.4230687596039707,
      "learning_rate": 2.7753564453412467e-05,
      "loss": 0.7025,
      "num_tokens": 4328763014.0,
      "step": 1417
    },
    {
      "epoch": 2.0762044330463456,
      "grad_norm": 0.49876048732424827,
      "learning_rate": 2.773738372000893e-05,
      "loss": 0.7117,
      "num_tokens": 4331931985.0,
      "step": 1418
    },
    {
      "epoch": 2.0776699029126213,
      "grad_norm": 0.4608663773168124,
      "learning_rate": 2.7721197824427654e-05,
      "loss": 0.7195,
      "num_tokens": 4334890165.0,
      "step": 1419
    },
    {
      "epoch": 2.079135372778897,
      "grad_norm": 0.4396921963271726,
      "learning_rate": 2.7705006781231816e-05,
      "loss": 0.718,
      "num_tokens": 4337979803.0,
      "step": 1420
    },
    {
      "epoch": 2.080600842645173,
      "grad_norm": 0.4800781680275101,
      "learning_rate": 2.7688810604989192e-05,
      "loss": 0.7168,
      "num_tokens": 4340877628.0,
      "step": 1421
    },
    {
      "epoch": 2.082066312511449,
      "grad_norm": 0.42113088750392697,
      "learning_rate": 2.7672609310272225e-05,
      "loss": 0.7167,
      "num_tokens": 4343880829.0,
      "step": 1422
    },
    {
      "epoch": 2.083531782377725,
      "grad_norm": 0.6151064757874548,
      "learning_rate": 2.7656402911657916e-05,
      "loss": 0.7201,
      "num_tokens": 4347124292.0,
      "step": 1423
    },
    {
      "epoch": 2.084997252244001,
      "grad_norm": 0.3903084005491772,
      "learning_rate": 2.764019142372788e-05,
      "loss": 0.6929,
      "num_tokens": 4350317771.0,
      "step": 1424
    },
    {
      "epoch": 2.0864627221102765,
      "grad_norm": 0.4241896334842566,
      "learning_rate": 2.7623974861068325e-05,
      "loss": 0.7061,
      "num_tokens": 4353409749.0,
      "step": 1425
    },
    {
      "epoch": 2.0879281919765527,
      "grad_norm": 0.5307331126530576,
      "learning_rate": 2.760775323826999e-05,
      "loss": 0.6962,
      "num_tokens": 4356315287.0,
      "step": 1426
    },
    {
      "epoch": 2.0893936618428284,
      "grad_norm": 0.45170129375128054,
      "learning_rate": 2.75915265699282e-05,
      "loss": 0.7093,
      "num_tokens": 4359469850.0,
      "step": 1427
    },
    {
      "epoch": 2.090859131709104,
      "grad_norm": 0.4296285528738621,
      "learning_rate": 2.7575294870642788e-05,
      "loss": 0.7104,
      "num_tokens": 4362633581.0,
      "step": 1428
    },
    {
      "epoch": 2.0923246015753802,
      "grad_norm": 0.46034044236783916,
      "learning_rate": 2.7559058155018156e-05,
      "loss": 0.7056,
      "num_tokens": 4365811768.0,
      "step": 1429
    },
    {
      "epoch": 2.093790071441656,
      "grad_norm": 0.48075567907737804,
      "learning_rate": 2.7542816437663174e-05,
      "loss": 0.7083,
      "num_tokens": 4368898502.0,
      "step": 1430
    },
    {
      "epoch": 2.0952555413079317,
      "grad_norm": 0.4018431277987484,
      "learning_rate": 2.752656973319124e-05,
      "loss": 0.7116,
      "num_tokens": 4371745065.0,
      "step": 1431
    },
    {
      "epoch": 2.096721011174208,
      "grad_norm": 0.3996716354520948,
      "learning_rate": 2.751031805622024e-05,
      "loss": 0.7054,
      "num_tokens": 4374797049.0,
      "step": 1432
    },
    {
      "epoch": 2.0981864810404836,
      "grad_norm": 0.48972108168234985,
      "learning_rate": 2.7494061421372516e-05,
      "loss": 0.7097,
      "num_tokens": 4377865245.0,
      "step": 1433
    },
    {
      "epoch": 2.0996519509067593,
      "grad_norm": 0.4190529698139894,
      "learning_rate": 2.7477799843274883e-05,
      "loss": 0.7062,
      "num_tokens": 4381044465.0,
      "step": 1434
    },
    {
      "epoch": 2.1011174207730354,
      "grad_norm": 0.3814217685185149,
      "learning_rate": 2.7461533336558602e-05,
      "loss": 0.71,
      "num_tokens": 4383879629.0,
      "step": 1435
    },
    {
      "epoch": 2.102582890639311,
      "grad_norm": 0.42597411237294686,
      "learning_rate": 2.7445261915859368e-05,
      "loss": 0.7172,
      "num_tokens": 4386979223.0,
      "step": 1436
    },
    {
      "epoch": 2.1040483605055873,
      "grad_norm": 0.4907231609040234,
      "learning_rate": 2.74289855958173e-05,
      "loss": 0.7013,
      "num_tokens": 4389983995.0,
      "step": 1437
    },
    {
      "epoch": 2.105513830371863,
      "grad_norm": 0.4435422173519564,
      "learning_rate": 2.7412704391076914e-05,
      "loss": 0.7137,
      "num_tokens": 4393091651.0,
      "step": 1438
    },
    {
      "epoch": 2.1069793002381387,
      "grad_norm": 0.40065940305539516,
      "learning_rate": 2.7396418316287147e-05,
      "loss": 0.7041,
      "num_tokens": 4396246794.0,
      "step": 1439
    },
    {
      "epoch": 2.108444770104415,
      "grad_norm": 0.559896009581261,
      "learning_rate": 2.7380127386101286e-05,
      "loss": 0.7418,
      "num_tokens": 4399129558.0,
      "step": 1440
    },
    {
      "epoch": 2.1099102399706906,
      "grad_norm": 0.4947013484164558,
      "learning_rate": 2.736383161517701e-05,
      "loss": 0.7082,
      "num_tokens": 4402121775.0,
      "step": 1441
    },
    {
      "epoch": 2.1113757098369663,
      "grad_norm": 0.4353515988323889,
      "learning_rate": 2.7347531018176328e-05,
      "loss": 0.7096,
      "num_tokens": 4405205601.0,
      "step": 1442
    },
    {
      "epoch": 2.1128411797032425,
      "grad_norm": 0.6034462994223836,
      "learning_rate": 2.7331225609765627e-05,
      "loss": 0.6952,
      "num_tokens": 4408553106.0,
      "step": 1443
    },
    {
      "epoch": 2.114306649569518,
      "grad_norm": 0.46473933307350507,
      "learning_rate": 2.731491540461559e-05,
      "loss": 0.7298,
      "num_tokens": 4411510861.0,
      "step": 1444
    },
    {
      "epoch": 2.115772119435794,
      "grad_norm": 0.6234698195596777,
      "learning_rate": 2.729860041740123e-05,
      "loss": 0.702,
      "num_tokens": 4414567370.0,
      "step": 1445
    },
    {
      "epoch": 2.11723758930207,
      "grad_norm": 0.4869956682434465,
      "learning_rate": 2.7282280662801875e-05,
      "loss": 0.7135,
      "num_tokens": 4417650261.0,
      "step": 1446
    },
    {
      "epoch": 2.118703059168346,
      "grad_norm": 0.5516773152750696,
      "learning_rate": 2.7265956155501127e-05,
      "loss": 0.7194,
      "num_tokens": 4420436912.0,
      "step": 1447
    },
    {
      "epoch": 2.120168529034622,
      "grad_norm": 0.424115026151848,
      "learning_rate": 2.724962691018685e-05,
      "loss": 0.7025,
      "num_tokens": 4423466190.0,
      "step": 1448
    },
    {
      "epoch": 2.1216339989008977,
      "grad_norm": 0.5130395934123798,
      "learning_rate": 2.7233292941551206e-05,
      "loss": 0.6927,
      "num_tokens": 4426574882.0,
      "step": 1449
    },
    {
      "epoch": 2.1230994687671734,
      "grad_norm": 0.39643696321888394,
      "learning_rate": 2.7216954264290585e-05,
      "loss": 0.721,
      "num_tokens": 4429442758.0,
      "step": 1450
    },
    {
      "epoch": 2.1245649386334495,
      "grad_norm": 0.4834044492577929,
      "learning_rate": 2.7200610893105603e-05,
      "loss": 0.6992,
      "num_tokens": 4432645594.0,
      "step": 1451
    },
    {
      "epoch": 2.1260304084997252,
      "grad_norm": 0.4545083425022002,
      "learning_rate": 2.718426284270113e-05,
      "loss": 0.7086,
      "num_tokens": 4435626799.0,
      "step": 1452
    },
    {
      "epoch": 2.127495878366001,
      "grad_norm": 0.49992934503077,
      "learning_rate": 2.7167910127786225e-05,
      "loss": 0.7033,
      "num_tokens": 4438856763.0,
      "step": 1453
    },
    {
      "epoch": 2.128961348232277,
      "grad_norm": 0.4287812255041137,
      "learning_rate": 2.715155276307415e-05,
      "loss": 0.7388,
      "num_tokens": 4441772945.0,
      "step": 1454
    },
    {
      "epoch": 2.130426818098553,
      "grad_norm": 0.472994373049365,
      "learning_rate": 2.713519076328234e-05,
      "loss": 0.7067,
      "num_tokens": 4444951065.0,
      "step": 1455
    },
    {
      "epoch": 2.1318922879648285,
      "grad_norm": 0.5468504578822707,
      "learning_rate": 2.7118824143132418e-05,
      "loss": 0.7021,
      "num_tokens": 4448191737.0,
      "step": 1456
    },
    {
      "epoch": 2.1333577578311047,
      "grad_norm": 0.4750744629594362,
      "learning_rate": 2.7102452917350155e-05,
      "loss": 0.7084,
      "num_tokens": 4451274622.0,
      "step": 1457
    },
    {
      "epoch": 2.1348232276973804,
      "grad_norm": 0.43552327613943814,
      "learning_rate": 2.708607710066546e-05,
      "loss": 0.715,
      "num_tokens": 4454363885.0,
      "step": 1458
    },
    {
      "epoch": 2.136288697563656,
      "grad_norm": 0.5272211124276347,
      "learning_rate": 2.706969670781239e-05,
      "loss": 0.7019,
      "num_tokens": 4457392721.0,
      "step": 1459
    },
    {
      "epoch": 2.1377541674299323,
      "grad_norm": 0.45237717868001254,
      "learning_rate": 2.70533117535291e-05,
      "loss": 0.7175,
      "num_tokens": 4460383747.0,
      "step": 1460
    },
    {
      "epoch": 2.139219637296208,
      "grad_norm": 0.5471667785200469,
      "learning_rate": 2.7036922252557864e-05,
      "loss": 0.7017,
      "num_tokens": 4463452054.0,
      "step": 1461
    },
    {
      "epoch": 2.140685107162484,
      "grad_norm": 0.40507183699805566,
      "learning_rate": 2.7020528219645025e-05,
      "loss": 0.6975,
      "num_tokens": 4466204417.0,
      "step": 1462
    },
    {
      "epoch": 2.14215057702876,
      "grad_norm": 0.5545717789795022,
      "learning_rate": 2.7004129669541043e-05,
      "loss": 0.6959,
      "num_tokens": 4469346772.0,
      "step": 1463
    },
    {
      "epoch": 2.1436160468950356,
      "grad_norm": 0.42769524467855713,
      "learning_rate": 2.6987726617000402e-05,
      "loss": 0.7149,
      "num_tokens": 4472292685.0,
      "step": 1464
    },
    {
      "epoch": 2.1450815167613118,
      "grad_norm": 0.4629481148897773,
      "learning_rate": 2.6971319076781655e-05,
      "loss": 0.7093,
      "num_tokens": 4475373653.0,
      "step": 1465
    },
    {
      "epoch": 2.1465469866275875,
      "grad_norm": 0.46920658665945847,
      "learning_rate": 2.6954907063647398e-05,
      "loss": 0.7189,
      "num_tokens": 4478564827.0,
      "step": 1466
    },
    {
      "epoch": 2.148012456493863,
      "grad_norm": 0.4349684646217136,
      "learning_rate": 2.6938490592364237e-05,
      "loss": 0.7033,
      "num_tokens": 4481389987.0,
      "step": 1467
    },
    {
      "epoch": 2.1494779263601393,
      "grad_norm": 0.4112801751522899,
      "learning_rate": 2.6922069677702804e-05,
      "loss": 0.718,
      "num_tokens": 4484338026.0,
      "step": 1468
    },
    {
      "epoch": 2.150943396226415,
      "grad_norm": 0.47651020160405994,
      "learning_rate": 2.690564433443771e-05,
      "loss": 0.718,
      "num_tokens": 4487434030.0,
      "step": 1469
    },
    {
      "epoch": 2.1524088660926908,
      "grad_norm": 0.4867578362602603,
      "learning_rate": 2.6889214577347576e-05,
      "loss": 0.7135,
      "num_tokens": 4490399652.0,
      "step": 1470
    },
    {
      "epoch": 2.153874335958967,
      "grad_norm": 0.35700012689753946,
      "learning_rate": 2.6872780421214973e-05,
      "loss": 0.7085,
      "num_tokens": 4493401596.0,
      "step": 1471
    },
    {
      "epoch": 2.1553398058252426,
      "grad_norm": 0.5453609352937585,
      "learning_rate": 2.685634188082644e-05,
      "loss": 0.7004,
      "num_tokens": 4496308185.0,
      "step": 1472
    },
    {
      "epoch": 2.156805275691519,
      "grad_norm": 0.4391571082705138,
      "learning_rate": 2.6839898970972456e-05,
      "loss": 0.7134,
      "num_tokens": 4499304149.0,
      "step": 1473
    },
    {
      "epoch": 2.1582707455577945,
      "grad_norm": 0.3989838507080752,
      "learning_rate": 2.6823451706447437e-05,
      "loss": 0.7002,
      "num_tokens": 4502305220.0,
      "step": 1474
    },
    {
      "epoch": 2.1597362154240702,
      "grad_norm": 0.5076727658096067,
      "learning_rate": 2.680700010204972e-05,
      "loss": 0.6906,
      "num_tokens": 4505395280.0,
      "step": 1475
    },
    {
      "epoch": 2.1612016852903464,
      "grad_norm": 0.46482056059113497,
      "learning_rate": 2.6790544172581524e-05,
      "loss": 0.6913,
      "num_tokens": 4508448842.0,
      "step": 1476
    },
    {
      "epoch": 2.162667155156622,
      "grad_norm": 0.39315894360191994,
      "learning_rate": 2.6774083932848988e-05,
      "loss": 0.7341,
      "num_tokens": 4511449747.0,
      "step": 1477
    },
    {
      "epoch": 2.164132625022898,
      "grad_norm": 0.4096665179245095,
      "learning_rate": 2.6757619397662112e-05,
      "loss": 0.6871,
      "num_tokens": 4514622716.0,
      "step": 1478
    },
    {
      "epoch": 2.165598094889174,
      "grad_norm": 0.391551317163508,
      "learning_rate": 2.674115058183477e-05,
      "loss": 0.7054,
      "num_tokens": 4517658281.0,
      "step": 1479
    },
    {
      "epoch": 2.1670635647554497,
      "grad_norm": 0.46982096394357575,
      "learning_rate": 2.672467750018469e-05,
      "loss": 0.6988,
      "num_tokens": 4520863503.0,
      "step": 1480
    },
    {
      "epoch": 2.1685290346217254,
      "grad_norm": 0.4046832299726708,
      "learning_rate": 2.670820016753342e-05,
      "loss": 0.7074,
      "num_tokens": 4523857466.0,
      "step": 1481
    },
    {
      "epoch": 2.1699945044880016,
      "grad_norm": 0.4515367388867664,
      "learning_rate": 2.6691718598706343e-05,
      "loss": 0.7136,
      "num_tokens": 4526887968.0,
      "step": 1482
    },
    {
      "epoch": 2.1714599743542773,
      "grad_norm": 0.3740509512074513,
      "learning_rate": 2.6675232808532666e-05,
      "loss": 0.7192,
      "num_tokens": 4529893348.0,
      "step": 1483
    },
    {
      "epoch": 2.172925444220553,
      "grad_norm": 0.5568171670414099,
      "learning_rate": 2.6658742811845377e-05,
      "loss": 0.7144,
      "num_tokens": 4533121756.0,
      "step": 1484
    },
    {
      "epoch": 2.174390914086829,
      "grad_norm": 0.4454143865159811,
      "learning_rate": 2.6642248623481256e-05,
      "loss": 0.7195,
      "num_tokens": 4536019859.0,
      "step": 1485
    },
    {
      "epoch": 2.175856383953105,
      "grad_norm": 0.5022680642208118,
      "learning_rate": 2.6625750258280852e-05,
      "loss": 0.7024,
      "num_tokens": 4539487786.0,
      "step": 1486
    },
    {
      "epoch": 2.177321853819381,
      "grad_norm": 0.4566353701294357,
      "learning_rate": 2.660924773108847e-05,
      "loss": 0.7182,
      "num_tokens": 4542315820.0,
      "step": 1487
    },
    {
      "epoch": 2.1787873236856568,
      "grad_norm": 0.4824052964454861,
      "learning_rate": 2.6592741056752172e-05,
      "loss": 0.7224,
      "num_tokens": 4545143690.0,
      "step": 1488
    },
    {
      "epoch": 2.1802527935519325,
      "grad_norm": 0.46144642592245877,
      "learning_rate": 2.6576230250123737e-05,
      "loss": 0.7191,
      "num_tokens": 4547967569.0,
      "step": 1489
    },
    {
      "epoch": 2.1817182634182086,
      "grad_norm": 0.4499005684772337,
      "learning_rate": 2.655971532605867e-05,
      "loss": 0.6857,
      "num_tokens": 4551101569.0,
      "step": 1490
    },
    {
      "epoch": 2.1831837332844843,
      "grad_norm": 0.42158175585861707,
      "learning_rate": 2.6543196299416178e-05,
      "loss": 0.7023,
      "num_tokens": 4554025894.0,
      "step": 1491
    },
    {
      "epoch": 2.18464920315076,
      "grad_norm": 0.46260129774360087,
      "learning_rate": 2.6526673185059156e-05,
      "loss": 0.7041,
      "num_tokens": 4557073969.0,
      "step": 1492
    },
    {
      "epoch": 2.186114673017036,
      "grad_norm": 0.38271699280905513,
      "learning_rate": 2.651014599785418e-05,
      "loss": 0.7101,
      "num_tokens": 4560149156.0,
      "step": 1493
    },
    {
      "epoch": 2.187580142883312,
      "grad_norm": 0.5187488912099569,
      "learning_rate": 2.6493614752671496e-05,
      "loss": 0.6977,
      "num_tokens": 4563138977.0,
      "step": 1494
    },
    {
      "epoch": 2.1890456127495876,
      "grad_norm": 0.43906544785007634,
      "learning_rate": 2.6477079464384994e-05,
      "loss": 0.724,
      "num_tokens": 4566257165.0,
      "step": 1495
    },
    {
      "epoch": 2.190511082615864,
      "grad_norm": 0.49988045967845707,
      "learning_rate": 2.64605401478722e-05,
      "loss": 0.7375,
      "num_tokens": 4569269646.0,
      "step": 1496
    },
    {
      "epoch": 2.1919765524821395,
      "grad_norm": 0.42744981173953817,
      "learning_rate": 2.6443996818014275e-05,
      "loss": 0.7084,
      "num_tokens": 4572346437.0,
      "step": 1497
    },
    {
      "epoch": 2.1934420223484157,
      "grad_norm": 0.5026916773678158,
      "learning_rate": 2.642744948969598e-05,
      "loss": 0.7178,
      "num_tokens": 4575512790.0,
      "step": 1498
    },
    {
      "epoch": 2.1949074922146914,
      "grad_norm": 0.4347661171067446,
      "learning_rate": 2.641089817780568e-05,
      "loss": 0.7091,
      "num_tokens": 4578404555.0,
      "step": 1499
    },
    {
      "epoch": 2.196372962080967,
      "grad_norm": 0.46608475793445053,
      "learning_rate": 2.6394342897235313e-05,
      "loss": 0.7034,
      "num_tokens": 4581373173.0,
      "step": 1500
    },
    {
      "epoch": 2.1978384319472433,
      "grad_norm": 0.4215216463291551,
      "learning_rate": 2.637778366288041e-05,
      "loss": 0.7131,
      "num_tokens": 4584220543.0,
      "step": 1501
    },
    {
      "epoch": 2.199303901813519,
      "grad_norm": 0.4236248838351213,
      "learning_rate": 2.6361220489640035e-05,
      "loss": 0.6885,
      "num_tokens": 4587205615.0,
      "step": 1502
    },
    {
      "epoch": 2.2007693716797947,
      "grad_norm": 0.3776607862012483,
      "learning_rate": 2.6344653392416812e-05,
      "loss": 0.7009,
      "num_tokens": 4590391814.0,
      "step": 1503
    },
    {
      "epoch": 2.202234841546071,
      "grad_norm": 0.4931212910627832,
      "learning_rate": 2.6328082386116896e-05,
      "loss": 0.7129,
      "num_tokens": 4593328534.0,
      "step": 1504
    },
    {
      "epoch": 2.2037003114123466,
      "grad_norm": 0.5261525985739094,
      "learning_rate": 2.631150748564994e-05,
      "loss": 0.6845,
      "num_tokens": 4596808432.0,
      "step": 1505
    },
    {
      "epoch": 2.2051657812786223,
      "grad_norm": 0.34419896421973223,
      "learning_rate": 2.629492870592913e-05,
      "loss": 0.7047,
      "num_tokens": 4599888220.0,
      "step": 1506
    },
    {
      "epoch": 2.2066312511448984,
      "grad_norm": 0.62009634919208,
      "learning_rate": 2.627834606187112e-05,
      "loss": 0.7181,
      "num_tokens": 4603018773.0,
      "step": 1507
    },
    {
      "epoch": 2.208096721011174,
      "grad_norm": 0.40849523678650623,
      "learning_rate": 2.6261759568396043e-05,
      "loss": 0.7003,
      "num_tokens": 4605994869.0,
      "step": 1508
    },
    {
      "epoch": 2.20956219087745,
      "grad_norm": 0.5019592793522145,
      "learning_rate": 2.624516924042751e-05,
      "loss": 0.7096,
      "num_tokens": 4609322052.0,
      "step": 1509
    },
    {
      "epoch": 2.211027660743726,
      "grad_norm": 0.4703685006069305,
      "learning_rate": 2.622857509289256e-05,
      "loss": 0.7021,
      "num_tokens": 4612192892.0,
      "step": 1510
    },
    {
      "epoch": 2.2124931306100017,
      "grad_norm": 0.47818667062498516,
      "learning_rate": 2.6211977140721696e-05,
      "loss": 0.7056,
      "num_tokens": 4615287610.0,
      "step": 1511
    },
    {
      "epoch": 2.213958600476278,
      "grad_norm": 0.4575909375597416,
      "learning_rate": 2.6195375398848823e-05,
      "loss": 0.7253,
      "num_tokens": 4618348670.0,
      "step": 1512
    },
    {
      "epoch": 2.2154240703425536,
      "grad_norm": 0.49874073229048155,
      "learning_rate": 2.6178769882211258e-05,
      "loss": 0.7091,
      "num_tokens": 4621590688.0,
      "step": 1513
    },
    {
      "epoch": 2.2168895402088293,
      "grad_norm": 0.4614959372438202,
      "learning_rate": 2.6162160605749716e-05,
      "loss": 0.7075,
      "num_tokens": 4624811553.0,
      "step": 1514
    },
    {
      "epoch": 2.2183550100751055,
      "grad_norm": 0.5292662513011022,
      "learning_rate": 2.614554758440831e-05,
      "loss": 0.6965,
      "num_tokens": 4628256482.0,
      "step": 1515
    },
    {
      "epoch": 2.219820479941381,
      "grad_norm": 0.46423765460771643,
      "learning_rate": 2.6128930833134494e-05,
      "loss": 0.6853,
      "num_tokens": 4631310222.0,
      "step": 1516
    },
    {
      "epoch": 2.221285949807657,
      "grad_norm": 0.47402951890962386,
      "learning_rate": 2.6112310366879102e-05,
      "loss": 0.719,
      "num_tokens": 4634292178.0,
      "step": 1517
    },
    {
      "epoch": 2.222751419673933,
      "grad_norm": 0.47173853329743476,
      "learning_rate": 2.6095686200596308e-05,
      "loss": 0.7069,
      "num_tokens": 4637265516.0,
      "step": 1518
    },
    {
      "epoch": 2.224216889540209,
      "grad_norm": 0.40933048509366193,
      "learning_rate": 2.6079058349243598e-05,
      "loss": 0.6891,
      "num_tokens": 4640375282.0,
      "step": 1519
    },
    {
      "epoch": 2.2256823594064845,
      "grad_norm": 0.4621371773273731,
      "learning_rate": 2.6062426827781793e-05,
      "loss": 0.7071,
      "num_tokens": 4643398067.0,
      "step": 1520
    },
    {
      "epoch": 2.2271478292727607,
      "grad_norm": 0.43312626084354955,
      "learning_rate": 2.6045791651175002e-05,
      "loss": 0.7168,
      "num_tokens": 4646284713.0,
      "step": 1521
    },
    {
      "epoch": 2.2286132991390364,
      "grad_norm": 0.4514745879032844,
      "learning_rate": 2.6029152834390635e-05,
      "loss": 0.7185,
      "num_tokens": 4649232533.0,
      "step": 1522
    },
    {
      "epoch": 2.2300787690053125,
      "grad_norm": 0.4268540691537877,
      "learning_rate": 2.6012510392399362e-05,
      "loss": 0.6973,
      "num_tokens": 4652360771.0,
      "step": 1523
    },
    {
      "epoch": 2.2315442388715883,
      "grad_norm": 0.4563440088845739,
      "learning_rate": 2.599586434017513e-05,
      "loss": 0.6998,
      "num_tokens": 4655269763.0,
      "step": 1524
    },
    {
      "epoch": 2.233009708737864,
      "grad_norm": 0.38187485764544354,
      "learning_rate": 2.597921469269513e-05,
      "loss": 0.695,
      "num_tokens": 4658393410.0,
      "step": 1525
    },
    {
      "epoch": 2.23447517860414,
      "grad_norm": 0.42925310462392985,
      "learning_rate": 2.5962561464939784e-05,
      "loss": 0.6866,
      "num_tokens": 4661531663.0,
      "step": 1526
    },
    {
      "epoch": 2.235940648470416,
      "grad_norm": 0.4136933855087547,
      "learning_rate": 2.5945904671892734e-05,
      "loss": 0.7121,
      "num_tokens": 4664562036.0,
      "step": 1527
    },
    {
      "epoch": 2.2374061183366916,
      "grad_norm": 0.4750444842705771,
      "learning_rate": 2.592924432854083e-05,
      "loss": 0.7163,
      "num_tokens": 4667274575.0,
      "step": 1528
    },
    {
      "epoch": 2.2388715882029677,
      "grad_norm": 0.3936626194420731,
      "learning_rate": 2.5912580449874128e-05,
      "loss": 0.7105,
      "num_tokens": 4670268030.0,
      "step": 1529
    },
    {
      "epoch": 2.2403370580692434,
      "grad_norm": 0.49449552868472935,
      "learning_rate": 2.5895913050885853e-05,
      "loss": 0.7003,
      "num_tokens": 4673271758.0,
      "step": 1530
    },
    {
      "epoch": 2.241802527935519,
      "grad_norm": 0.4138204838924853,
      "learning_rate": 2.5879242146572393e-05,
      "loss": 0.709,
      "num_tokens": 4676157079.0,
      "step": 1531
    },
    {
      "epoch": 2.2432679978017953,
      "grad_norm": 0.4813252943006639,
      "learning_rate": 2.5862567751933318e-05,
      "loss": 0.7008,
      "num_tokens": 4679380199.0,
      "step": 1532
    },
    {
      "epoch": 2.244733467668071,
      "grad_norm": 0.46217047253176013,
      "learning_rate": 2.5845889881971297e-05,
      "loss": 0.7148,
      "num_tokens": 4682210999.0,
      "step": 1533
    },
    {
      "epoch": 2.2461989375343467,
      "grad_norm": 0.4262143569993668,
      "learning_rate": 2.5829208551692154e-05,
      "loss": 0.702,
      "num_tokens": 4685386780.0,
      "step": 1534
    },
    {
      "epoch": 2.247664407400623,
      "grad_norm": 0.5126445835024396,
      "learning_rate": 2.5812523776104817e-05,
      "loss": 0.7073,
      "num_tokens": 4688314825.0,
      "step": 1535
    },
    {
      "epoch": 2.2491298772668986,
      "grad_norm": 0.3705290058926664,
      "learning_rate": 2.5795835570221327e-05,
      "loss": 0.7214,
      "num_tokens": 4691458716.0,
      "step": 1536
    },
    {
      "epoch": 2.2505953471331743,
      "grad_norm": 0.40531584365768175,
      "learning_rate": 2.5779143949056785e-05,
      "loss": 0.6937,
      "num_tokens": 4694633835.0,
      "step": 1537
    },
    {
      "epoch": 2.2520608169994505,
      "grad_norm": 0.47780126223279634,
      "learning_rate": 2.5762448927629383e-05,
      "loss": 0.6917,
      "num_tokens": 4697932335.0,
      "step": 1538
    },
    {
      "epoch": 2.253526286865726,
      "grad_norm": 0.4019289723190445,
      "learning_rate": 2.5745750520960377e-05,
      "loss": 0.6984,
      "num_tokens": 4700874782.0,
      "step": 1539
    },
    {
      "epoch": 2.2549917567320024,
      "grad_norm": 0.4392304072433179,
      "learning_rate": 2.5729048744074065e-05,
      "loss": 0.7176,
      "num_tokens": 4703963192.0,
      "step": 1540
    },
    {
      "epoch": 2.256457226598278,
      "grad_norm": 0.3942424069978266,
      "learning_rate": 2.5712343611997763e-05,
      "loss": 0.6814,
      "num_tokens": 4706901606.0,
      "step": 1541
    },
    {
      "epoch": 2.257922696464554,
      "grad_norm": 0.631590921534017,
      "learning_rate": 2.569563513976182e-05,
      "loss": 0.6993,
      "num_tokens": 4709676244.0,
      "step": 1542
    },
    {
      "epoch": 2.25938816633083,
      "grad_norm": 0.4449125280535514,
      "learning_rate": 2.5678923342399586e-05,
      "loss": 0.7284,
      "num_tokens": 4712611226.0,
      "step": 1543
    },
    {
      "epoch": 2.2608536361971057,
      "grad_norm": 0.5971225221636886,
      "learning_rate": 2.5662208234947415e-05,
      "loss": 0.6987,
      "num_tokens": 4715552972.0,
      "step": 1544
    },
    {
      "epoch": 2.2623191060633814,
      "grad_norm": 0.5233618672343574,
      "learning_rate": 2.564548983244461e-05,
      "loss": 0.7031,
      "num_tokens": 4718899810.0,
      "step": 1545
    },
    {
      "epoch": 2.2637845759296575,
      "grad_norm": 0.5113970821870213,
      "learning_rate": 2.5628768149933468e-05,
      "loss": 0.7278,
      "num_tokens": 4721818354.0,
      "step": 1546
    },
    {
      "epoch": 2.2652500457959333,
      "grad_norm": 0.6096047752396083,
      "learning_rate": 2.5612043202459224e-05,
      "loss": 0.6757,
      "num_tokens": 4724906045.0,
      "step": 1547
    },
    {
      "epoch": 2.2667155156622094,
      "grad_norm": 0.4600765189742594,
      "learning_rate": 2.559531500507005e-05,
      "loss": 0.689,
      "num_tokens": 4728105245.0,
      "step": 1548
    },
    {
      "epoch": 2.268180985528485,
      "grad_norm": 0.6494951961737271,
      "learning_rate": 2.557858357281705e-05,
      "loss": 0.6973,
      "num_tokens": 4731210768.0,
      "step": 1549
    },
    {
      "epoch": 2.269646455394761,
      "grad_norm": 0.4601967886926085,
      "learning_rate": 2.556184892075423e-05,
      "loss": 0.7006,
      "num_tokens": 4734185598.0,
      "step": 1550
    },
    {
      "epoch": 2.271111925261037,
      "grad_norm": 0.5967838727220792,
      "learning_rate": 2.5545111063938496e-05,
      "loss": 0.7,
      "num_tokens": 4737151000.0,
      "step": 1551
    },
    {
      "epoch": 2.2725773951273127,
      "grad_norm": 0.5146575347552436,
      "learning_rate": 2.552837001742963e-05,
      "loss": 0.692,
      "num_tokens": 4740369524.0,
      "step": 1552
    },
    {
      "epoch": 2.2740428649935884,
      "grad_norm": 0.4668041001226483,
      "learning_rate": 2.5511625796290314e-05,
      "loss": 0.7003,
      "num_tokens": 4743511147.0,
      "step": 1553
    },
    {
      "epoch": 2.2755083348598646,
      "grad_norm": 0.5075830399790876,
      "learning_rate": 2.5494878415586038e-05,
      "loss": 0.7105,
      "num_tokens": 4746688378.0,
      "step": 1554
    },
    {
      "epoch": 2.2769738047261403,
      "grad_norm": 0.3995088345431428,
      "learning_rate": 2.5478127890385174e-05,
      "loss": 0.6921,
      "num_tokens": 4749826435.0,
      "step": 1555
    },
    {
      "epoch": 2.278439274592416,
      "grad_norm": 0.5519960268752788,
      "learning_rate": 2.5461374235758907e-05,
      "loss": 0.7085,
      "num_tokens": 4752719778.0,
      "step": 1556
    },
    {
      "epoch": 2.279904744458692,
      "grad_norm": 0.4405535560249131,
      "learning_rate": 2.5444617466781246e-05,
      "loss": 0.7091,
      "num_tokens": 4755800998.0,
      "step": 1557
    },
    {
      "epoch": 2.281370214324968,
      "grad_norm": 0.5356627528981739,
      "learning_rate": 2.5427857598528988e-05,
      "loss": 0.706,
      "num_tokens": 4758878194.0,
      "step": 1558
    },
    {
      "epoch": 2.2828356841912436,
      "grad_norm": 0.5159859616617177,
      "learning_rate": 2.541109464608173e-05,
      "loss": 0.6924,
      "num_tokens": 4762149343.0,
      "step": 1559
    },
    {
      "epoch": 2.2843011540575198,
      "grad_norm": 0.4477858834521287,
      "learning_rate": 2.5394328624521846e-05,
      "loss": 0.7112,
      "num_tokens": 4765138501.0,
      "step": 1560
    },
    {
      "epoch": 2.2857666239237955,
      "grad_norm": 0.527033830404785,
      "learning_rate": 2.537755954893446e-05,
      "loss": 0.6973,
      "num_tokens": 4768263380.0,
      "step": 1561
    },
    {
      "epoch": 2.287232093790071,
      "grad_norm": 0.38730883261663307,
      "learning_rate": 2.5360787434407454e-05,
      "loss": 0.6989,
      "num_tokens": 4771231518.0,
      "step": 1562
    },
    {
      "epoch": 2.2886975636563474,
      "grad_norm": 0.5710216011318371,
      "learning_rate": 2.534401229603144e-05,
      "loss": 0.707,
      "num_tokens": 4774301939.0,
      "step": 1563
    },
    {
      "epoch": 2.290163033522623,
      "grad_norm": 0.42658426396905924,
      "learning_rate": 2.532723414889976e-05,
      "loss": 0.7184,
      "num_tokens": 4777108125.0,
      "step": 1564
    },
    {
      "epoch": 2.2916285033888992,
      "grad_norm": 0.48031632185990397,
      "learning_rate": 2.531045300810844e-05,
      "loss": 0.6995,
      "num_tokens": 4780285349.0,
      "step": 1565
    },
    {
      "epoch": 2.293093973255175,
      "grad_norm": 0.4910101340705258,
      "learning_rate": 2.529366888875622e-05,
      "loss": 0.6913,
      "num_tokens": 4783304659.0,
      "step": 1566
    },
    {
      "epoch": 2.2945594431214507,
      "grad_norm": 0.40827077273428575,
      "learning_rate": 2.527688180594451e-05,
      "loss": 0.7106,
      "num_tokens": 4786294098.0,
      "step": 1567
    },
    {
      "epoch": 2.296024912987727,
      "grad_norm": 0.5490585576812276,
      "learning_rate": 2.5260091774777406e-05,
      "loss": 0.7009,
      "num_tokens": 4789405194.0,
      "step": 1568
    },
    {
      "epoch": 2.2974903828540025,
      "grad_norm": 0.42890809911903044,
      "learning_rate": 2.5243298810361622e-05,
      "loss": 0.7098,
      "num_tokens": 4792473881.0,
      "step": 1569
    },
    {
      "epoch": 2.2989558527202782,
      "grad_norm": 0.5078086718368822,
      "learning_rate": 2.5226502927806546e-05,
      "loss": 0.6987,
      "num_tokens": 4795759964.0,
      "step": 1570
    },
    {
      "epoch": 2.3004213225865544,
      "grad_norm": 0.47328390655665037,
      "learning_rate": 2.520970414222417e-05,
      "loss": 0.7131,
      "num_tokens": 4798846170.0,
      "step": 1571
    },
    {
      "epoch": 2.30188679245283,
      "grad_norm": 0.422901477677502,
      "learning_rate": 2.5192902468729093e-05,
      "loss": 0.7225,
      "num_tokens": 4801812460.0,
      "step": 1572
    },
    {
      "epoch": 2.3033522623191063,
      "grad_norm": 0.49037466698442217,
      "learning_rate": 2.5176097922438537e-05,
      "loss": 0.6935,
      "num_tokens": 4804899800.0,
      "step": 1573
    },
    {
      "epoch": 2.304817732185382,
      "grad_norm": 0.41520483193175606,
      "learning_rate": 2.5159290518472297e-05,
      "loss": 0.6959,
      "num_tokens": 4807988317.0,
      "step": 1574
    },
    {
      "epoch": 2.3062832020516577,
      "grad_norm": 0.4440102539104402,
      "learning_rate": 2.514248027195273e-05,
      "loss": 0.701,
      "num_tokens": 4811072056.0,
      "step": 1575
    },
    {
      "epoch": 2.307748671917934,
      "grad_norm": 0.48360650396831395,
      "learning_rate": 2.512566719800475e-05,
      "loss": 0.7159,
      "num_tokens": 4814157145.0,
      "step": 1576
    },
    {
      "epoch": 2.3092141417842096,
      "grad_norm": 0.46976552211687617,
      "learning_rate": 2.5108851311755835e-05,
      "loss": 0.6882,
      "num_tokens": 4817345667.0,
      "step": 1577
    },
    {
      "epoch": 2.3106796116504853,
      "grad_norm": 0.4134040896116298,
      "learning_rate": 2.5092032628335986e-05,
      "loss": 0.692,
      "num_tokens": 4820549194.0,
      "step": 1578
    },
    {
      "epoch": 2.3121450815167615,
      "grad_norm": 0.4826129593980683,
      "learning_rate": 2.50752111628777e-05,
      "loss": 0.7106,
      "num_tokens": 4823537588.0,
      "step": 1579
    },
    {
      "epoch": 2.313610551383037,
      "grad_norm": 0.4011606785911005,
      "learning_rate": 2.5058386930516002e-05,
      "loss": 0.7172,
      "num_tokens": 4826543198.0,
      "step": 1580
    },
    {
      "epoch": 2.315076021249313,
      "grad_norm": 0.4201363852877886,
      "learning_rate": 2.50415599463884e-05,
      "loss": 0.6857,
      "num_tokens": 4829772695.0,
      "step": 1581
    },
    {
      "epoch": 2.316541491115589,
      "grad_norm": 0.40708394860164904,
      "learning_rate": 2.502473022563487e-05,
      "loss": 0.7065,
      "num_tokens": 4832793879.0,
      "step": 1582
    },
    {
      "epoch": 2.3180069609818648,
      "grad_norm": 0.43947573586368893,
      "learning_rate": 2.500789778339786e-05,
      "loss": 0.716,
      "num_tokens": 4835940572.0,
      "step": 1583
    },
    {
      "epoch": 2.3194724308481405,
      "grad_norm": 0.386665759483699,
      "learning_rate": 2.4991062634822257e-05,
      "loss": 0.7022,
      "num_tokens": 4839149225.0,
      "step": 1584
    },
    {
      "epoch": 2.3209379007144166,
      "grad_norm": 0.41471932244545606,
      "learning_rate": 2.497422479505539e-05,
      "loss": 0.7021,
      "num_tokens": 4841830529.0,
      "step": 1585
    },
    {
      "epoch": 2.3224033705806923,
      "grad_norm": 0.4244088347430158,
      "learning_rate": 2.495738427924701e-05,
      "loss": 0.712,
      "num_tokens": 4844732046.0,
      "step": 1586
    },
    {
      "epoch": 2.323868840446968,
      "grad_norm": 0.3936269678643711,
      "learning_rate": 2.4940541102549274e-05,
      "loss": 0.699,
      "num_tokens": 4847695548.0,
      "step": 1587
    },
    {
      "epoch": 2.3253343103132442,
      "grad_norm": 0.47417714097432423,
      "learning_rate": 2.492369528011673e-05,
      "loss": 0.7058,
      "num_tokens": 4850741647.0,
      "step": 1588
    },
    {
      "epoch": 2.32679978017952,
      "grad_norm": 0.4345149690779481,
      "learning_rate": 2.4906846827106313e-05,
      "loss": 0.6876,
      "num_tokens": 4853614086.0,
      "step": 1589
    },
    {
      "epoch": 2.328265250045796,
      "grad_norm": 0.4304813030682135,
      "learning_rate": 2.488999575867731e-05,
      "loss": 0.7022,
      "num_tokens": 4856680013.0,
      "step": 1590
    },
    {
      "epoch": 2.329730719912072,
      "grad_norm": 0.4593925584550838,
      "learning_rate": 2.4873142089991388e-05,
      "loss": 0.699,
      "num_tokens": 4859686497.0,
      "step": 1591
    },
    {
      "epoch": 2.3311961897783475,
      "grad_norm": 0.46948684905366594,
      "learning_rate": 2.4856285836212525e-05,
      "loss": 0.6895,
      "num_tokens": 4862695328.0,
      "step": 1592
    },
    {
      "epoch": 2.3326616596446237,
      "grad_norm": 0.37813928354894105,
      "learning_rate": 2.4839427012507047e-05,
      "loss": 0.7088,
      "num_tokens": 4865676776.0,
      "step": 1593
    },
    {
      "epoch": 2.3341271295108994,
      "grad_norm": 0.4251925757993123,
      "learning_rate": 2.482256563404357e-05,
      "loss": 0.6918,
      "num_tokens": 4868658389.0,
      "step": 1594
    },
    {
      "epoch": 2.335592599377175,
      "grad_norm": 0.4184711349236048,
      "learning_rate": 2.4805701715993042e-05,
      "loss": 0.6904,
      "num_tokens": 4871595173.0,
      "step": 1595
    },
    {
      "epoch": 2.3370580692434513,
      "grad_norm": 0.42153381504455484,
      "learning_rate": 2.4788835273528658e-05,
      "loss": 0.6986,
      "num_tokens": 4874800681.0,
      "step": 1596
    },
    {
      "epoch": 2.338523539109727,
      "grad_norm": 0.4317675602123301,
      "learning_rate": 2.47719663218259e-05,
      "loss": 0.7039,
      "num_tokens": 4877773000.0,
      "step": 1597
    },
    {
      "epoch": 2.339989008976003,
      "grad_norm": 0.46909322199986475,
      "learning_rate": 2.475509487606252e-05,
      "loss": 0.6894,
      "num_tokens": 4880941312.0,
      "step": 1598
    },
    {
      "epoch": 2.341454478842279,
      "grad_norm": 0.44854738056695603,
      "learning_rate": 2.4738220951418504e-05,
      "loss": 0.683,
      "num_tokens": 4884092210.0,
      "step": 1599
    },
    {
      "epoch": 2.3429199487085546,
      "grad_norm": 0.5999481479828603,
      "learning_rate": 2.472134456307606e-05,
      "loss": 0.7192,
      "num_tokens": 4887082679.0,
      "step": 1600
    },
    {
      "epoch": 2.3443854185748307,
      "grad_norm": 0.44290837918437054,
      "learning_rate": 2.4704465726219623e-05,
      "loss": 0.6946,
      "num_tokens": 4890236856.0,
      "step": 1601
    },
    {
      "epoch": 2.3458508884411065,
      "grad_norm": 0.7516365469189337,
      "learning_rate": 2.4687584456035834e-05,
      "loss": 0.7009,
      "num_tokens": 4893460982.0,
      "step": 1602
    },
    {
      "epoch": 2.347316358307382,
      "grad_norm": 0.5728080190258974,
      "learning_rate": 2.4670700767713517e-05,
      "loss": 0.7053,
      "num_tokens": 4896576079.0,
      "step": 1603
    },
    {
      "epoch": 2.3487818281736583,
      "grad_norm": 0.7198878208333858,
      "learning_rate": 2.4653814676443665e-05,
      "loss": 0.7091,
      "num_tokens": 4899471390.0,
      "step": 1604
    },
    {
      "epoch": 2.350247298039934,
      "grad_norm": 0.623252980083619,
      "learning_rate": 2.4636926197419445e-05,
      "loss": 0.7154,
      "num_tokens": 4902446164.0,
      "step": 1605
    },
    {
      "epoch": 2.3517127679062098,
      "grad_norm": 0.6355916390669742,
      "learning_rate": 2.462003534583617e-05,
      "loss": 0.6829,
      "num_tokens": 4905752212.0,
      "step": 1606
    },
    {
      "epoch": 2.353178237772486,
      "grad_norm": 0.5479898931986629,
      "learning_rate": 2.460314213689128e-05,
      "loss": 0.6986,
      "num_tokens": 4908606830.0,
      "step": 1607
    },
    {
      "epoch": 2.3546437076387616,
      "grad_norm": 0.6508796583730337,
      "learning_rate": 2.4586246585784352e-05,
      "loss": 0.7,
      "num_tokens": 4911688117.0,
      "step": 1608
    },
    {
      "epoch": 2.3561091775050373,
      "grad_norm": 0.537044909721635,
      "learning_rate": 2.456934870771705e-05,
      "loss": 0.7032,
      "num_tokens": 4914736118.0,
      "step": 1609
    },
    {
      "epoch": 2.3575746473713135,
      "grad_norm": 0.7071834570416751,
      "learning_rate": 2.455244851789315e-05,
      "loss": 0.7085,
      "num_tokens": 4917919508.0,
      "step": 1610
    },
    {
      "epoch": 2.359040117237589,
      "grad_norm": 0.559881708320018,
      "learning_rate": 2.453554603151849e-05,
      "loss": 0.6853,
      "num_tokens": 4920880512.0,
      "step": 1611
    },
    {
      "epoch": 2.360505587103865,
      "grad_norm": 0.7560345234773478,
      "learning_rate": 2.4518641263800996e-05,
      "loss": 0.6977,
      "num_tokens": 4924049676.0,
      "step": 1612
    },
    {
      "epoch": 2.361971056970141,
      "grad_norm": 0.6308929617902254,
      "learning_rate": 2.4501734229950623e-05,
      "loss": 0.6943,
      "num_tokens": 4927340029.0,
      "step": 1613
    },
    {
      "epoch": 2.363436526836417,
      "grad_norm": 0.6843937841346347,
      "learning_rate": 2.4484824945179382e-05,
      "loss": 0.6966,
      "num_tokens": 4930386666.0,
      "step": 1614
    },
    {
      "epoch": 2.364901996702693,
      "grad_norm": 0.6054878100395716,
      "learning_rate": 2.446791342470131e-05,
      "loss": 0.6846,
      "num_tokens": 4933479391.0,
      "step": 1615
    },
    {
      "epoch": 2.3663674665689687,
      "grad_norm": 0.6008603662918548,
      "learning_rate": 2.4450999683732443e-05,
      "loss": 0.7068,
      "num_tokens": 4936559134.0,
      "step": 1616
    },
    {
      "epoch": 2.3678329364352444,
      "grad_norm": 0.5441142453503164,
      "learning_rate": 2.443408373749082e-05,
      "loss": 0.6959,
      "num_tokens": 4939407688.0,
      "step": 1617
    },
    {
      "epoch": 2.3692984063015206,
      "grad_norm": 0.6613507741763264,
      "learning_rate": 2.441716560119647e-05,
      "loss": 0.6766,
      "num_tokens": 4942401419.0,
      "step": 1618
    },
    {
      "epoch": 2.3707638761677963,
      "grad_norm": 0.5502094135369707,
      "learning_rate": 2.440024529007138e-05,
      "loss": 0.6917,
      "num_tokens": 4945238043.0,
      "step": 1619
    },
    {
      "epoch": 2.372229346034072,
      "grad_norm": 0.7294829761149649,
      "learning_rate": 2.438332281933951e-05,
      "loss": 0.7021,
      "num_tokens": 4948276175.0,
      "step": 1620
    },
    {
      "epoch": 2.373694815900348,
      "grad_norm": 0.5920730310224475,
      "learning_rate": 2.4366398204226747e-05,
      "loss": 0.6899,
      "num_tokens": 4951371074.0,
      "step": 1621
    },
    {
      "epoch": 2.375160285766624,
      "grad_norm": 0.6548788248575416,
      "learning_rate": 2.4349471459960935e-05,
      "loss": 0.7138,
      "num_tokens": 4954345236.0,
      "step": 1622
    },
    {
      "epoch": 2.3766257556329,
      "grad_norm": 0.5823421828811941,
      "learning_rate": 2.433254260177179e-05,
      "loss": 0.6959,
      "num_tokens": 4957507241.0,
      "step": 1623
    },
    {
      "epoch": 2.3780912254991757,
      "grad_norm": 0.5657737679124532,
      "learning_rate": 2.4315611644890962e-05,
      "loss": 0.6917,
      "num_tokens": 4960675310.0,
      "step": 1624
    },
    {
      "epoch": 2.3795566953654514,
      "grad_norm": 0.500301905045223,
      "learning_rate": 2.4298678604551987e-05,
      "loss": 0.7202,
      "num_tokens": 4963633293.0,
      "step": 1625
    },
    {
      "epoch": 2.3810221652317276,
      "grad_norm": 0.6258679615658008,
      "learning_rate": 2.4281743495990265e-05,
      "loss": 0.7014,
      "num_tokens": 4966820549.0,
      "step": 1626
    },
    {
      "epoch": 2.3824876350980033,
      "grad_norm": 0.43036139990685973,
      "learning_rate": 2.4264806334443063e-05,
      "loss": 0.6837,
      "num_tokens": 4969868328.0,
      "step": 1627
    },
    {
      "epoch": 2.383953104964279,
      "grad_norm": 0.707392943102859,
      "learning_rate": 2.4247867135149492e-05,
      "loss": 0.6939,
      "num_tokens": 4972953368.0,
      "step": 1628
    },
    {
      "epoch": 2.385418574830555,
      "grad_norm": 0.5701379155261812,
      "learning_rate": 2.42309259133505e-05,
      "loss": 0.7108,
      "num_tokens": 4976035208.0,
      "step": 1629
    },
    {
      "epoch": 2.386884044696831,
      "grad_norm": 0.6526350427642641,
      "learning_rate": 2.4213982684288854e-05,
      "loss": 0.6972,
      "num_tokens": 4979018875.0,
      "step": 1630
    },
    {
      "epoch": 2.3883495145631066,
      "grad_norm": 0.5851435564536693,
      "learning_rate": 2.4197037463209125e-05,
      "loss": 0.7061,
      "num_tokens": 4982171477.0,
      "step": 1631
    },
    {
      "epoch": 2.389814984429383,
      "grad_norm": 0.515465607738937,
      "learning_rate": 2.4180090265357677e-05,
      "loss": 0.7094,
      "num_tokens": 4985252380.0,
      "step": 1632
    },
    {
      "epoch": 2.3912804542956585,
      "grad_norm": 0.5460414437995168,
      "learning_rate": 2.4163141105982654e-05,
      "loss": 0.7081,
      "num_tokens": 4988211886.0,
      "step": 1633
    },
    {
      "epoch": 2.392745924161934,
      "grad_norm": 0.5601054959392333,
      "learning_rate": 2.4146190000333963e-05,
      "loss": 0.7204,
      "num_tokens": 4991042932.0,
      "step": 1634
    },
    {
      "epoch": 2.3942113940282104,
      "grad_norm": 0.4462370193759463,
      "learning_rate": 2.412923696366326e-05,
      "loss": 0.7007,
      "num_tokens": 4994114342.0,
      "step": 1635
    },
    {
      "epoch": 2.395676863894486,
      "grad_norm": 0.6403216886203645,
      "learning_rate": 2.411228201122395e-05,
      "loss": 0.7231,
      "num_tokens": 4997190930.0,
      "step": 1636
    },
    {
      "epoch": 2.397142333760762,
      "grad_norm": 0.5212004733261729,
      "learning_rate": 2.4095325158271147e-05,
      "loss": 0.6925,
      "num_tokens": 5000483987.0,
      "step": 1637
    },
    {
      "epoch": 2.398607803627038,
      "grad_norm": 0.5929047369321417,
      "learning_rate": 2.407836642006168e-05,
      "loss": 0.7,
      "num_tokens": 5003424040.0,
      "step": 1638
    },
    {
      "epoch": 2.4000732734933137,
      "grad_norm": 0.562573654438875,
      "learning_rate": 2.4061405811854077e-05,
      "loss": 0.6756,
      "num_tokens": 5006555950.0,
      "step": 1639
    },
    {
      "epoch": 2.40153874335959,
      "grad_norm": 0.5650852948311534,
      "learning_rate": 2.4044443348908552e-05,
      "loss": 0.7066,
      "num_tokens": 5009595176.0,
      "step": 1640
    },
    {
      "epoch": 2.4030042132258655,
      "grad_norm": 0.49521276928984426,
      "learning_rate": 2.4027479046486973e-05,
      "loss": 0.6868,
      "num_tokens": 5012782271.0,
      "step": 1641
    },
    {
      "epoch": 2.4044696830921413,
      "grad_norm": 0.5544305326399391,
      "learning_rate": 2.401051291985288e-05,
      "loss": 0.7119,
      "num_tokens": 5015810449.0,
      "step": 1642
    },
    {
      "epoch": 2.4059351529584174,
      "grad_norm": 0.4821641040968821,
      "learning_rate": 2.3993544984271445e-05,
      "loss": 0.7057,
      "num_tokens": 5018756729.0,
      "step": 1643
    },
    {
      "epoch": 2.407400622824693,
      "grad_norm": 0.6098504832682118,
      "learning_rate": 2.3976575255009472e-05,
      "loss": 0.6941,
      "num_tokens": 5021596367.0,
      "step": 1644
    },
    {
      "epoch": 2.408866092690969,
      "grad_norm": 0.5536130451663451,
      "learning_rate": 2.3959603747335365e-05,
      "loss": 0.7022,
      "num_tokens": 5024755560.0,
      "step": 1645
    },
    {
      "epoch": 2.410331562557245,
      "grad_norm": 0.5351561881399429,
      "learning_rate": 2.3942630476519156e-05,
      "loss": 0.6997,
      "num_tokens": 5027873348.0,
      "step": 1646
    },
    {
      "epoch": 2.4117970324235207,
      "grad_norm": 0.534630250382874,
      "learning_rate": 2.392565545783244e-05,
      "loss": 0.7111,
      "num_tokens": 5030963769.0,
      "step": 1647
    },
    {
      "epoch": 2.413262502289797,
      "grad_norm": 0.5299766340975229,
      "learning_rate": 2.3908678706548386e-05,
      "loss": 0.6982,
      "num_tokens": 5034006383.0,
      "step": 1648
    },
    {
      "epoch": 2.4147279721560726,
      "grad_norm": 0.46516869549737344,
      "learning_rate": 2.3891700237941737e-05,
      "loss": 0.692,
      "num_tokens": 5037105465.0,
      "step": 1649
    },
    {
      "epoch": 2.4161934420223483,
      "grad_norm": 0.6530463712273142,
      "learning_rate": 2.3874720067288767e-05,
      "loss": 0.7158,
      "num_tokens": 5039962408.0,
      "step": 1650
    },
    {
      "epoch": 2.4176589118886245,
      "grad_norm": 0.5623690248547457,
      "learning_rate": 2.3857738209867284e-05,
      "loss": 0.6917,
      "num_tokens": 5043197782.0,
      "step": 1651
    },
    {
      "epoch": 2.4191243817549,
      "grad_norm": 0.6112477508786739,
      "learning_rate": 2.3840754680956618e-05,
      "loss": 0.7081,
      "num_tokens": 5046333919.0,
      "step": 1652
    },
    {
      "epoch": 2.420589851621176,
      "grad_norm": 0.6650381935697949,
      "learning_rate": 2.3823769495837606e-05,
      "loss": 0.6894,
      "num_tokens": 5049411444.0,
      "step": 1653
    },
    {
      "epoch": 2.422055321487452,
      "grad_norm": 0.4987206742555979,
      "learning_rate": 2.3806782669792558e-05,
      "loss": 0.6974,
      "num_tokens": 5052550110.0,
      "step": 1654
    },
    {
      "epoch": 2.4235207913537278,
      "grad_norm": 0.5586768913963881,
      "learning_rate": 2.3789794218105286e-05,
      "loss": 0.7135,
      "num_tokens": 5055481069.0,
      "step": 1655
    },
    {
      "epoch": 2.4249862612200035,
      "grad_norm": 0.4232436970320549,
      "learning_rate": 2.3772804156061038e-05,
      "loss": 0.6691,
      "num_tokens": 5058621117.0,
      "step": 1656
    },
    {
      "epoch": 2.4264517310862797,
      "grad_norm": 0.5546340569433037,
      "learning_rate": 2.3755812498946536e-05,
      "loss": 0.6892,
      "num_tokens": 5061339517.0,
      "step": 1657
    },
    {
      "epoch": 2.4279172009525554,
      "grad_norm": 0.4114748367994068,
      "learning_rate": 2.3738819262049916e-05,
      "loss": 0.7053,
      "num_tokens": 5064247144.0,
      "step": 1658
    },
    {
      "epoch": 2.429382670818831,
      "grad_norm": 0.47357288393711305,
      "learning_rate": 2.3721824460660745e-05,
      "loss": 0.7086,
      "num_tokens": 5067353206.0,
      "step": 1659
    },
    {
      "epoch": 2.4308481406851072,
      "grad_norm": 0.44768639244794906,
      "learning_rate": 2.370482811007e-05,
      "loss": 0.7034,
      "num_tokens": 5070369815.0,
      "step": 1660
    },
    {
      "epoch": 2.432313610551383,
      "grad_norm": 0.4271374396606263,
      "learning_rate": 2.368783022557005e-05,
      "loss": 0.7017,
      "num_tokens": 5073526215.0,
      "step": 1661
    },
    {
      "epoch": 2.4337790804176587,
      "grad_norm": 0.4502350928500912,
      "learning_rate": 2.3670830822454635e-05,
      "loss": 0.7076,
      "num_tokens": 5076585741.0,
      "step": 1662
    },
    {
      "epoch": 2.435244550283935,
      "grad_norm": 0.43177961736615933,
      "learning_rate": 2.365382991601888e-05,
      "loss": 0.6956,
      "num_tokens": 5079608999.0,
      "step": 1663
    },
    {
      "epoch": 2.4367100201502105,
      "grad_norm": 0.3908636891324485,
      "learning_rate": 2.363682752155925e-05,
      "loss": 0.7123,
      "num_tokens": 5082616950.0,
      "step": 1664
    },
    {
      "epoch": 2.4381754900164867,
      "grad_norm": 0.4679917280263215,
      "learning_rate": 2.361982365437354e-05,
      "loss": 0.7066,
      "num_tokens": 5085895965.0,
      "step": 1665
    },
    {
      "epoch": 2.4396409598827624,
      "grad_norm": 0.4532241232858204,
      "learning_rate": 2.36028183297609e-05,
      "loss": 0.7074,
      "num_tokens": 5088877120.0,
      "step": 1666
    },
    {
      "epoch": 2.441106429749038,
      "grad_norm": 0.4028034497483124,
      "learning_rate": 2.3585811563021754e-05,
      "loss": 0.7031,
      "num_tokens": 5091909265.0,
      "step": 1667
    },
    {
      "epoch": 2.4425718996153143,
      "grad_norm": 0.4667322255552734,
      "learning_rate": 2.356880336945785e-05,
      "loss": 0.6878,
      "num_tokens": 5095085879.0,
      "step": 1668
    },
    {
      "epoch": 2.44403736948159,
      "grad_norm": 0.3659754752343338,
      "learning_rate": 2.3551793764372213e-05,
      "loss": 0.6904,
      "num_tokens": 5098130086.0,
      "step": 1669
    },
    {
      "epoch": 2.4455028393478657,
      "grad_norm": 0.4789051098208179,
      "learning_rate": 2.353478276306913e-05,
      "loss": 0.6954,
      "num_tokens": 5101240176.0,
      "step": 1670
    },
    {
      "epoch": 2.446968309214142,
      "grad_norm": 0.4326893420828402,
      "learning_rate": 2.351777038085416e-05,
      "loss": 0.709,
      "num_tokens": 5104380830.0,
      "step": 1671
    },
    {
      "epoch": 2.4484337790804176,
      "grad_norm": 0.4736160950258135,
      "learning_rate": 2.3500756633034085e-05,
      "loss": 0.6838,
      "num_tokens": 5107118370.0,
      "step": 1672
    },
    {
      "epoch": 2.4498992489466938,
      "grad_norm": 0.36776651798126647,
      "learning_rate": 2.3483741534916935e-05,
      "loss": 0.6816,
      "num_tokens": 5110175639.0,
      "step": 1673
    },
    {
      "epoch": 2.4513647188129695,
      "grad_norm": 0.39959284191242234,
      "learning_rate": 2.3466725101811943e-05,
      "loss": 0.7102,
      "num_tokens": 5113498309.0,
      "step": 1674
    },
    {
      "epoch": 2.452830188679245,
      "grad_norm": 0.4932618347964198,
      "learning_rate": 2.344970734902954e-05,
      "loss": 0.6927,
      "num_tokens": 5116775774.0,
      "step": 1675
    },
    {
      "epoch": 2.4542956585455213,
      "grad_norm": 0.3618976055016676,
      "learning_rate": 2.3432688291881364e-05,
      "loss": 0.7032,
      "num_tokens": 5119803544.0,
      "step": 1676
    },
    {
      "epoch": 2.455761128411797,
      "grad_norm": 0.45775564343348013,
      "learning_rate": 2.3415667945680193e-05,
      "loss": 0.6973,
      "num_tokens": 5122860585.0,
      "step": 1677
    },
    {
      "epoch": 2.4572265982780728,
      "grad_norm": 0.4286152454645037,
      "learning_rate": 2.3398646325740004e-05,
      "loss": 0.7159,
      "num_tokens": 5125736092.0,
      "step": 1678
    },
    {
      "epoch": 2.458692068144349,
      "grad_norm": 0.4261392179623483,
      "learning_rate": 2.338162344737589e-05,
      "loss": 0.7173,
      "num_tokens": 5128728650.0,
      "step": 1679
    },
    {
      "epoch": 2.4601575380106246,
      "grad_norm": 0.42639584811684483,
      "learning_rate": 2.3364599325904096e-05,
      "loss": 0.6852,
      "num_tokens": 5131903574.0,
      "step": 1680
    },
    {
      "epoch": 2.4616230078769004,
      "grad_norm": 0.4571805749246758,
      "learning_rate": 2.3347573976641967e-05,
      "loss": 0.7089,
      "num_tokens": 5135062335.0,
      "step": 1681
    },
    {
      "epoch": 2.4630884777431765,
      "grad_norm": 0.4213473887983809,
      "learning_rate": 2.3330547414907968e-05,
      "loss": 0.6843,
      "num_tokens": 5137902574.0,
      "step": 1682
    },
    {
      "epoch": 2.4645539476094522,
      "grad_norm": 0.48983112933732914,
      "learning_rate": 2.3313519656021636e-05,
      "loss": 0.6974,
      "num_tokens": 5140844939.0,
      "step": 1683
    },
    {
      "epoch": 2.466019417475728,
      "grad_norm": 0.3893923429199581,
      "learning_rate": 2.3296490715303622e-05,
      "loss": 0.7073,
      "num_tokens": 5143718465.0,
      "step": 1684
    },
    {
      "epoch": 2.467484887342004,
      "grad_norm": 0.41429606875502134,
      "learning_rate": 2.32794606080756e-05,
      "loss": 0.7016,
      "num_tokens": 5146873373.0,
      "step": 1685
    },
    {
      "epoch": 2.46895035720828,
      "grad_norm": 0.3842309561278893,
      "learning_rate": 2.3262429349660312e-05,
      "loss": 0.7035,
      "num_tokens": 5149840982.0,
      "step": 1686
    },
    {
      "epoch": 2.4704158270745555,
      "grad_norm": 0.42592416558034174,
      "learning_rate": 2.324539695538154e-05,
      "loss": 0.7002,
      "num_tokens": 5152910877.0,
      "step": 1687
    },
    {
      "epoch": 2.4718812969408317,
      "grad_norm": 0.3741186707997887,
      "learning_rate": 2.322836344056408e-05,
      "loss": 0.7026,
      "num_tokens": 5156090266.0,
      "step": 1688
    },
    {
      "epoch": 2.4733467668071074,
      "grad_norm": 0.3770154200690991,
      "learning_rate": 2.3211328820533737e-05,
      "loss": 0.7015,
      "num_tokens": 5159062745.0,
      "step": 1689
    },
    {
      "epoch": 2.4748122366733836,
      "grad_norm": 0.4557063885567201,
      "learning_rate": 2.3194293110617302e-05,
      "loss": 0.6826,
      "num_tokens": 5162059475.0,
      "step": 1690
    },
    {
      "epoch": 2.4762777065396593,
      "grad_norm": 0.37808215438769005,
      "learning_rate": 2.3177256326142578e-05,
      "loss": 0.6901,
      "num_tokens": 5165172317.0,
      "step": 1691
    },
    {
      "epoch": 2.477743176405935,
      "grad_norm": 0.4257996289001036,
      "learning_rate": 2.3160218482438296e-05,
      "loss": 0.6911,
      "num_tokens": 5168235557.0,
      "step": 1692
    },
    {
      "epoch": 2.479208646272211,
      "grad_norm": 0.4490498053557484,
      "learning_rate": 2.314317959483416e-05,
      "loss": 0.7011,
      "num_tokens": 5171380314.0,
      "step": 1693
    },
    {
      "epoch": 2.480674116138487,
      "grad_norm": 0.39807884170215335,
      "learning_rate": 2.312613967866081e-05,
      "loss": 0.6953,
      "num_tokens": 5174354500.0,
      "step": 1694
    },
    {
      "epoch": 2.4821395860047626,
      "grad_norm": 0.43858665354529874,
      "learning_rate": 2.310909874924982e-05,
      "loss": 0.6969,
      "num_tokens": 5177534841.0,
      "step": 1695
    },
    {
      "epoch": 2.4836050558710387,
      "grad_norm": 0.39430986308107313,
      "learning_rate": 2.3092056821933655e-05,
      "loss": 0.6832,
      "num_tokens": 5180601199.0,
      "step": 1696
    },
    {
      "epoch": 2.4850705257373145,
      "grad_norm": 0.4398628004232293,
      "learning_rate": 2.3075013912045695e-05,
      "loss": 0.698,
      "num_tokens": 5183935316.0,
      "step": 1697
    },
    {
      "epoch": 2.4865359956035906,
      "grad_norm": 0.39109478350342175,
      "learning_rate": 2.305797003492021e-05,
      "loss": 0.6905,
      "num_tokens": 5186949660.0,
      "step": 1698
    },
    {
      "epoch": 2.4880014654698663,
      "grad_norm": 0.5184648301653152,
      "learning_rate": 2.304092520589232e-05,
      "loss": 0.7016,
      "num_tokens": 5190005585.0,
      "step": 1699
    },
    {
      "epoch": 2.489466935336142,
      "grad_norm": 0.39290488353167197,
      "learning_rate": 2.3023879440298006e-05,
      "loss": 0.7048,
      "num_tokens": 5192907436.0,
      "step": 1700
    },
    {
      "epoch": 2.490932405202418,
      "grad_norm": 0.477744882231662,
      "learning_rate": 2.3006832753474105e-05,
      "loss": 0.7205,
      "num_tokens": 5195806349.0,
      "step": 1701
    },
    {
      "epoch": 2.492397875068694,
      "grad_norm": 0.4152970304027692,
      "learning_rate": 2.2989785160758268e-05,
      "loss": 0.7154,
      "num_tokens": 5198886653.0,
      "step": 1702
    },
    {
      "epoch": 2.4938633449349696,
      "grad_norm": 0.490282393976516,
      "learning_rate": 2.2972736677488972e-05,
      "loss": 0.7165,
      "num_tokens": 5201814735.0,
      "step": 1703
    },
    {
      "epoch": 2.495328814801246,
      "grad_norm": 0.427275582242966,
      "learning_rate": 2.2955687319005495e-05,
      "loss": 0.6924,
      "num_tokens": 5204874261.0,
      "step": 1704
    },
    {
      "epoch": 2.4967942846675215,
      "grad_norm": 0.44740514325574476,
      "learning_rate": 2.2938637100647893e-05,
      "loss": 0.6948,
      "num_tokens": 5207905406.0,
      "step": 1705
    },
    {
      "epoch": 2.4982597545337972,
      "grad_norm": 0.40128117085505505,
      "learning_rate": 2.2921586037757004e-05,
      "loss": 0.6953,
      "num_tokens": 5210908059.0,
      "step": 1706
    },
    {
      "epoch": 2.4997252244000734,
      "grad_norm": 0.4319433461674659,
      "learning_rate": 2.2904534145674414e-05,
      "loss": 0.7169,
      "num_tokens": 5214098635.0,
      "step": 1707
    },
    {
      "epoch": 2.501190694266349,
      "grad_norm": 0.42759555494130985,
      "learning_rate": 2.2887481439742473e-05,
      "loss": 0.7144,
      "num_tokens": 5217162954.0,
      "step": 1708
    },
    {
      "epoch": 2.502656164132625,
      "grad_norm": 0.4252207273843664,
      "learning_rate": 2.2870427935304257e-05,
      "loss": 0.6839,
      "num_tokens": 5220274712.0,
      "step": 1709
    },
    {
      "epoch": 2.504121633998901,
      "grad_norm": 0.44018302759285016,
      "learning_rate": 2.2853373647703547e-05,
      "loss": 0.6866,
      "num_tokens": 5222996767.0,
      "step": 1710
    },
    {
      "epoch": 2.5055871038651767,
      "grad_norm": 0.35725160423238894,
      "learning_rate": 2.2836318592284847e-05,
      "loss": 0.713,
      "num_tokens": 5225987737.0,
      "step": 1711
    },
    {
      "epoch": 2.5070525737314524,
      "grad_norm": 0.45228040489835597,
      "learning_rate": 2.281926278439335e-05,
      "loss": 0.6976,
      "num_tokens": 5229086541.0,
      "step": 1712
    },
    {
      "epoch": 2.5085180435977286,
      "grad_norm": 0.3852933112978403,
      "learning_rate": 2.2802206239374906e-05,
      "loss": 0.6985,
      "num_tokens": 5232170082.0,
      "step": 1713
    },
    {
      "epoch": 2.5099835134640043,
      "grad_norm": 0.4734746106179736,
      "learning_rate": 2.2785148972576052e-05,
      "loss": 0.7017,
      "num_tokens": 5235463301.0,
      "step": 1714
    },
    {
      "epoch": 2.51144898333028,
      "grad_norm": 0.4309946167740658,
      "learning_rate": 2.276809099934396e-05,
      "loss": 0.7017,
      "num_tokens": 5238456703.0,
      "step": 1715
    },
    {
      "epoch": 2.512914453196556,
      "grad_norm": 0.4516134152131566,
      "learning_rate": 2.275103233502645e-05,
      "loss": 0.7238,
      "num_tokens": 5241540472.0,
      "step": 1716
    },
    {
      "epoch": 2.514379923062832,
      "grad_norm": 0.42861159158040885,
      "learning_rate": 2.2733972994971944e-05,
      "loss": 0.723,
      "num_tokens": 5244701257.0,
      "step": 1717
    },
    {
      "epoch": 2.515845392929108,
      "grad_norm": 0.4321770221433927,
      "learning_rate": 2.2716912994529493e-05,
      "loss": 0.6975,
      "num_tokens": 5247814451.0,
      "step": 1718
    },
    {
      "epoch": 2.5173108627953837,
      "grad_norm": 0.4773590465118812,
      "learning_rate": 2.2699852349048737e-05,
      "loss": 0.6926,
      "num_tokens": 5250944383.0,
      "step": 1719
    },
    {
      "epoch": 2.51877633266166,
      "grad_norm": 0.4210834166454949,
      "learning_rate": 2.2682791073879885e-05,
      "loss": 0.6955,
      "num_tokens": 5254017490.0,
      "step": 1720
    },
    {
      "epoch": 2.5202418025279356,
      "grad_norm": 0.46231394185287894,
      "learning_rate": 2.2665729184373728e-05,
      "loss": 0.6885,
      "num_tokens": 5257067037.0,
      "step": 1721
    },
    {
      "epoch": 2.5217072723942113,
      "grad_norm": 0.413395191394943,
      "learning_rate": 2.2648666695881594e-05,
      "loss": 0.6932,
      "num_tokens": 5259987698.0,
      "step": 1722
    },
    {
      "epoch": 2.5231727422604875,
      "grad_norm": 0.4302812291449578,
      "learning_rate": 2.263160362375536e-05,
      "loss": 0.69,
      "num_tokens": 5262896645.0,
      "step": 1723
    },
    {
      "epoch": 2.524638212126763,
      "grad_norm": 0.41170690795787557,
      "learning_rate": 2.2614539983347425e-05,
      "loss": 0.7038,
      "num_tokens": 5266013894.0,
      "step": 1724
    },
    {
      "epoch": 2.526103681993039,
      "grad_norm": 0.47109836332524,
      "learning_rate": 2.2597475790010706e-05,
      "loss": 0.6931,
      "num_tokens": 5268965193.0,
      "step": 1725
    },
    {
      "epoch": 2.527569151859315,
      "grad_norm": 0.3831334459134184,
      "learning_rate": 2.2580411059098615e-05,
      "loss": 0.6842,
      "num_tokens": 5272026640.0,
      "step": 1726
    },
    {
      "epoch": 2.529034621725591,
      "grad_norm": 0.5457375082552042,
      "learning_rate": 2.256334580596503e-05,
      "loss": 0.7046,
      "num_tokens": 5274948614.0,
      "step": 1727
    },
    {
      "epoch": 2.5305000915918665,
      "grad_norm": 0.41519285498910474,
      "learning_rate": 2.2546280045964327e-05,
      "loss": 0.6981,
      "num_tokens": 5277909807.0,
      "step": 1728
    },
    {
      "epoch": 2.5319655614581427,
      "grad_norm": 0.46697041524216876,
      "learning_rate": 2.2529213794451317e-05,
      "loss": 0.6927,
      "num_tokens": 5281150736.0,
      "step": 1729
    },
    {
      "epoch": 2.5334310313244184,
      "grad_norm": 0.5287569033034807,
      "learning_rate": 2.251214706678127e-05,
      "loss": 0.6949,
      "num_tokens": 5283921672.0,
      "step": 1730
    },
    {
      "epoch": 2.534896501190694,
      "grad_norm": 0.3837027940308178,
      "learning_rate": 2.249507987830986e-05,
      "loss": 0.6931,
      "num_tokens": 5287039342.0,
      "step": 1731
    },
    {
      "epoch": 2.5363619710569703,
      "grad_norm": 0.5449338532038152,
      "learning_rate": 2.2478012244393207e-05,
      "loss": 0.692,
      "num_tokens": 5290142128.0,
      "step": 1732
    },
    {
      "epoch": 2.537827440923246,
      "grad_norm": 0.4387671705981872,
      "learning_rate": 2.246094418038781e-05,
      "loss": 0.7046,
      "num_tokens": 5292818568.0,
      "step": 1733
    },
    {
      "epoch": 2.5392929107895217,
      "grad_norm": 0.4243455517852554,
      "learning_rate": 2.2443875701650555e-05,
      "loss": 0.7018,
      "num_tokens": 5295912168.0,
      "step": 1734
    },
    {
      "epoch": 2.540758380655798,
      "grad_norm": 0.47978551005261766,
      "learning_rate": 2.242680682353871e-05,
      "loss": 0.7027,
      "num_tokens": 5299081450.0,
      "step": 1735
    },
    {
      "epoch": 2.5422238505220736,
      "grad_norm": 0.38962393793596006,
      "learning_rate": 2.2409737561409905e-05,
      "loss": 0.6863,
      "num_tokens": 5302223258.0,
      "step": 1736
    },
    {
      "epoch": 2.5436893203883493,
      "grad_norm": 0.40803281708973477,
      "learning_rate": 2.2392667930622105e-05,
      "loss": 0.6944,
      "num_tokens": 5305234427.0,
      "step": 1737
    },
    {
      "epoch": 2.5451547902546254,
      "grad_norm": 0.42172256957755233,
      "learning_rate": 2.2375597946533604e-05,
      "loss": 0.6916,
      "num_tokens": 5308139897.0,
      "step": 1738
    },
    {
      "epoch": 2.546620260120901,
      "grad_norm": 0.3851222393626624,
      "learning_rate": 2.2358527624503033e-05,
      "loss": 0.7006,
      "num_tokens": 5311130614.0,
      "step": 1739
    },
    {
      "epoch": 2.548085729987177,
      "grad_norm": 0.4159949505676457,
      "learning_rate": 2.234145697988932e-05,
      "loss": 0.694,
      "num_tokens": 5314286173.0,
      "step": 1740
    },
    {
      "epoch": 2.549551199853453,
      "grad_norm": 0.3953443808609083,
      "learning_rate": 2.2324386028051655e-05,
      "loss": 0.6774,
      "num_tokens": 5317176163.0,
      "step": 1741
    },
    {
      "epoch": 2.5510166697197287,
      "grad_norm": 0.35991533607107334,
      "learning_rate": 2.230731478434955e-05,
      "loss": 0.6973,
      "num_tokens": 5320262176.0,
      "step": 1742
    },
    {
      "epoch": 2.552482139586005,
      "grad_norm": 0.4835340817744485,
      "learning_rate": 2.229024326414275e-05,
      "loss": 0.6875,
      "num_tokens": 5323407802.0,
      "step": 1743
    },
    {
      "epoch": 2.5539476094522806,
      "grad_norm": 0.38014384561880427,
      "learning_rate": 2.227317148279125e-05,
      "loss": 0.7047,
      "num_tokens": 5326330881.0,
      "step": 1744
    },
    {
      "epoch": 2.5554130793185568,
      "grad_norm": 0.4086776774011518,
      "learning_rate": 2.2256099455655295e-05,
      "loss": 0.7168,
      "num_tokens": 5329366826.0,
      "step": 1745
    },
    {
      "epoch": 2.5568785491848325,
      "grad_norm": 0.4002383270441421,
      "learning_rate": 2.2239027198095344e-05,
      "loss": 0.6983,
      "num_tokens": 5332498326.0,
      "step": 1746
    },
    {
      "epoch": 2.558344019051108,
      "grad_norm": 0.41980371497316715,
      "learning_rate": 2.2221954725472053e-05,
      "loss": 0.6841,
      "num_tokens": 5335455858.0,
      "step": 1747
    },
    {
      "epoch": 2.5598094889173844,
      "grad_norm": 0.38530416294195546,
      "learning_rate": 2.2204882053146278e-05,
      "loss": 0.694,
      "num_tokens": 5338458041.0,
      "step": 1748
    },
    {
      "epoch": 2.56127495878366,
      "grad_norm": 0.4452669570570983,
      "learning_rate": 2.2187809196479076e-05,
      "loss": 0.6905,
      "num_tokens": 5341521467.0,
      "step": 1749
    },
    {
      "epoch": 2.562740428649936,
      "grad_norm": 0.3812132328470332,
      "learning_rate": 2.217073617083163e-05,
      "loss": 0.714,
      "num_tokens": 5344442054.0,
      "step": 1750
    },
    {
      "epoch": 2.564205898516212,
      "grad_norm": 0.45277523551930565,
      "learning_rate": 2.2153662991565313e-05,
      "loss": 0.6977,
      "num_tokens": 5347643496.0,
      "step": 1751
    },
    {
      "epoch": 2.5656713683824877,
      "grad_norm": 0.4420879986386117,
      "learning_rate": 2.2136589674041603e-05,
      "loss": 0.685,
      "num_tokens": 5350782441.0,
      "step": 1752
    },
    {
      "epoch": 2.5671368382487634,
      "grad_norm": 0.37851210046543343,
      "learning_rate": 2.2119516233622127e-05,
      "loss": 0.6912,
      "num_tokens": 5353838752.0,
      "step": 1753
    },
    {
      "epoch": 2.5686023081150395,
      "grad_norm": 0.4195134406106909,
      "learning_rate": 2.2102442685668612e-05,
      "loss": 0.7112,
      "num_tokens": 5356932787.0,
      "step": 1754
    },
    {
      "epoch": 2.5700677779813152,
      "grad_norm": 0.3928975146385568,
      "learning_rate": 2.2085369045542887e-05,
      "loss": 0.7135,
      "num_tokens": 5359720144.0,
      "step": 1755
    },
    {
      "epoch": 2.571533247847591,
      "grad_norm": 0.4119050901772562,
      "learning_rate": 2.206829532860686e-05,
      "loss": 0.7014,
      "num_tokens": 5362939231.0,
      "step": 1756
    },
    {
      "epoch": 2.572998717713867,
      "grad_norm": 0.3538439833749421,
      "learning_rate": 2.20512215502225e-05,
      "loss": 0.7209,
      "num_tokens": 5366051538.0,
      "step": 1757
    },
    {
      "epoch": 2.574464187580143,
      "grad_norm": 0.40435964804803937,
      "learning_rate": 2.2034147725751858e-05,
      "loss": 0.6833,
      "num_tokens": 5368924107.0,
      "step": 1758
    },
    {
      "epoch": 2.5759296574464186,
      "grad_norm": 0.38603606136999563,
      "learning_rate": 2.2017073870556986e-05,
      "loss": 0.7111,
      "num_tokens": 5371984696.0,
      "step": 1759
    },
    {
      "epoch": 2.5773951273126947,
      "grad_norm": 0.41242163844300006,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.7063,
      "num_tokens": 5374816634.0,
      "step": 1760
    },
    {
      "epoch": 2.5788605971789704,
      "grad_norm": 0.35169022483313644,
      "learning_rate": 2.1982926129443026e-05,
      "loss": 0.6735,
      "num_tokens": 5377841465.0,
      "step": 1761
    },
    {
      "epoch": 2.580326067045246,
      "grad_norm": 0.36207685864554656,
      "learning_rate": 2.1965852274248155e-05,
      "loss": 0.6847,
      "num_tokens": 5380959479.0,
      "step": 1762
    },
    {
      "epoch": 2.5817915369115223,
      "grad_norm": 0.37897119335917434,
      "learning_rate": 2.194877844977751e-05,
      "loss": 0.6991,
      "num_tokens": 5383841763.0,
      "step": 1763
    },
    {
      "epoch": 2.583257006777798,
      "grad_norm": 0.3815462715395564,
      "learning_rate": 2.1931704671393154e-05,
      "loss": 0.6963,
      "num_tokens": 5386711891.0,
      "step": 1764
    },
    {
      "epoch": 2.5847224766440737,
      "grad_norm": 0.352353635425442,
      "learning_rate": 2.1914630954457122e-05,
      "loss": 0.7225,
      "num_tokens": 5389859131.0,
      "step": 1765
    },
    {
      "epoch": 2.58618794651035,
      "grad_norm": 0.40112907509807566,
      "learning_rate": 2.1897557314331393e-05,
      "loss": 0.7244,
      "num_tokens": 5392961727.0,
      "step": 1766
    },
    {
      "epoch": 2.5876534163766256,
      "grad_norm": 0.4226198297504579,
      "learning_rate": 2.1880483766377886e-05,
      "loss": 0.6996,
      "num_tokens": 5395811060.0,
      "step": 1767
    },
    {
      "epoch": 2.5891188862429018,
      "grad_norm": 0.39483269218651457,
      "learning_rate": 2.1863410325958406e-05,
      "loss": 0.6961,
      "num_tokens": 5398810836.0,
      "step": 1768
    },
    {
      "epoch": 2.5905843561091775,
      "grad_norm": 0.4029231323130329,
      "learning_rate": 2.18463370084347e-05,
      "loss": 0.6979,
      "num_tokens": 5401936970.0,
      "step": 1769
    },
    {
      "epoch": 2.5920498259754536,
      "grad_norm": 0.37324675287376885,
      "learning_rate": 2.1829263829168372e-05,
      "loss": 0.6857,
      "num_tokens": 5404893409.0,
      "step": 1770
    },
    {
      "epoch": 2.5935152958417294,
      "grad_norm": 0.3943844612362309,
      "learning_rate": 2.1812190803520936e-05,
      "loss": 0.7134,
      "num_tokens": 5407687620.0,
      "step": 1771
    },
    {
      "epoch": 2.594980765708005,
      "grad_norm": 0.3984205688295169,
      "learning_rate": 2.1795117946853724e-05,
      "loss": 0.7132,
      "num_tokens": 5410805330.0,
      "step": 1772
    },
    {
      "epoch": 2.5964462355742812,
      "grad_norm": 0.3795056158888877,
      "learning_rate": 2.177804527452796e-05,
      "loss": 0.7003,
      "num_tokens": 5413818981.0,
      "step": 1773
    },
    {
      "epoch": 2.597911705440557,
      "grad_norm": 0.3540162489850375,
      "learning_rate": 2.1760972801904666e-05,
      "loss": 0.7204,
      "num_tokens": 5416739527.0,
      "step": 1774
    },
    {
      "epoch": 2.5993771753068327,
      "grad_norm": 0.3900492395667007,
      "learning_rate": 2.1743900544344718e-05,
      "loss": 0.69,
      "num_tokens": 5419965309.0,
      "step": 1775
    },
    {
      "epoch": 2.600842645173109,
      "grad_norm": 0.35087922048268094,
      "learning_rate": 2.1726828517208753e-05,
      "loss": 0.7029,
      "num_tokens": 5422932178.0,
      "step": 1776
    },
    {
      "epoch": 2.6023081150393845,
      "grad_norm": 0.3462316482646767,
      "learning_rate": 2.1709756735857264e-05,
      "loss": 0.7047,
      "num_tokens": 5426243883.0,
      "step": 1777
    },
    {
      "epoch": 2.6037735849056602,
      "grad_norm": 0.37080165293383144,
      "learning_rate": 2.1692685215650457e-05,
      "loss": 0.6986,
      "num_tokens": 5429327654.0,
      "step": 1778
    },
    {
      "epoch": 2.6052390547719364,
      "grad_norm": 0.3509337018028498,
      "learning_rate": 2.167561397194835e-05,
      "loss": 0.6917,
      "num_tokens": 5432430222.0,
      "step": 1779
    },
    {
      "epoch": 2.606704524638212,
      "grad_norm": 0.3808514092369401,
      "learning_rate": 2.1658543020110687e-05,
      "loss": 0.692,
      "num_tokens": 5435391517.0,
      "step": 1780
    },
    {
      "epoch": 2.608169994504488,
      "grad_norm": 0.3866933141864262,
      "learning_rate": 2.1641472375496973e-05,
      "loss": 0.7037,
      "num_tokens": 5438286691.0,
      "step": 1781
    },
    {
      "epoch": 2.609635464370764,
      "grad_norm": 0.39788507690008107,
      "learning_rate": 2.16244020534664e-05,
      "loss": 0.7102,
      "num_tokens": 5441318897.0,
      "step": 1782
    },
    {
      "epoch": 2.6111009342370397,
      "grad_norm": 0.4216778813061623,
      "learning_rate": 2.1607332069377904e-05,
      "loss": 0.6946,
      "num_tokens": 5444477122.0,
      "step": 1783
    },
    {
      "epoch": 2.6125664041033154,
      "grad_norm": 0.3993955932403623,
      "learning_rate": 2.15902624385901e-05,
      "loss": 0.6913,
      "num_tokens": 5447574224.0,
      "step": 1784
    },
    {
      "epoch": 2.6140318739695916,
      "grad_norm": 0.4038131848765513,
      "learning_rate": 2.15731931764613e-05,
      "loss": 0.7098,
      "num_tokens": 5450607876.0,
      "step": 1785
    },
    {
      "epoch": 2.6154973438358673,
      "grad_norm": 0.44514487786936924,
      "learning_rate": 2.1556124298349454e-05,
      "loss": 0.714,
      "num_tokens": 5453702027.0,
      "step": 1786
    },
    {
      "epoch": 2.616962813702143,
      "grad_norm": 0.4219765937577303,
      "learning_rate": 2.1539055819612202e-05,
      "loss": 0.6972,
      "num_tokens": 5456774227.0,
      "step": 1787
    },
    {
      "epoch": 2.618428283568419,
      "grad_norm": 0.40362396332004413,
      "learning_rate": 2.1521987755606795e-05,
      "loss": 0.7139,
      "num_tokens": 5459875307.0,
      "step": 1788
    },
    {
      "epoch": 2.619893753434695,
      "grad_norm": 0.4551726443459678,
      "learning_rate": 2.1504920121690147e-05,
      "loss": 0.6956,
      "num_tokens": 5462919027.0,
      "step": 1789
    },
    {
      "epoch": 2.6213592233009706,
      "grad_norm": 0.4075526551445301,
      "learning_rate": 2.148785293321874e-05,
      "loss": 0.7113,
      "num_tokens": 5466043971.0,
      "step": 1790
    },
    {
      "epoch": 2.6228246931672468,
      "grad_norm": 0.4018130600523857,
      "learning_rate": 2.1470786205548692e-05,
      "loss": 0.7056,
      "num_tokens": 5468859575.0,
      "step": 1791
    },
    {
      "epoch": 2.6242901630335225,
      "grad_norm": 0.45756838576783526,
      "learning_rate": 2.145371995403568e-05,
      "loss": 0.7036,
      "num_tokens": 5472004094.0,
      "step": 1792
    },
    {
      "epoch": 2.6257556328997986,
      "grad_norm": 0.3928693030898483,
      "learning_rate": 2.143665419403498e-05,
      "loss": 0.6883,
      "num_tokens": 5474896038.0,
      "step": 1793
    },
    {
      "epoch": 2.6272211027660743,
      "grad_norm": 0.41922446579264494,
      "learning_rate": 2.1419588940901394e-05,
      "loss": 0.7026,
      "num_tokens": 5478119123.0,
      "step": 1794
    },
    {
      "epoch": 2.6286865726323505,
      "grad_norm": 0.3672995555651936,
      "learning_rate": 2.14025242099893e-05,
      "loss": 0.6952,
      "num_tokens": 5481086729.0,
      "step": 1795
    },
    {
      "epoch": 2.630152042498626,
      "grad_norm": 0.4107603458910209,
      "learning_rate": 2.138546001665258e-05,
      "loss": 0.711,
      "num_tokens": 5484401277.0,
      "step": 1796
    },
    {
      "epoch": 2.631617512364902,
      "grad_norm": 0.37399207291427383,
      "learning_rate": 2.1368396376244652e-05,
      "loss": 0.6819,
      "num_tokens": 5487447872.0,
      "step": 1797
    },
    {
      "epoch": 2.633082982231178,
      "grad_norm": 0.3861560819726002,
      "learning_rate": 2.1351333304118415e-05,
      "loss": 0.6765,
      "num_tokens": 5490382272.0,
      "step": 1798
    },
    {
      "epoch": 2.634548452097454,
      "grad_norm": 0.3865982944097399,
      "learning_rate": 2.133427081562628e-05,
      "loss": 0.7088,
      "num_tokens": 5493473157.0,
      "step": 1799
    },
    {
      "epoch": 2.6360139219637295,
      "grad_norm": 0.3922135141945683,
      "learning_rate": 2.1317208926120117e-05,
      "loss": 0.6707,
      "num_tokens": 5496846991.0,
      "step": 1800
    },
    {
      "epoch": 2.6374793918300057,
      "grad_norm": 0.4133871026661409,
      "learning_rate": 2.1300147650951266e-05,
      "loss": 0.7022,
      "num_tokens": 5499843470.0,
      "step": 1801
    },
    {
      "epoch": 2.6389448616962814,
      "grad_norm": 0.39347843996125587,
      "learning_rate": 2.1283087005470512e-05,
      "loss": 0.6822,
      "num_tokens": 5503094514.0,
      "step": 1802
    },
    {
      "epoch": 2.640410331562557,
      "grad_norm": 0.35761310039194905,
      "learning_rate": 2.1266027005028062e-05,
      "loss": 0.6841,
      "num_tokens": 5506150519.0,
      "step": 1803
    },
    {
      "epoch": 2.6418758014288333,
      "grad_norm": 0.35598576319078123,
      "learning_rate": 2.124896766497356e-05,
      "loss": 0.6694,
      "num_tokens": 5509417285.0,
      "step": 1804
    },
    {
      "epoch": 2.643341271295109,
      "grad_norm": 0.4052146642496105,
      "learning_rate": 2.1231909000656044e-05,
      "loss": 0.6916,
      "num_tokens": 5512551490.0,
      "step": 1805
    },
    {
      "epoch": 2.6448067411613847,
      "grad_norm": 0.3494288417198552,
      "learning_rate": 2.1214851027423954e-05,
      "loss": 0.7086,
      "num_tokens": 5515598834.0,
      "step": 1806
    },
    {
      "epoch": 2.646272211027661,
      "grad_norm": 0.39636487396000797,
      "learning_rate": 2.1197793760625097e-05,
      "loss": 0.6926,
      "num_tokens": 5518888111.0,
      "step": 1807
    },
    {
      "epoch": 2.6477376808939366,
      "grad_norm": 0.353916515149785,
      "learning_rate": 2.118073721560666e-05,
      "loss": 0.7119,
      "num_tokens": 5522004678.0,
      "step": 1808
    },
    {
      "epoch": 2.6492031507602123,
      "grad_norm": 0.3332388046949698,
      "learning_rate": 2.1163681407715155e-05,
      "loss": 0.6994,
      "num_tokens": 5525014561.0,
      "step": 1809
    },
    {
      "epoch": 2.6506686206264884,
      "grad_norm": 0.3926047733788625,
      "learning_rate": 2.114662635229646e-05,
      "loss": 0.68,
      "num_tokens": 5528249058.0,
      "step": 1810
    },
    {
      "epoch": 2.652134090492764,
      "grad_norm": 0.32867142274955646,
      "learning_rate": 2.112957206469575e-05,
      "loss": 0.7051,
      "num_tokens": 5531216271.0,
      "step": 1811
    },
    {
      "epoch": 2.65359956035904,
      "grad_norm": 0.3901645980901081,
      "learning_rate": 2.1112518560257536e-05,
      "loss": 0.6957,
      "num_tokens": 5534103824.0,
      "step": 1812
    },
    {
      "epoch": 2.655065030225316,
      "grad_norm": 0.3529613171246676,
      "learning_rate": 2.1095465854325595e-05,
      "loss": 0.7047,
      "num_tokens": 5537041946.0,
      "step": 1813
    },
    {
      "epoch": 2.6565305000915918,
      "grad_norm": 0.4390300160654016,
      "learning_rate": 2.107841396224301e-05,
      "loss": 0.7053,
      "num_tokens": 5540194317.0,
      "step": 1814
    },
    {
      "epoch": 2.6579959699578675,
      "grad_norm": 0.33388418473388254,
      "learning_rate": 2.1061362899352113e-05,
      "loss": 0.6875,
      "num_tokens": 5543143168.0,
      "step": 1815
    },
    {
      "epoch": 2.6594614398241436,
      "grad_norm": 0.4324288965771675,
      "learning_rate": 2.1044312680994517e-05,
      "loss": 0.697,
      "num_tokens": 5546325640.0,
      "step": 1816
    },
    {
      "epoch": 2.6609269096904193,
      "grad_norm": 0.36462957981234934,
      "learning_rate": 2.1027263322511034e-05,
      "loss": 0.6901,
      "num_tokens": 5549427067.0,
      "step": 1817
    },
    {
      "epoch": 2.6623923795566955,
      "grad_norm": 0.49675386738857186,
      "learning_rate": 2.101021483924174e-05,
      "loss": 0.69,
      "num_tokens": 5552496473.0,
      "step": 1818
    },
    {
      "epoch": 2.663857849422971,
      "grad_norm": 0.3456516159140831,
      "learning_rate": 2.09931672465259e-05,
      "loss": 0.7108,
      "num_tokens": 5555449379.0,
      "step": 1819
    },
    {
      "epoch": 2.6653233192892474,
      "grad_norm": 0.44287645498830064,
      "learning_rate": 2.0976120559702e-05,
      "loss": 0.6944,
      "num_tokens": 5558723986.0,
      "step": 1820
    },
    {
      "epoch": 2.666788789155523,
      "grad_norm": 0.37872291517836526,
      "learning_rate": 2.0959074794107685e-05,
      "loss": 0.7063,
      "num_tokens": 5561709036.0,
      "step": 1821
    },
    {
      "epoch": 2.668254259021799,
      "grad_norm": 0.40924023003541127,
      "learning_rate": 2.0942029965079795e-05,
      "loss": 0.6852,
      "num_tokens": 5564909277.0,
      "step": 1822
    },
    {
      "epoch": 2.669719728888075,
      "grad_norm": 0.35888719145064746,
      "learning_rate": 2.09249860879543e-05,
      "loss": 0.6826,
      "num_tokens": 5567730994.0,
      "step": 1823
    },
    {
      "epoch": 2.6711851987543507,
      "grad_norm": 0.41781673870284436,
      "learning_rate": 2.0907943178066347e-05,
      "loss": 0.7062,
      "num_tokens": 5570652587.0,
      "step": 1824
    },
    {
      "epoch": 2.6726506686206264,
      "grad_norm": 0.3846080508440028,
      "learning_rate": 2.0890901250750182e-05,
      "loss": 0.6894,
      "num_tokens": 5573812149.0,
      "step": 1825
    },
    {
      "epoch": 2.6741161384869025,
      "grad_norm": 0.3963136899058783,
      "learning_rate": 2.0873860321339198e-05,
      "loss": 0.7051,
      "num_tokens": 5576743662.0,
      "step": 1826
    },
    {
      "epoch": 2.6755816083531783,
      "grad_norm": 0.3377951819800646,
      "learning_rate": 2.085682040516585e-05,
      "loss": 0.6921,
      "num_tokens": 5579923593.0,
      "step": 1827
    },
    {
      "epoch": 2.677047078219454,
      "grad_norm": 0.40164867568003276,
      "learning_rate": 2.083978151756171e-05,
      "loss": 0.6812,
      "num_tokens": 5582995374.0,
      "step": 1828
    },
    {
      "epoch": 2.67851254808573,
      "grad_norm": 0.37262522384475627,
      "learning_rate": 2.0822743673857424e-05,
      "loss": 0.6843,
      "num_tokens": 5586071446.0,
      "step": 1829
    },
    {
      "epoch": 2.679978017952006,
      "grad_norm": 0.3707030834020138,
      "learning_rate": 2.08057068893827e-05,
      "loss": 0.7068,
      "num_tokens": 5589461796.0,
      "step": 1830
    },
    {
      "epoch": 2.6814434878182816,
      "grad_norm": 0.3968189777496934,
      "learning_rate": 2.078867117946627e-05,
      "loss": 0.7003,
      "num_tokens": 5592591289.0,
      "step": 1831
    },
    {
      "epoch": 2.6829089576845577,
      "grad_norm": 0.36790201680087165,
      "learning_rate": 2.0771636559435925e-05,
      "loss": 0.6868,
      "num_tokens": 5595949016.0,
      "step": 1832
    },
    {
      "epoch": 2.6843744275508334,
      "grad_norm": 0.33842322862070695,
      "learning_rate": 2.075460304461846e-05,
      "loss": 0.7103,
      "num_tokens": 5599161111.0,
      "step": 1833
    },
    {
      "epoch": 2.685839897417109,
      "grad_norm": 0.4620805082759224,
      "learning_rate": 2.0737570650339697e-05,
      "loss": 0.7053,
      "num_tokens": 5602338595.0,
      "step": 1834
    },
    {
      "epoch": 2.6873053672833853,
      "grad_norm": 0.38798298967771305,
      "learning_rate": 2.0720539391924404e-05,
      "loss": 0.6827,
      "num_tokens": 5605594684.0,
      "step": 1835
    },
    {
      "epoch": 2.688770837149661,
      "grad_norm": 0.41064092560633614,
      "learning_rate": 2.0703509284696387e-05,
      "loss": 0.7185,
      "num_tokens": 5608642184.0,
      "step": 1836
    },
    {
      "epoch": 2.6902363070159367,
      "grad_norm": 0.3895576422553758,
      "learning_rate": 2.0686480343978367e-05,
      "loss": 0.7058,
      "num_tokens": 5611508375.0,
      "step": 1837
    },
    {
      "epoch": 2.691701776882213,
      "grad_norm": 0.33863716781512265,
      "learning_rate": 2.0669452585092045e-05,
      "loss": 0.6907,
      "num_tokens": 5614632883.0,
      "step": 1838
    },
    {
      "epoch": 2.6931672467484886,
      "grad_norm": 0.38078708994935273,
      "learning_rate": 2.065242602335804e-05,
      "loss": 0.7108,
      "num_tokens": 5617714705.0,
      "step": 1839
    },
    {
      "epoch": 2.6946327166147643,
      "grad_norm": 0.37187279184922245,
      "learning_rate": 2.0635400674095917e-05,
      "loss": 0.6899,
      "num_tokens": 5620987663.0,
      "step": 1840
    },
    {
      "epoch": 2.6960981864810405,
      "grad_norm": 0.4285827455168878,
      "learning_rate": 2.0618376552624112e-05,
      "loss": 0.6825,
      "num_tokens": 5623945959.0,
      "step": 1841
    },
    {
      "epoch": 2.697563656347316,
      "grad_norm": 0.3398789045751483,
      "learning_rate": 2.060135367426e-05,
      "loss": 0.7081,
      "num_tokens": 5626818341.0,
      "step": 1842
    },
    {
      "epoch": 2.6990291262135924,
      "grad_norm": 0.4522142093140537,
      "learning_rate": 2.0584332054319813e-05,
      "loss": 0.6995,
      "num_tokens": 5629863945.0,
      "step": 1843
    },
    {
      "epoch": 2.700494596079868,
      "grad_norm": 0.3718638179894877,
      "learning_rate": 2.0567311708118648e-05,
      "loss": 0.6896,
      "num_tokens": 5632785397.0,
      "step": 1844
    },
    {
      "epoch": 2.7019600659461442,
      "grad_norm": 0.47562345054329325,
      "learning_rate": 2.0550292650970465e-05,
      "loss": 0.716,
      "num_tokens": 5635752125.0,
      "step": 1845
    },
    {
      "epoch": 2.70342553581242,
      "grad_norm": 0.34135866770631185,
      "learning_rate": 2.0533274898188063e-05,
      "loss": 0.7054,
      "num_tokens": 5638668172.0,
      "step": 1846
    },
    {
      "epoch": 2.7048910056786957,
      "grad_norm": 0.4531980762127052,
      "learning_rate": 2.0516258465083078e-05,
      "loss": 0.6767,
      "num_tokens": 5641784699.0,
      "step": 1847
    },
    {
      "epoch": 2.706356475544972,
      "grad_norm": 0.3765670349169559,
      "learning_rate": 2.049924336696592e-05,
      "loss": 0.6985,
      "num_tokens": 5644911110.0,
      "step": 1848
    },
    {
      "epoch": 2.7078219454112475,
      "grad_norm": 0.4503306914335617,
      "learning_rate": 2.048222961914585e-05,
      "loss": 0.7102,
      "num_tokens": 5647943900.0,
      "step": 1849
    },
    {
      "epoch": 2.7092874152775233,
      "grad_norm": 0.4077906759209094,
      "learning_rate": 2.0465217236930872e-05,
      "loss": 0.6954,
      "num_tokens": 5650905017.0,
      "step": 1850
    },
    {
      "epoch": 2.7107528851437994,
      "grad_norm": 0.39058169235816986,
      "learning_rate": 2.0448206235627793e-05,
      "loss": 0.6723,
      "num_tokens": 5654030374.0,
      "step": 1851
    },
    {
      "epoch": 2.712218355010075,
      "grad_norm": 0.35262572872567305,
      "learning_rate": 2.0431196630542152e-05,
      "loss": 0.6993,
      "num_tokens": 5656926120.0,
      "step": 1852
    },
    {
      "epoch": 2.713683824876351,
      "grad_norm": 0.4320281864467972,
      "learning_rate": 2.041418843697826e-05,
      "loss": 0.7013,
      "num_tokens": 5660071546.0,
      "step": 1853
    },
    {
      "epoch": 2.715149294742627,
      "grad_norm": 0.34058345320231265,
      "learning_rate": 2.039718167023911e-05,
      "loss": 0.6831,
      "num_tokens": 5663124417.0,
      "step": 1854
    },
    {
      "epoch": 2.7166147646089027,
      "grad_norm": 0.42107414267544746,
      "learning_rate": 2.0380176345626464e-05,
      "loss": 0.7094,
      "num_tokens": 5666231747.0,
      "step": 1855
    },
    {
      "epoch": 2.7180802344751784,
      "grad_norm": 0.34553029213117037,
      "learning_rate": 2.0363172478440755e-05,
      "loss": 0.6903,
      "num_tokens": 5669198435.0,
      "step": 1856
    },
    {
      "epoch": 2.7195457043414546,
      "grad_norm": 0.47405852912933555,
      "learning_rate": 2.0346170083981128e-05,
      "loss": 0.7004,
      "num_tokens": 5672072496.0,
      "step": 1857
    },
    {
      "epoch": 2.7210111742077303,
      "grad_norm": 0.37886933445204674,
      "learning_rate": 2.032916917754537e-05,
      "loss": 0.6853,
      "num_tokens": 5675283184.0,
      "step": 1858
    },
    {
      "epoch": 2.722476644074006,
      "grad_norm": 0.4496114130762444,
      "learning_rate": 2.0312169774429962e-05,
      "loss": 0.6776,
      "num_tokens": 5678586334.0,
      "step": 1859
    },
    {
      "epoch": 2.723942113940282,
      "grad_norm": 0.4045274483148738,
      "learning_rate": 2.029517188993e-05,
      "loss": 0.7075,
      "num_tokens": 5681679719.0,
      "step": 1860
    },
    {
      "epoch": 2.725407583806558,
      "grad_norm": 0.46062080862925814,
      "learning_rate": 2.0278175539339268e-05,
      "loss": 0.6849,
      "num_tokens": 5684783448.0,
      "step": 1861
    },
    {
      "epoch": 2.7268730536728336,
      "grad_norm": 0.3912130015570455,
      "learning_rate": 2.0261180737950093e-05,
      "loss": 0.6839,
      "num_tokens": 5687697997.0,
      "step": 1862
    },
    {
      "epoch": 2.7283385235391098,
      "grad_norm": 0.4263455334127441,
      "learning_rate": 2.0244187501053473e-05,
      "loss": 0.6986,
      "num_tokens": 5690641954.0,
      "step": 1863
    },
    {
      "epoch": 2.7298039934053855,
      "grad_norm": 0.42211758936665417,
      "learning_rate": 2.0227195843938964e-05,
      "loss": 0.7131,
      "num_tokens": 5693682210.0,
      "step": 1864
    },
    {
      "epoch": 2.731269463271661,
      "grad_norm": 0.42200797281752933,
      "learning_rate": 2.021020578189472e-05,
      "loss": 0.7104,
      "num_tokens": 5696710995.0,
      "step": 1865
    },
    {
      "epoch": 2.7327349331379374,
      "grad_norm": 0.4083485467131711,
      "learning_rate": 2.0193217330207444e-05,
      "loss": 0.6877,
      "num_tokens": 5699915350.0,
      "step": 1866
    },
    {
      "epoch": 2.734200403004213,
      "grad_norm": 0.4368797719595614,
      "learning_rate": 2.0176230504162406e-05,
      "loss": 0.7094,
      "num_tokens": 5702878393.0,
      "step": 1867
    },
    {
      "epoch": 2.7356658728704892,
      "grad_norm": 0.35656118083370464,
      "learning_rate": 2.0159245319043384e-05,
      "loss": 0.697,
      "num_tokens": 5705795546.0,
      "step": 1868
    },
    {
      "epoch": 2.737131342736765,
      "grad_norm": 0.3790384057967855,
      "learning_rate": 2.014226179013272e-05,
      "loss": 0.679,
      "num_tokens": 5708817354.0,
      "step": 1869
    },
    {
      "epoch": 2.738596812603041,
      "grad_norm": 0.3697500515685886,
      "learning_rate": 2.0125279932711236e-05,
      "loss": 0.6821,
      "num_tokens": 5711896696.0,
      "step": 1870
    },
    {
      "epoch": 2.740062282469317,
      "grad_norm": 0.3864257412733465,
      "learning_rate": 2.0108299762058272e-05,
      "loss": 0.7067,
      "num_tokens": 5715207251.0,
      "step": 1871
    },
    {
      "epoch": 2.7415277523355925,
      "grad_norm": 0.3450370998987545,
      "learning_rate": 2.009132129345162e-05,
      "loss": 0.6932,
      "num_tokens": 5718113176.0,
      "step": 1872
    },
    {
      "epoch": 2.7429932222018687,
      "grad_norm": 0.4143403147958288,
      "learning_rate": 2.007434454216757e-05,
      "loss": 0.6894,
      "num_tokens": 5721111377.0,
      "step": 1873
    },
    {
      "epoch": 2.7444586920681444,
      "grad_norm": 0.36533215122779955,
      "learning_rate": 2.0057369523480846e-05,
      "loss": 0.6892,
      "num_tokens": 5724135704.0,
      "step": 1874
    },
    {
      "epoch": 2.74592416193442,
      "grad_norm": 0.43584583309456276,
      "learning_rate": 2.0040396252664644e-05,
      "loss": 0.6832,
      "num_tokens": 5727201969.0,
      "step": 1875
    },
    {
      "epoch": 2.7473896318006963,
      "grad_norm": 0.37863836415657054,
      "learning_rate": 2.002342474499054e-05,
      "loss": 0.6821,
      "num_tokens": 5730277975.0,
      "step": 1876
    },
    {
      "epoch": 2.748855101666972,
      "grad_norm": 0.40322304847383567,
      "learning_rate": 2.0006455015728564e-05,
      "loss": 0.7056,
      "num_tokens": 5733137079.0,
      "step": 1877
    },
    {
      "epoch": 2.7503205715332477,
      "grad_norm": 0.35187076225357466,
      "learning_rate": 1.998948708014712e-05,
      "loss": 0.689,
      "num_tokens": 5736379579.0,
      "step": 1878
    },
    {
      "epoch": 2.751786041399524,
      "grad_norm": 0.4319404264863892,
      "learning_rate": 1.997252095351303e-05,
      "loss": 0.6988,
      "num_tokens": 5739469536.0,
      "step": 1879
    },
    {
      "epoch": 2.7532515112657996,
      "grad_norm": 0.38328696671974516,
      "learning_rate": 1.9955556651091454e-05,
      "loss": 0.6815,
      "num_tokens": 5742433345.0,
      "step": 1880
    },
    {
      "epoch": 2.7547169811320753,
      "grad_norm": 0.45781096018396555,
      "learning_rate": 1.993859418814593e-05,
      "loss": 0.7084,
      "num_tokens": 5745325060.0,
      "step": 1881
    },
    {
      "epoch": 2.7561824509983515,
      "grad_norm": 0.3699073595019125,
      "learning_rate": 1.9921633579938322e-05,
      "loss": 0.6812,
      "num_tokens": 5748362798.0,
      "step": 1882
    },
    {
      "epoch": 2.757647920864627,
      "grad_norm": 0.41551368502428265,
      "learning_rate": 1.9904674841728856e-05,
      "loss": 0.7089,
      "num_tokens": 5751412711.0,
      "step": 1883
    },
    {
      "epoch": 2.759113390730903,
      "grad_norm": 0.37668531385344006,
      "learning_rate": 1.988771798877605e-05,
      "loss": 0.6855,
      "num_tokens": 5754600002.0,
      "step": 1884
    },
    {
      "epoch": 2.760578860597179,
      "grad_norm": 0.4048852438805406,
      "learning_rate": 1.9870763036336743e-05,
      "loss": 0.7192,
      "num_tokens": 5757610235.0,
      "step": 1885
    },
    {
      "epoch": 2.7620443304634548,
      "grad_norm": 0.39428828621174894,
      "learning_rate": 1.9853809999666046e-05,
      "loss": 0.6845,
      "num_tokens": 5760854231.0,
      "step": 1886
    },
    {
      "epoch": 2.7635098003297305,
      "grad_norm": 0.4023068092442269,
      "learning_rate": 1.983685889401735e-05,
      "loss": 0.6885,
      "num_tokens": 5763905463.0,
      "step": 1887
    },
    {
      "epoch": 2.7649752701960066,
      "grad_norm": 0.3736588372317174,
      "learning_rate": 1.9819909734642332e-05,
      "loss": 0.7003,
      "num_tokens": 5766912652.0,
      "step": 1888
    },
    {
      "epoch": 2.7664407400622824,
      "grad_norm": 0.4326364510277143,
      "learning_rate": 1.980296253679088e-05,
      "loss": 0.6911,
      "num_tokens": 5769856498.0,
      "step": 1889
    },
    {
      "epoch": 2.767906209928558,
      "grad_norm": 0.37565820521286686,
      "learning_rate": 1.9786017315711158e-05,
      "loss": 0.6775,
      "num_tokens": 5772896066.0,
      "step": 1890
    },
    {
      "epoch": 2.7693716797948342,
      "grad_norm": 0.43593184614905583,
      "learning_rate": 1.9769074086649508e-05,
      "loss": 0.705,
      "num_tokens": 5775979870.0,
      "step": 1891
    },
    {
      "epoch": 2.77083714966111,
      "grad_norm": 0.35682828413699147,
      "learning_rate": 1.975213286485052e-05,
      "loss": 0.6851,
      "num_tokens": 5778932355.0,
      "step": 1892
    },
    {
      "epoch": 2.772302619527386,
      "grad_norm": 0.38432362196373404,
      "learning_rate": 1.973519366555694e-05,
      "loss": 0.6881,
      "num_tokens": 5782202904.0,
      "step": 1893
    },
    {
      "epoch": 2.773768089393662,
      "grad_norm": 0.35040578237667497,
      "learning_rate": 1.9718256504009744e-05,
      "loss": 0.6661,
      "num_tokens": 5785224805.0,
      "step": 1894
    },
    {
      "epoch": 2.775233559259938,
      "grad_norm": 0.37705310585506896,
      "learning_rate": 1.9701321395448022e-05,
      "loss": 0.6659,
      "num_tokens": 5788305409.0,
      "step": 1895
    },
    {
      "epoch": 2.7766990291262137,
      "grad_norm": 0.36256851603745943,
      "learning_rate": 1.9684388355109044e-05,
      "loss": 0.6916,
      "num_tokens": 5791701268.0,
      "step": 1896
    },
    {
      "epoch": 2.7781644989924894,
      "grad_norm": 0.40394284613345866,
      "learning_rate": 1.9667457398228214e-05,
      "loss": 0.6936,
      "num_tokens": 5794655552.0,
      "step": 1897
    },
    {
      "epoch": 2.7796299688587656,
      "grad_norm": 0.3346330967971263,
      "learning_rate": 1.9650528540039077e-05,
      "loss": 0.6799,
      "num_tokens": 5797605044.0,
      "step": 1898
    },
    {
      "epoch": 2.7810954387250413,
      "grad_norm": 0.43005260249305177,
      "learning_rate": 1.9633601795773255e-05,
      "loss": 0.6869,
      "num_tokens": 5800634529.0,
      "step": 1899
    },
    {
      "epoch": 2.782560908591317,
      "grad_norm": 0.3601654604379806,
      "learning_rate": 1.9616677180660498e-05,
      "loss": 0.6967,
      "num_tokens": 5803531440.0,
      "step": 1900
    },
    {
      "epoch": 2.784026378457593,
      "grad_norm": 0.37389592472348404,
      "learning_rate": 1.9599754709928626e-05,
      "loss": 0.6903,
      "num_tokens": 5806750210.0,
      "step": 1901
    },
    {
      "epoch": 2.785491848323869,
      "grad_norm": 0.3291190453167584,
      "learning_rate": 1.9582834398803543e-05,
      "loss": 0.6988,
      "num_tokens": 5809871378.0,
      "step": 1902
    },
    {
      "epoch": 2.7869573181901446,
      "grad_norm": 0.4093555448024012,
      "learning_rate": 1.9565916262509187e-05,
      "loss": 0.6841,
      "num_tokens": 5812906445.0,
      "step": 1903
    },
    {
      "epoch": 2.7884227880564207,
      "grad_norm": 0.33756077789977046,
      "learning_rate": 1.954900031626757e-05,
      "loss": 0.6983,
      "num_tokens": 5816048697.0,
      "step": 1904
    },
    {
      "epoch": 2.7898882579226965,
      "grad_norm": 0.40976163975175744,
      "learning_rate": 1.9532086575298693e-05,
      "loss": 0.685,
      "num_tokens": 5819070232.0,
      "step": 1905
    },
    {
      "epoch": 2.791353727788972,
      "grad_norm": 0.35350587358661595,
      "learning_rate": 1.9515175054820627e-05,
      "loss": 0.6828,
      "num_tokens": 5822151339.0,
      "step": 1906
    },
    {
      "epoch": 2.7928191976552483,
      "grad_norm": 0.36815085340035886,
      "learning_rate": 1.949826577004938e-05,
      "loss": 0.6746,
      "num_tokens": 5824986913.0,
      "step": 1907
    },
    {
      "epoch": 2.794284667521524,
      "grad_norm": 0.35416890338570767,
      "learning_rate": 1.9481358736199013e-05,
      "loss": 0.6866,
      "num_tokens": 5828083147.0,
      "step": 1908
    },
    {
      "epoch": 2.7957501373877998,
      "grad_norm": 0.4006922627003764,
      "learning_rate": 1.946445396848151e-05,
      "loss": 0.6878,
      "num_tokens": 5830979362.0,
      "step": 1909
    },
    {
      "epoch": 2.797215607254076,
      "grad_norm": 0.32327354560521077,
      "learning_rate": 1.9447551482106853e-05,
      "loss": 0.6986,
      "num_tokens": 5834092904.0,
      "step": 1910
    },
    {
      "epoch": 2.7986810771203516,
      "grad_norm": 0.3705055243280932,
      "learning_rate": 1.943065129228295e-05,
      "loss": 0.6937,
      "num_tokens": 5837060766.0,
      "step": 1911
    },
    {
      "epoch": 2.8001465469866273,
      "grad_norm": 0.38551516967273763,
      "learning_rate": 1.9413753414215657e-05,
      "loss": 0.7102,
      "num_tokens": 5840018691.0,
      "step": 1912
    },
    {
      "epoch": 2.8016120168529035,
      "grad_norm": 0.35223400051176984,
      "learning_rate": 1.9396857863108724e-05,
      "loss": 0.7185,
      "num_tokens": 5842919140.0,
      "step": 1913
    },
    {
      "epoch": 2.8030774867191792,
      "grad_norm": 0.35570064466087653,
      "learning_rate": 1.937996465416384e-05,
      "loss": 0.6994,
      "num_tokens": 5845823397.0,
      "step": 1914
    },
    {
      "epoch": 2.804542956585455,
      "grad_norm": 0.3761013512147621,
      "learning_rate": 1.9363073802580557e-05,
      "loss": 0.7083,
      "num_tokens": 5848883525.0,
      "step": 1915
    },
    {
      "epoch": 2.806008426451731,
      "grad_norm": 0.363491056932675,
      "learning_rate": 1.9346185323556344e-05,
      "loss": 0.6864,
      "num_tokens": 5852063416.0,
      "step": 1916
    },
    {
      "epoch": 2.807473896318007,
      "grad_norm": 0.3532499632175771,
      "learning_rate": 1.932929923228649e-05,
      "loss": 0.6975,
      "num_tokens": 5855151078.0,
      "step": 1917
    },
    {
      "epoch": 2.808939366184283,
      "grad_norm": 0.346419028538593,
      "learning_rate": 1.931241554396417e-05,
      "loss": 0.7022,
      "num_tokens": 5858223609.0,
      "step": 1918
    },
    {
      "epoch": 2.8104048360505587,
      "grad_norm": 0.3598536867518936,
      "learning_rate": 1.9295534273780373e-05,
      "loss": 0.6932,
      "num_tokens": 5861434296.0,
      "step": 1919
    },
    {
      "epoch": 2.811870305916835,
      "grad_norm": 0.33538570146648855,
      "learning_rate": 1.9278655436923944e-05,
      "loss": 0.6935,
      "num_tokens": 5864616857.0,
      "step": 1920
    },
    {
      "epoch": 2.8133357757831106,
      "grad_norm": 0.3407397523661818,
      "learning_rate": 1.92617790485815e-05,
      "loss": 0.7038,
      "num_tokens": 5867683769.0,
      "step": 1921
    },
    {
      "epoch": 2.8148012456493863,
      "grad_norm": 0.3474651121837004,
      "learning_rate": 1.9244905123937484e-05,
      "loss": 0.6706,
      "num_tokens": 5870808683.0,
      "step": 1922
    },
    {
      "epoch": 2.8162667155156624,
      "grad_norm": 0.3535642898357413,
      "learning_rate": 1.9228033678174107e-05,
      "loss": 0.668,
      "num_tokens": 5874097110.0,
      "step": 1923
    },
    {
      "epoch": 2.817732185381938,
      "grad_norm": 0.40673062592993187,
      "learning_rate": 1.921116472647135e-05,
      "loss": 0.6682,
      "num_tokens": 5877249775.0,
      "step": 1924
    },
    {
      "epoch": 2.819197655248214,
      "grad_norm": 0.3230262762962952,
      "learning_rate": 1.9194298284006964e-05,
      "loss": 0.6922,
      "num_tokens": 5880316687.0,
      "step": 1925
    },
    {
      "epoch": 2.82066312511449,
      "grad_norm": 0.4042774708173968,
      "learning_rate": 1.917743436595643e-05,
      "loss": 0.6857,
      "num_tokens": 5883403526.0,
      "step": 1926
    },
    {
      "epoch": 2.8221285949807657,
      "grad_norm": 0.3053629983996727,
      "learning_rate": 1.9160572987492966e-05,
      "loss": 0.7041,
      "num_tokens": 5886376312.0,
      "step": 1927
    },
    {
      "epoch": 2.8235940648470415,
      "grad_norm": 0.37037867053268025,
      "learning_rate": 1.9143714163787477e-05,
      "loss": 0.6912,
      "num_tokens": 5889510427.0,
      "step": 1928
    },
    {
      "epoch": 2.8250595347133176,
      "grad_norm": 0.36391403257767957,
      "learning_rate": 1.912685791000862e-05,
      "loss": 0.695,
      "num_tokens": 5892878037.0,
      "step": 1929
    },
    {
      "epoch": 2.8265250045795933,
      "grad_norm": 0.35445728795796444,
      "learning_rate": 1.9110004241322695e-05,
      "loss": 0.6884,
      "num_tokens": 5895947896.0,
      "step": 1930
    },
    {
      "epoch": 2.827990474445869,
      "grad_norm": 0.3449886754584593,
      "learning_rate": 1.90931531728937e-05,
      "loss": 0.7061,
      "num_tokens": 5898861553.0,
      "step": 1931
    },
    {
      "epoch": 2.829455944312145,
      "grad_norm": 0.3570554916034151,
      "learning_rate": 1.9076304719883273e-05,
      "loss": 0.6845,
      "num_tokens": 5901914756.0,
      "step": 1932
    },
    {
      "epoch": 2.830921414178421,
      "grad_norm": 0.35699459082912244,
      "learning_rate": 1.9059458897450738e-05,
      "loss": 0.6829,
      "num_tokens": 5904657656.0,
      "step": 1933
    },
    {
      "epoch": 2.8323868840446966,
      "grad_norm": 0.3651182972677928,
      "learning_rate": 1.9042615720752997e-05,
      "loss": 0.6986,
      "num_tokens": 5907802966.0,
      "step": 1934
    },
    {
      "epoch": 2.833852353910973,
      "grad_norm": 0.3402541539487093,
      "learning_rate": 1.9025775204944617e-05,
      "loss": 0.6946,
      "num_tokens": 5910862409.0,
      "step": 1935
    },
    {
      "epoch": 2.8353178237772485,
      "grad_norm": 0.33018596619363494,
      "learning_rate": 1.9008937365177752e-05,
      "loss": 0.6739,
      "num_tokens": 5913906448.0,
      "step": 1936
    },
    {
      "epoch": 2.836783293643524,
      "grad_norm": 0.3910154366124698,
      "learning_rate": 1.8992102216602153e-05,
      "loss": 0.7064,
      "num_tokens": 5917064374.0,
      "step": 1937
    },
    {
      "epoch": 2.8382487635098004,
      "grad_norm": 0.37557458167084157,
      "learning_rate": 1.8975269774365133e-05,
      "loss": 0.7019,
      "num_tokens": 5920305901.0,
      "step": 1938
    },
    {
      "epoch": 2.839714233376076,
      "grad_norm": 0.32694739275000034,
      "learning_rate": 1.895844005361161e-05,
      "loss": 0.6813,
      "num_tokens": 5923514490.0,
      "step": 1939
    },
    {
      "epoch": 2.841179703242352,
      "grad_norm": 0.35576440389585745,
      "learning_rate": 1.8941613069484e-05,
      "loss": 0.6651,
      "num_tokens": 5926727802.0,
      "step": 1940
    },
    {
      "epoch": 2.842645173108628,
      "grad_norm": 0.3168482585406132,
      "learning_rate": 1.892478883712231e-05,
      "loss": 0.6801,
      "num_tokens": 5929957229.0,
      "step": 1941
    },
    {
      "epoch": 2.8441106429749037,
      "grad_norm": 0.3821920714964589,
      "learning_rate": 1.8907967371664023e-05,
      "loss": 0.6867,
      "num_tokens": 5932848511.0,
      "step": 1942
    },
    {
      "epoch": 2.84557611284118,
      "grad_norm": 0.392316780120927,
      "learning_rate": 1.889114868824417e-05,
      "loss": 0.689,
      "num_tokens": 5935814222.0,
      "step": 1943
    },
    {
      "epoch": 2.8470415827074556,
      "grad_norm": 0.3658397844712153,
      "learning_rate": 1.8874332801995258e-05,
      "loss": 0.701,
      "num_tokens": 5938491297.0,
      "step": 1944
    },
    {
      "epoch": 2.8485070525737317,
      "grad_norm": 0.38956888785417565,
      "learning_rate": 1.8857519728047284e-05,
      "loss": 0.6845,
      "num_tokens": 5941700614.0,
      "step": 1945
    },
    {
      "epoch": 2.8499725224400074,
      "grad_norm": 0.34260603257937944,
      "learning_rate": 1.884070948152771e-05,
      "loss": 0.7164,
      "num_tokens": 5944705336.0,
      "step": 1946
    },
    {
      "epoch": 2.851437992306283,
      "grad_norm": 0.32792243192687826,
      "learning_rate": 1.8823902077561472e-05,
      "loss": 0.6882,
      "num_tokens": 5947620870.0,
      "step": 1947
    },
    {
      "epoch": 2.8529034621725593,
      "grad_norm": 0.37292528363446426,
      "learning_rate": 1.880709753127091e-05,
      "loss": 0.6987,
      "num_tokens": 5950601305.0,
      "step": 1948
    },
    {
      "epoch": 2.854368932038835,
      "grad_norm": 0.2923390203809084,
      "learning_rate": 1.8790295857775844e-05,
      "loss": 0.7039,
      "num_tokens": 5953704643.0,
      "step": 1949
    },
    {
      "epoch": 2.8558344019051107,
      "grad_norm": 0.3581424940987141,
      "learning_rate": 1.877349707219346e-05,
      "loss": 0.6787,
      "num_tokens": 5956684637.0,
      "step": 1950
    },
    {
      "epoch": 2.857299871771387,
      "grad_norm": 0.3570182333752147,
      "learning_rate": 1.8756701189638384e-05,
      "loss": 0.6913,
      "num_tokens": 5959658750.0,
      "step": 1951
    },
    {
      "epoch": 2.8587653416376626,
      "grad_norm": 0.34086946700270954,
      "learning_rate": 1.87399082252226e-05,
      "loss": 0.6794,
      "num_tokens": 5962764476.0,
      "step": 1952
    },
    {
      "epoch": 2.8602308115039383,
      "grad_norm": 0.3629568152454884,
      "learning_rate": 1.8723118194055492e-05,
      "loss": 0.6734,
      "num_tokens": 5965903691.0,
      "step": 1953
    },
    {
      "epoch": 2.8616962813702145,
      "grad_norm": 0.35540443474987515,
      "learning_rate": 1.8706331111243784e-05,
      "loss": 0.6666,
      "num_tokens": 5969137440.0,
      "step": 1954
    },
    {
      "epoch": 2.86316175123649,
      "grad_norm": 0.34529871490478364,
      "learning_rate": 1.8689546991891563e-05,
      "loss": 0.6929,
      "num_tokens": 5972245303.0,
      "step": 1955
    },
    {
      "epoch": 2.864627221102766,
      "grad_norm": 0.3537956442922439,
      "learning_rate": 1.8672765851100246e-05,
      "loss": 0.6868,
      "num_tokens": 5975304556.0,
      "step": 1956
    },
    {
      "epoch": 2.866092690969042,
      "grad_norm": 0.3412992197419074,
      "learning_rate": 1.8655987703968564e-05,
      "loss": 0.6976,
      "num_tokens": 5978383642.0,
      "step": 1957
    },
    {
      "epoch": 2.867558160835318,
      "grad_norm": 0.3625105789656368,
      "learning_rate": 1.8639212565592548e-05,
      "loss": 0.688,
      "num_tokens": 5981742498.0,
      "step": 1958
    },
    {
      "epoch": 2.8690236307015935,
      "grad_norm": 0.35385510737549086,
      "learning_rate": 1.8622440451065545e-05,
      "loss": 0.7053,
      "num_tokens": 5984854190.0,
      "step": 1959
    },
    {
      "epoch": 2.8704891005678697,
      "grad_norm": 0.36185960735870565,
      "learning_rate": 1.860567137547816e-05,
      "loss": 0.6856,
      "num_tokens": 5987966447.0,
      "step": 1960
    },
    {
      "epoch": 2.8719545704341454,
      "grad_norm": 0.333693003472386,
      "learning_rate": 1.8588905353918275e-05,
      "loss": 0.6979,
      "num_tokens": 5991227331.0,
      "step": 1961
    },
    {
      "epoch": 2.873420040300421,
      "grad_norm": 0.37576610367029845,
      "learning_rate": 1.8572142401471018e-05,
      "loss": 0.6902,
      "num_tokens": 5994234249.0,
      "step": 1962
    },
    {
      "epoch": 2.8748855101666972,
      "grad_norm": 0.3426411176108004,
      "learning_rate": 1.8555382533218763e-05,
      "loss": 0.6835,
      "num_tokens": 5997329850.0,
      "step": 1963
    },
    {
      "epoch": 2.876350980032973,
      "grad_norm": 0.3663376674511217,
      "learning_rate": 1.8538625764241095e-05,
      "loss": 0.6777,
      "num_tokens": 6000224490.0,
      "step": 1964
    },
    {
      "epoch": 2.8778164498992487,
      "grad_norm": 0.33963546920435034,
      "learning_rate": 1.8521872109614835e-05,
      "loss": 0.6739,
      "num_tokens": 6003499083.0,
      "step": 1965
    },
    {
      "epoch": 2.879281919765525,
      "grad_norm": 0.3559222916732817,
      "learning_rate": 1.8505121584413964e-05,
      "loss": 0.6932,
      "num_tokens": 6006607123.0,
      "step": 1966
    },
    {
      "epoch": 2.8807473896318005,
      "grad_norm": 0.37887861916783216,
      "learning_rate": 1.8488374203709692e-05,
      "loss": 0.7119,
      "num_tokens": 6009692730.0,
      "step": 1967
    },
    {
      "epoch": 2.8822128594980767,
      "grad_norm": 0.3509608164424842,
      "learning_rate": 1.8471629982570376e-05,
      "loss": 0.702,
      "num_tokens": 6013038845.0,
      "step": 1968
    },
    {
      "epoch": 2.8836783293643524,
      "grad_norm": 0.3841466536592505,
      "learning_rate": 1.845488893606151e-05,
      "loss": 0.6878,
      "num_tokens": 6015935749.0,
      "step": 1969
    },
    {
      "epoch": 2.8851437992306286,
      "grad_norm": 0.36735969129977003,
      "learning_rate": 1.8438151079245775e-05,
      "loss": 0.6906,
      "num_tokens": 6018830876.0,
      "step": 1970
    },
    {
      "epoch": 2.8866092690969043,
      "grad_norm": 0.38382536465969336,
      "learning_rate": 1.8421416427182954e-05,
      "loss": 0.6766,
      "num_tokens": 6022137368.0,
      "step": 1971
    },
    {
      "epoch": 2.88807473896318,
      "grad_norm": 0.33074056182233375,
      "learning_rate": 1.8404684994929955e-05,
      "loss": 0.7241,
      "num_tokens": 6025242862.0,
      "step": 1972
    },
    {
      "epoch": 2.889540208829456,
      "grad_norm": 0.46333707425506526,
      "learning_rate": 1.838795679754078e-05,
      "loss": 0.7069,
      "num_tokens": 6028378641.0,
      "step": 1973
    },
    {
      "epoch": 2.891005678695732,
      "grad_norm": 0.4007889857057646,
      "learning_rate": 1.837123185006654e-05,
      "loss": 0.699,
      "num_tokens": 6031696806.0,
      "step": 1974
    },
    {
      "epoch": 2.8924711485620076,
      "grad_norm": 0.41913592586509835,
      "learning_rate": 1.8354510167555398e-05,
      "loss": 0.6846,
      "num_tokens": 6034870308.0,
      "step": 1975
    },
    {
      "epoch": 2.8939366184282838,
      "grad_norm": 0.4263626095747065,
      "learning_rate": 1.8337791765052597e-05,
      "loss": 0.6857,
      "num_tokens": 6037896141.0,
      "step": 1976
    },
    {
      "epoch": 2.8954020882945595,
      "grad_norm": 0.42329805523919944,
      "learning_rate": 1.8321076657600416e-05,
      "loss": 0.6789,
      "num_tokens": 6040892132.0,
      "step": 1977
    },
    {
      "epoch": 2.896867558160835,
      "grad_norm": 0.4092090936670306,
      "learning_rate": 1.8304364860238193e-05,
      "loss": 0.6651,
      "num_tokens": 6043928449.0,
      "step": 1978
    },
    {
      "epoch": 2.8983330280271113,
      "grad_norm": 0.3751279328037842,
      "learning_rate": 1.8287656388002246e-05,
      "loss": 0.6877,
      "num_tokens": 6046908766.0,
      "step": 1979
    },
    {
      "epoch": 2.899798497893387,
      "grad_norm": 0.38055453401862305,
      "learning_rate": 1.8270951255925948e-05,
      "loss": 0.7023,
      "num_tokens": 6050153868.0,
      "step": 1980
    },
    {
      "epoch": 2.9012639677596628,
      "grad_norm": 0.312525578205032,
      "learning_rate": 1.8254249479039625e-05,
      "loss": 0.6967,
      "num_tokens": 6053352635.0,
      "step": 1981
    },
    {
      "epoch": 2.902729437625939,
      "grad_norm": 0.38618325070426335,
      "learning_rate": 1.8237551072370626e-05,
      "loss": 0.6889,
      "num_tokens": 6056252968.0,
      "step": 1982
    },
    {
      "epoch": 2.9041949074922147,
      "grad_norm": 0.339837013539051,
      "learning_rate": 1.8220856050943224e-05,
      "loss": 0.6977,
      "num_tokens": 6059508239.0,
      "step": 1983
    },
    {
      "epoch": 2.9056603773584904,
      "grad_norm": 0.34698121123570486,
      "learning_rate": 1.8204164429778686e-05,
      "loss": 0.693,
      "num_tokens": 6062613842.0,
      "step": 1984
    },
    {
      "epoch": 2.9071258472247665,
      "grad_norm": 0.38807600758497707,
      "learning_rate": 1.8187476223895185e-05,
      "loss": 0.6826,
      "num_tokens": 6065697777.0,
      "step": 1985
    },
    {
      "epoch": 2.9085913170910422,
      "grad_norm": 0.3276864806259998,
      "learning_rate": 1.8170791448307848e-05,
      "loss": 0.7038,
      "num_tokens": 6068962297.0,
      "step": 1986
    },
    {
      "epoch": 2.910056786957318,
      "grad_norm": 0.40051535514351794,
      "learning_rate": 1.8154110118028705e-05,
      "loss": 0.7096,
      "num_tokens": 6072009212.0,
      "step": 1987
    },
    {
      "epoch": 2.911522256823594,
      "grad_norm": 0.3323727306371662,
      "learning_rate": 1.813743224806669e-05,
      "loss": 0.7101,
      "num_tokens": 6075007107.0,
      "step": 1988
    },
    {
      "epoch": 2.91298772668987,
      "grad_norm": 0.4334141586499341,
      "learning_rate": 1.8120757853427606e-05,
      "loss": 0.7052,
      "num_tokens": 6078205322.0,
      "step": 1989
    },
    {
      "epoch": 2.9144531965561455,
      "grad_norm": 0.3140357045600773,
      "learning_rate": 1.810408694911415e-05,
      "loss": 0.7002,
      "num_tokens": 6081104034.0,
      "step": 1990
    },
    {
      "epoch": 2.9159186664224217,
      "grad_norm": 0.38700248973075596,
      "learning_rate": 1.8087419550125874e-05,
      "loss": 0.6748,
      "num_tokens": 6083958702.0,
      "step": 1991
    },
    {
      "epoch": 2.9173841362886974,
      "grad_norm": 0.3490434040026488,
      "learning_rate": 1.8070755671459175e-05,
      "loss": 0.6891,
      "num_tokens": 6087077720.0,
      "step": 1992
    },
    {
      "epoch": 2.9188496061549736,
      "grad_norm": 0.3498098134972655,
      "learning_rate": 1.8054095328107272e-05,
      "loss": 0.6696,
      "num_tokens": 6090072360.0,
      "step": 1993
    },
    {
      "epoch": 2.9203150760212493,
      "grad_norm": 0.3949658650311826,
      "learning_rate": 1.8037438535060225e-05,
      "loss": 0.6876,
      "num_tokens": 6092950922.0,
      "step": 1994
    },
    {
      "epoch": 2.9217805458875254,
      "grad_norm": 0.3161783163196648,
      "learning_rate": 1.8020785307304872e-05,
      "loss": 0.7002,
      "num_tokens": 6095918763.0,
      "step": 1995
    },
    {
      "epoch": 2.923246015753801,
      "grad_norm": 0.38556120222194146,
      "learning_rate": 1.8004135659824875e-05,
      "loss": 0.6808,
      "num_tokens": 6098822898.0,
      "step": 1996
    },
    {
      "epoch": 2.924711485620077,
      "grad_norm": 0.3587199835861605,
      "learning_rate": 1.798748960760064e-05,
      "loss": 0.6744,
      "num_tokens": 6101945080.0,
      "step": 1997
    },
    {
      "epoch": 2.926176955486353,
      "grad_norm": 0.3271502966881546,
      "learning_rate": 1.7970847165609374e-05,
      "loss": 0.6721,
      "num_tokens": 6104777350.0,
      "step": 1998
    },
    {
      "epoch": 2.9276424253526288,
      "grad_norm": 0.4228446267314715,
      "learning_rate": 1.7954208348824997e-05,
      "loss": 0.6942,
      "num_tokens": 6107646001.0,
      "step": 1999
    },
    {
      "epoch": 2.9291078952189045,
      "grad_norm": 0.3133220781095222,
      "learning_rate": 1.7937573172218212e-05,
      "loss": 0.6926,
      "num_tokens": 6110615904.0,
      "step": 2000
    },
    {
      "epoch": 2.9305733650851806,
      "grad_norm": 0.3775104048144457,
      "learning_rate": 1.79209416507564e-05,
      "loss": 0.6886,
      "num_tokens": 6113656397.0,
      "step": 2001
    },
    {
      "epoch": 2.9320388349514563,
      "grad_norm": 0.35694742680102304,
      "learning_rate": 1.7904313799403698e-05,
      "loss": 0.6917,
      "num_tokens": 6116749959.0,
      "step": 2002
    },
    {
      "epoch": 2.933504304817732,
      "grad_norm": 0.40851525346760326,
      "learning_rate": 1.78876896331209e-05,
      "loss": 0.708,
      "num_tokens": 6119604315.0,
      "step": 2003
    },
    {
      "epoch": 2.934969774684008,
      "grad_norm": 0.3761625313848583,
      "learning_rate": 1.787106916686551e-05,
      "loss": 0.6972,
      "num_tokens": 6122878239.0,
      "step": 2004
    },
    {
      "epoch": 2.936435244550284,
      "grad_norm": 0.38867963173723463,
      "learning_rate": 1.78544524155917e-05,
      "loss": 0.7012,
      "num_tokens": 6125973777.0,
      "step": 2005
    },
    {
      "epoch": 2.9379007144165596,
      "grad_norm": 0.34962605390690465,
      "learning_rate": 1.783783939425029e-05,
      "loss": 0.7014,
      "num_tokens": 6129142680.0,
      "step": 2006
    },
    {
      "epoch": 2.939366184282836,
      "grad_norm": 0.3546333441676004,
      "learning_rate": 1.7821230117788754e-05,
      "loss": 0.6779,
      "num_tokens": 6132180228.0,
      "step": 2007
    },
    {
      "epoch": 2.9408316541491115,
      "grad_norm": 0.3276804886197234,
      "learning_rate": 1.7804624601151186e-05,
      "loss": 0.7016,
      "num_tokens": 6135365645.0,
      "step": 2008
    },
    {
      "epoch": 2.9422971240153872,
      "grad_norm": 0.38118667908784354,
      "learning_rate": 1.778802285927831e-05,
      "loss": 0.7014,
      "num_tokens": 6138301537.0,
      "step": 2009
    },
    {
      "epoch": 2.9437625938816634,
      "grad_norm": 0.3030655580025739,
      "learning_rate": 1.777142490710744e-05,
      "loss": 0.6984,
      "num_tokens": 6141476876.0,
      "step": 2010
    },
    {
      "epoch": 2.945228063747939,
      "grad_norm": 0.3238988627748957,
      "learning_rate": 1.77548307595725e-05,
      "loss": 0.6728,
      "num_tokens": 6144538754.0,
      "step": 2011
    },
    {
      "epoch": 2.946693533614215,
      "grad_norm": 0.35759059886743016,
      "learning_rate": 1.7738240431603963e-05,
      "loss": 0.6748,
      "num_tokens": 6147628620.0,
      "step": 2012
    },
    {
      "epoch": 2.948159003480491,
      "grad_norm": 0.35907742703413315,
      "learning_rate": 1.772165393812889e-05,
      "loss": 0.6958,
      "num_tokens": 6150645389.0,
      "step": 2013
    },
    {
      "epoch": 2.9496244733467667,
      "grad_norm": 0.30403079460782306,
      "learning_rate": 1.770507129407087e-05,
      "loss": 0.6859,
      "num_tokens": 6153921611.0,
      "step": 2014
    },
    {
      "epoch": 2.9510899432130424,
      "grad_norm": 0.37598170318169066,
      "learning_rate": 1.7688492514350068e-05,
      "loss": 0.6762,
      "num_tokens": 6156983485.0,
      "step": 2015
    },
    {
      "epoch": 2.9525554130793186,
      "grad_norm": 0.31697226414770346,
      "learning_rate": 1.7671917613883113e-05,
      "loss": 0.6914,
      "num_tokens": 6160061482.0,
      "step": 2016
    },
    {
      "epoch": 2.9540208829455943,
      "grad_norm": 0.37827740261772685,
      "learning_rate": 1.7655346607583194e-05,
      "loss": 0.6735,
      "num_tokens": 6163266924.0,
      "step": 2017
    },
    {
      "epoch": 2.9554863528118704,
      "grad_norm": 0.33509526288167346,
      "learning_rate": 1.763877951035997e-05,
      "loss": 0.701,
      "num_tokens": 6166246755.0,
      "step": 2018
    },
    {
      "epoch": 2.956951822678146,
      "grad_norm": 0.37339587042726974,
      "learning_rate": 1.7622216337119603e-05,
      "loss": 0.7046,
      "num_tokens": 6169065787.0,
      "step": 2019
    },
    {
      "epoch": 2.9584172925444223,
      "grad_norm": 0.34883979666738685,
      "learning_rate": 1.7605657102764696e-05,
      "loss": 0.6978,
      "num_tokens": 6172068175.0,
      "step": 2020
    },
    {
      "epoch": 2.959882762410698,
      "grad_norm": 0.36818699995555954,
      "learning_rate": 1.7589101822194333e-05,
      "loss": 0.6866,
      "num_tokens": 6175135136.0,
      "step": 2021
    },
    {
      "epoch": 2.9613482322769737,
      "grad_norm": 0.3541876725547167,
      "learning_rate": 1.7572550510304023e-05,
      "loss": 0.6726,
      "num_tokens": 6178472990.0,
      "step": 2022
    },
    {
      "epoch": 2.96281370214325,
      "grad_norm": 0.3502493600297295,
      "learning_rate": 1.7556003181985738e-05,
      "loss": 0.6948,
      "num_tokens": 6181375729.0,
      "step": 2023
    },
    {
      "epoch": 2.9642791720095256,
      "grad_norm": 0.3074632576421066,
      "learning_rate": 1.7539459852127808e-05,
      "loss": 0.6878,
      "num_tokens": 6184304808.0,
      "step": 2024
    },
    {
      "epoch": 2.9657446418758013,
      "grad_norm": 0.35466937291046186,
      "learning_rate": 1.7522920535615012e-05,
      "loss": 0.6838,
      "num_tokens": 6187334838.0,
      "step": 2025
    },
    {
      "epoch": 2.9672101117420775,
      "grad_norm": 0.3562676360141291,
      "learning_rate": 1.7506385247328506e-05,
      "loss": 0.6932,
      "num_tokens": 6190255504.0,
      "step": 2026
    },
    {
      "epoch": 2.968675581608353,
      "grad_norm": 0.303503744960109,
      "learning_rate": 1.7489854002145827e-05,
      "loss": 0.6974,
      "num_tokens": 6193288467.0,
      "step": 2027
    },
    {
      "epoch": 2.970141051474629,
      "grad_norm": 0.32562598526387065,
      "learning_rate": 1.747332681494085e-05,
      "loss": 0.6909,
      "num_tokens": 6196569499.0,
      "step": 2028
    },
    {
      "epoch": 2.971606521340905,
      "grad_norm": 0.33373255373100813,
      "learning_rate": 1.7456803700583834e-05,
      "loss": 0.6863,
      "num_tokens": 6199619526.0,
      "step": 2029
    },
    {
      "epoch": 2.973071991207181,
      "grad_norm": 0.3198568304440413,
      "learning_rate": 1.7440284673941333e-05,
      "loss": 0.6825,
      "num_tokens": 6202637993.0,
      "step": 2030
    },
    {
      "epoch": 2.9745374610734565,
      "grad_norm": 0.33763520976388306,
      "learning_rate": 1.742376974987627e-05,
      "loss": 0.672,
      "num_tokens": 6206081580.0,
      "step": 2031
    },
    {
      "epoch": 2.9760029309397327,
      "grad_norm": 0.3070274246440076,
      "learning_rate": 1.740725894324783e-05,
      "loss": 0.7014,
      "num_tokens": 6209300732.0,
      "step": 2032
    },
    {
      "epoch": 2.9774684008060084,
      "grad_norm": 0.34349860040159286,
      "learning_rate": 1.7390752268911536e-05,
      "loss": 0.6913,
      "num_tokens": 6212277976.0,
      "step": 2033
    },
    {
      "epoch": 2.978933870672284,
      "grad_norm": 0.4143081879824312,
      "learning_rate": 1.7374249741719157e-05,
      "loss": 0.6789,
      "num_tokens": 6215416573.0,
      "step": 2034
    },
    {
      "epoch": 2.9803993405385603,
      "grad_norm": 0.37676670897175035,
      "learning_rate": 1.7357751376518753e-05,
      "loss": 0.6819,
      "num_tokens": 6218531261.0,
      "step": 2035
    },
    {
      "epoch": 2.981864810404836,
      "grad_norm": 0.36030950039593773,
      "learning_rate": 1.7341257188154625e-05,
      "loss": 0.6899,
      "num_tokens": 6221342276.0,
      "step": 2036
    },
    {
      "epoch": 2.9833302802711117,
      "grad_norm": 0.3505728891330053,
      "learning_rate": 1.7324767191467346e-05,
      "loss": 0.6976,
      "num_tokens": 6224625446.0,
      "step": 2037
    },
    {
      "epoch": 2.984795750137388,
      "grad_norm": 0.3465075304757723,
      "learning_rate": 1.730828140129366e-05,
      "loss": 0.6875,
      "num_tokens": 6228127319.0,
      "step": 2038
    },
    {
      "epoch": 2.9862612200036636,
      "grad_norm": 0.3715512861827909,
      "learning_rate": 1.7291799832466593e-05,
      "loss": 0.6915,
      "num_tokens": 6231009456.0,
      "step": 2039
    },
    {
      "epoch": 2.9877266898699393,
      "grad_norm": 0.34542116605574746,
      "learning_rate": 1.727532249981531e-05,
      "loss": 0.6912,
      "num_tokens": 6234280381.0,
      "step": 2040
    },
    {
      "epoch": 2.9891921597362154,
      "grad_norm": 0.36454647320853734,
      "learning_rate": 1.7258849418165233e-05,
      "loss": 0.6982,
      "num_tokens": 6237307926.0,
      "step": 2041
    },
    {
      "epoch": 2.990657629602491,
      "grad_norm": 0.4067892269676075,
      "learning_rate": 1.7242380602337883e-05,
      "loss": 0.6788,
      "num_tokens": 6240224395.0,
      "step": 2042
    },
    {
      "epoch": 2.9921230994687673,
      "grad_norm": 0.3676539478700517,
      "learning_rate": 1.7225916067151018e-05,
      "loss": 0.6815,
      "num_tokens": 6243319959.0,
      "step": 2043
    },
    {
      "epoch": 2.993588569335043,
      "grad_norm": 0.3914655111034985,
      "learning_rate": 1.7209455827418482e-05,
      "loss": 0.6983,
      "num_tokens": 6246489593.0,
      "step": 2044
    },
    {
      "epoch": 2.995054039201319,
      "grad_norm": 0.40855158212096393,
      "learning_rate": 1.7192999897950287e-05,
      "loss": 0.6935,
      "num_tokens": 6249433834.0,
      "step": 2045
    },
    {
      "epoch": 2.996519509067595,
      "grad_norm": 0.35085973761937456,
      "learning_rate": 1.717654829355256e-05,
      "loss": 0.6889,
      "num_tokens": 6252616084.0,
      "step": 2046
    },
    {
      "epoch": 2.9979849789338706,
      "grad_norm": 0.356228228313392,
      "learning_rate": 1.7160101029027546e-05,
      "loss": 0.6708,
      "num_tokens": 6255602719.0,
      "step": 2047
    },
    {
      "epoch": 2.9994504488001468,
      "grad_norm": 0.33861350832732645,
      "learning_rate": 1.7143658119173566e-05,
      "loss": 0.7048,
      "num_tokens": 6258454926.0,
      "step": 2048
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.5031557411193512,
      "learning_rate": 1.7127219578785036e-05,
      "loss": 0.7186,
      "num_tokens": 6259067011.0,
      "step": 2049
    },
    {
      "epoch": 3.0014654698662757,
      "grad_norm": 0.4224423264160683,
      "learning_rate": 1.7110785422652437e-05,
      "loss": 0.6814,
      "num_tokens": 6262206132.0,
      "step": 2050
    },
    {
      "epoch": 3.002930939732552,
      "grad_norm": 0.33422512397609005,
      "learning_rate": 1.7094355665562295e-05,
      "loss": 0.6667,
      "num_tokens": 6265118234.0,
      "step": 2051
    },
    {
      "epoch": 3.0043964095988276,
      "grad_norm": 0.3297955699437232,
      "learning_rate": 1.707793032229721e-05,
      "loss": 0.6822,
      "num_tokens": 6268176043.0,
      "step": 2052
    },
    {
      "epoch": 3.0058618794651033,
      "grad_norm": 0.323332935817789,
      "learning_rate": 1.706150940763577e-05,
      "loss": 0.6894,
      "num_tokens": 6271240226.0,
      "step": 2053
    },
    {
      "epoch": 3.0073273493313795,
      "grad_norm": 0.3350721172132651,
      "learning_rate": 1.7045092936352614e-05,
      "loss": 0.6912,
      "num_tokens": 6274072923.0,
      "step": 2054
    },
    {
      "epoch": 3.008792819197655,
      "grad_norm": 0.3351606184640879,
      "learning_rate": 1.7028680923218354e-05,
      "loss": 0.6783,
      "num_tokens": 6277083057.0,
      "step": 2055
    },
    {
      "epoch": 3.0102582890639313,
      "grad_norm": 0.3660419633573275,
      "learning_rate": 1.701227338299961e-05,
      "loss": 0.6935,
      "num_tokens": 6280156156.0,
      "step": 2056
    },
    {
      "epoch": 3.011723758930207,
      "grad_norm": 0.3312951438971161,
      "learning_rate": 1.6995870330458966e-05,
      "loss": 0.6802,
      "num_tokens": 6282982212.0,
      "step": 2057
    },
    {
      "epoch": 3.0131892287964828,
      "grad_norm": 0.33722951689594466,
      "learning_rate": 1.6979471780354977e-05,
      "loss": 0.6915,
      "num_tokens": 6285921397.0,
      "step": 2058
    },
    {
      "epoch": 3.014654698662759,
      "grad_norm": 0.3505891508562686,
      "learning_rate": 1.6963077747442148e-05,
      "loss": 0.6774,
      "num_tokens": 6288870625.0,
      "step": 2059
    },
    {
      "epoch": 3.0161201685290346,
      "grad_norm": 0.30274043418187113,
      "learning_rate": 1.694668824647091e-05,
      "loss": 0.6595,
      "num_tokens": 6291850259.0,
      "step": 2060
    },
    {
      "epoch": 3.0175856383953104,
      "grad_norm": 0.32988057400965043,
      "learning_rate": 1.6930303292187614e-05,
      "loss": 0.6675,
      "num_tokens": 6294686873.0,
      "step": 2061
    },
    {
      "epoch": 3.0190511082615865,
      "grad_norm": 0.34627987479833816,
      "learning_rate": 1.6913922899334546e-05,
      "loss": 0.6872,
      "num_tokens": 6297683748.0,
      "step": 2062
    },
    {
      "epoch": 3.0205165781278622,
      "grad_norm": 0.3434529888154081,
      "learning_rate": 1.6897547082649854e-05,
      "loss": 0.6844,
      "num_tokens": 6300936948.0,
      "step": 2063
    },
    {
      "epoch": 3.021982047994138,
      "grad_norm": 0.3539262401078821,
      "learning_rate": 1.6881175856867595e-05,
      "loss": 0.6832,
      "num_tokens": 6304278502.0,
      "step": 2064
    },
    {
      "epoch": 3.023447517860414,
      "grad_norm": 0.3311790003944426,
      "learning_rate": 1.6864809236717668e-05,
      "loss": 0.6843,
      "num_tokens": 6307095446.0,
      "step": 2065
    },
    {
      "epoch": 3.02491298772669,
      "grad_norm": 0.36789046890051574,
      "learning_rate": 1.684844723692586e-05,
      "loss": 0.6806,
      "num_tokens": 6310047278.0,
      "step": 2066
    },
    {
      "epoch": 3.0263784575929655,
      "grad_norm": 0.3470119623491541,
      "learning_rate": 1.6832089872213777e-05,
      "loss": 0.6914,
      "num_tokens": 6313076626.0,
      "step": 2067
    },
    {
      "epoch": 3.0278439274592417,
      "grad_norm": 0.33487147311455095,
      "learning_rate": 1.6815737157298877e-05,
      "loss": 0.6916,
      "num_tokens": 6316073842.0,
      "step": 2068
    },
    {
      "epoch": 3.0293093973255174,
      "grad_norm": 0.3230807977427358,
      "learning_rate": 1.6799389106894403e-05,
      "loss": 0.6868,
      "num_tokens": 6319301453.0,
      "step": 2069
    },
    {
      "epoch": 3.0307748671917936,
      "grad_norm": 0.3180722086456074,
      "learning_rate": 1.6783045735709428e-05,
      "loss": 0.6921,
      "num_tokens": 6322401931.0,
      "step": 2070
    },
    {
      "epoch": 3.0322403370580693,
      "grad_norm": 0.33670671117367174,
      "learning_rate": 1.6766707058448803e-05,
      "loss": 0.671,
      "num_tokens": 6325585536.0,
      "step": 2071
    },
    {
      "epoch": 3.033705806924345,
      "grad_norm": 0.31408224949763724,
      "learning_rate": 1.6750373089813154e-05,
      "loss": 0.6832,
      "num_tokens": 6328722959.0,
      "step": 2072
    },
    {
      "epoch": 3.035171276790621,
      "grad_norm": 0.3642184167764613,
      "learning_rate": 1.6734043844498876e-05,
      "loss": 0.6788,
      "num_tokens": 6331680020.0,
      "step": 2073
    },
    {
      "epoch": 3.036636746656897,
      "grad_norm": 0.32962595743150236,
      "learning_rate": 1.6717719337198127e-05,
      "loss": 0.6519,
      "num_tokens": 6334828135.0,
      "step": 2074
    },
    {
      "epoch": 3.0381022165231726,
      "grad_norm": 0.3232330604857763,
      "learning_rate": 1.670139958259877e-05,
      "loss": 0.6785,
      "num_tokens": 6337839080.0,
      "step": 2075
    },
    {
      "epoch": 3.0395676863894487,
      "grad_norm": 0.3250695566597705,
      "learning_rate": 1.668508459538442e-05,
      "loss": 0.6935,
      "num_tokens": 6340754944.0,
      "step": 2076
    },
    {
      "epoch": 3.0410331562557245,
      "grad_norm": 0.3221266200601106,
      "learning_rate": 1.666877439023438e-05,
      "loss": 0.6901,
      "num_tokens": 6343819200.0,
      "step": 2077
    },
    {
      "epoch": 3.042498626122,
      "grad_norm": 0.3296352686180719,
      "learning_rate": 1.6652468981823685e-05,
      "loss": 0.6816,
      "num_tokens": 6347125703.0,
      "step": 2078
    },
    {
      "epoch": 3.0439640959882763,
      "grad_norm": 0.31356938928695366,
      "learning_rate": 1.6636168384823003e-05,
      "loss": 0.6911,
      "num_tokens": 6350092301.0,
      "step": 2079
    },
    {
      "epoch": 3.045429565854552,
      "grad_norm": 0.40497840693334636,
      "learning_rate": 1.661987261389872e-05,
      "loss": 0.6837,
      "num_tokens": 6353382573.0,
      "step": 2080
    },
    {
      "epoch": 3.046895035720828,
      "grad_norm": 0.3694915380389864,
      "learning_rate": 1.6603581683712855e-05,
      "loss": 0.6703,
      "num_tokens": 6356579080.0,
      "step": 2081
    },
    {
      "epoch": 3.048360505587104,
      "grad_norm": 0.327444954656064,
      "learning_rate": 1.6587295608923088e-05,
      "loss": 0.6476,
      "num_tokens": 6359627012.0,
      "step": 2082
    },
    {
      "epoch": 3.0498259754533796,
      "grad_norm": 0.39278171655316585,
      "learning_rate": 1.6571014404182702e-05,
      "loss": 0.6722,
      "num_tokens": 6362720294.0,
      "step": 2083
    },
    {
      "epoch": 3.051291445319656,
      "grad_norm": 0.3442239662301714,
      "learning_rate": 1.655473808414064e-05,
      "loss": 0.7035,
      "num_tokens": 6365757347.0,
      "step": 2084
    },
    {
      "epoch": 3.0527569151859315,
      "grad_norm": 0.3972954608234565,
      "learning_rate": 1.6538466663441407e-05,
      "loss": 0.674,
      "num_tokens": 6368761836.0,
      "step": 2085
    },
    {
      "epoch": 3.0542223850522072,
      "grad_norm": 0.3375812877720009,
      "learning_rate": 1.6522200156725126e-05,
      "loss": 0.679,
      "num_tokens": 6371778786.0,
      "step": 2086
    },
    {
      "epoch": 3.0556878549184834,
      "grad_norm": 0.4437058608617698,
      "learning_rate": 1.6505938578627486e-05,
      "loss": 0.6882,
      "num_tokens": 6374853527.0,
      "step": 2087
    },
    {
      "epoch": 3.057153324784759,
      "grad_norm": 0.42743427608449025,
      "learning_rate": 1.648968194377977e-05,
      "loss": 0.6872,
      "num_tokens": 6378049482.0,
      "step": 2088
    },
    {
      "epoch": 3.058618794651035,
      "grad_norm": 0.3538406193531048,
      "learning_rate": 1.647343026680876e-05,
      "loss": 0.6726,
      "num_tokens": 6381104689.0,
      "step": 2089
    },
    {
      "epoch": 3.060084264517311,
      "grad_norm": 0.38825422674801074,
      "learning_rate": 1.645718356233683e-05,
      "loss": 0.6723,
      "num_tokens": 6384326919.0,
      "step": 2090
    },
    {
      "epoch": 3.0615497343835867,
      "grad_norm": 0.3440679508909914,
      "learning_rate": 1.644094184498185e-05,
      "loss": 0.6862,
      "num_tokens": 6387468403.0,
      "step": 2091
    },
    {
      "epoch": 3.0630152042498624,
      "grad_norm": 0.3543595185449668,
      "learning_rate": 1.6424705129357214e-05,
      "loss": 0.6829,
      "num_tokens": 6390672195.0,
      "step": 2092
    },
    {
      "epoch": 3.0644806741161386,
      "grad_norm": 0.3415495368643185,
      "learning_rate": 1.6408473430071814e-05,
      "loss": 0.6558,
      "num_tokens": 6393826183.0,
      "step": 2093
    },
    {
      "epoch": 3.0659461439824143,
      "grad_norm": 0.33432570703215225,
      "learning_rate": 1.639224676173002e-05,
      "loss": 0.6706,
      "num_tokens": 6396915085.0,
      "step": 2094
    },
    {
      "epoch": 3.0674116138486904,
      "grad_norm": 0.35526691787573805,
      "learning_rate": 1.6376025138931688e-05,
      "loss": 0.6661,
      "num_tokens": 6399857711.0,
      "step": 2095
    },
    {
      "epoch": 3.068877083714966,
      "grad_norm": 0.3647927096302436,
      "learning_rate": 1.6359808576272124e-05,
      "loss": 0.6579,
      "num_tokens": 6402908286.0,
      "step": 2096
    },
    {
      "epoch": 3.070342553581242,
      "grad_norm": 0.3258342159765199,
      "learning_rate": 1.6343597088342093e-05,
      "loss": 0.6921,
      "num_tokens": 6405979136.0,
      "step": 2097
    },
    {
      "epoch": 3.071808023447518,
      "grad_norm": 0.35594299253606976,
      "learning_rate": 1.6327390689727784e-05,
      "loss": 0.6772,
      "num_tokens": 6408767712.0,
      "step": 2098
    },
    {
      "epoch": 3.0732734933137937,
      "grad_norm": 0.34662255840313994,
      "learning_rate": 1.6311189395010814e-05,
      "loss": 0.669,
      "num_tokens": 6411914456.0,
      "step": 2099
    },
    {
      "epoch": 3.0747389631800695,
      "grad_norm": 0.3758124609997256,
      "learning_rate": 1.629499321876819e-05,
      "loss": 0.6908,
      "num_tokens": 6414927012.0,
      "step": 2100
    },
    {
      "epoch": 3.0762044330463456,
      "grad_norm": 0.3420976452784462,
      "learning_rate": 1.6278802175572352e-05,
      "loss": 0.678,
      "num_tokens": 6417999351.0,
      "step": 2101
    },
    {
      "epoch": 3.0776699029126213,
      "grad_norm": 0.3836269349315167,
      "learning_rate": 1.6262616279991074e-05,
      "loss": 0.6719,
      "num_tokens": 6420917759.0,
      "step": 2102
    },
    {
      "epoch": 3.079135372778897,
      "grad_norm": 0.3188623340869601,
      "learning_rate": 1.624643554658754e-05,
      "loss": 0.6833,
      "num_tokens": 6423957561.0,
      "step": 2103
    },
    {
      "epoch": 3.080600842645173,
      "grad_norm": 0.4172948146299732,
      "learning_rate": 1.6230259989920254e-05,
      "loss": 0.691,
      "num_tokens": 6426877110.0,
      "step": 2104
    },
    {
      "epoch": 3.082066312511449,
      "grad_norm": 0.34842503086110554,
      "learning_rate": 1.62140896245431e-05,
      "loss": 0.6812,
      "num_tokens": 6430061820.0,
      "step": 2105
    },
    {
      "epoch": 3.083531782377725,
      "grad_norm": 0.38123910517037546,
      "learning_rate": 1.6197924465005247e-05,
      "loss": 0.6688,
      "num_tokens": 6433416768.0,
      "step": 2106
    },
    {
      "epoch": 3.084997252244001,
      "grad_norm": 0.3791992936226009,
      "learning_rate": 1.6181764525851216e-05,
      "loss": 0.6792,
      "num_tokens": 6436431491.0,
      "step": 2107
    },
    {
      "epoch": 3.0864627221102765,
      "grad_norm": 0.40668472630323815,
      "learning_rate": 1.6165609821620815e-05,
      "loss": 0.6841,
      "num_tokens": 6439566711.0,
      "step": 2108
    },
    {
      "epoch": 3.0879281919765527,
      "grad_norm": 0.359017610469667,
      "learning_rate": 1.614946036684916e-05,
      "loss": 0.6677,
      "num_tokens": 6442767207.0,
      "step": 2109
    },
    {
      "epoch": 3.0893936618428284,
      "grad_norm": 0.38278484880385877,
      "learning_rate": 1.6133316176066603e-05,
      "loss": 0.6836,
      "num_tokens": 6445713428.0,
      "step": 2110
    },
    {
      "epoch": 3.090859131709104,
      "grad_norm": 0.38137948980533215,
      "learning_rate": 1.6117177263798805e-05,
      "loss": 0.6844,
      "num_tokens": 6448675661.0,
      "step": 2111
    },
    {
      "epoch": 3.0923246015753802,
      "grad_norm": 0.41628865869698056,
      "learning_rate": 1.6101043644566657e-05,
      "loss": 0.6807,
      "num_tokens": 6451842076.0,
      "step": 2112
    },
    {
      "epoch": 3.093790071441656,
      "grad_norm": 0.3883440551540658,
      "learning_rate": 1.6084915332886283e-05,
      "loss": 0.6741,
      "num_tokens": 6454854709.0,
      "step": 2113
    },
    {
      "epoch": 3.0952555413079317,
      "grad_norm": 0.40255646083210295,
      "learning_rate": 1.606879234326904e-05,
      "loss": 0.6677,
      "num_tokens": 6458010447.0,
      "step": 2114
    },
    {
      "epoch": 3.096721011174208,
      "grad_norm": 0.39704534371332556,
      "learning_rate": 1.6052674690221502e-05,
      "loss": 0.6613,
      "num_tokens": 6461066162.0,
      "step": 2115
    },
    {
      "epoch": 3.0981864810404836,
      "grad_norm": 0.3886346324544583,
      "learning_rate": 1.603656238824542e-05,
      "loss": 0.6941,
      "num_tokens": 6464197889.0,
      "step": 2116
    },
    {
      "epoch": 3.0996519509067593,
      "grad_norm": 0.38252522020183527,
      "learning_rate": 1.6020455451837756e-05,
      "loss": 0.6782,
      "num_tokens": 6467157698.0,
      "step": 2117
    },
    {
      "epoch": 3.1011174207730354,
      "grad_norm": 0.36211673855804943,
      "learning_rate": 1.6004353895490625e-05,
      "loss": 0.6642,
      "num_tokens": 6470210237.0,
      "step": 2118
    },
    {
      "epoch": 3.102582890639311,
      "grad_norm": 0.3856766803522645,
      "learning_rate": 1.5988257733691316e-05,
      "loss": 0.6758,
      "num_tokens": 6473268077.0,
      "step": 2119
    },
    {
      "epoch": 3.1040483605055873,
      "grad_norm": 0.3481586993218587,
      "learning_rate": 1.597216698092225e-05,
      "loss": 0.6709,
      "num_tokens": 6476528160.0,
      "step": 2120
    },
    {
      "epoch": 3.105513830371863,
      "grad_norm": 0.3590460316913547,
      "learning_rate": 1.5956081651660996e-05,
      "loss": 0.6675,
      "num_tokens": 6479578714.0,
      "step": 2121
    },
    {
      "epoch": 3.1069793002381387,
      "grad_norm": 0.33615313011749387,
      "learning_rate": 1.594000176038023e-05,
      "loss": 0.6663,
      "num_tokens": 6482705944.0,
      "step": 2122
    },
    {
      "epoch": 3.108444770104415,
      "grad_norm": 0.3236756249977203,
      "learning_rate": 1.5923927321547746e-05,
      "loss": 0.6808,
      "num_tokens": 6485871488.0,
      "step": 2123
    },
    {
      "epoch": 3.1099102399706906,
      "grad_norm": 0.3695252183278405,
      "learning_rate": 1.5907858349626417e-05,
      "loss": 0.6799,
      "num_tokens": 6489058166.0,
      "step": 2124
    },
    {
      "epoch": 3.1113757098369663,
      "grad_norm": 0.32963711587246763,
      "learning_rate": 1.5891794859074218e-05,
      "loss": 0.6847,
      "num_tokens": 6492103149.0,
      "step": 2125
    },
    {
      "epoch": 3.1128411797032425,
      "grad_norm": 0.41839204547089254,
      "learning_rate": 1.5875736864344173e-05,
      "loss": 0.6871,
      "num_tokens": 6495176950.0,
      "step": 2126
    },
    {
      "epoch": 3.114306649569518,
      "grad_norm": 0.3512569925923776,
      "learning_rate": 1.5859684379884386e-05,
      "loss": 0.6808,
      "num_tokens": 6498176489.0,
      "step": 2127
    },
    {
      "epoch": 3.115772119435794,
      "grad_norm": 0.36063105420447306,
      "learning_rate": 1.5843637420137964e-05,
      "loss": 0.6768,
      "num_tokens": 6501396030.0,
      "step": 2128
    },
    {
      "epoch": 3.11723758930207,
      "grad_norm": 0.32621973050160546,
      "learning_rate": 1.5827595999543083e-05,
      "loss": 0.6923,
      "num_tokens": 6504690118.0,
      "step": 2129
    },
    {
      "epoch": 3.118703059168346,
      "grad_norm": 0.35149157548091964,
      "learning_rate": 1.581156013253291e-05,
      "loss": 0.6793,
      "num_tokens": 6507700325.0,
      "step": 2130
    },
    {
      "epoch": 3.120168529034622,
      "grad_norm": 0.3335168983206523,
      "learning_rate": 1.5795529833535624e-05,
      "loss": 0.6809,
      "num_tokens": 6510775395.0,
      "step": 2131
    },
    {
      "epoch": 3.1216339989008977,
      "grad_norm": 0.37697926199316717,
      "learning_rate": 1.5779505116974397e-05,
      "loss": 0.6794,
      "num_tokens": 6513730601.0,
      "step": 2132
    },
    {
      "epoch": 3.1230994687671734,
      "grad_norm": 0.3394315115708865,
      "learning_rate": 1.5763485997267374e-05,
      "loss": 0.6643,
      "num_tokens": 6516914142.0,
      "step": 2133
    },
    {
      "epoch": 3.1245649386334495,
      "grad_norm": 0.3729353334078359,
      "learning_rate": 1.574747248882767e-05,
      "loss": 0.6672,
      "num_tokens": 6520170174.0,
      "step": 2134
    },
    {
      "epoch": 3.1260304084997252,
      "grad_norm": 0.3185867953721918,
      "learning_rate": 1.573146460606333e-05,
      "loss": 0.6844,
      "num_tokens": 6523208772.0,
      "step": 2135
    },
    {
      "epoch": 3.127495878366001,
      "grad_norm": 0.351362358762775,
      "learning_rate": 1.5715462363377373e-05,
      "loss": 0.6655,
      "num_tokens": 6526316228.0,
      "step": 2136
    },
    {
      "epoch": 3.128961348232277,
      "grad_norm": 0.3100315556746897,
      "learning_rate": 1.569946577516771e-05,
      "loss": 0.655,
      "num_tokens": 6529481065.0,
      "step": 2137
    },
    {
      "epoch": 3.130426818098553,
      "grad_norm": 0.3353223156940941,
      "learning_rate": 1.5683474855827184e-05,
      "loss": 0.6568,
      "num_tokens": 6532737114.0,
      "step": 2138
    },
    {
      "epoch": 3.1318922879648285,
      "grad_norm": 0.33532223811041706,
      "learning_rate": 1.5667489619743526e-05,
      "loss": 0.6734,
      "num_tokens": 6535935843.0,
      "step": 2139
    },
    {
      "epoch": 3.1333577578311047,
      "grad_norm": 0.30606383231230366,
      "learning_rate": 1.565151008129938e-05,
      "loss": 0.6701,
      "num_tokens": 6538881922.0,
      "step": 2140
    },
    {
      "epoch": 3.1348232276973804,
      "grad_norm": 0.3307091758533152,
      "learning_rate": 1.5635536254872215e-05,
      "loss": 0.6826,
      "num_tokens": 6541895270.0,
      "step": 2141
    },
    {
      "epoch": 3.136288697563656,
      "grad_norm": 0.37348146903267926,
      "learning_rate": 1.561956815483441e-05,
      "loss": 0.6797,
      "num_tokens": 6544947524.0,
      "step": 2142
    },
    {
      "epoch": 3.1377541674299323,
      "grad_norm": 0.33153329253197217,
      "learning_rate": 1.5603605795553152e-05,
      "loss": 0.6729,
      "num_tokens": 6548143802.0,
      "step": 2143
    },
    {
      "epoch": 3.139219637296208,
      "grad_norm": 0.32796179700895456,
      "learning_rate": 1.5587649191390503e-05,
      "loss": 0.6777,
      "num_tokens": 6551071640.0,
      "step": 2144
    },
    {
      "epoch": 3.140685107162484,
      "grad_norm": 0.3464505620525035,
      "learning_rate": 1.5571698356703304e-05,
      "loss": 0.692,
      "num_tokens": 6553747257.0,
      "step": 2145
    },
    {
      "epoch": 3.14215057702876,
      "grad_norm": 0.37393892056616373,
      "learning_rate": 1.5555753305843247e-05,
      "loss": 0.6665,
      "num_tokens": 6556941452.0,
      "step": 2146
    },
    {
      "epoch": 3.1436160468950356,
      "grad_norm": 0.3572027370281763,
      "learning_rate": 1.5539814053156774e-05,
      "loss": 0.686,
      "num_tokens": 6560015505.0,
      "step": 2147
    },
    {
      "epoch": 3.1450815167613118,
      "grad_norm": 0.358789914330423,
      "learning_rate": 1.5523880612985158e-05,
      "loss": 0.6911,
      "num_tokens": 6563094871.0,
      "step": 2148
    },
    {
      "epoch": 3.1465469866275875,
      "grad_norm": 0.35403223539985357,
      "learning_rate": 1.550795299966441e-05,
      "loss": 0.6885,
      "num_tokens": 6566200574.0,
      "step": 2149
    },
    {
      "epoch": 3.148012456493863,
      "grad_norm": 0.3435954417692675,
      "learning_rate": 1.549203122752531e-05,
      "loss": 0.6803,
      "num_tokens": 6569444759.0,
      "step": 2150
    },
    {
      "epoch": 3.1494779263601393,
      "grad_norm": 0.40243043510734866,
      "learning_rate": 1.5476115310893374e-05,
      "loss": 0.6894,
      "num_tokens": 6572517975.0,
      "step": 2151
    },
    {
      "epoch": 3.150943396226415,
      "grad_norm": 0.3551990424124929,
      "learning_rate": 1.546020526408886e-05,
      "loss": 0.6778,
      "num_tokens": 6575689139.0,
      "step": 2152
    },
    {
      "epoch": 3.1524088660926908,
      "grad_norm": 0.38906533579795216,
      "learning_rate": 1.544430110142674e-05,
      "loss": 0.6802,
      "num_tokens": 6578720664.0,
      "step": 2153
    },
    {
      "epoch": 3.153874335958967,
      "grad_norm": 0.3334290684495408,
      "learning_rate": 1.5428402837216698e-05,
      "loss": 0.6581,
      "num_tokens": 6581812094.0,
      "step": 2154
    },
    {
      "epoch": 3.1553398058252426,
      "grad_norm": 0.3580881405786349,
      "learning_rate": 1.541251048576309e-05,
      "loss": 0.6662,
      "num_tokens": 6584681433.0,
      "step": 2155
    },
    {
      "epoch": 3.156805275691519,
      "grad_norm": 0.33514221294765534,
      "learning_rate": 1.5396624061364986e-05,
      "loss": 0.6768,
      "num_tokens": 6587922659.0,
      "step": 2156
    },
    {
      "epoch": 3.1582707455577945,
      "grad_norm": 0.37803928258551445,
      "learning_rate": 1.538074357831609e-05,
      "loss": 0.6958,
      "num_tokens": 6591068884.0,
      "step": 2157
    },
    {
      "epoch": 3.1597362154240702,
      "grad_norm": 0.3396774408281775,
      "learning_rate": 1.5364869050904784e-05,
      "loss": 0.6746,
      "num_tokens": 6594124495.0,
      "step": 2158
    },
    {
      "epoch": 3.1612016852903464,
      "grad_norm": 0.3978873284791709,
      "learning_rate": 1.5349000493414064e-05,
      "loss": 0.6887,
      "num_tokens": 6597349994.0,
      "step": 2159
    },
    {
      "epoch": 3.162667155156622,
      "grad_norm": 0.3549172474043443,
      "learning_rate": 1.5333137920121603e-05,
      "loss": 0.6834,
      "num_tokens": 6600381722.0,
      "step": 2160
    },
    {
      "epoch": 3.164132625022898,
      "grad_norm": 0.4270583315333125,
      "learning_rate": 1.531728134529963e-05,
      "loss": 0.6929,
      "num_tokens": 6603341483.0,
      "step": 2161
    },
    {
      "epoch": 3.165598094889174,
      "grad_norm": 0.36780095211023944,
      "learning_rate": 1.530143078321503e-05,
      "loss": 0.6703,
      "num_tokens": 6606571024.0,
      "step": 2162
    },
    {
      "epoch": 3.1670635647554497,
      "grad_norm": 0.36676550355380316,
      "learning_rate": 1.528558624812923e-05,
      "loss": 0.6731,
      "num_tokens": 6609643142.0,
      "step": 2163
    },
    {
      "epoch": 3.1685290346217254,
      "grad_norm": 0.4319633664953416,
      "learning_rate": 1.5269747754298282e-05,
      "loss": 0.6894,
      "num_tokens": 6612827609.0,
      "step": 2164
    },
    {
      "epoch": 3.1699945044880016,
      "grad_norm": 0.32463116295474903,
      "learning_rate": 1.525391531597276e-05,
      "loss": 0.6818,
      "num_tokens": 6616032672.0,
      "step": 2165
    },
    {
      "epoch": 3.1714599743542773,
      "grad_norm": 0.37076375695490305,
      "learning_rate": 1.5238088947397821e-05,
      "loss": 0.6664,
      "num_tokens": 6619006262.0,
      "step": 2166
    },
    {
      "epoch": 3.172925444220553,
      "grad_norm": 0.3090841548748596,
      "learning_rate": 1.5222268662813147e-05,
      "loss": 0.6942,
      "num_tokens": 6622019797.0,
      "step": 2167
    },
    {
      "epoch": 3.174390914086829,
      "grad_norm": 0.39116074570869747,
      "learning_rate": 1.5206454476452943e-05,
      "loss": 0.6796,
      "num_tokens": 6625136846.0,
      "step": 2168
    },
    {
      "epoch": 3.175856383953105,
      "grad_norm": 0.33154282508782873,
      "learning_rate": 1.5190646402545932e-05,
      "loss": 0.6762,
      "num_tokens": 6628316094.0,
      "step": 2169
    },
    {
      "epoch": 3.177321853819381,
      "grad_norm": 0.3709285273692325,
      "learning_rate": 1.5174844455315337e-05,
      "loss": 0.6626,
      "num_tokens": 6631297128.0,
      "step": 2170
    },
    {
      "epoch": 3.1787873236856568,
      "grad_norm": 0.34144743802179606,
      "learning_rate": 1.5159048648978871e-05,
      "loss": 0.6745,
      "num_tokens": 6634506548.0,
      "step": 2171
    },
    {
      "epoch": 3.1802527935519325,
      "grad_norm": 0.3489259524261533,
      "learning_rate": 1.514325899774872e-05,
      "loss": 0.6913,
      "num_tokens": 6637528234.0,
      "step": 2172
    },
    {
      "epoch": 3.1817182634182086,
      "grad_norm": 0.37937038611210316,
      "learning_rate": 1.5127475515831518e-05,
      "loss": 0.6732,
      "num_tokens": 6640722083.0,
      "step": 2173
    },
    {
      "epoch": 3.1831837332844843,
      "grad_norm": 0.36287961485271486,
      "learning_rate": 1.5111698217428385e-05,
      "loss": 0.6968,
      "num_tokens": 6643702838.0,
      "step": 2174
    },
    {
      "epoch": 3.18464920315076,
      "grad_norm": 0.34096981213879085,
      "learning_rate": 1.5095927116734843e-05,
      "loss": 0.6967,
      "num_tokens": 6646717972.0,
      "step": 2175
    },
    {
      "epoch": 3.186114673017036,
      "grad_norm": 0.3358867655740543,
      "learning_rate": 1.5080162227940833e-05,
      "loss": 0.6856,
      "num_tokens": 6649613437.0,
      "step": 2176
    },
    {
      "epoch": 3.187580142883312,
      "grad_norm": 0.3395614003074484,
      "learning_rate": 1.5064403565230748e-05,
      "loss": 0.6881,
      "num_tokens": 6652592234.0,
      "step": 2177
    },
    {
      "epoch": 3.1890456127495876,
      "grad_norm": 0.375174015105079,
      "learning_rate": 1.5048651142783329e-05,
      "loss": 0.6751,
      "num_tokens": 6655710912.0,
      "step": 2178
    },
    {
      "epoch": 3.190511082615864,
      "grad_norm": 0.32186965779647103,
      "learning_rate": 1.5032904974771745e-05,
      "loss": 0.6874,
      "num_tokens": 6658867016.0,
      "step": 2179
    },
    {
      "epoch": 3.1919765524821395,
      "grad_norm": 0.3462342411753982,
      "learning_rate": 1.501716507536351e-05,
      "loss": 0.6901,
      "num_tokens": 6661876024.0,
      "step": 2180
    },
    {
      "epoch": 3.1934420223484157,
      "grad_norm": 0.31813234056432127,
      "learning_rate": 1.500143145872052e-05,
      "loss": 0.693,
      "num_tokens": 6665051133.0,
      "step": 2181
    },
    {
      "epoch": 3.1949074922146914,
      "grad_norm": 0.33435101040820614,
      "learning_rate": 1.4985704138998985e-05,
      "loss": 0.6772,
      "num_tokens": 6668097226.0,
      "step": 2182
    },
    {
      "epoch": 3.196372962080967,
      "grad_norm": 0.3261309598554135,
      "learning_rate": 1.4969983130349484e-05,
      "loss": 0.7026,
      "num_tokens": 6671169908.0,
      "step": 2183
    },
    {
      "epoch": 3.1978384319472433,
      "grad_norm": 0.3273073177324331,
      "learning_rate": 1.4954268446916898e-05,
      "loss": 0.702,
      "num_tokens": 6674136747.0,
      "step": 2184
    },
    {
      "epoch": 3.199303901813519,
      "grad_norm": 0.3377485054824053,
      "learning_rate": 1.4938560102840427e-05,
      "loss": 0.6635,
      "num_tokens": 6677220274.0,
      "step": 2185
    },
    {
      "epoch": 3.2007693716797947,
      "grad_norm": 0.34524027547480124,
      "learning_rate": 1.4922858112253557e-05,
      "loss": 0.6862,
      "num_tokens": 6680431798.0,
      "step": 2186
    },
    {
      "epoch": 3.202234841546071,
      "grad_norm": 0.3507686079267351,
      "learning_rate": 1.4907162489284069e-05,
      "loss": 0.6934,
      "num_tokens": 6683435069.0,
      "step": 2187
    },
    {
      "epoch": 3.2037003114123466,
      "grad_norm": 0.32999794628298756,
      "learning_rate": 1.4891473248053996e-05,
      "loss": 0.6973,
      "num_tokens": 6686566828.0,
      "step": 2188
    },
    {
      "epoch": 3.2051657812786223,
      "grad_norm": 0.3136046924382425,
      "learning_rate": 1.4875790402679661e-05,
      "loss": 0.6854,
      "num_tokens": 6689782224.0,
      "step": 2189
    },
    {
      "epoch": 3.2066312511448984,
      "grad_norm": 0.32515951032017804,
      "learning_rate": 1.4860113967271596e-05,
      "loss": 0.6856,
      "num_tokens": 6692846309.0,
      "step": 2190
    },
    {
      "epoch": 3.208096721011174,
      "grad_norm": 0.31650135833595877,
      "learning_rate": 1.4844443955934598e-05,
      "loss": 0.668,
      "num_tokens": 6695809367.0,
      "step": 2191
    },
    {
      "epoch": 3.20956219087745,
      "grad_norm": 0.40281201083599827,
      "learning_rate": 1.4828780382767653e-05,
      "loss": 0.6798,
      "num_tokens": 6698745000.0,
      "step": 2192
    },
    {
      "epoch": 3.211027660743726,
      "grad_norm": 0.3463172416490409,
      "learning_rate": 1.4813123261863985e-05,
      "loss": 0.6805,
      "num_tokens": 6701836956.0,
      "step": 2193
    },
    {
      "epoch": 3.2124931306100017,
      "grad_norm": 0.38024279603097016,
      "learning_rate": 1.4797472607310987e-05,
      "loss": 0.673,
      "num_tokens": 6704810294.0,
      "step": 2194
    },
    {
      "epoch": 3.213958600476278,
      "grad_norm": 0.33588207172315354,
      "learning_rate": 1.4781828433190254e-05,
      "loss": 0.6856,
      "num_tokens": 6707824469.0,
      "step": 2195
    },
    {
      "epoch": 3.2154240703425536,
      "grad_norm": 0.3703923094204619,
      "learning_rate": 1.4766190753577529e-05,
      "loss": 0.6889,
      "num_tokens": 6710885465.0,
      "step": 2196
    },
    {
      "epoch": 3.2168895402088293,
      "grad_norm": 0.3703740975686388,
      "learning_rate": 1.4750559582542736e-05,
      "loss": 0.6688,
      "num_tokens": 6713851628.0,
      "step": 2197
    },
    {
      "epoch": 3.2183550100751055,
      "grad_norm": 0.3144649156416061,
      "learning_rate": 1.4734934934149916e-05,
      "loss": 0.6801,
      "num_tokens": 6716764443.0,
      "step": 2198
    },
    {
      "epoch": 3.219820479941381,
      "grad_norm": 0.38900381526559935,
      "learning_rate": 1.4719316822457273e-05,
      "loss": 0.678,
      "num_tokens": 6719999859.0,
      "step": 2199
    },
    {
      "epoch": 3.221285949807657,
      "grad_norm": 0.33524956255487665,
      "learning_rate": 1.4703705261517099e-05,
      "loss": 0.6844,
      "num_tokens": 6722962041.0,
      "step": 2200
    },
    {
      "epoch": 3.222751419673933,
      "grad_norm": 0.3484951221426812,
      "learning_rate": 1.4688100265375811e-05,
      "loss": 0.6928,
      "num_tokens": 6725972779.0,
      "step": 2201
    },
    {
      "epoch": 3.224216889540209,
      "grad_norm": 0.329299307887011,
      "learning_rate": 1.4672501848073912e-05,
      "loss": 0.6747,
      "num_tokens": 6729014712.0,
      "step": 2202
    },
    {
      "epoch": 3.2256823594064845,
      "grad_norm": 0.3329634018053515,
      "learning_rate": 1.4656910023645995e-05,
      "loss": 0.6745,
      "num_tokens": 6732201336.0,
      "step": 2203
    },
    {
      "epoch": 3.2271478292727607,
      "grad_norm": 0.322502127550236,
      "learning_rate": 1.4641324806120696e-05,
      "loss": 0.6781,
      "num_tokens": 6735071700.0,
      "step": 2204
    },
    {
      "epoch": 3.2286132991390364,
      "grad_norm": 0.3185411588236215,
      "learning_rate": 1.4625746209520749e-05,
      "loss": 0.6826,
      "num_tokens": 6738066161.0,
      "step": 2205
    },
    {
      "epoch": 3.2300787690053125,
      "grad_norm": 0.31764081750882994,
      "learning_rate": 1.4610174247862876e-05,
      "loss": 0.6809,
      "num_tokens": 6740974558.0,
      "step": 2206
    },
    {
      "epoch": 3.2315442388715883,
      "grad_norm": 0.29053229419035065,
      "learning_rate": 1.4594608935157885e-05,
      "loss": 0.6694,
      "num_tokens": 6744063295.0,
      "step": 2207
    },
    {
      "epoch": 3.233009708737864,
      "grad_norm": 0.29376965575001546,
      "learning_rate": 1.4579050285410566e-05,
      "loss": 0.6819,
      "num_tokens": 6747030380.0,
      "step": 2208
    },
    {
      "epoch": 3.23447517860414,
      "grad_norm": 0.31032965858496087,
      "learning_rate": 1.456349831261973e-05,
      "loss": 0.7037,
      "num_tokens": 6749760356.0,
      "step": 2209
    },
    {
      "epoch": 3.235940648470416,
      "grad_norm": 0.293484953578673,
      "learning_rate": 1.4547953030778162e-05,
      "loss": 0.6918,
      "num_tokens": 6752687510.0,
      "step": 2210
    },
    {
      "epoch": 3.2374061183366916,
      "grad_norm": 0.3403444945496453,
      "learning_rate": 1.4532414453872644e-05,
      "loss": 0.6764,
      "num_tokens": 6755756932.0,
      "step": 2211
    },
    {
      "epoch": 3.2388715882029677,
      "grad_norm": 0.29676807109039155,
      "learning_rate": 1.4516882595883922e-05,
      "loss": 0.667,
      "num_tokens": 6758615049.0,
      "step": 2212
    },
    {
      "epoch": 3.2403370580692434,
      "grad_norm": 0.35634955344277586,
      "learning_rate": 1.4501357470786701e-05,
      "loss": 0.6778,
      "num_tokens": 6761802248.0,
      "step": 2213
    },
    {
      "epoch": 3.241802527935519,
      "grad_norm": 0.3307800128661071,
      "learning_rate": 1.448583909254961e-05,
      "loss": 0.6829,
      "num_tokens": 6765019175.0,
      "step": 2214
    },
    {
      "epoch": 3.2432679978017953,
      "grad_norm": 0.3741646783456708,
      "learning_rate": 1.4470327475135237e-05,
      "loss": 0.6812,
      "num_tokens": 6768091832.0,
      "step": 2215
    },
    {
      "epoch": 3.244733467668071,
      "grad_norm": 0.32907694948207883,
      "learning_rate": 1.4454822632500057e-05,
      "loss": 0.67,
      "num_tokens": 6771247479.0,
      "step": 2216
    },
    {
      "epoch": 3.2461989375343467,
      "grad_norm": 0.36233687617378973,
      "learning_rate": 1.4439324578594449e-05,
      "loss": 0.6689,
      "num_tokens": 6774254048.0,
      "step": 2217
    },
    {
      "epoch": 3.247664407400623,
      "grad_norm": 0.36024035839932406,
      "learning_rate": 1.4423833327362733e-05,
      "loss": 0.6776,
      "num_tokens": 6777318945.0,
      "step": 2218
    },
    {
      "epoch": 3.2491298772668986,
      "grad_norm": 0.33130001635500456,
      "learning_rate": 1.4408348892743038e-05,
      "loss": 0.6778,
      "num_tokens": 6780468893.0,
      "step": 2219
    },
    {
      "epoch": 3.2505953471331743,
      "grad_norm": 0.3529350825995393,
      "learning_rate": 1.4392871288667415e-05,
      "loss": 0.6866,
      "num_tokens": 6783364596.0,
      "step": 2220
    },
    {
      "epoch": 3.2520608169994505,
      "grad_norm": 0.31452154488881984,
      "learning_rate": 1.437740052906174e-05,
      "loss": 0.6714,
      "num_tokens": 6786094906.0,
      "step": 2221
    },
    {
      "epoch": 3.253526286865726,
      "grad_norm": 0.33598379488235836,
      "learning_rate": 1.4361936627845743e-05,
      "loss": 0.6772,
      "num_tokens": 6789169180.0,
      "step": 2222
    },
    {
      "epoch": 3.2549917567320024,
      "grad_norm": 0.351336948034945,
      "learning_rate": 1.434647959893297e-05,
      "loss": 0.6863,
      "num_tokens": 6792383107.0,
      "step": 2223
    },
    {
      "epoch": 3.256457226598278,
      "grad_norm": 0.3585054378367536,
      "learning_rate": 1.4331029456230815e-05,
      "loss": 0.6828,
      "num_tokens": 6795587375.0,
      "step": 2224
    },
    {
      "epoch": 3.257922696464554,
      "grad_norm": 0.3414079607236522,
      "learning_rate": 1.431558621364043e-05,
      "loss": 0.6763,
      "num_tokens": 6798888543.0,
      "step": 2225
    },
    {
      "epoch": 3.25938816633083,
      "grad_norm": 0.3327998028330822,
      "learning_rate": 1.4300149885056796e-05,
      "loss": 0.6807,
      "num_tokens": 6801804347.0,
      "step": 2226
    },
    {
      "epoch": 3.2608536361971057,
      "grad_norm": 0.3167086853747828,
      "learning_rate": 1.428472048436866e-05,
      "loss": 0.667,
      "num_tokens": 6804831921.0,
      "step": 2227
    },
    {
      "epoch": 3.2623191060633814,
      "grad_norm": 0.3213085026847831,
      "learning_rate": 1.426929802545853e-05,
      "loss": 0.6764,
      "num_tokens": 6807934073.0,
      "step": 2228
    },
    {
      "epoch": 3.2637845759296575,
      "grad_norm": 0.3155746741670374,
      "learning_rate": 1.4253882522202678e-05,
      "loss": 0.682,
      "num_tokens": 6810736900.0,
      "step": 2229
    },
    {
      "epoch": 3.2652500457959333,
      "grad_norm": 0.3429565517590045,
      "learning_rate": 1.4238473988471114e-05,
      "loss": 0.6845,
      "num_tokens": 6813960313.0,
      "step": 2230
    },
    {
      "epoch": 3.2667155156622094,
      "grad_norm": 0.31215201273172505,
      "learning_rate": 1.4223072438127568e-05,
      "loss": 0.687,
      "num_tokens": 6816901576.0,
      "step": 2231
    },
    {
      "epoch": 3.268180985528485,
      "grad_norm": 0.3655182025342804,
      "learning_rate": 1.4207677885029515e-05,
      "loss": 0.6779,
      "num_tokens": 6819882404.0,
      "step": 2232
    },
    {
      "epoch": 3.269646455394761,
      "grad_norm": 0.343901315186695,
      "learning_rate": 1.4192290343028092e-05,
      "loss": 0.6814,
      "num_tokens": 6823063949.0,
      "step": 2233
    },
    {
      "epoch": 3.271111925261037,
      "grad_norm": 0.3649952117226686,
      "learning_rate": 1.4176909825968169e-05,
      "loss": 0.6695,
      "num_tokens": 6826207270.0,
      "step": 2234
    },
    {
      "epoch": 3.2725773951273127,
      "grad_norm": 0.3150627365726366,
      "learning_rate": 1.416153634768827e-05,
      "loss": 0.6821,
      "num_tokens": 6829268598.0,
      "step": 2235
    },
    {
      "epoch": 3.2740428649935884,
      "grad_norm": 0.3448130724101046,
      "learning_rate": 1.4146169922020593e-05,
      "loss": 0.6811,
      "num_tokens": 6832319359.0,
      "step": 2236
    },
    {
      "epoch": 3.2755083348598646,
      "grad_norm": 0.3204323429495144,
      "learning_rate": 1.413081056279098e-05,
      "loss": 0.679,
      "num_tokens": 6835395884.0,
      "step": 2237
    },
    {
      "epoch": 3.2769738047261403,
      "grad_norm": 0.3449419329255864,
      "learning_rate": 1.4115458283818954e-05,
      "loss": 0.684,
      "num_tokens": 6838550890.0,
      "step": 2238
    },
    {
      "epoch": 3.278439274592416,
      "grad_norm": 0.29644439843354614,
      "learning_rate": 1.4100113098917607e-05,
      "loss": 0.684,
      "num_tokens": 6841579138.0,
      "step": 2239
    },
    {
      "epoch": 3.279904744458692,
      "grad_norm": 0.294865551902728,
      "learning_rate": 1.4084775021893694e-05,
      "loss": 0.6981,
      "num_tokens": 6844736173.0,
      "step": 2240
    },
    {
      "epoch": 3.281370214324968,
      "grad_norm": 0.34089358510716355,
      "learning_rate": 1.4069444066547569e-05,
      "loss": 0.6879,
      "num_tokens": 6847846050.0,
      "step": 2241
    },
    {
      "epoch": 3.2828356841912436,
      "grad_norm": 0.29380175599429265,
      "learning_rate": 1.4054120246673155e-05,
      "loss": 0.6705,
      "num_tokens": 6851128281.0,
      "step": 2242
    },
    {
      "epoch": 3.2843011540575198,
      "grad_norm": 0.3099267246707363,
      "learning_rate": 1.4038803576057985e-05,
      "loss": 0.6539,
      "num_tokens": 6854044981.0,
      "step": 2243
    },
    {
      "epoch": 3.2857666239237955,
      "grad_norm": 0.3317476191532662,
      "learning_rate": 1.4023494068483126e-05,
      "loss": 0.6929,
      "num_tokens": 6857093702.0,
      "step": 2244
    },
    {
      "epoch": 3.287232093790071,
      "grad_norm": 0.3342460252328149,
      "learning_rate": 1.4008191737723231e-05,
      "loss": 0.6808,
      "num_tokens": 6860160096.0,
      "step": 2245
    },
    {
      "epoch": 3.2886975636563474,
      "grad_norm": 0.31467113768075833,
      "learning_rate": 1.3992896597546488e-05,
      "loss": 0.6666,
      "num_tokens": 6863407765.0,
      "step": 2246
    },
    {
      "epoch": 3.290163033522623,
      "grad_norm": 0.3233439638387,
      "learning_rate": 1.3977608661714597e-05,
      "loss": 0.6777,
      "num_tokens": 6866334946.0,
      "step": 2247
    },
    {
      "epoch": 3.2916285033888992,
      "grad_norm": 0.3252819294460103,
      "learning_rate": 1.3962327943982801e-05,
      "loss": 0.7073,
      "num_tokens": 6869325918.0,
      "step": 2248
    },
    {
      "epoch": 3.293093973255175,
      "grad_norm": 0.33384092195596515,
      "learning_rate": 1.3947054458099834e-05,
      "loss": 0.6836,
      "num_tokens": 6872365514.0,
      "step": 2249
    },
    {
      "epoch": 3.2945594431214507,
      "grad_norm": 0.3030650556123793,
      "learning_rate": 1.3931788217807922e-05,
      "loss": 0.699,
      "num_tokens": 6875384995.0,
      "step": 2250
    },
    {
      "epoch": 3.296024912987727,
      "grad_norm": 0.30004336519608776,
      "learning_rate": 1.3916529236842778e-05,
      "loss": 0.6613,
      "num_tokens": 6878390591.0,
      "step": 2251
    },
    {
      "epoch": 3.2974903828540025,
      "grad_norm": 0.3235219113121467,
      "learning_rate": 1.390127752893359e-05,
      "loss": 0.6699,
      "num_tokens": 6881471440.0,
      "step": 2252
    },
    {
      "epoch": 3.2989558527202782,
      "grad_norm": 0.3392067338882081,
      "learning_rate": 1.3886033107802978e-05,
      "loss": 0.6939,
      "num_tokens": 6884511463.0,
      "step": 2253
    },
    {
      "epoch": 3.3004213225865544,
      "grad_norm": 0.3615886904528167,
      "learning_rate": 1.3870795987167038e-05,
      "loss": 0.6648,
      "num_tokens": 6887523540.0,
      "step": 2254
    },
    {
      "epoch": 3.30188679245283,
      "grad_norm": 0.34664615274847677,
      "learning_rate": 1.3855566180735274e-05,
      "loss": 0.6865,
      "num_tokens": 6890635083.0,
      "step": 2255
    },
    {
      "epoch": 3.3033522623191063,
      "grad_norm": 0.36629244917730075,
      "learning_rate": 1.3840343702210613e-05,
      "loss": 0.6852,
      "num_tokens": 6893632234.0,
      "step": 2256
    },
    {
      "epoch": 3.304817732185382,
      "grad_norm": 0.36467042089702134,
      "learning_rate": 1.3825128565289397e-05,
      "loss": 0.6884,
      "num_tokens": 6896381691.0,
      "step": 2257
    },
    {
      "epoch": 3.3062832020516577,
      "grad_norm": 0.3308979470901881,
      "learning_rate": 1.380992078366135e-05,
      "loss": 0.6677,
      "num_tokens": 6899525654.0,
      "step": 2258
    },
    {
      "epoch": 3.307748671917934,
      "grad_norm": 0.34347012678799804,
      "learning_rate": 1.379472037100959e-05,
      "loss": 0.6549,
      "num_tokens": 6902535258.0,
      "step": 2259
    },
    {
      "epoch": 3.3092141417842096,
      "grad_norm": 0.33596519839731187,
      "learning_rate": 1.3779527341010604e-05,
      "loss": 0.7001,
      "num_tokens": 6905336734.0,
      "step": 2260
    },
    {
      "epoch": 3.3106796116504853,
      "grad_norm": 0.31491775577342085,
      "learning_rate": 1.3764341707334228e-05,
      "loss": 0.6823,
      "num_tokens": 6908403416.0,
      "step": 2261
    },
    {
      "epoch": 3.3121450815167615,
      "grad_norm": 0.2886369860619765,
      "learning_rate": 1.374916348364364e-05,
      "loss": 0.6919,
      "num_tokens": 6911201668.0,
      "step": 2262
    },
    {
      "epoch": 3.313610551383037,
      "grad_norm": 0.3340409051853242,
      "learning_rate": 1.3733992683595383e-05,
      "loss": 0.6796,
      "num_tokens": 6914253734.0,
      "step": 2263
    },
    {
      "epoch": 3.315076021249313,
      "grad_norm": 0.29590315770498204,
      "learning_rate": 1.3718829320839264e-05,
      "loss": 0.6678,
      "num_tokens": 6917493336.0,
      "step": 2264
    },
    {
      "epoch": 3.316541491115589,
      "grad_norm": 0.3194743323388789,
      "learning_rate": 1.3703673409018454e-05,
      "loss": 0.6881,
      "num_tokens": 6920427891.0,
      "step": 2265
    },
    {
      "epoch": 3.3180069609818648,
      "grad_norm": 0.3250552601122571,
      "learning_rate": 1.3688524961769396e-05,
      "loss": 0.6652,
      "num_tokens": 6923427226.0,
      "step": 2266
    },
    {
      "epoch": 3.3194724308481405,
      "grad_norm": 0.3056619755452292,
      "learning_rate": 1.3673383992721805e-05,
      "loss": 0.6851,
      "num_tokens": 6926285418.0,
      "step": 2267
    },
    {
      "epoch": 3.3209379007144166,
      "grad_norm": 0.3689544438387145,
      "learning_rate": 1.365825051549868e-05,
      "loss": 0.6944,
      "num_tokens": 6929304850.0,
      "step": 2268
    },
    {
      "epoch": 3.3224033705806923,
      "grad_norm": 0.3307073899899554,
      "learning_rate": 1.3643124543716306e-05,
      "loss": 0.6833,
      "num_tokens": 6932252100.0,
      "step": 2269
    },
    {
      "epoch": 3.323868840446968,
      "grad_norm": 0.322901940669808,
      "learning_rate": 1.362800609098415e-05,
      "loss": 0.6703,
      "num_tokens": 6935438106.0,
      "step": 2270
    },
    {
      "epoch": 3.3253343103132442,
      "grad_norm": 0.3295309240327121,
      "learning_rate": 1.3612895170904978e-05,
      "loss": 0.6671,
      "num_tokens": 6938392196.0,
      "step": 2271
    },
    {
      "epoch": 3.32679978017952,
      "grad_norm": 0.3130084215237398,
      "learning_rate": 1.3597791797074747e-05,
      "loss": 0.6879,
      "num_tokens": 6941452086.0,
      "step": 2272
    },
    {
      "epoch": 3.328265250045796,
      "grad_norm": 0.31682734957264447,
      "learning_rate": 1.3582695983082628e-05,
      "loss": 0.6954,
      "num_tokens": 6944669372.0,
      "step": 2273
    },
    {
      "epoch": 3.329730719912072,
      "grad_norm": 0.3219980580465248,
      "learning_rate": 1.356760774251099e-05,
      "loss": 0.6681,
      "num_tokens": 6947734641.0,
      "step": 2274
    },
    {
      "epoch": 3.3311961897783475,
      "grad_norm": 0.3010604462563766,
      "learning_rate": 1.3552527088935398e-05,
      "loss": 0.6838,
      "num_tokens": 6950914445.0,
      "step": 2275
    },
    {
      "epoch": 3.3326616596446237,
      "grad_norm": 0.34640704965083374,
      "learning_rate": 1.353745403592457e-05,
      "loss": 0.6705,
      "num_tokens": 6954093926.0,
      "step": 2276
    },
    {
      "epoch": 3.3341271295108994,
      "grad_norm": 0.2888482138610229,
      "learning_rate": 1.3522388597040414e-05,
      "loss": 0.651,
      "num_tokens": 6957087740.0,
      "step": 2277
    },
    {
      "epoch": 3.335592599377175,
      "grad_norm": 0.31745608301724065,
      "learning_rate": 1.3507330785837951e-05,
      "loss": 0.6707,
      "num_tokens": 6960164577.0,
      "step": 2278
    },
    {
      "epoch": 3.3370580692434513,
      "grad_norm": 0.32897975693350096,
      "learning_rate": 1.3492280615865379e-05,
      "loss": 0.6619,
      "num_tokens": 6963272086.0,
      "step": 2279
    },
    {
      "epoch": 3.338523539109727,
      "grad_norm": 0.4181102507337895,
      "learning_rate": 1.347723810066399e-05,
      "loss": 0.6743,
      "num_tokens": 6966349528.0,
      "step": 2280
    },
    {
      "epoch": 3.339989008976003,
      "grad_norm": 0.36363650305439443,
      "learning_rate": 1.3462203253768208e-05,
      "loss": 0.7085,
      "num_tokens": 6969187044.0,
      "step": 2281
    },
    {
      "epoch": 3.341454478842279,
      "grad_norm": 0.4097173039169629,
      "learning_rate": 1.3447176088705534e-05,
      "loss": 0.6689,
      "num_tokens": 6972410726.0,
      "step": 2282
    },
    {
      "epoch": 3.3429199487085546,
      "grad_norm": 0.34231706961005426,
      "learning_rate": 1.3432156618996595e-05,
      "loss": 0.6737,
      "num_tokens": 6975349627.0,
      "step": 2283
    },
    {
      "epoch": 3.3443854185748307,
      "grad_norm": 0.3729967767997746,
      "learning_rate": 1.3417144858155047e-05,
      "loss": 0.6815,
      "num_tokens": 6978431669.0,
      "step": 2284
    },
    {
      "epoch": 3.3458508884411065,
      "grad_norm": 0.39461721416163326,
      "learning_rate": 1.3402140819687657e-05,
      "loss": 0.6717,
      "num_tokens": 6981741050.0,
      "step": 2285
    },
    {
      "epoch": 3.347316358307382,
      "grad_norm": 0.3784022594140698,
      "learning_rate": 1.338714451709421e-05,
      "loss": 0.6783,
      "num_tokens": 6984720001.0,
      "step": 2286
    },
    {
      "epoch": 3.3487818281736583,
      "grad_norm": 0.3981178606238951,
      "learning_rate": 1.3372155963867543e-05,
      "loss": 0.6753,
      "num_tokens": 6987675448.0,
      "step": 2287
    },
    {
      "epoch": 3.350247298039934,
      "grad_norm": 0.355069531791707,
      "learning_rate": 1.3357175173493521e-05,
      "loss": 0.6695,
      "num_tokens": 6990732502.0,
      "step": 2288
    },
    {
      "epoch": 3.3517127679062098,
      "grad_norm": 0.3193229119423912,
      "learning_rate": 1.3342202159451028e-05,
      "loss": 0.667,
      "num_tokens": 6993940342.0,
      "step": 2289
    },
    {
      "epoch": 3.353178237772486,
      "grad_norm": 0.3056923199213101,
      "learning_rate": 1.3327236935211932e-05,
      "loss": 0.677,
      "num_tokens": 6997146027.0,
      "step": 2290
    },
    {
      "epoch": 3.3546437076387616,
      "grad_norm": 0.29958428572163304,
      "learning_rate": 1.3312279514241131e-05,
      "loss": 0.6651,
      "num_tokens": 7000152649.0,
      "step": 2291
    },
    {
      "epoch": 3.3561091775050373,
      "grad_norm": 0.3467557574344487,
      "learning_rate": 1.3297329909996448e-05,
      "loss": 0.6646,
      "num_tokens": 7003532383.0,
      "step": 2292
    },
    {
      "epoch": 3.3575746473713135,
      "grad_norm": 0.3267572385991316,
      "learning_rate": 1.3282388135928728e-05,
      "loss": 0.6885,
      "num_tokens": 7006607278.0,
      "step": 2293
    },
    {
      "epoch": 3.359040117237589,
      "grad_norm": 0.3206475661589553,
      "learning_rate": 1.3267454205481733e-05,
      "loss": 0.687,
      "num_tokens": 7009382798.0,
      "step": 2294
    },
    {
      "epoch": 3.360505587103865,
      "grad_norm": 0.32368929997365975,
      "learning_rate": 1.3252528132092187e-05,
      "loss": 0.6833,
      "num_tokens": 7012298136.0,
      "step": 2295
    },
    {
      "epoch": 3.361971056970141,
      "grad_norm": 0.3094890614542978,
      "learning_rate": 1.323760992918972e-05,
      "loss": 0.699,
      "num_tokens": 7015232330.0,
      "step": 2296
    },
    {
      "epoch": 3.363436526836417,
      "grad_norm": 0.3487273707901496,
      "learning_rate": 1.3222699610196937e-05,
      "loss": 0.6736,
      "num_tokens": 7018172998.0,
      "step": 2297
    },
    {
      "epoch": 3.364901996702693,
      "grad_norm": 0.3515686272653105,
      "learning_rate": 1.3207797188529266e-05,
      "loss": 0.6887,
      "num_tokens": 7021184101.0,
      "step": 2298
    },
    {
      "epoch": 3.3663674665689687,
      "grad_norm": 0.29650476465713105,
      "learning_rate": 1.3192902677595103e-05,
      "loss": 0.6703,
      "num_tokens": 7024244543.0,
      "step": 2299
    },
    {
      "epoch": 3.3678329364352444,
      "grad_norm": 0.33592640509617094,
      "learning_rate": 1.3178016090795698e-05,
      "loss": 0.6844,
      "num_tokens": 7027028689.0,
      "step": 2300
    },
    {
      "epoch": 3.3692984063015206,
      "grad_norm": 0.3194273158667683,
      "learning_rate": 1.3163137441525149e-05,
      "loss": 0.6698,
      "num_tokens": 7030264147.0,
      "step": 2301
    },
    {
      "epoch": 3.3707638761677963,
      "grad_norm": 0.3753814485627654,
      "learning_rate": 1.3148266743170455e-05,
      "loss": 0.6836,
      "num_tokens": 7033089490.0,
      "step": 2302
    },
    {
      "epoch": 3.372229346034072,
      "grad_norm": 0.34706312204489564,
      "learning_rate": 1.3133404009111435e-05,
      "loss": 0.6832,
      "num_tokens": 7036143714.0,
      "step": 2303
    },
    {
      "epoch": 3.373694815900348,
      "grad_norm": 0.331858292708562,
      "learning_rate": 1.3118549252720749e-05,
      "loss": 0.6656,
      "num_tokens": 7039128864.0,
      "step": 2304
    },
    {
      "epoch": 3.375160285766624,
      "grad_norm": 0.3721720300610116,
      "learning_rate": 1.3103702487363874e-05,
      "loss": 0.6817,
      "num_tokens": 7042456485.0,
      "step": 2305
    },
    {
      "epoch": 3.3766257556329,
      "grad_norm": 0.3209218557529914,
      "learning_rate": 1.3088863726399104e-05,
      "loss": 0.6604,
      "num_tokens": 7045541868.0,
      "step": 2306
    },
    {
      "epoch": 3.3780912254991757,
      "grad_norm": 0.34855914856981385,
      "learning_rate": 1.307403298317752e-05,
      "loss": 0.6556,
      "num_tokens": 7048531803.0,
      "step": 2307
    },
    {
      "epoch": 3.3795566953654514,
      "grad_norm": 0.3174212365677777,
      "learning_rate": 1.305921027104301e-05,
      "loss": 0.6729,
      "num_tokens": 7051528726.0,
      "step": 2308
    },
    {
      "epoch": 3.3810221652317276,
      "grad_norm": 0.39004927569528114,
      "learning_rate": 1.3044395603332218e-05,
      "loss": 0.6715,
      "num_tokens": 7054592015.0,
      "step": 2309
    },
    {
      "epoch": 3.3824876350980033,
      "grad_norm": 0.3151158737092963,
      "learning_rate": 1.3029588993374555e-05,
      "loss": 0.6728,
      "num_tokens": 7057649264.0,
      "step": 2310
    },
    {
      "epoch": 3.383953104964279,
      "grad_norm": 0.3654467531907248,
      "learning_rate": 1.3014790454492188e-05,
      "loss": 0.6908,
      "num_tokens": 7060582662.0,
      "step": 2311
    },
    {
      "epoch": 3.385418574830555,
      "grad_norm": 0.3082319094156999,
      "learning_rate": 1.3000000000000006e-05,
      "loss": 0.666,
      "num_tokens": 7063631182.0,
      "step": 2312
    },
    {
      "epoch": 3.386884044696831,
      "grad_norm": 0.39086602685852867,
      "learning_rate": 1.2985217643205635e-05,
      "loss": 0.6874,
      "num_tokens": 7066681921.0,
      "step": 2313
    },
    {
      "epoch": 3.3883495145631066,
      "grad_norm": 0.35380598996053286,
      "learning_rate": 1.2970443397409429e-05,
      "loss": 0.6687,
      "num_tokens": 7069724670.0,
      "step": 2314
    },
    {
      "epoch": 3.389814984429383,
      "grad_norm": 0.3958640733466791,
      "learning_rate": 1.295567727590441e-05,
      "loss": 0.6802,
      "num_tokens": 7072597507.0,
      "step": 2315
    },
    {
      "epoch": 3.3912804542956585,
      "grad_norm": 0.3614978076753916,
      "learning_rate": 1.2940919291976328e-05,
      "loss": 0.6904,
      "num_tokens": 7075664618.0,
      "step": 2316
    },
    {
      "epoch": 3.392745924161934,
      "grad_norm": 0.3723945975632481,
      "learning_rate": 1.2926169458903586e-05,
      "loss": 0.6831,
      "num_tokens": 7078742799.0,
      "step": 2317
    },
    {
      "epoch": 3.3942113940282104,
      "grad_norm": 0.3735268200488837,
      "learning_rate": 1.291142778995726e-05,
      "loss": 0.6679,
      "num_tokens": 7081768854.0,
      "step": 2318
    },
    {
      "epoch": 3.395676863894486,
      "grad_norm": 0.2986545121820524,
      "learning_rate": 1.2896694298401084e-05,
      "loss": 0.6657,
      "num_tokens": 7085080146.0,
      "step": 2319
    },
    {
      "epoch": 3.397142333760762,
      "grad_norm": 0.3623872259314723,
      "learning_rate": 1.288196899749143e-05,
      "loss": 0.6667,
      "num_tokens": 7087831548.0,
      "step": 2320
    },
    {
      "epoch": 3.398607803627038,
      "grad_norm": 0.32018000226353216,
      "learning_rate": 1.2867251900477293e-05,
      "loss": 0.6823,
      "num_tokens": 7090654567.0,
      "step": 2321
    },
    {
      "epoch": 3.4000732734933137,
      "grad_norm": 0.31583539861420573,
      "learning_rate": 1.2852543020600321e-05,
      "loss": 0.6719,
      "num_tokens": 7093790563.0,
      "step": 2322
    },
    {
      "epoch": 3.40153874335959,
      "grad_norm": 0.2811197944488693,
      "learning_rate": 1.2837842371094713e-05,
      "loss": 0.6674,
      "num_tokens": 7096858180.0,
      "step": 2323
    },
    {
      "epoch": 3.4030042132258655,
      "grad_norm": 0.32856016218185036,
      "learning_rate": 1.2823149965187311e-05,
      "loss": 0.6946,
      "num_tokens": 7100064381.0,
      "step": 2324
    },
    {
      "epoch": 3.4044696830921413,
      "grad_norm": 0.2769950924241463,
      "learning_rate": 1.2808465816097522e-05,
      "loss": 0.6804,
      "num_tokens": 7102999829.0,
      "step": 2325
    },
    {
      "epoch": 3.4059351529584174,
      "grad_norm": 0.34330362374469703,
      "learning_rate": 1.2793789937037317e-05,
      "loss": 0.6695,
      "num_tokens": 7106216312.0,
      "step": 2326
    },
    {
      "epoch": 3.407400622824693,
      "grad_norm": 0.2856780301131587,
      "learning_rate": 1.2779122341211232e-05,
      "loss": 0.6643,
      "num_tokens": 7109321156.0,
      "step": 2327
    },
    {
      "epoch": 3.408866092690969,
      "grad_norm": 0.30505365461867,
      "learning_rate": 1.2764463041816368e-05,
      "loss": 0.6877,
      "num_tokens": 7112020274.0,
      "step": 2328
    },
    {
      "epoch": 3.410331562557245,
      "grad_norm": 0.29084219421018226,
      "learning_rate": 1.2749812052042318e-05,
      "loss": 0.6564,
      "num_tokens": 7115113898.0,
      "step": 2329
    },
    {
      "epoch": 3.4117970324235207,
      "grad_norm": 0.33118526188579545,
      "learning_rate": 1.2735169385071249e-05,
      "loss": 0.6818,
      "num_tokens": 7118094291.0,
      "step": 2330
    },
    {
      "epoch": 3.413262502289797,
      "grad_norm": 0.27904179651743855,
      "learning_rate": 1.2720535054077802e-05,
      "loss": 0.6641,
      "num_tokens": 7121203622.0,
      "step": 2331
    },
    {
      "epoch": 3.4147279721560726,
      "grad_norm": 0.3406836944996538,
      "learning_rate": 1.2705909072229136e-05,
      "loss": 0.6671,
      "num_tokens": 7124401148.0,
      "step": 2332
    },
    {
      "epoch": 3.4161934420223483,
      "grad_norm": 0.30258100250347625,
      "learning_rate": 1.2691291452684897e-05,
      "loss": 0.6485,
      "num_tokens": 7127684983.0,
      "step": 2333
    },
    {
      "epoch": 3.4176589118886245,
      "grad_norm": 0.34755552490240693,
      "learning_rate": 1.2676682208597195e-05,
      "loss": 0.6836,
      "num_tokens": 7130540166.0,
      "step": 2334
    },
    {
      "epoch": 3.4191243817549,
      "grad_norm": 0.28949444078834957,
      "learning_rate": 1.266208135311061e-05,
      "loss": 0.6498,
      "num_tokens": 7133701750.0,
      "step": 2335
    },
    {
      "epoch": 3.420589851621176,
      "grad_norm": 0.3960667964491369,
      "learning_rate": 1.2647488899362202e-05,
      "loss": 0.6699,
      "num_tokens": 7136741270.0,
      "step": 2336
    },
    {
      "epoch": 3.422055321487452,
      "grad_norm": 0.3280653859184688,
      "learning_rate": 1.2632904860481411e-05,
      "loss": 0.6993,
      "num_tokens": 7139718746.0,
      "step": 2337
    },
    {
      "epoch": 3.4235207913537278,
      "grad_norm": 0.3576538292986628,
      "learning_rate": 1.2618329249590169e-05,
      "loss": 0.6735,
      "num_tokens": 7142678424.0,
      "step": 2338
    },
    {
      "epoch": 3.4249862612200035,
      "grad_norm": 0.32212034405204354,
      "learning_rate": 1.2603762079802786e-05,
      "loss": 0.6802,
      "num_tokens": 7145654006.0,
      "step": 2339
    },
    {
      "epoch": 3.4264517310862797,
      "grad_norm": 0.30672463488346974,
      "learning_rate": 1.258920336422599e-05,
      "loss": 0.6578,
      "num_tokens": 7149145769.0,
      "step": 2340
    },
    {
      "epoch": 3.4279172009525554,
      "grad_norm": 0.31720751614403514,
      "learning_rate": 1.2574653115958898e-05,
      "loss": 0.6972,
      "num_tokens": 7152094042.0,
      "step": 2341
    },
    {
      "epoch": 3.429382670818831,
      "grad_norm": 0.3450719675940326,
      "learning_rate": 1.2560111348093018e-05,
      "loss": 0.6594,
      "num_tokens": 7155080596.0,
      "step": 2342
    },
    {
      "epoch": 3.4308481406851072,
      "grad_norm": 0.328726912495686,
      "learning_rate": 1.254557807371222e-05,
      "loss": 0.6911,
      "num_tokens": 7158240618.0,
      "step": 2343
    },
    {
      "epoch": 3.432313610551383,
      "grad_norm": 0.3068959105484197,
      "learning_rate": 1.2531053305892722e-05,
      "loss": 0.6576,
      "num_tokens": 7161595894.0,
      "step": 2344
    },
    {
      "epoch": 3.4337790804176587,
      "grad_norm": 0.3230000235002226,
      "learning_rate": 1.2516537057703119e-05,
      "loss": 0.6728,
      "num_tokens": 7164862625.0,
      "step": 2345
    },
    {
      "epoch": 3.435244550283935,
      "grad_norm": 0.3214012284629574,
      "learning_rate": 1.25020293422043e-05,
      "loss": 0.6818,
      "num_tokens": 7167991999.0,
      "step": 2346
    },
    {
      "epoch": 3.4367100201502105,
      "grad_norm": 0.3171803346470209,
      "learning_rate": 1.2487530172449515e-05,
      "loss": 0.6823,
      "num_tokens": 7171136798.0,
      "step": 2347
    },
    {
      "epoch": 3.4381754900164867,
      "grad_norm": 0.3027568047694281,
      "learning_rate": 1.2473039561484304e-05,
      "loss": 0.6846,
      "num_tokens": 7174491603.0,
      "step": 2348
    },
    {
      "epoch": 3.4396409598827624,
      "grad_norm": 0.31532837114211915,
      "learning_rate": 1.2458557522346505e-05,
      "loss": 0.6769,
      "num_tokens": 7177585508.0,
      "step": 2349
    },
    {
      "epoch": 3.441106429749038,
      "grad_norm": 0.3139428931768803,
      "learning_rate": 1.2444084068066253e-05,
      "loss": 0.673,
      "num_tokens": 7180847162.0,
      "step": 2350
    },
    {
      "epoch": 3.4425718996153143,
      "grad_norm": 0.29392171220174806,
      "learning_rate": 1.2429619211665953e-05,
      "loss": 0.6568,
      "num_tokens": 7184024099.0,
      "step": 2351
    },
    {
      "epoch": 3.44403736948159,
      "grad_norm": 0.32066426359223765,
      "learning_rate": 1.241516296616027e-05,
      "loss": 0.6988,
      "num_tokens": 7187054010.0,
      "step": 2352
    },
    {
      "epoch": 3.4455028393478657,
      "grad_norm": 0.30459298795233797,
      "learning_rate": 1.2400715344556141e-05,
      "loss": 0.6726,
      "num_tokens": 7190059261.0,
      "step": 2353
    },
    {
      "epoch": 3.446968309214142,
      "grad_norm": 0.33554619800277674,
      "learning_rate": 1.2386276359852718e-05,
      "loss": 0.644,
      "num_tokens": 7193071246.0,
      "step": 2354
    },
    {
      "epoch": 3.4484337790804176,
      "grad_norm": 0.301480308805606,
      "learning_rate": 1.23718460250414e-05,
      "loss": 0.6552,
      "num_tokens": 7196267637.0,
      "step": 2355
    },
    {
      "epoch": 3.4498992489466938,
      "grad_norm": 0.3305896278720203,
      "learning_rate": 1.2357424353105793e-05,
      "loss": 0.6692,
      "num_tokens": 7199245819.0,
      "step": 2356
    },
    {
      "epoch": 3.4513647188129695,
      "grad_norm": 0.3000067324698446,
      "learning_rate": 1.2343011357021718e-05,
      "loss": 0.6819,
      "num_tokens": 7202174867.0,
      "step": 2357
    },
    {
      "epoch": 3.452830188679245,
      "grad_norm": 0.331108974525821,
      "learning_rate": 1.232860704975717e-05,
      "loss": 0.6832,
      "num_tokens": 7204906384.0,
      "step": 2358
    },
    {
      "epoch": 3.4542956585455213,
      "grad_norm": 0.30822440479392427,
      "learning_rate": 1.2314211444272365e-05,
      "loss": 0.6863,
      "num_tokens": 7208019892.0,
      "step": 2359
    },
    {
      "epoch": 3.455761128411797,
      "grad_norm": 0.31457760557042735,
      "learning_rate": 1.2299824553519644e-05,
      "loss": 0.6889,
      "num_tokens": 7210873575.0,
      "step": 2360
    },
    {
      "epoch": 3.4572265982780728,
      "grad_norm": 0.278716029425566,
      "learning_rate": 1.2285446390443539e-05,
      "loss": 0.6714,
      "num_tokens": 7214120061.0,
      "step": 2361
    },
    {
      "epoch": 3.458692068144349,
      "grad_norm": 0.285286088938773,
      "learning_rate": 1.227107696798072e-05,
      "loss": 0.6666,
      "num_tokens": 7217238963.0,
      "step": 2362
    },
    {
      "epoch": 3.4601575380106246,
      "grad_norm": 0.28648517598618223,
      "learning_rate": 1.2256716299059986e-05,
      "loss": 0.6695,
      "num_tokens": 7220460767.0,
      "step": 2363
    },
    {
      "epoch": 3.4616230078769004,
      "grad_norm": 0.3052140797961659,
      "learning_rate": 1.2242364396602269e-05,
      "loss": 0.6835,
      "num_tokens": 7223662508.0,
      "step": 2364
    },
    {
      "epoch": 3.4630884777431765,
      "grad_norm": 0.2901735753817344,
      "learning_rate": 1.222802127352061e-05,
      "loss": 0.6598,
      "num_tokens": 7226556205.0,
      "step": 2365
    },
    {
      "epoch": 3.4645539476094522,
      "grad_norm": 0.31674110575252773,
      "learning_rate": 1.2213686942720141e-05,
      "loss": 0.6534,
      "num_tokens": 7229653626.0,
      "step": 2366
    },
    {
      "epoch": 3.466019417475728,
      "grad_norm": 0.3014376617658758,
      "learning_rate": 1.2199361417098107e-05,
      "loss": 0.6705,
      "num_tokens": 7232741107.0,
      "step": 2367
    },
    {
      "epoch": 3.467484887342004,
      "grad_norm": 0.31605317001348443,
      "learning_rate": 1.2185044709543816e-05,
      "loss": 0.6674,
      "num_tokens": 7235890740.0,
      "step": 2368
    },
    {
      "epoch": 3.46895035720828,
      "grad_norm": 0.32170857203914577,
      "learning_rate": 1.2170736832938634e-05,
      "loss": 0.6867,
      "num_tokens": 7239034835.0,
      "step": 2369
    },
    {
      "epoch": 3.4704158270745555,
      "grad_norm": 0.34563186676956076,
      "learning_rate": 1.2156437800155998e-05,
      "loss": 0.6765,
      "num_tokens": 7241847331.0,
      "step": 2370
    },
    {
      "epoch": 3.4718812969408317,
      "grad_norm": 0.3511241479423847,
      "learning_rate": 1.2142147624061375e-05,
      "loss": 0.6682,
      "num_tokens": 7244759779.0,
      "step": 2371
    },
    {
      "epoch": 3.4733467668071074,
      "grad_norm": 0.3399586216182509,
      "learning_rate": 1.212786631751226e-05,
      "loss": 0.6627,
      "num_tokens": 7247840593.0,
      "step": 2372
    },
    {
      "epoch": 3.4748122366733836,
      "grad_norm": 0.29951225287561806,
      "learning_rate": 1.2113593893358202e-05,
      "loss": 0.6753,
      "num_tokens": 7251033643.0,
      "step": 2373
    },
    {
      "epoch": 3.4762777065396593,
      "grad_norm": 0.34699683368497464,
      "learning_rate": 1.2099330364440698e-05,
      "loss": 0.6834,
      "num_tokens": 7254183466.0,
      "step": 2374
    },
    {
      "epoch": 3.477743176405935,
      "grad_norm": 0.30406767950367236,
      "learning_rate": 1.20850757435933e-05,
      "loss": 0.6811,
      "num_tokens": 7257176008.0,
      "step": 2375
    },
    {
      "epoch": 3.479208646272211,
      "grad_norm": 0.3147466196484179,
      "learning_rate": 1.2070830043641513e-05,
      "loss": 0.6594,
      "num_tokens": 7260275012.0,
      "step": 2376
    },
    {
      "epoch": 3.480674116138487,
      "grad_norm": 0.34258867428165707,
      "learning_rate": 1.2056593277402821e-05,
      "loss": 0.6711,
      "num_tokens": 7263432607.0,
      "step": 2377
    },
    {
      "epoch": 3.4821395860047626,
      "grad_norm": 0.31976791048027403,
      "learning_rate": 1.2042365457686673e-05,
      "loss": 0.6647,
      "num_tokens": 7266396860.0,
      "step": 2378
    },
    {
      "epoch": 3.4836050558710387,
      "grad_norm": 0.2988654469084843,
      "learning_rate": 1.202814659729447e-05,
      "loss": 0.6572,
      "num_tokens": 7269705808.0,
      "step": 2379
    },
    {
      "epoch": 3.4850705257373145,
      "grad_norm": 0.2983857252247023,
      "learning_rate": 1.2013936709019537e-05,
      "loss": 0.6885,
      "num_tokens": 7273062104.0,
      "step": 2380
    },
    {
      "epoch": 3.4865359956035906,
      "grad_norm": 0.3258846155478573,
      "learning_rate": 1.1999735805647165e-05,
      "loss": 0.6684,
      "num_tokens": 7276142321.0,
      "step": 2381
    },
    {
      "epoch": 3.4880014654698663,
      "grad_norm": 0.30700234880976096,
      "learning_rate": 1.1985543899954512e-05,
      "loss": 0.6663,
      "num_tokens": 7279211511.0,
      "step": 2382
    },
    {
      "epoch": 3.489466935336142,
      "grad_norm": 0.29632272611490373,
      "learning_rate": 1.197136100471066e-05,
      "loss": 0.6895,
      "num_tokens": 7282053955.0,
      "step": 2383
    },
    {
      "epoch": 3.490932405202418,
      "grad_norm": 0.3172831140223887,
      "learning_rate": 1.1957187132676602e-05,
      "loss": 0.68,
      "num_tokens": 7285150067.0,
      "step": 2384
    },
    {
      "epoch": 3.492397875068694,
      "grad_norm": 0.3228199075746972,
      "learning_rate": 1.1943022296605191e-05,
      "loss": 0.6647,
      "num_tokens": 7288194947.0,
      "step": 2385
    },
    {
      "epoch": 3.4938633449349696,
      "grad_norm": 0.3149587595933546,
      "learning_rate": 1.1928866509241157e-05,
      "loss": 0.6756,
      "num_tokens": 7291377348.0,
      "step": 2386
    },
    {
      "epoch": 3.495328814801246,
      "grad_norm": 0.3406084635985525,
      "learning_rate": 1.1914719783321082e-05,
      "loss": 0.668,
      "num_tokens": 7294474689.0,
      "step": 2387
    },
    {
      "epoch": 3.4967942846675215,
      "grad_norm": 0.31293647254906365,
      "learning_rate": 1.190058213157341e-05,
      "loss": 0.6849,
      "num_tokens": 7297689793.0,
      "step": 2388
    },
    {
      "epoch": 3.4982597545337972,
      "grad_norm": 0.3087122610479668,
      "learning_rate": 1.1886453566718398e-05,
      "loss": 0.6907,
      "num_tokens": 7300621503.0,
      "step": 2389
    },
    {
      "epoch": 3.4997252244000734,
      "grad_norm": 0.31231545309716685,
      "learning_rate": 1.1872334101468162e-05,
      "loss": 0.6629,
      "num_tokens": 7303762136.0,
      "step": 2390
    },
    {
      "epoch": 3.501190694266349,
      "grad_norm": 0.34081748392374056,
      "learning_rate": 1.1858223748526582e-05,
      "loss": 0.6739,
      "num_tokens": 7306830187.0,
      "step": 2391
    },
    {
      "epoch": 3.502656164132625,
      "grad_norm": 0.2956876890680326,
      "learning_rate": 1.1844122520589388e-05,
      "loss": 0.6908,
      "num_tokens": 7309979745.0,
      "step": 2392
    },
    {
      "epoch": 3.504121633998901,
      "grad_norm": 0.33504728375560827,
      "learning_rate": 1.1830030430344066e-05,
      "loss": 0.6715,
      "num_tokens": 7313217919.0,
      "step": 2393
    },
    {
      "epoch": 3.5055871038651767,
      "grad_norm": 0.3228847026551916,
      "learning_rate": 1.1815947490469898e-05,
      "loss": 0.6847,
      "num_tokens": 7316187662.0,
      "step": 2394
    },
    {
      "epoch": 3.5070525737314524,
      "grad_norm": 0.3190855505102043,
      "learning_rate": 1.1801873713637929e-05,
      "loss": 0.6568,
      "num_tokens": 7318928023.0,
      "step": 2395
    },
    {
      "epoch": 3.5085180435977286,
      "grad_norm": 0.33169212129944275,
      "learning_rate": 1.1787809112510953e-05,
      "loss": 0.697,
      "num_tokens": 7321736489.0,
      "step": 2396
    },
    {
      "epoch": 3.5099835134640043,
      "grad_norm": 0.3186227794810296,
      "learning_rate": 1.1773753699743508e-05,
      "loss": 0.6561,
      "num_tokens": 7324627649.0,
      "step": 2397
    },
    {
      "epoch": 3.51144898333028,
      "grad_norm": 0.32800268371295815,
      "learning_rate": 1.1759707487981884e-05,
      "loss": 0.6846,
      "num_tokens": 7327823174.0,
      "step": 2398
    },
    {
      "epoch": 3.512914453196556,
      "grad_norm": 0.3117434465143724,
      "learning_rate": 1.1745670489864071e-05,
      "loss": 0.6916,
      "num_tokens": 7330682225.0,
      "step": 2399
    },
    {
      "epoch": 3.514379923062832,
      "grad_norm": 0.30612016979345497,
      "learning_rate": 1.1731642718019778e-05,
      "loss": 0.6662,
      "num_tokens": 7333644351.0,
      "step": 2400
    },
    {
      "epoch": 3.515845392929108,
      "grad_norm": 0.29859776607888855,
      "learning_rate": 1.1717624185070412e-05,
      "loss": 0.6857,
      "num_tokens": 7336580978.0,
      "step": 2401
    },
    {
      "epoch": 3.5173108627953837,
      "grad_norm": 0.2996909880032615,
      "learning_rate": 1.1703614903629066e-05,
      "loss": 0.6651,
      "num_tokens": 7339585122.0,
      "step": 2402
    },
    {
      "epoch": 3.51877633266166,
      "grad_norm": 0.3009302885592069,
      "learning_rate": 1.1689614886300503e-05,
      "loss": 0.6741,
      "num_tokens": 7342597151.0,
      "step": 2403
    },
    {
      "epoch": 3.5202418025279356,
      "grad_norm": 0.3100609929259285,
      "learning_rate": 1.1675624145681177e-05,
      "loss": 0.6752,
      "num_tokens": 7345771108.0,
      "step": 2404
    },
    {
      "epoch": 3.5217072723942113,
      "grad_norm": 0.308843106369383,
      "learning_rate": 1.1661642694359148e-05,
      "loss": 0.6908,
      "num_tokens": 7349000205.0,
      "step": 2405
    },
    {
      "epoch": 3.5231727422604875,
      "grad_norm": 0.3209577737539571,
      "learning_rate": 1.1647670544914169e-05,
      "loss": 0.6782,
      "num_tokens": 7352112966.0,
      "step": 2406
    },
    {
      "epoch": 3.524638212126763,
      "grad_norm": 0.3110953464933069,
      "learning_rate": 1.1633707709917592e-05,
      "loss": 0.6953,
      "num_tokens": 7354978452.0,
      "step": 2407
    },
    {
      "epoch": 3.526103681993039,
      "grad_norm": 0.3279755786806314,
      "learning_rate": 1.1619754201932395e-05,
      "loss": 0.6656,
      "num_tokens": 7358149721.0,
      "step": 2408
    },
    {
      "epoch": 3.527569151859315,
      "grad_norm": 0.2974767408723198,
      "learning_rate": 1.1605810033513167e-05,
      "loss": 0.6735,
      "num_tokens": 7361086351.0,
      "step": 2409
    },
    {
      "epoch": 3.529034621725591,
      "grad_norm": 0.31000145223830883,
      "learning_rate": 1.1591875217206097e-05,
      "loss": 0.6777,
      "num_tokens": 7364153661.0,
      "step": 2410
    },
    {
      "epoch": 3.5305000915918665,
      "grad_norm": 0.3091034998688398,
      "learning_rate": 1.1577949765548942e-05,
      "loss": 0.6523,
      "num_tokens": 7367416224.0,
      "step": 2411
    },
    {
      "epoch": 3.5319655614581427,
      "grad_norm": 0.34048417431596073,
      "learning_rate": 1.1564033691071064e-05,
      "loss": 0.6784,
      "num_tokens": 7370570256.0,
      "step": 2412
    },
    {
      "epoch": 3.5334310313244184,
      "grad_norm": 0.3040636017422998,
      "learning_rate": 1.1550127006293359e-05,
      "loss": 0.6653,
      "num_tokens": 7373561120.0,
      "step": 2413
    },
    {
      "epoch": 3.534896501190694,
      "grad_norm": 0.3352297732946215,
      "learning_rate": 1.1536229723728293e-05,
      "loss": 0.6723,
      "num_tokens": 7376555616.0,
      "step": 2414
    },
    {
      "epoch": 3.5363619710569703,
      "grad_norm": 0.32107690588335847,
      "learning_rate": 1.152234185587986e-05,
      "loss": 0.6742,
      "num_tokens": 7379513107.0,
      "step": 2415
    },
    {
      "epoch": 3.537827440923246,
      "grad_norm": 0.3583952252603583,
      "learning_rate": 1.1508463415243591e-05,
      "loss": 0.6693,
      "num_tokens": 7382653405.0,
      "step": 2416
    },
    {
      "epoch": 3.5392929107895217,
      "grad_norm": 0.31521458709650757,
      "learning_rate": 1.1494594414306524e-05,
      "loss": 0.6672,
      "num_tokens": 7385621914.0,
      "step": 2417
    },
    {
      "epoch": 3.540758380655798,
      "grad_norm": 0.3245981497297923,
      "learning_rate": 1.1480734865547232e-05,
      "loss": 0.6776,
      "num_tokens": 7388830318.0,
      "step": 2418
    },
    {
      "epoch": 3.5422238505220736,
      "grad_norm": 0.31857414184122734,
      "learning_rate": 1.146688478143574e-05,
      "loss": 0.6751,
      "num_tokens": 7392015572.0,
      "step": 2419
    },
    {
      "epoch": 3.5436893203883493,
      "grad_norm": 0.2989734199140742,
      "learning_rate": 1.1453044174433599e-05,
      "loss": 0.6681,
      "num_tokens": 7394975426.0,
      "step": 2420
    },
    {
      "epoch": 3.5451547902546254,
      "grad_norm": 0.3106820756724418,
      "learning_rate": 1.1439213056993807e-05,
      "loss": 0.6805,
      "num_tokens": 7397879200.0,
      "step": 2421
    },
    {
      "epoch": 3.546620260120901,
      "grad_norm": 0.29264808049099134,
      "learning_rate": 1.1425391441560833e-05,
      "loss": 0.6845,
      "num_tokens": 7401004071.0,
      "step": 2422
    },
    {
      "epoch": 3.548085729987177,
      "grad_norm": 0.3355958791831573,
      "learning_rate": 1.1411579340570596e-05,
      "loss": 0.6744,
      "num_tokens": 7404079357.0,
      "step": 2423
    },
    {
      "epoch": 3.549551199853453,
      "grad_norm": 0.2741205355097215,
      "learning_rate": 1.1397776766450455e-05,
      "loss": 0.6916,
      "num_tokens": 7407225537.0,
      "step": 2424
    },
    {
      "epoch": 3.5510166697197287,
      "grad_norm": 0.3292161958600772,
      "learning_rate": 1.1383983731619189e-05,
      "loss": 0.659,
      "num_tokens": 7410238758.0,
      "step": 2425
    },
    {
      "epoch": 3.552482139586005,
      "grad_norm": 0.30334474024653624,
      "learning_rate": 1.1370200248487005e-05,
      "loss": 0.6636,
      "num_tokens": 7413382124.0,
      "step": 2426
    },
    {
      "epoch": 3.5539476094522806,
      "grad_norm": 0.3122897834918295,
      "learning_rate": 1.1356426329455523e-05,
      "loss": 0.6712,
      "num_tokens": 7416631877.0,
      "step": 2427
    },
    {
      "epoch": 3.5554130793185568,
      "grad_norm": 0.28251662334328553,
      "learning_rate": 1.1342661986917727e-05,
      "loss": 0.6758,
      "num_tokens": 7419706477.0,
      "step": 2428
    },
    {
      "epoch": 3.5568785491848325,
      "grad_norm": 0.3302209485935383,
      "learning_rate": 1.1328907233258017e-05,
      "loss": 0.6882,
      "num_tokens": 7422693437.0,
      "step": 2429
    },
    {
      "epoch": 3.558344019051108,
      "grad_norm": 0.27725619159336623,
      "learning_rate": 1.1315162080852153e-05,
      "loss": 0.6865,
      "num_tokens": 7425882887.0,
      "step": 2430
    },
    {
      "epoch": 3.5598094889173844,
      "grad_norm": 0.3052826145787302,
      "learning_rate": 1.1301426542067253e-05,
      "loss": 0.6623,
      "num_tokens": 7428948520.0,
      "step": 2431
    },
    {
      "epoch": 3.56127495878366,
      "grad_norm": 0.31759531740749314,
      "learning_rate": 1.1287700629261788e-05,
      "loss": 0.6733,
      "num_tokens": 7432021257.0,
      "step": 2432
    },
    {
      "epoch": 3.562740428649936,
      "grad_norm": 0.29637352126862826,
      "learning_rate": 1.1273984354785573e-05,
      "loss": 0.6781,
      "num_tokens": 7435379613.0,
      "step": 2433
    },
    {
      "epoch": 3.564205898516212,
      "grad_norm": 0.29905448620210534,
      "learning_rate": 1.1260277730979732e-05,
      "loss": 0.6955,
      "num_tokens": 7438327137.0,
      "step": 2434
    },
    {
      "epoch": 3.5656713683824877,
      "grad_norm": 0.32013836190467143,
      "learning_rate": 1.1246580770176747e-05,
      "loss": 0.6859,
      "num_tokens": 7441549909.0,
      "step": 2435
    },
    {
      "epoch": 3.5671368382487634,
      "grad_norm": 0.2640364992423809,
      "learning_rate": 1.123289348470035e-05,
      "loss": 0.6578,
      "num_tokens": 7444871233.0,
      "step": 2436
    },
    {
      "epoch": 3.5686023081150395,
      "grad_norm": 0.2914409791179919,
      "learning_rate": 1.121921588686561e-05,
      "loss": 0.6975,
      "num_tokens": 7448054549.0,
      "step": 2437
    },
    {
      "epoch": 3.5700677779813152,
      "grad_norm": 0.263412748692145,
      "learning_rate": 1.1205547988978872e-05,
      "loss": 0.6695,
      "num_tokens": 7451107463.0,
      "step": 2438
    },
    {
      "epoch": 3.571533247847591,
      "grad_norm": 0.2846600980766393,
      "learning_rate": 1.1191889803337742e-05,
      "loss": 0.6721,
      "num_tokens": 7454246500.0,
      "step": 2439
    },
    {
      "epoch": 3.572998717713867,
      "grad_norm": 0.3014154428012928,
      "learning_rate": 1.1178241342231084e-05,
      "loss": 0.6636,
      "num_tokens": 7457131636.0,
      "step": 2440
    },
    {
      "epoch": 3.574464187580143,
      "grad_norm": 0.2978661191119342,
      "learning_rate": 1.1164602617939048e-05,
      "loss": 0.6708,
      "num_tokens": 7460246918.0,
      "step": 2441
    },
    {
      "epoch": 3.5759296574464186,
      "grad_norm": 0.30076357611194116,
      "learning_rate": 1.1150973642732966e-05,
      "loss": 0.6549,
      "num_tokens": 7463576084.0,
      "step": 2442
    },
    {
      "epoch": 3.5773951273126947,
      "grad_norm": 0.2863629321850154,
      "learning_rate": 1.1137354428875451e-05,
      "loss": 0.6756,
      "num_tokens": 7466606469.0,
      "step": 2443
    },
    {
      "epoch": 3.5788605971789704,
      "grad_norm": 0.3082620695239256,
      "learning_rate": 1.11237449886203e-05,
      "loss": 0.6931,
      "num_tokens": 7469575060.0,
      "step": 2444
    },
    {
      "epoch": 3.580326067045246,
      "grad_norm": 0.32662559781020256,
      "learning_rate": 1.1110145334212537e-05,
      "loss": 0.6819,
      "num_tokens": 7472543091.0,
      "step": 2445
    },
    {
      "epoch": 3.5817915369115223,
      "grad_norm": 0.3066812971474287,
      "learning_rate": 1.1096555477888364e-05,
      "loss": 0.6869,
      "num_tokens": 7475404389.0,
      "step": 2446
    },
    {
      "epoch": 3.583257006777798,
      "grad_norm": 0.3185435722184899,
      "learning_rate": 1.108297543187518e-05,
      "loss": 0.6643,
      "num_tokens": 7478495716.0,
      "step": 2447
    },
    {
      "epoch": 3.5847224766440737,
      "grad_norm": 0.29316325614853733,
      "learning_rate": 1.1069405208391548e-05,
      "loss": 0.684,
      "num_tokens": 7481638522.0,
      "step": 2448
    },
    {
      "epoch": 3.58618794651035,
      "grad_norm": 0.33682574734278803,
      "learning_rate": 1.1055844819647212e-05,
      "loss": 0.6668,
      "num_tokens": 7484687624.0,
      "step": 2449
    },
    {
      "epoch": 3.5876534163766256,
      "grad_norm": 0.3218032931995025,
      "learning_rate": 1.1042294277843029e-05,
      "loss": 0.6593,
      "num_tokens": 7487937718.0,
      "step": 2450
    },
    {
      "epoch": 3.5891188862429018,
      "grad_norm": 0.2893656187391702,
      "learning_rate": 1.1028753595171043e-05,
      "loss": 0.6679,
      "num_tokens": 7491168376.0,
      "step": 2451
    },
    {
      "epoch": 3.5905843561091775,
      "grad_norm": 0.29788363342234875,
      "learning_rate": 1.1015222783814394e-05,
      "loss": 0.6789,
      "num_tokens": 7494038205.0,
      "step": 2452
    },
    {
      "epoch": 3.5920498259754536,
      "grad_norm": 0.30472083432686636,
      "learning_rate": 1.100170185594735e-05,
      "loss": 0.667,
      "num_tokens": 7497146229.0,
      "step": 2453
    },
    {
      "epoch": 3.5935152958417294,
      "grad_norm": 0.2826467265773505,
      "learning_rate": 1.0988190823735296e-05,
      "loss": 0.6531,
      "num_tokens": 7500386224.0,
      "step": 2454
    },
    {
      "epoch": 3.594980765708005,
      "grad_norm": 0.3026899390097473,
      "learning_rate": 1.0974689699334698e-05,
      "loss": 0.6546,
      "num_tokens": 7503485044.0,
      "step": 2455
    },
    {
      "epoch": 3.5964462355742812,
      "grad_norm": 0.31469939831440097,
      "learning_rate": 1.0961198494893103e-05,
      "loss": 0.6753,
      "num_tokens": 7506564051.0,
      "step": 2456
    },
    {
      "epoch": 3.597911705440557,
      "grad_norm": 0.2710537488846357,
      "learning_rate": 1.094771722254917e-05,
      "loss": 0.6941,
      "num_tokens": 7509809911.0,
      "step": 2457
    },
    {
      "epoch": 3.5993771753068327,
      "grad_norm": 0.30738783109323087,
      "learning_rate": 1.093424589443258e-05,
      "loss": 0.6907,
      "num_tokens": 7512748299.0,
      "step": 2458
    },
    {
      "epoch": 3.600842645173109,
      "grad_norm": 0.30891779221321897,
      "learning_rate": 1.0920784522664082e-05,
      "loss": 0.6777,
      "num_tokens": 7515576216.0,
      "step": 2459
    },
    {
      "epoch": 3.6023081150393845,
      "grad_norm": 0.30926510200579055,
      "learning_rate": 1.0907333119355474e-05,
      "loss": 0.6856,
      "num_tokens": 7518771846.0,
      "step": 2460
    },
    {
      "epoch": 3.6037735849056602,
      "grad_norm": 0.29555837648463,
      "learning_rate": 1.0893891696609574e-05,
      "loss": 0.6648,
      "num_tokens": 7521650681.0,
      "step": 2461
    },
    {
      "epoch": 3.6052390547719364,
      "grad_norm": 0.29179950038306074,
      "learning_rate": 1.0880460266520217e-05,
      "loss": 0.6712,
      "num_tokens": 7524771409.0,
      "step": 2462
    },
    {
      "epoch": 3.606704524638212,
      "grad_norm": 0.2949680669282753,
      "learning_rate": 1.086703884117227e-05,
      "loss": 0.6651,
      "num_tokens": 7527884436.0,
      "step": 2463
    },
    {
      "epoch": 3.608169994504488,
      "grad_norm": 0.3113228417189421,
      "learning_rate": 1.0853627432641574e-05,
      "loss": 0.6759,
      "num_tokens": 7531128415.0,
      "step": 2464
    },
    {
      "epoch": 3.609635464370764,
      "grad_norm": 0.2990122120864179,
      "learning_rate": 1.0840226052994956e-05,
      "loss": 0.6881,
      "num_tokens": 7533996387.0,
      "step": 2465
    },
    {
      "epoch": 3.6111009342370397,
      "grad_norm": 0.31610036410732795,
      "learning_rate": 1.082683471429025e-05,
      "loss": 0.6611,
      "num_tokens": 7537218327.0,
      "step": 2466
    },
    {
      "epoch": 3.6125664041033154,
      "grad_norm": 0.34876853132124036,
      "learning_rate": 1.0813453428576217e-05,
      "loss": 0.6779,
      "num_tokens": 7540367016.0,
      "step": 2467
    },
    {
      "epoch": 3.6140318739695916,
      "grad_norm": 0.2868164608089626,
      "learning_rate": 1.0800082207892606e-05,
      "loss": 0.6631,
      "num_tokens": 7543395617.0,
      "step": 2468
    },
    {
      "epoch": 3.6154973438358673,
      "grad_norm": 0.35731004208776446,
      "learning_rate": 1.0786721064270082e-05,
      "loss": 0.6811,
      "num_tokens": 7546296839.0,
      "step": 2469
    },
    {
      "epoch": 3.616962813702143,
      "grad_norm": 0.3502379833139834,
      "learning_rate": 1.0773370009730271e-05,
      "loss": 0.6819,
      "num_tokens": 7549352517.0,
      "step": 2470
    },
    {
      "epoch": 3.618428283568419,
      "grad_norm": 0.33556830058437254,
      "learning_rate": 1.0760029056285688e-05,
      "loss": 0.6779,
      "num_tokens": 7552249430.0,
      "step": 2471
    },
    {
      "epoch": 3.619893753434695,
      "grad_norm": 0.3368153165093765,
      "learning_rate": 1.0746698215939804e-05,
      "loss": 0.6802,
      "num_tokens": 7555371589.0,
      "step": 2472
    },
    {
      "epoch": 3.6213592233009706,
      "grad_norm": 0.28041658644882733,
      "learning_rate": 1.0733377500686938e-05,
      "loss": 0.6748,
      "num_tokens": 7558335168.0,
      "step": 2473
    },
    {
      "epoch": 3.6228246931672468,
      "grad_norm": 0.33339076221456904,
      "learning_rate": 1.0720066922512344e-05,
      "loss": 0.6723,
      "num_tokens": 7561172497.0,
      "step": 2474
    },
    {
      "epoch": 3.6242901630335225,
      "grad_norm": 0.2751221705904446,
      "learning_rate": 1.0706766493392134e-05,
      "loss": 0.6884,
      "num_tokens": 7564065905.0,
      "step": 2475
    },
    {
      "epoch": 3.6257556328997986,
      "grad_norm": 0.31540842089342086,
      "learning_rate": 1.0693476225293291e-05,
      "loss": 0.6844,
      "num_tokens": 7567103046.0,
      "step": 2476
    },
    {
      "epoch": 3.6272211027660743,
      "grad_norm": 0.30369311040829944,
      "learning_rate": 1.0680196130173657e-05,
      "loss": 0.6697,
      "num_tokens": 7570251950.0,
      "step": 2477
    },
    {
      "epoch": 3.6286865726323505,
      "grad_norm": 0.29919512986829705,
      "learning_rate": 1.066692621998192e-05,
      "loss": 0.6719,
      "num_tokens": 7573398511.0,
      "step": 2478
    },
    {
      "epoch": 3.630152042498626,
      "grad_norm": 0.29264263824919257,
      "learning_rate": 1.0653666506657594e-05,
      "loss": 0.6817,
      "num_tokens": 7576314315.0,
      "step": 2479
    },
    {
      "epoch": 3.631617512364902,
      "grad_norm": 0.31460981691262546,
      "learning_rate": 1.0640417002131056e-05,
      "loss": 0.6869,
      "num_tokens": 7579252136.0,
      "step": 2480
    },
    {
      "epoch": 3.633082982231178,
      "grad_norm": 0.291306390631656,
      "learning_rate": 1.0627177718323442e-05,
      "loss": 0.6741,
      "num_tokens": 7582322163.0,
      "step": 2481
    },
    {
      "epoch": 3.634548452097454,
      "grad_norm": 0.2832672676801926,
      "learning_rate": 1.0613948667146739e-05,
      "loss": 0.6797,
      "num_tokens": 7585103430.0,
      "step": 2482
    },
    {
      "epoch": 3.6360139219637295,
      "grad_norm": 0.34654531627354435,
      "learning_rate": 1.0600729860503705e-05,
      "loss": 0.676,
      "num_tokens": 7588264433.0,
      "step": 2483
    },
    {
      "epoch": 3.6374793918300057,
      "grad_norm": 0.30732383194519486,
      "learning_rate": 1.0587521310287883e-05,
      "loss": 0.6926,
      "num_tokens": 7591342894.0,
      "step": 2484
    },
    {
      "epoch": 3.6389448616962814,
      "grad_norm": 0.31001511455355674,
      "learning_rate": 1.0574323028383582e-05,
      "loss": 0.6609,
      "num_tokens": 7594464048.0,
      "step": 2485
    },
    {
      "epoch": 3.640410331562557,
      "grad_norm": 0.30941668234959,
      "learning_rate": 1.05611350266659e-05,
      "loss": 0.6832,
      "num_tokens": 7597610959.0,
      "step": 2486
    },
    {
      "epoch": 3.6418758014288333,
      "grad_norm": 0.31385530730910494,
      "learning_rate": 1.0547957317000641e-05,
      "loss": 0.6648,
      "num_tokens": 7600612628.0,
      "step": 2487
    },
    {
      "epoch": 3.643341271295109,
      "grad_norm": 0.306982708913995,
      "learning_rate": 1.0534789911244392e-05,
      "loss": 0.6981,
      "num_tokens": 7603397486.0,
      "step": 2488
    },
    {
      "epoch": 3.6448067411613847,
      "grad_norm": 0.3005703017667938,
      "learning_rate": 1.0521632821244442e-05,
      "loss": 0.6543,
      "num_tokens": 7606346131.0,
      "step": 2489
    },
    {
      "epoch": 3.646272211027661,
      "grad_norm": 0.28183605119915317,
      "learning_rate": 1.0508486058838803e-05,
      "loss": 0.6716,
      "num_tokens": 7609593654.0,
      "step": 2490
    },
    {
      "epoch": 3.6477376808939366,
      "grad_norm": 0.3359298387773411,
      "learning_rate": 1.0495349635856205e-05,
      "loss": 0.6876,
      "num_tokens": 7612537919.0,
      "step": 2491
    },
    {
      "epoch": 3.6492031507602123,
      "grad_norm": 0.2796982497662434,
      "learning_rate": 1.0482223564116064e-05,
      "loss": 0.6667,
      "num_tokens": 7615614940.0,
      "step": 2492
    },
    {
      "epoch": 3.6506686206264884,
      "grad_norm": 0.2967855074456007,
      "learning_rate": 1.0469107855428482e-05,
      "loss": 0.6852,
      "num_tokens": 7618601798.0,
      "step": 2493
    },
    {
      "epoch": 3.652134090492764,
      "grad_norm": 0.30041487183434823,
      "learning_rate": 1.0456002521594261e-05,
      "loss": 0.6717,
      "num_tokens": 7621712576.0,
      "step": 2494
    },
    {
      "epoch": 3.65359956035904,
      "grad_norm": 0.29115252309196177,
      "learning_rate": 1.0442907574404826e-05,
      "loss": 0.6879,
      "num_tokens": 7624793878.0,
      "step": 2495
    },
    {
      "epoch": 3.655065030225316,
      "grad_norm": 0.28438649447782227,
      "learning_rate": 1.0429823025642292e-05,
      "loss": 0.6843,
      "num_tokens": 7627788658.0,
      "step": 2496
    },
    {
      "epoch": 3.6565305000915918,
      "grad_norm": 0.31406830667174906,
      "learning_rate": 1.0416748887079409e-05,
      "loss": 0.6833,
      "num_tokens": 7630923206.0,
      "step": 2497
    },
    {
      "epoch": 3.6579959699578675,
      "grad_norm": 0.27413164244969046,
      "learning_rate": 1.0403685170479549e-05,
      "loss": 0.6821,
      "num_tokens": 7634064556.0,
      "step": 2498
    },
    {
      "epoch": 3.6594614398241436,
      "grad_norm": 0.29466619893687507,
      "learning_rate": 1.0390631887596711e-05,
      "loss": 0.6666,
      "num_tokens": 7636916978.0,
      "step": 2499
    },
    {
      "epoch": 3.6609269096904193,
      "grad_norm": 0.2872211689665231,
      "learning_rate": 1.0377589050175537e-05,
      "loss": 0.6653,
      "num_tokens": 7639909487.0,
      "step": 2500
    },
    {
      "epoch": 3.6623923795566955,
      "grad_norm": 0.2927302764129567,
      "learning_rate": 1.0364556669951206e-05,
      "loss": 0.69,
      "num_tokens": 7643011121.0,
      "step": 2501
    },
    {
      "epoch": 3.663857849422971,
      "grad_norm": 0.2831122315463431,
      "learning_rate": 1.0351534758649556e-05,
      "loss": 0.6845,
      "num_tokens": 7646324069.0,
      "step": 2502
    },
    {
      "epoch": 3.6653233192892474,
      "grad_norm": 0.29366516489507044,
      "learning_rate": 1.0338523327986967e-05,
      "loss": 0.6732,
      "num_tokens": 7649483651.0,
      "step": 2503
    },
    {
      "epoch": 3.666788789155523,
      "grad_norm": 0.2555832022964601,
      "learning_rate": 1.0325522389670396e-05,
      "loss": 0.6706,
      "num_tokens": 7652300475.0,
      "step": 2504
    },
    {
      "epoch": 3.668254259021799,
      "grad_norm": 0.29264155225706545,
      "learning_rate": 1.0312531955397365e-05,
      "loss": 0.6793,
      "num_tokens": 7655265789.0,
      "step": 2505
    },
    {
      "epoch": 3.669719728888075,
      "grad_norm": 0.2891594987700585,
      "learning_rate": 1.0299552036855936e-05,
      "loss": 0.6791,
      "num_tokens": 7658195221.0,
      "step": 2506
    },
    {
      "epoch": 3.6711851987543507,
      "grad_norm": 0.2918687059484172,
      "learning_rate": 1.028658264572472e-05,
      "loss": 0.68,
      "num_tokens": 7661239568.0,
      "step": 2507
    },
    {
      "epoch": 3.6726506686206264,
      "grad_norm": 0.28918619491959674,
      "learning_rate": 1.0273623793672855e-05,
      "loss": 0.6808,
      "num_tokens": 7664256989.0,
      "step": 2508
    },
    {
      "epoch": 3.6741161384869025,
      "grad_norm": 0.2843455615943383,
      "learning_rate": 1.0260675492359988e-05,
      "loss": 0.6761,
      "num_tokens": 7667611927.0,
      "step": 2509
    },
    {
      "epoch": 3.6755816083531783,
      "grad_norm": 0.26475771415499977,
      "learning_rate": 1.0247737753436276e-05,
      "loss": 0.6853,
      "num_tokens": 7670963744.0,
      "step": 2510
    },
    {
      "epoch": 3.677047078219454,
      "grad_norm": 0.26538236612581934,
      "learning_rate": 1.0234810588542388e-05,
      "loss": 0.6824,
      "num_tokens": 7673908057.0,
      "step": 2511
    },
    {
      "epoch": 3.67851254808573,
      "grad_norm": 0.2567292074101507,
      "learning_rate": 1.022189400930944e-05,
      "loss": 0.6852,
      "num_tokens": 7676966949.0,
      "step": 2512
    },
    {
      "epoch": 3.679978017952006,
      "grad_norm": 0.2891340531331051,
      "learning_rate": 1.0208988027359075e-05,
      "loss": 0.6922,
      "num_tokens": 7679820437.0,
      "step": 2513
    },
    {
      "epoch": 3.6814434878182816,
      "grad_norm": 0.2880442655690957,
      "learning_rate": 1.0196092654303365e-05,
      "loss": 0.6794,
      "num_tokens": 7682844348.0,
      "step": 2514
    },
    {
      "epoch": 3.6829089576845577,
      "grad_norm": 0.29895084942396183,
      "learning_rate": 1.0183207901744847e-05,
      "loss": 0.6711,
      "num_tokens": 7685760633.0,
      "step": 2515
    },
    {
      "epoch": 3.6843744275508334,
      "grad_norm": 0.26905366572962364,
      "learning_rate": 1.0170333781276494e-05,
      "loss": 0.6623,
      "num_tokens": 7688843262.0,
      "step": 2516
    },
    {
      "epoch": 3.685839897417109,
      "grad_norm": 0.28872555424826135,
      "learning_rate": 1.0157470304481744e-05,
      "loss": 0.6651,
      "num_tokens": 7692110176.0,
      "step": 2517
    },
    {
      "epoch": 3.6873053672833853,
      "grad_norm": 0.2914233824247382,
      "learning_rate": 1.014461748293441e-05,
      "loss": 0.6824,
      "num_tokens": 7695282037.0,
      "step": 2518
    },
    {
      "epoch": 3.688770837149661,
      "grad_norm": 0.2730801279675292,
      "learning_rate": 1.013177532819876e-05,
      "loss": 0.6576,
      "num_tokens": 7698127309.0,
      "step": 2519
    },
    {
      "epoch": 3.6902363070159367,
      "grad_norm": 0.2674863229120733,
      "learning_rate": 1.0118943851829446e-05,
      "loss": 0.6676,
      "num_tokens": 7701199759.0,
      "step": 2520
    },
    {
      "epoch": 3.691701776882213,
      "grad_norm": 0.29227892466431765,
      "learning_rate": 1.0106123065371514e-05,
      "loss": 0.6686,
      "num_tokens": 7704332832.0,
      "step": 2521
    },
    {
      "epoch": 3.6931672467484886,
      "grad_norm": 0.2773566555537987,
      "learning_rate": 1.0093312980360385e-05,
      "loss": 0.6634,
      "num_tokens": 7707543670.0,
      "step": 2522
    },
    {
      "epoch": 3.6946327166147643,
      "grad_norm": 0.32721425206433763,
      "learning_rate": 1.0080513608321865e-05,
      "loss": 0.689,
      "num_tokens": 7710868371.0,
      "step": 2523
    },
    {
      "epoch": 3.6960981864810405,
      "grad_norm": 0.3006430914753327,
      "learning_rate": 1.0067724960772109e-05,
      "loss": 0.6819,
      "num_tokens": 7713747537.0,
      "step": 2524
    },
    {
      "epoch": 3.697563656347316,
      "grad_norm": 0.3150755745836954,
      "learning_rate": 1.0054947049217647e-05,
      "loss": 0.6927,
      "num_tokens": 7716859228.0,
      "step": 2525
    },
    {
      "epoch": 3.6990291262135924,
      "grad_norm": 0.2972694992312387,
      "learning_rate": 1.0042179885155303e-05,
      "loss": 0.6625,
      "num_tokens": 7719677400.0,
      "step": 2526
    },
    {
      "epoch": 3.700494596079868,
      "grad_norm": 0.31489504056310763,
      "learning_rate": 1.002942348007228e-05,
      "loss": 0.6872,
      "num_tokens": 7722783491.0,
      "step": 2527
    },
    {
      "epoch": 3.7019600659461442,
      "grad_norm": 0.3223807469437241,
      "learning_rate": 1.0016677845446072e-05,
      "loss": 0.6906,
      "num_tokens": 7725820317.0,
      "step": 2528
    },
    {
      "epoch": 3.70342553581242,
      "grad_norm": 0.3321982269495967,
      "learning_rate": 1.0003942992744489e-05,
      "loss": 0.6602,
      "num_tokens": 7729025141.0,
      "step": 2529
    },
    {
      "epoch": 3.7048910056786957,
      "grad_norm": 0.3077796869089496,
      "learning_rate": 9.991218933425637e-06,
      "loss": 0.6722,
      "num_tokens": 7731945509.0,
      "step": 2530
    },
    {
      "epoch": 3.706356475544972,
      "grad_norm": 0.28365465149349534,
      "learning_rate": 9.978505678937934e-06,
      "loss": 0.6643,
      "num_tokens": 7735129106.0,
      "step": 2531
    },
    {
      "epoch": 3.7078219454112475,
      "grad_norm": 0.31121861394148304,
      "learning_rate": 9.965803240720028e-06,
      "loss": 0.6653,
      "num_tokens": 7738178206.0,
      "step": 2532
    },
    {
      "epoch": 3.7092874152775233,
      "grad_norm": 0.2974056411017123,
      "learning_rate": 9.953111630200886e-06,
      "loss": 0.6745,
      "num_tokens": 7741122687.0,
      "step": 2533
    },
    {
      "epoch": 3.7107528851437994,
      "grad_norm": 0.27072787689200517,
      "learning_rate": 9.940430858799705e-06,
      "loss": 0.6716,
      "num_tokens": 7744424145.0,
      "step": 2534
    },
    {
      "epoch": 3.712218355010075,
      "grad_norm": 0.30139556517477184,
      "learning_rate": 9.927760937925932e-06,
      "loss": 0.6839,
      "num_tokens": 7747397341.0,
      "step": 2535
    },
    {
      "epoch": 3.713683824876351,
      "grad_norm": 0.30327818503478193,
      "learning_rate": 9.915101878979261e-06,
      "loss": 0.6774,
      "num_tokens": 7750702233.0,
      "step": 2536
    },
    {
      "epoch": 3.715149294742627,
      "grad_norm": 0.25574910927636846,
      "learning_rate": 9.9024536933496e-06,
      "loss": 0.6892,
      "num_tokens": 7753571529.0,
      "step": 2537
    },
    {
      "epoch": 3.7166147646089027,
      "grad_norm": 0.29541167592518347,
      "learning_rate": 9.889816392417078e-06,
      "loss": 0.6732,
      "num_tokens": 7756710866.0,
      "step": 2538
    },
    {
      "epoch": 3.7180802344751784,
      "grad_norm": 0.297537095831004,
      "learning_rate": 9.877189987552053e-06,
      "loss": 0.6689,
      "num_tokens": 7759680899.0,
      "step": 2539
    },
    {
      "epoch": 3.7195457043414546,
      "grad_norm": 0.2915933035084471,
      "learning_rate": 9.864574490115035e-06,
      "loss": 0.6853,
      "num_tokens": 7763040808.0,
      "step": 2540
    },
    {
      "epoch": 3.7210111742077303,
      "grad_norm": 0.29951436246811586,
      "learning_rate": 9.851969911456759e-06,
      "loss": 0.67,
      "num_tokens": 7766170276.0,
      "step": 2541
    },
    {
      "epoch": 3.722476644074006,
      "grad_norm": 0.2810187376628007,
      "learning_rate": 9.839376262918117e-06,
      "loss": 0.6673,
      "num_tokens": 7769305589.0,
      "step": 2542
    },
    {
      "epoch": 3.723942113940282,
      "grad_norm": 0.2948032027989562,
      "learning_rate": 9.826793555830174e-06,
      "loss": 0.6759,
      "num_tokens": 7772396648.0,
      "step": 2543
    },
    {
      "epoch": 3.725407583806558,
      "grad_norm": 0.30046004929788206,
      "learning_rate": 9.81422180151414e-06,
      "loss": 0.6976,
      "num_tokens": 7775346899.0,
      "step": 2544
    },
    {
      "epoch": 3.7268730536728336,
      "grad_norm": 0.3019977849771077,
      "learning_rate": 9.801661011281394e-06,
      "loss": 0.6644,
      "num_tokens": 7778381715.0,
      "step": 2545
    },
    {
      "epoch": 3.7283385235391098,
      "grad_norm": 0.27941845967068596,
      "learning_rate": 9.789111196433414e-06,
      "loss": 0.683,
      "num_tokens": 7781592312.0,
      "step": 2546
    },
    {
      "epoch": 3.7298039934053855,
      "grad_norm": 0.292500854200043,
      "learning_rate": 9.77657236826184e-06,
      "loss": 0.6943,
      "num_tokens": 7784621078.0,
      "step": 2547
    },
    {
      "epoch": 3.731269463271661,
      "grad_norm": 0.29110615446361826,
      "learning_rate": 9.764044538048407e-06,
      "loss": 0.6792,
      "num_tokens": 7787659503.0,
      "step": 2548
    },
    {
      "epoch": 3.7327349331379374,
      "grad_norm": 0.2659996585450849,
      "learning_rate": 9.751527717064942e-06,
      "loss": 0.6758,
      "num_tokens": 7790621153.0,
      "step": 2549
    },
    {
      "epoch": 3.734200403004213,
      "grad_norm": 0.29733080179630234,
      "learning_rate": 9.739021916573402e-06,
      "loss": 0.6848,
      "num_tokens": 7793720104.0,
      "step": 2550
    },
    {
      "epoch": 3.7356658728704892,
      "grad_norm": 0.28355092022416384,
      "learning_rate": 9.726527147825799e-06,
      "loss": 0.6688,
      "num_tokens": 7796633974.0,
      "step": 2551
    },
    {
      "epoch": 3.737131342736765,
      "grad_norm": 0.2975335758827942,
      "learning_rate": 9.71404342206423e-06,
      "loss": 0.6579,
      "num_tokens": 7799587608.0,
      "step": 2552
    },
    {
      "epoch": 3.738596812603041,
      "grad_norm": 0.3070679896608045,
      "learning_rate": 9.701570750520855e-06,
      "loss": 0.6789,
      "num_tokens": 7802427390.0,
      "step": 2553
    },
    {
      "epoch": 3.740062282469317,
      "grad_norm": 0.29657927091874214,
      "learning_rate": 9.689109144417889e-06,
      "loss": 0.6724,
      "num_tokens": 7805660850.0,
      "step": 2554
    },
    {
      "epoch": 3.7415277523355925,
      "grad_norm": 0.30825631635089884,
      "learning_rate": 9.676658614967584e-06,
      "loss": 0.6689,
      "num_tokens": 7808748317.0,
      "step": 2555
    },
    {
      "epoch": 3.7429932222018687,
      "grad_norm": 0.28548605008654426,
      "learning_rate": 9.664219173372245e-06,
      "loss": 0.6787,
      "num_tokens": 7811664244.0,
      "step": 2556
    },
    {
      "epoch": 3.7444586920681444,
      "grad_norm": 0.28015007793959407,
      "learning_rate": 9.651790830824168e-06,
      "loss": 0.6911,
      "num_tokens": 7814645384.0,
      "step": 2557
    },
    {
      "epoch": 3.74592416193442,
      "grad_norm": 0.2722052814509414,
      "learning_rate": 9.639373598505702e-06,
      "loss": 0.6695,
      "num_tokens": 7817794683.0,
      "step": 2558
    },
    {
      "epoch": 3.7473896318006963,
      "grad_norm": 0.2763262908984455,
      "learning_rate": 9.626967487589167e-06,
      "loss": 0.6853,
      "num_tokens": 7820752026.0,
      "step": 2559
    },
    {
      "epoch": 3.748855101666972,
      "grad_norm": 0.2889935609300939,
      "learning_rate": 9.614572509236895e-06,
      "loss": 0.7023,
      "num_tokens": 7823674673.0,
      "step": 2560
    },
    {
      "epoch": 3.7503205715332477,
      "grad_norm": 0.2935161488609743,
      "learning_rate": 9.602188674601188e-06,
      "loss": 0.6813,
      "num_tokens": 7826649529.0,
      "step": 2561
    },
    {
      "epoch": 3.751786041399524,
      "grad_norm": 0.27044019496273075,
      "learning_rate": 9.589815994824345e-06,
      "loss": 0.6623,
      "num_tokens": 7829642816.0,
      "step": 2562
    },
    {
      "epoch": 3.7532515112657996,
      "grad_norm": 0.28101090033973114,
      "learning_rate": 9.577454481038596e-06,
      "loss": 0.664,
      "num_tokens": 7832603851.0,
      "step": 2563
    },
    {
      "epoch": 3.7547169811320753,
      "grad_norm": 0.2960425904130176,
      "learning_rate": 9.565104144366151e-06,
      "loss": 0.6756,
      "num_tokens": 7835760789.0,
      "step": 2564
    },
    {
      "epoch": 3.7561824509983515,
      "grad_norm": 0.3035027146822713,
      "learning_rate": 9.552764995919153e-06,
      "loss": 0.6712,
      "num_tokens": 7838844373.0,
      "step": 2565
    },
    {
      "epoch": 3.757647920864627,
      "grad_norm": 0.26592183053554036,
      "learning_rate": 9.540437046799679e-06,
      "loss": 0.6613,
      "num_tokens": 7841749140.0,
      "step": 2566
    },
    {
      "epoch": 3.759113390730903,
      "grad_norm": 0.26646455789193796,
      "learning_rate": 9.528120308099726e-06,
      "loss": 0.6749,
      "num_tokens": 7844907280.0,
      "step": 2567
    },
    {
      "epoch": 3.760578860597179,
      "grad_norm": 0.25245195035740703,
      "learning_rate": 9.515814790901214e-06,
      "loss": 0.6744,
      "num_tokens": 7847925479.0,
      "step": 2568
    },
    {
      "epoch": 3.7620443304634548,
      "grad_norm": 0.328246444084621,
      "learning_rate": 9.503520506275949e-06,
      "loss": 0.6635,
      "num_tokens": 7851023636.0,
      "step": 2569
    },
    {
      "epoch": 3.7635098003297305,
      "grad_norm": 0.2806172477929138,
      "learning_rate": 9.491237465285662e-06,
      "loss": 0.6685,
      "num_tokens": 7854072023.0,
      "step": 2570
    },
    {
      "epoch": 3.7649752701960066,
      "grad_norm": 0.2850073594375248,
      "learning_rate": 9.478965678981927e-06,
      "loss": 0.6604,
      "num_tokens": 7856935884.0,
      "step": 2571
    },
    {
      "epoch": 3.7664407400622824,
      "grad_norm": 0.25297965713546156,
      "learning_rate": 9.466705158406227e-06,
      "loss": 0.6623,
      "num_tokens": 7860276372.0,
      "step": 2572
    },
    {
      "epoch": 3.767906209928558,
      "grad_norm": 0.3036682880571809,
      "learning_rate": 9.45445591458989e-06,
      "loss": 0.6746,
      "num_tokens": 7863415985.0,
      "step": 2573
    },
    {
      "epoch": 3.7693716797948342,
      "grad_norm": 0.28536263786229527,
      "learning_rate": 9.442217958554104e-06,
      "loss": 0.6833,
      "num_tokens": 7866513850.0,
      "step": 2574
    },
    {
      "epoch": 3.77083714966111,
      "grad_norm": 0.28249421324657004,
      "learning_rate": 9.429991301309889e-06,
      "loss": 0.6783,
      "num_tokens": 7869742061.0,
      "step": 2575
    },
    {
      "epoch": 3.772302619527386,
      "grad_norm": 0.3116289323411193,
      "learning_rate": 9.417775953858127e-06,
      "loss": 0.6598,
      "num_tokens": 7872835156.0,
      "step": 2576
    },
    {
      "epoch": 3.773768089393662,
      "grad_norm": 0.24902493548603707,
      "learning_rate": 9.405571927189487e-06,
      "loss": 0.6721,
      "num_tokens": 7875927673.0,
      "step": 2577
    },
    {
      "epoch": 3.775233559259938,
      "grad_norm": 0.296037283137463,
      "learning_rate": 9.393379232284483e-06,
      "loss": 0.6797,
      "num_tokens": 7879081655.0,
      "step": 2578
    },
    {
      "epoch": 3.7766990291262137,
      "grad_norm": 0.2744546560702928,
      "learning_rate": 9.381197880113419e-06,
      "loss": 0.6789,
      "num_tokens": 7882219855.0,
      "step": 2579
    },
    {
      "epoch": 3.7781644989924894,
      "grad_norm": 0.2955318821509196,
      "learning_rate": 9.369027881636396e-06,
      "loss": 0.6674,
      "num_tokens": 7885255085.0,
      "step": 2580
    },
    {
      "epoch": 3.7796299688587656,
      "grad_norm": 0.2887703794558178,
      "learning_rate": 9.3568692478033e-06,
      "loss": 0.6983,
      "num_tokens": 7887839264.0,
      "step": 2581
    },
    {
      "epoch": 3.7810954387250413,
      "grad_norm": 0.2872755939280624,
      "learning_rate": 9.344721989553789e-06,
      "loss": 0.6654,
      "num_tokens": 7890916670.0,
      "step": 2582
    },
    {
      "epoch": 3.782560908591317,
      "grad_norm": 0.3037807160168398,
      "learning_rate": 9.332586117817285e-06,
      "loss": 0.6529,
      "num_tokens": 7894014031.0,
      "step": 2583
    },
    {
      "epoch": 3.784026378457593,
      "grad_norm": 0.32136928938617815,
      "learning_rate": 9.320461643512982e-06,
      "loss": 0.6852,
      "num_tokens": 7896855265.0,
      "step": 2584
    },
    {
      "epoch": 3.785491848323869,
      "grad_norm": 0.29581564476892874,
      "learning_rate": 9.308348577549788e-06,
      "loss": 0.6755,
      "num_tokens": 7899835481.0,
      "step": 2585
    },
    {
      "epoch": 3.7869573181901446,
      "grad_norm": 0.29806016747146313,
      "learning_rate": 9.296246930826374e-06,
      "loss": 0.6835,
      "num_tokens": 7902829335.0,
      "step": 2586
    },
    {
      "epoch": 3.7884227880564207,
      "grad_norm": 0.28195275753103555,
      "learning_rate": 9.284156714231127e-06,
      "loss": 0.7039,
      "num_tokens": 7905715629.0,
      "step": 2587
    },
    {
      "epoch": 3.7898882579226965,
      "grad_norm": 0.3118629792590862,
      "learning_rate": 9.272077938642147e-06,
      "loss": 0.6948,
      "num_tokens": 7908678897.0,
      "step": 2588
    },
    {
      "epoch": 3.791353727788972,
      "grad_norm": 0.28072850622434986,
      "learning_rate": 9.260010614927241e-06,
      "loss": 0.6752,
      "num_tokens": 7911765069.0,
      "step": 2589
    },
    {
      "epoch": 3.7928191976552483,
      "grad_norm": 0.29140090401129704,
      "learning_rate": 9.247954753943916e-06,
      "loss": 0.6711,
      "num_tokens": 7914928632.0,
      "step": 2590
    },
    {
      "epoch": 3.794284667521524,
      "grad_norm": 0.31322006725359225,
      "learning_rate": 9.23591036653936e-06,
      "loss": 0.683,
      "num_tokens": 7917996260.0,
      "step": 2591
    },
    {
      "epoch": 3.7957501373877998,
      "grad_norm": 0.2892032627857932,
      "learning_rate": 9.22387746355044e-06,
      "loss": 0.6614,
      "num_tokens": 7920989003.0,
      "step": 2592
    },
    {
      "epoch": 3.797215607254076,
      "grad_norm": 0.27280792997736203,
      "learning_rate": 9.211856055803701e-06,
      "loss": 0.6776,
      "num_tokens": 7923981971.0,
      "step": 2593
    },
    {
      "epoch": 3.7986810771203516,
      "grad_norm": 0.27115764240706564,
      "learning_rate": 9.199846154115316e-06,
      "loss": 0.6526,
      "num_tokens": 7927088409.0,
      "step": 2594
    },
    {
      "epoch": 3.8001465469866273,
      "grad_norm": 0.2761122233967579,
      "learning_rate": 9.187847769291142e-06,
      "loss": 0.6627,
      "num_tokens": 7930403463.0,
      "step": 2595
    },
    {
      "epoch": 3.8016120168529035,
      "grad_norm": 0.2857152570343955,
      "learning_rate": 9.175860912126649e-06,
      "loss": 0.6693,
      "num_tokens": 7933485282.0,
      "step": 2596
    },
    {
      "epoch": 3.8030774867191792,
      "grad_norm": 0.3010174165815394,
      "learning_rate": 9.163885593406944e-06,
      "loss": 0.6823,
      "num_tokens": 7936530184.0,
      "step": 2597
    },
    {
      "epoch": 3.804542956585455,
      "grad_norm": 0.2939410946823734,
      "learning_rate": 9.15192182390675e-06,
      "loss": 0.6777,
      "num_tokens": 7939588216.0,
      "step": 2598
    },
    {
      "epoch": 3.806008426451731,
      "grad_norm": 0.295328243524195,
      "learning_rate": 9.139969614390396e-06,
      "loss": 0.6677,
      "num_tokens": 7942700811.0,
      "step": 2599
    },
    {
      "epoch": 3.807473896318007,
      "grad_norm": 0.3287908728101239,
      "learning_rate": 9.12802897561181e-06,
      "loss": 0.6736,
      "num_tokens": 7945771185.0,
      "step": 2600
    },
    {
      "epoch": 3.808939366184283,
      "grad_norm": 0.2752500052547036,
      "learning_rate": 9.116099918314532e-06,
      "loss": 0.6718,
      "num_tokens": 7948830238.0,
      "step": 2601
    },
    {
      "epoch": 3.8104048360505587,
      "grad_norm": 0.30811031253723264,
      "learning_rate": 9.104182453231635e-06,
      "loss": 0.6967,
      "num_tokens": 7951881977.0,
      "step": 2602
    },
    {
      "epoch": 3.811870305916835,
      "grad_norm": 0.30485405634493695,
      "learning_rate": 9.092276591085812e-06,
      "loss": 0.6686,
      "num_tokens": 7954773548.0,
      "step": 2603
    },
    {
      "epoch": 3.8133357757831106,
      "grad_norm": 0.2848357175627662,
      "learning_rate": 9.080382342589282e-06,
      "loss": 0.6815,
      "num_tokens": 7957798664.0,
      "step": 2604
    },
    {
      "epoch": 3.8148012456493863,
      "grad_norm": 0.30458103130061936,
      "learning_rate": 9.068499718443829e-06,
      "loss": 0.6881,
      "num_tokens": 7960839633.0,
      "step": 2605
    },
    {
      "epoch": 3.8162667155156624,
      "grad_norm": 0.29631685637194255,
      "learning_rate": 9.056628729340767e-06,
      "loss": 0.6658,
      "num_tokens": 7963647591.0,
      "step": 2606
    },
    {
      "epoch": 3.817732185381938,
      "grad_norm": 0.2761036544596787,
      "learning_rate": 9.044769385960968e-06,
      "loss": 0.6928,
      "num_tokens": 7966823472.0,
      "step": 2607
    },
    {
      "epoch": 3.819197655248214,
      "grad_norm": 0.2957423191406887,
      "learning_rate": 9.032921698974785e-06,
      "loss": 0.6577,
      "num_tokens": 7969964587.0,
      "step": 2608
    },
    {
      "epoch": 3.82066312511449,
      "grad_norm": 0.2751329376583761,
      "learning_rate": 9.021085679042124e-06,
      "loss": 0.6849,
      "num_tokens": 7972915043.0,
      "step": 2609
    },
    {
      "epoch": 3.8221285949807657,
      "grad_norm": 0.2961086114512179,
      "learning_rate": 9.009261336812369e-06,
      "loss": 0.6778,
      "num_tokens": 7975774983.0,
      "step": 2610
    },
    {
      "epoch": 3.8235940648470415,
      "grad_norm": 0.24370241441125515,
      "learning_rate": 8.9974486829244e-06,
      "loss": 0.6878,
      "num_tokens": 7978807654.0,
      "step": 2611
    },
    {
      "epoch": 3.8250595347133176,
      "grad_norm": 0.29066016590327115,
      "learning_rate": 8.98564772800659e-06,
      "loss": 0.6765,
      "num_tokens": 7981864652.0,
      "step": 2612
    },
    {
      "epoch": 3.8265250045795933,
      "grad_norm": 0.29122387613682843,
      "learning_rate": 8.973858482676776e-06,
      "loss": 0.6656,
      "num_tokens": 7984885876.0,
      "step": 2613
    },
    {
      "epoch": 3.827990474445869,
      "grad_norm": 0.2958314936624019,
      "learning_rate": 8.962080957542256e-06,
      "loss": 0.685,
      "num_tokens": 7987866785.0,
      "step": 2614
    },
    {
      "epoch": 3.829455944312145,
      "grad_norm": 0.2902433712908369,
      "learning_rate": 8.95031516319981e-06,
      "loss": 0.6681,
      "num_tokens": 7990952752.0,
      "step": 2615
    },
    {
      "epoch": 3.830921414178421,
      "grad_norm": 0.2823962411543531,
      "learning_rate": 8.938561110235617e-06,
      "loss": 0.6765,
      "num_tokens": 7994212607.0,
      "step": 2616
    },
    {
      "epoch": 3.8323868840446966,
      "grad_norm": 0.2572386320017982,
      "learning_rate": 8.92681880922534e-06,
      "loss": 0.6701,
      "num_tokens": 7997329609.0,
      "step": 2617
    },
    {
      "epoch": 3.833852353910973,
      "grad_norm": 0.2894794057598482,
      "learning_rate": 8.915088270734038e-06,
      "loss": 0.644,
      "num_tokens": 8000307583.0,
      "step": 2618
    },
    {
      "epoch": 3.8353178237772485,
      "grad_norm": 0.2787709802110485,
      "learning_rate": 8.903369505316194e-06,
      "loss": 0.6661,
      "num_tokens": 8003354856.0,
      "step": 2619
    },
    {
      "epoch": 3.836783293643524,
      "grad_norm": 0.3280560367382823,
      "learning_rate": 8.891662523515694e-06,
      "loss": 0.6755,
      "num_tokens": 8006625836.0,
      "step": 2620
    },
    {
      "epoch": 3.8382487635098004,
      "grad_norm": 0.31177549594042875,
      "learning_rate": 8.879967335865844e-06,
      "loss": 0.6797,
      "num_tokens": 8009820820.0,
      "step": 2621
    },
    {
      "epoch": 3.839714233376076,
      "grad_norm": 0.2886129992099771,
      "learning_rate": 8.868283952889302e-06,
      "loss": 0.6638,
      "num_tokens": 8012838990.0,
      "step": 2622
    },
    {
      "epoch": 3.841179703242352,
      "grad_norm": 0.2674566208355911,
      "learning_rate": 8.856612385098133e-06,
      "loss": 0.6743,
      "num_tokens": 8015640257.0,
      "step": 2623
    },
    {
      "epoch": 3.842645173108628,
      "grad_norm": 0.2766055524635886,
      "learning_rate": 8.844952642993767e-06,
      "loss": 0.6741,
      "num_tokens": 8018877173.0,
      "step": 2624
    },
    {
      "epoch": 3.8441106429749037,
      "grad_norm": 0.2785310173598924,
      "learning_rate": 8.833304737066983e-06,
      "loss": 0.6822,
      "num_tokens": 8021783938.0,
      "step": 2625
    },
    {
      "epoch": 3.84557611284118,
      "grad_norm": 0.28413870221980897,
      "learning_rate": 8.821668677797917e-06,
      "loss": 0.6631,
      "num_tokens": 8025014715.0,
      "step": 2626
    },
    {
      "epoch": 3.8470415827074556,
      "grad_norm": 0.26530037810688034,
      "learning_rate": 8.810044475656048e-06,
      "loss": 0.6821,
      "num_tokens": 8028230011.0,
      "step": 2627
    },
    {
      "epoch": 3.8485070525737317,
      "grad_norm": 0.2675244971737039,
      "learning_rate": 8.798432141100178e-06,
      "loss": 0.6878,
      "num_tokens": 8031281072.0,
      "step": 2628
    },
    {
      "epoch": 3.8499725224400074,
      "grad_norm": 0.25790948540107633,
      "learning_rate": 8.786831684578452e-06,
      "loss": 0.6791,
      "num_tokens": 8034183257.0,
      "step": 2629
    },
    {
      "epoch": 3.851437992306283,
      "grad_norm": 0.27899670873736865,
      "learning_rate": 8.7752431165283e-06,
      "loss": 0.6709,
      "num_tokens": 8037175102.0,
      "step": 2630
    },
    {
      "epoch": 3.8529034621725593,
      "grad_norm": 0.28037591661662437,
      "learning_rate": 8.763666447376471e-06,
      "loss": 0.6643,
      "num_tokens": 8040361089.0,
      "step": 2631
    },
    {
      "epoch": 3.854368932038835,
      "grad_norm": 0.262876543930627,
      "learning_rate": 8.752101687539017e-06,
      "loss": 0.678,
      "num_tokens": 8043264778.0,
      "step": 2632
    },
    {
      "epoch": 3.8558344019051107,
      "grad_norm": 0.2775689870643678,
      "learning_rate": 8.740548847421244e-06,
      "loss": 0.6828,
      "num_tokens": 8046122589.0,
      "step": 2633
    },
    {
      "epoch": 3.857299871771387,
      "grad_norm": 0.2572539992890923,
      "learning_rate": 8.72900793741777e-06,
      "loss": 0.675,
      "num_tokens": 8049338598.0,
      "step": 2634
    },
    {
      "epoch": 3.8587653416376626,
      "grad_norm": 0.27585388716884945,
      "learning_rate": 8.717478967912456e-06,
      "loss": 0.6782,
      "num_tokens": 8052485229.0,
      "step": 2635
    },
    {
      "epoch": 3.8602308115039383,
      "grad_norm": 0.2755765636373551,
      "learning_rate": 8.705961949278424e-06,
      "loss": 0.6761,
      "num_tokens": 8055196581.0,
      "step": 2636
    },
    {
      "epoch": 3.8616962813702145,
      "grad_norm": 0.2966791818815388,
      "learning_rate": 8.694456891878042e-06,
      "loss": 0.6621,
      "num_tokens": 8058332186.0,
      "step": 2637
    },
    {
      "epoch": 3.86316175123649,
      "grad_norm": 0.27662932774519583,
      "learning_rate": 8.682963806062934e-06,
      "loss": 0.6679,
      "num_tokens": 8061156469.0,
      "step": 2638
    },
    {
      "epoch": 3.864627221102766,
      "grad_norm": 0.2852690323028242,
      "learning_rate": 8.671482702173914e-06,
      "loss": 0.6835,
      "num_tokens": 8064292607.0,
      "step": 2639
    },
    {
      "epoch": 3.866092690969042,
      "grad_norm": 0.2996124478406475,
      "learning_rate": 8.660013590541058e-06,
      "loss": 0.663,
      "num_tokens": 8067650995.0,
      "step": 2640
    },
    {
      "epoch": 3.867558160835318,
      "grad_norm": 0.2865163220228784,
      "learning_rate": 8.648556481483626e-06,
      "loss": 0.6796,
      "num_tokens": 8070566911.0,
      "step": 2641
    },
    {
      "epoch": 3.8690236307015935,
      "grad_norm": 0.29368594627340305,
      "learning_rate": 8.637111385310081e-06,
      "loss": 0.6766,
      "num_tokens": 8073547075.0,
      "step": 2642
    },
    {
      "epoch": 3.8704891005678697,
      "grad_norm": 0.2564187908110803,
      "learning_rate": 8.62567831231809e-06,
      "loss": 0.6833,
      "num_tokens": 8076700238.0,
      "step": 2643
    },
    {
      "epoch": 3.8719545704341454,
      "grad_norm": 0.29680717760487557,
      "learning_rate": 8.614257272794485e-06,
      "loss": 0.6826,
      "num_tokens": 8079645729.0,
      "step": 2644
    },
    {
      "epoch": 3.873420040300421,
      "grad_norm": 0.28371769892819854,
      "learning_rate": 8.602848277015283e-06,
      "loss": 0.6769,
      "num_tokens": 8082524907.0,
      "step": 2645
    },
    {
      "epoch": 3.8748855101666972,
      "grad_norm": 0.271141989777032,
      "learning_rate": 8.591451335245676e-06,
      "loss": 0.6709,
      "num_tokens": 8085568230.0,
      "step": 2646
    },
    {
      "epoch": 3.876350980032973,
      "grad_norm": 0.27165224538294513,
      "learning_rate": 8.580066457739969e-06,
      "loss": 0.6628,
      "num_tokens": 8088553239.0,
      "step": 2647
    },
    {
      "epoch": 3.8778164498992487,
      "grad_norm": 0.27534220747304594,
      "learning_rate": 8.568693654741663e-06,
      "loss": 0.7004,
      "num_tokens": 8091671276.0,
      "step": 2648
    },
    {
      "epoch": 3.879281919765525,
      "grad_norm": 0.29122425588477363,
      "learning_rate": 8.557332936483363e-06,
      "loss": 0.6572,
      "num_tokens": 8094769582.0,
      "step": 2649
    },
    {
      "epoch": 3.8807473896318005,
      "grad_norm": 0.27091399049704856,
      "learning_rate": 8.545984313186807e-06,
      "loss": 0.6852,
      "num_tokens": 8097715309.0,
      "step": 2650
    },
    {
      "epoch": 3.8822128594980767,
      "grad_norm": 0.2635147918111219,
      "learning_rate": 8.534647795062854e-06,
      "loss": 0.6714,
      "num_tokens": 8100806666.0,
      "step": 2651
    },
    {
      "epoch": 3.8836783293643524,
      "grad_norm": 0.26563613792863067,
      "learning_rate": 8.523323392311481e-06,
      "loss": 0.6852,
      "num_tokens": 8103660070.0,
      "step": 2652
    },
    {
      "epoch": 3.8851437992306286,
      "grad_norm": 0.28845707342552124,
      "learning_rate": 8.512011115121734e-06,
      "loss": 0.6589,
      "num_tokens": 8106896808.0,
      "step": 2653
    },
    {
      "epoch": 3.8866092690969043,
      "grad_norm": 0.28105316591364743,
      "learning_rate": 8.500710973671785e-06,
      "loss": 0.6917,
      "num_tokens": 8109830260.0,
      "step": 2654
    },
    {
      "epoch": 3.88807473896318,
      "grad_norm": 0.25233624958217415,
      "learning_rate": 8.489422978128866e-06,
      "loss": 0.672,
      "num_tokens": 8113078918.0,
      "step": 2655
    },
    {
      "epoch": 3.889540208829456,
      "grad_norm": 0.28944307034709277,
      "learning_rate": 8.478147138649285e-06,
      "loss": 0.6737,
      "num_tokens": 8116115483.0,
      "step": 2656
    },
    {
      "epoch": 3.891005678695732,
      "grad_norm": 0.26756053575908045,
      "learning_rate": 8.466883465378411e-06,
      "loss": 0.6781,
      "num_tokens": 8118972483.0,
      "step": 2657
    },
    {
      "epoch": 3.8924711485620076,
      "grad_norm": 0.29212649347832187,
      "learning_rate": 8.45563196845067e-06,
      "loss": 0.6887,
      "num_tokens": 8121794379.0,
      "step": 2658
    },
    {
      "epoch": 3.8939366184282838,
      "grad_norm": 0.26520099862955143,
      "learning_rate": 8.444392657989528e-06,
      "loss": 0.6698,
      "num_tokens": 8124854157.0,
      "step": 2659
    },
    {
      "epoch": 3.8954020882945595,
      "grad_norm": 0.27292468852138146,
      "learning_rate": 8.433165544107507e-06,
      "loss": 0.6614,
      "num_tokens": 8128044483.0,
      "step": 2660
    },
    {
      "epoch": 3.896867558160835,
      "grad_norm": 0.29125596078158467,
      "learning_rate": 8.42195063690611e-06,
      "loss": 0.6709,
      "num_tokens": 8130977878.0,
      "step": 2661
    },
    {
      "epoch": 3.8983330280271113,
      "grad_norm": 0.2713717567242872,
      "learning_rate": 8.410747946475904e-06,
      "loss": 0.6718,
      "num_tokens": 8134115439.0,
      "step": 2662
    },
    {
      "epoch": 3.899798497893387,
      "grad_norm": 0.252182453619379,
      "learning_rate": 8.399557482896438e-06,
      "loss": 0.6925,
      "num_tokens": 8137020312.0,
      "step": 2663
    },
    {
      "epoch": 3.9012639677596628,
      "grad_norm": 0.2631827515113229,
      "learning_rate": 8.388379256236269e-06,
      "loss": 0.6796,
      "num_tokens": 8139916933.0,
      "step": 2664
    },
    {
      "epoch": 3.902729437625939,
      "grad_norm": 0.27019753372277,
      "learning_rate": 8.377213276552938e-06,
      "loss": 0.6622,
      "num_tokens": 8143257060.0,
      "step": 2665
    },
    {
      "epoch": 3.9041949074922147,
      "grad_norm": 0.267678346606212,
      "learning_rate": 8.366059553892981e-06,
      "loss": 0.6825,
      "num_tokens": 8146165938.0,
      "step": 2666
    },
    {
      "epoch": 3.9056603773584904,
      "grad_norm": 0.26550430172064926,
      "learning_rate": 8.35491809829188e-06,
      "loss": 0.6765,
      "num_tokens": 8149232621.0,
      "step": 2667
    },
    {
      "epoch": 3.9071258472247665,
      "grad_norm": 0.27956038242878367,
      "learning_rate": 8.343788919774104e-06,
      "loss": 0.6714,
      "num_tokens": 8152266089.0,
      "step": 2668
    },
    {
      "epoch": 3.9085913170910422,
      "grad_norm": 0.2886528899403416,
      "learning_rate": 8.332672028353068e-06,
      "loss": 0.6697,
      "num_tokens": 8155351182.0,
      "step": 2669
    },
    {
      "epoch": 3.910056786957318,
      "grad_norm": 0.2848214444533986,
      "learning_rate": 8.321567434031129e-06,
      "loss": 0.6638,
      "num_tokens": 8158759182.0,
      "step": 2670
    },
    {
      "epoch": 3.911522256823594,
      "grad_norm": 0.2826495909714783,
      "learning_rate": 8.310475146799578e-06,
      "loss": 0.6829,
      "num_tokens": 8161982305.0,
      "step": 2671
    },
    {
      "epoch": 3.91298772668987,
      "grad_norm": 0.2734369572083052,
      "learning_rate": 8.299395176638638e-06,
      "loss": 0.672,
      "num_tokens": 8164927054.0,
      "step": 2672
    },
    {
      "epoch": 3.9144531965561455,
      "grad_norm": 0.28618864419998835,
      "learning_rate": 8.288327533517448e-06,
      "loss": 0.6945,
      "num_tokens": 8168098906.0,
      "step": 2673
    },
    {
      "epoch": 3.9159186664224217,
      "grad_norm": 0.28473972828008887,
      "learning_rate": 8.277272227394058e-06,
      "loss": 0.6724,
      "num_tokens": 8171009114.0,
      "step": 2674
    },
    {
      "epoch": 3.9173841362886974,
      "grad_norm": 0.2955752577152905,
      "learning_rate": 8.266229268215414e-06,
      "loss": 0.6751,
      "num_tokens": 8173971784.0,
      "step": 2675
    },
    {
      "epoch": 3.9188496061549736,
      "grad_norm": 0.29317270873316065,
      "learning_rate": 8.255198665917348e-06,
      "loss": 0.6627,
      "num_tokens": 8177242688.0,
      "step": 2676
    },
    {
      "epoch": 3.9203150760212493,
      "grad_norm": 0.27711415016209684,
      "learning_rate": 8.244180430424598e-06,
      "loss": 0.6801,
      "num_tokens": 8180112205.0,
      "step": 2677
    },
    {
      "epoch": 3.9217805458875254,
      "grad_norm": 0.30285404368057467,
      "learning_rate": 8.233174571650733e-06,
      "loss": 0.6575,
      "num_tokens": 8183186080.0,
      "step": 2678
    },
    {
      "epoch": 3.923246015753801,
      "grad_norm": 0.30574464741790747,
      "learning_rate": 8.222181099498232e-06,
      "loss": 0.6727,
      "num_tokens": 8186184573.0,
      "step": 2679
    },
    {
      "epoch": 3.924711485620077,
      "grad_norm": 0.24377703140438342,
      "learning_rate": 8.2112000238584e-06,
      "loss": 0.685,
      "num_tokens": 8189185770.0,
      "step": 2680
    },
    {
      "epoch": 3.926176955486353,
      "grad_norm": 0.2908946105420151,
      "learning_rate": 8.200231354611392e-06,
      "loss": 0.6569,
      "num_tokens": 8192151854.0,
      "step": 2681
    },
    {
      "epoch": 3.9276424253526288,
      "grad_norm": 0.27033060870625786,
      "learning_rate": 8.189275101626208e-06,
      "loss": 0.6643,
      "num_tokens": 8195097159.0,
      "step": 2682
    },
    {
      "epoch": 3.9291078952189045,
      "grad_norm": 0.28739385192497846,
      "learning_rate": 8.178331274760679e-06,
      "loss": 0.6855,
      "num_tokens": 8197984058.0,
      "step": 2683
    },
    {
      "epoch": 3.9305733650851806,
      "grad_norm": 0.2814848220229086,
      "learning_rate": 8.167399883861435e-06,
      "loss": 0.6812,
      "num_tokens": 8200676091.0,
      "step": 2684
    },
    {
      "epoch": 3.9320388349514563,
      "grad_norm": 0.28565660559527944,
      "learning_rate": 8.156480938763946e-06,
      "loss": 0.6875,
      "num_tokens": 8203488571.0,
      "step": 2685
    },
    {
      "epoch": 3.933504304817732,
      "grad_norm": 0.28735350334703336,
      "learning_rate": 8.145574449292468e-06,
      "loss": 0.6608,
      "num_tokens": 8206355904.0,
      "step": 2686
    },
    {
      "epoch": 3.934969774684008,
      "grad_norm": 0.26294996728443243,
      "learning_rate": 8.134680425260042e-06,
      "loss": 0.6733,
      "num_tokens": 8209330366.0,
      "step": 2687
    },
    {
      "epoch": 3.936435244550284,
      "grad_norm": 0.28986874912566435,
      "learning_rate": 8.123798876468514e-06,
      "loss": 0.6805,
      "num_tokens": 8212358483.0,
      "step": 2688
    },
    {
      "epoch": 3.9379007144165596,
      "grad_norm": 0.2527607986748959,
      "learning_rate": 8.112929812708488e-06,
      "loss": 0.658,
      "num_tokens": 8215469086.0,
      "step": 2689
    },
    {
      "epoch": 3.939366184282836,
      "grad_norm": 0.2831970824250484,
      "learning_rate": 8.102073243759336e-06,
      "loss": 0.6821,
      "num_tokens": 8218712626.0,
      "step": 2690
    },
    {
      "epoch": 3.9408316541491115,
      "grad_norm": 0.30981527380054547,
      "learning_rate": 8.091229179389209e-06,
      "loss": 0.6875,
      "num_tokens": 8221504098.0,
      "step": 2691
    },
    {
      "epoch": 3.9422971240153872,
      "grad_norm": 0.2743504406539703,
      "learning_rate": 8.080397629354973e-06,
      "loss": 0.6848,
      "num_tokens": 8224313899.0,
      "step": 2692
    },
    {
      "epoch": 3.9437625938816634,
      "grad_norm": 0.30681434989527984,
      "learning_rate": 8.069578603402266e-06,
      "loss": 0.6704,
      "num_tokens": 8227316252.0,
      "step": 2693
    },
    {
      "epoch": 3.945228063747939,
      "grad_norm": 0.2643501452346204,
      "learning_rate": 8.058772111265436e-06,
      "loss": 0.679,
      "num_tokens": 8230398772.0,
      "step": 2694
    },
    {
      "epoch": 3.946693533614215,
      "grad_norm": 0.28777457740519835,
      "learning_rate": 8.047978162667566e-06,
      "loss": 0.6674,
      "num_tokens": 8233555152.0,
      "step": 2695
    },
    {
      "epoch": 3.948159003480491,
      "grad_norm": 0.296933485932885,
      "learning_rate": 8.037196767320442e-06,
      "loss": 0.6841,
      "num_tokens": 8236482699.0,
      "step": 2696
    },
    {
      "epoch": 3.9496244733467667,
      "grad_norm": 0.2551988686917353,
      "learning_rate": 8.026427934924576e-06,
      "loss": 0.6641,
      "num_tokens": 8239753512.0,
      "step": 2697
    },
    {
      "epoch": 3.9510899432130424,
      "grad_norm": 0.3147732459808143,
      "learning_rate": 8.015671675169145e-06,
      "loss": 0.6666,
      "num_tokens": 8242884192.0,
      "step": 2698
    },
    {
      "epoch": 3.9525554130793186,
      "grad_norm": 0.27216797620622435,
      "learning_rate": 8.004927997732044e-06,
      "loss": 0.6651,
      "num_tokens": 8246073370.0,
      "step": 2699
    },
    {
      "epoch": 3.9540208829455943,
      "grad_norm": 0.2983248033143554,
      "learning_rate": 7.994196912279832e-06,
      "loss": 0.6954,
      "num_tokens": 8248875273.0,
      "step": 2700
    },
    {
      "epoch": 3.9554863528118704,
      "grad_norm": 0.3097831974006926,
      "learning_rate": 7.983478428467738e-06,
      "loss": 0.6588,
      "num_tokens": 8251993853.0,
      "step": 2701
    },
    {
      "epoch": 3.956951822678146,
      "grad_norm": 0.28790171837437484,
      "learning_rate": 7.972772555939654e-06,
      "loss": 0.6894,
      "num_tokens": 8255279431.0,
      "step": 2702
    },
    {
      "epoch": 3.9584172925444223,
      "grad_norm": 0.32784542248287957,
      "learning_rate": 7.962079304328132e-06,
      "loss": 0.6848,
      "num_tokens": 8258356073.0,
      "step": 2703
    },
    {
      "epoch": 3.959882762410698,
      "grad_norm": 0.3085463965001779,
      "learning_rate": 7.951398683254354e-06,
      "loss": 0.6731,
      "num_tokens": 8261600675.0,
      "step": 2704
    },
    {
      "epoch": 3.9613482322769737,
      "grad_norm": 0.26266428658570223,
      "learning_rate": 7.940730702328165e-06,
      "loss": 0.6784,
      "num_tokens": 8264520375.0,
      "step": 2705
    },
    {
      "epoch": 3.96281370214325,
      "grad_norm": 0.292800762644355,
      "learning_rate": 7.930075371147996e-06,
      "loss": 0.6828,
      "num_tokens": 8267455973.0,
      "step": 2706
    },
    {
      "epoch": 3.9642791720095256,
      "grad_norm": 0.2855124226713279,
      "learning_rate": 7.91943269930094e-06,
      "loss": 0.6802,
      "num_tokens": 8270174899.0,
      "step": 2707
    },
    {
      "epoch": 3.9657446418758013,
      "grad_norm": 0.2591585490487283,
      "learning_rate": 7.908802696362666e-06,
      "loss": 0.6765,
      "num_tokens": 8273329226.0,
      "step": 2708
    },
    {
      "epoch": 3.9672101117420775,
      "grad_norm": 0.2522413383355082,
      "learning_rate": 7.898185371897465e-06,
      "loss": 0.6671,
      "num_tokens": 8276515983.0,
      "step": 2709
    },
    {
      "epoch": 3.968675581608353,
      "grad_norm": 0.2818748314538766,
      "learning_rate": 7.887580735458202e-06,
      "loss": 0.6861,
      "num_tokens": 8279707138.0,
      "step": 2710
    },
    {
      "epoch": 3.970141051474629,
      "grad_norm": 0.26344856502568376,
      "learning_rate": 7.876988796586355e-06,
      "loss": 0.6698,
      "num_tokens": 8282539039.0,
      "step": 2711
    },
    {
      "epoch": 3.971606521340905,
      "grad_norm": 0.2663220145376771,
      "learning_rate": 7.866409564811941e-06,
      "loss": 0.6541,
      "num_tokens": 8285552351.0,
      "step": 2712
    },
    {
      "epoch": 3.973071991207181,
      "grad_norm": 0.24150203982237475,
      "learning_rate": 7.855843049653577e-06,
      "loss": 0.6617,
      "num_tokens": 8288581585.0,
      "step": 2713
    },
    {
      "epoch": 3.9745374610734565,
      "grad_norm": 0.33394931652875476,
      "learning_rate": 7.84528926061842e-06,
      "loss": 0.6604,
      "num_tokens": 8291979499.0,
      "step": 2714
    },
    {
      "epoch": 3.9760029309397327,
      "grad_norm": 0.2589091075068548,
      "learning_rate": 7.834748207202171e-06,
      "loss": 0.6588,
      "num_tokens": 8295457376.0,
      "step": 2715
    },
    {
      "epoch": 3.9774684008060084,
      "grad_norm": 0.2774736726001636,
      "learning_rate": 7.824219898889093e-06,
      "loss": 0.6744,
      "num_tokens": 8298613024.0,
      "step": 2716
    },
    {
      "epoch": 3.978933870672284,
      "grad_norm": 0.259413832309309,
      "learning_rate": 7.813704345151966e-06,
      "loss": 0.6677,
      "num_tokens": 8301688640.0,
      "step": 2717
    },
    {
      "epoch": 3.9803993405385603,
      "grad_norm": 0.2595762142389086,
      "learning_rate": 7.8032015554521e-06,
      "loss": 0.6682,
      "num_tokens": 8304730302.0,
      "step": 2718
    },
    {
      "epoch": 3.981864810404836,
      "grad_norm": 0.28604162698048496,
      "learning_rate": 7.792711539239317e-06,
      "loss": 0.6835,
      "num_tokens": 8307781206.0,
      "step": 2719
    },
    {
      "epoch": 3.9833302802711117,
      "grad_norm": 0.27218237848258076,
      "learning_rate": 7.78223430595195e-06,
      "loss": 0.6598,
      "num_tokens": 8310588815.0,
      "step": 2720
    },
    {
      "epoch": 3.984795750137388,
      "grad_norm": 0.2666228702670444,
      "learning_rate": 7.771769865016821e-06,
      "loss": 0.6684,
      "num_tokens": 8313842554.0,
      "step": 2721
    },
    {
      "epoch": 3.9862612200036636,
      "grad_norm": 0.29323205318879547,
      "learning_rate": 7.761318225849266e-06,
      "loss": 0.651,
      "num_tokens": 8317098795.0,
      "step": 2722
    },
    {
      "epoch": 3.9877266898699393,
      "grad_norm": 0.27684218849403797,
      "learning_rate": 7.750879397853074e-06,
      "loss": 0.6646,
      "num_tokens": 8320447160.0,
      "step": 2723
    },
    {
      "epoch": 3.9891921597362154,
      "grad_norm": 0.2898667035902747,
      "learning_rate": 7.740453390420526e-06,
      "loss": 0.6761,
      "num_tokens": 8323494051.0,
      "step": 2724
    },
    {
      "epoch": 3.990657629602491,
      "grad_norm": 0.2894839920674067,
      "learning_rate": 7.73004021293236e-06,
      "loss": 0.6761,
      "num_tokens": 8326654926.0,
      "step": 2725
    },
    {
      "epoch": 3.9921230994687673,
      "grad_norm": 0.25343461168804654,
      "learning_rate": 7.71963987475777e-06,
      "loss": 0.6919,
      "num_tokens": 8329677946.0,
      "step": 2726
    },
    {
      "epoch": 3.993588569335043,
      "grad_norm": 0.29682184585993415,
      "learning_rate": 7.709252385254397e-06,
      "loss": 0.6849,
      "num_tokens": 8332626454.0,
      "step": 2727
    },
    {
      "epoch": 3.995054039201319,
      "grad_norm": 0.2911049582580313,
      "learning_rate": 7.698877753768342e-06,
      "loss": 0.6723,
      "num_tokens": 8335773538.0,
      "step": 2728
    },
    {
      "epoch": 3.996519509067595,
      "grad_norm": 0.26461608495369726,
      "learning_rate": 7.688515989634094e-06,
      "loss": 0.6755,
      "num_tokens": 8338733043.0,
      "step": 2729
    },
    {
      "epoch": 3.9979849789338706,
      "grad_norm": 0.3181464556943696,
      "learning_rate": 7.67816710217461e-06,
      "loss": 0.6518,
      "num_tokens": 8341589989.0,
      "step": 2730
    },
    {
      "epoch": 3.9994504488001468,
      "grad_norm": 0.30348912511808096,
      "learning_rate": 7.667831100701231e-06,
      "loss": 0.6718,
      "num_tokens": 8344725181.0,
      "step": 2731
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.30348912511808096,
      "learning_rate": 7.65750799451372e-06,
      "loss": 0.7002,
      "num_tokens": 8345433648.0,
      "step": 2732
    },
    {
      "epoch": 4.001465469866276,
      "grad_norm": 0.508376068031645,
      "learning_rate": 7.647197792900229e-06,
      "loss": 0.6838,
      "num_tokens": 8348399184.0,
      "step": 2733
    },
    {
      "epoch": 4.002930939732551,
      "grad_norm": 0.3144344002958086,
      "learning_rate": 7.636900505137303e-06,
      "loss": 0.6579,
      "num_tokens": 8351412987.0,
      "step": 2734
    },
    {
      "epoch": 4.004396409598828,
      "grad_norm": 0.3084916306748364,
      "learning_rate": 7.626616140489862e-06,
      "loss": 0.66,
      "num_tokens": 8354150387.0,
      "step": 2735
    },
    {
      "epoch": 4.005861879465104,
      "grad_norm": 0.27382554103019996,
      "learning_rate": 7.616344708211219e-06,
      "loss": 0.6728,
      "num_tokens": 8357185099.0,
      "step": 2736
    },
    {
      "epoch": 4.007327349331379,
      "grad_norm": 0.3215479218309215,
      "learning_rate": 7.606086217543012e-06,
      "loss": 0.6528,
      "num_tokens": 8360297951.0,
      "step": 2737
    },
    {
      "epoch": 4.008792819197655,
      "grad_norm": 0.29619854562154374,
      "learning_rate": 7.595840677715278e-06,
      "loss": 0.6689,
      "num_tokens": 8363408275.0,
      "step": 2738
    },
    {
      "epoch": 4.010258289063931,
      "grad_norm": 0.3249265795661397,
      "learning_rate": 7.585608097946378e-06,
      "loss": 0.6641,
      "num_tokens": 8366476373.0,
      "step": 2739
    },
    {
      "epoch": 4.011723758930207,
      "grad_norm": 0.330551640084049,
      "learning_rate": 7.5753884874430116e-06,
      "loss": 0.676,
      "num_tokens": 8369410282.0,
      "step": 2740
    },
    {
      "epoch": 4.013189228796483,
      "grad_norm": 0.2716452197475487,
      "learning_rate": 7.56518185540021e-06,
      "loss": 0.6631,
      "num_tokens": 8372631249.0,
      "step": 2741
    },
    {
      "epoch": 4.014654698662759,
      "grad_norm": 0.31013797919860586,
      "learning_rate": 7.554988211001349e-06,
      "loss": 0.6616,
      "num_tokens": 8375791215.0,
      "step": 2742
    },
    {
      "epoch": 4.016120168529034,
      "grad_norm": 0.25794433983908893,
      "learning_rate": 7.5448075634180774e-06,
      "loss": 0.6649,
      "num_tokens": 8378640904.0,
      "step": 2743
    },
    {
      "epoch": 4.01758563839531,
      "grad_norm": 0.2977701886352675,
      "learning_rate": 7.534639921810393e-06,
      "loss": 0.6713,
      "num_tokens": 8381717076.0,
      "step": 2744
    },
    {
      "epoch": 4.0190511082615865,
      "grad_norm": 0.2847131601267182,
      "learning_rate": 7.5244852953265645e-06,
      "loss": 0.6835,
      "num_tokens": 8384624240.0,
      "step": 2745
    },
    {
      "epoch": 4.020516578127863,
      "grad_norm": 0.25013769792327667,
      "learning_rate": 7.51434369310316e-06,
      "loss": 0.6633,
      "num_tokens": 8387770443.0,
      "step": 2746
    },
    {
      "epoch": 4.021982047994138,
      "grad_norm": 0.278537532010583,
      "learning_rate": 7.504215124265024e-06,
      "loss": 0.652,
      "num_tokens": 8390844504.0,
      "step": 2747
    },
    {
      "epoch": 4.023447517860414,
      "grad_norm": 0.24889276249252337,
      "learning_rate": 7.494099597925282e-06,
      "loss": 0.682,
      "num_tokens": 8393801605.0,
      "step": 2748
    },
    {
      "epoch": 4.02491298772669,
      "grad_norm": 0.266188825280029,
      "learning_rate": 7.4839971231853135e-06,
      "loss": 0.6382,
      "num_tokens": 8397032028.0,
      "step": 2749
    },
    {
      "epoch": 4.0263784575929655,
      "grad_norm": 0.25704780519505965,
      "learning_rate": 7.4739077091347736e-06,
      "loss": 0.6723,
      "num_tokens": 8400041175.0,
      "step": 2750
    },
    {
      "epoch": 4.027843927459242,
      "grad_norm": 0.25329141996211546,
      "learning_rate": 7.4638313648515394e-06,
      "loss": 0.6713,
      "num_tokens": 8402967808.0,
      "step": 2751
    },
    {
      "epoch": 4.029309397325518,
      "grad_norm": 0.2709114702585647,
      "learning_rate": 7.453768099401753e-06,
      "loss": 0.6779,
      "num_tokens": 8406118987.0,
      "step": 2752
    },
    {
      "epoch": 4.030774867191793,
      "grad_norm": 0.2783717176225457,
      "learning_rate": 7.443717921839777e-06,
      "loss": 0.6781,
      "num_tokens": 8409062428.0,
      "step": 2753
    },
    {
      "epoch": 4.032240337058069,
      "grad_norm": 0.2454565881005608,
      "learning_rate": 7.433680841208201e-06,
      "loss": 0.6689,
      "num_tokens": 8412253350.0,
      "step": 2754
    },
    {
      "epoch": 4.033705806924345,
      "grad_norm": 0.26045533324812575,
      "learning_rate": 7.4236568665378284e-06,
      "loss": 0.6532,
      "num_tokens": 8415584326.0,
      "step": 2755
    },
    {
      "epoch": 4.035171276790621,
      "grad_norm": 0.28096285529281234,
      "learning_rate": 7.413646006847673e-06,
      "loss": 0.6627,
      "num_tokens": 8418621493.0,
      "step": 2756
    },
    {
      "epoch": 4.036636746656897,
      "grad_norm": 0.23311111814616522,
      "learning_rate": 7.403648271144948e-06,
      "loss": 0.6571,
      "num_tokens": 8421767486.0,
      "step": 2757
    },
    {
      "epoch": 4.038102216523173,
      "grad_norm": 0.2762367998712191,
      "learning_rate": 7.393663668425053e-06,
      "loss": 0.691,
      "num_tokens": 8424680695.0,
      "step": 2758
    },
    {
      "epoch": 4.039567686389448,
      "grad_norm": 0.26476568984680576,
      "learning_rate": 7.383692207671591e-06,
      "loss": 0.6568,
      "num_tokens": 8427832208.0,
      "step": 2759
    },
    {
      "epoch": 4.0410331562557245,
      "grad_norm": 0.2675081083730221,
      "learning_rate": 7.373733897856305e-06,
      "loss": 0.6718,
      "num_tokens": 8430643530.0,
      "step": 2760
    },
    {
      "epoch": 4.042498626122001,
      "grad_norm": 0.28654333786659736,
      "learning_rate": 7.363788747939145e-06,
      "loss": 0.6726,
      "num_tokens": 8433629583.0,
      "step": 2761
    },
    {
      "epoch": 4.043964095988276,
      "grad_norm": 0.27260832203740254,
      "learning_rate": 7.3538567668681945e-06,
      "loss": 0.6528,
      "num_tokens": 8436798377.0,
      "step": 2762
    },
    {
      "epoch": 4.045429565854552,
      "grad_norm": 0.2781331400079584,
      "learning_rate": 7.343937963579695e-06,
      "loss": 0.6687,
      "num_tokens": 8439942152.0,
      "step": 2763
    },
    {
      "epoch": 4.046895035720828,
      "grad_norm": 0.29798966960144907,
      "learning_rate": 7.334032346998034e-06,
      "loss": 0.6613,
      "num_tokens": 8443170238.0,
      "step": 2764
    },
    {
      "epoch": 4.0483605055871035,
      "grad_norm": 0.2603043296538314,
      "learning_rate": 7.324139926035731e-06,
      "loss": 0.6812,
      "num_tokens": 8446493634.0,
      "step": 2765
    },
    {
      "epoch": 4.04982597545338,
      "grad_norm": 0.27220310310696,
      "learning_rate": 7.3142607095934305e-06,
      "loss": 0.6703,
      "num_tokens": 8449560608.0,
      "step": 2766
    },
    {
      "epoch": 4.051291445319656,
      "grad_norm": 0.26175798864093436,
      "learning_rate": 7.304394706559912e-06,
      "loss": 0.6653,
      "num_tokens": 8452731619.0,
      "step": 2767
    },
    {
      "epoch": 4.052756915185931,
      "grad_norm": 0.26363203450344047,
      "learning_rate": 7.294541925812045e-06,
      "loss": 0.6471,
      "num_tokens": 8456066377.0,
      "step": 2768
    },
    {
      "epoch": 4.054222385052207,
      "grad_norm": 0.2686557795862108,
      "learning_rate": 7.284702376214816e-06,
      "loss": 0.6627,
      "num_tokens": 8459107661.0,
      "step": 2769
    },
    {
      "epoch": 4.055687854918483,
      "grad_norm": 0.28977322650579684,
      "learning_rate": 7.274876066621299e-06,
      "loss": 0.6892,
      "num_tokens": 8462107866.0,
      "step": 2770
    },
    {
      "epoch": 4.0571533247847595,
      "grad_norm": 0.2869120462509204,
      "learning_rate": 7.2650630058726655e-06,
      "loss": 0.6562,
      "num_tokens": 8465169898.0,
      "step": 2771
    },
    {
      "epoch": 4.058618794651035,
      "grad_norm": 0.2384630582317283,
      "learning_rate": 7.255263202798146e-06,
      "loss": 0.6889,
      "num_tokens": 8468094534.0,
      "step": 2772
    },
    {
      "epoch": 4.060084264517311,
      "grad_norm": 0.2800094729451123,
      "learning_rate": 7.245476666215079e-06,
      "loss": 0.6715,
      "num_tokens": 8471037895.0,
      "step": 2773
    },
    {
      "epoch": 4.061549734383587,
      "grad_norm": 0.2678315392760878,
      "learning_rate": 7.23570340492882e-06,
      "loss": 0.6607,
      "num_tokens": 8474024951.0,
      "step": 2774
    },
    {
      "epoch": 4.063015204249862,
      "grad_norm": 0.27369581622668937,
      "learning_rate": 7.225943427732821e-06,
      "loss": 0.66,
      "num_tokens": 8477048924.0,
      "step": 2775
    },
    {
      "epoch": 4.064480674116139,
      "grad_norm": 0.2720273948972367,
      "learning_rate": 7.2161967434085615e-06,
      "loss": 0.6561,
      "num_tokens": 8480109703.0,
      "step": 2776
    },
    {
      "epoch": 4.065946143982415,
      "grad_norm": 0.2752478001555263,
      "learning_rate": 7.206463360725557e-06,
      "loss": 0.6685,
      "num_tokens": 8483388100.0,
      "step": 2777
    },
    {
      "epoch": 4.06741161384869,
      "grad_norm": 0.28040905587354104,
      "learning_rate": 7.1967432884413715e-06,
      "loss": 0.677,
      "num_tokens": 8486352183.0,
      "step": 2778
    },
    {
      "epoch": 4.068877083714966,
      "grad_norm": 0.2504779867523387,
      "learning_rate": 7.187036535301579e-06,
      "loss": 0.6693,
      "num_tokens": 8489535049.0,
      "step": 2779
    },
    {
      "epoch": 4.070342553581242,
      "grad_norm": 0.28816548136254017,
      "learning_rate": 7.17734311003977e-06,
      "loss": 0.6458,
      "num_tokens": 8492683871.0,
      "step": 2780
    },
    {
      "epoch": 4.071808023447518,
      "grad_norm": 0.25485690748991907,
      "learning_rate": 7.1676630213775535e-06,
      "loss": 0.6677,
      "num_tokens": 8495722829.0,
      "step": 2781
    },
    {
      "epoch": 4.073273493313794,
      "grad_norm": 0.25307589836860095,
      "learning_rate": 7.1579962780245345e-06,
      "loss": 0.6499,
      "num_tokens": 8498962950.0,
      "step": 2782
    },
    {
      "epoch": 4.07473896318007,
      "grad_norm": 0.26571228843267947,
      "learning_rate": 7.148342888678308e-06,
      "loss": 0.6564,
      "num_tokens": 8502227546.0,
      "step": 2783
    },
    {
      "epoch": 4.076204433046345,
      "grad_norm": 0.2691535135822236,
      "learning_rate": 7.138702862024452e-06,
      "loss": 0.6465,
      "num_tokens": 8505296443.0,
      "step": 2784
    },
    {
      "epoch": 4.077669902912621,
      "grad_norm": 0.27704830843265577,
      "learning_rate": 7.129076206736527e-06,
      "loss": 0.6756,
      "num_tokens": 8508376837.0,
      "step": 2785
    },
    {
      "epoch": 4.0791353727788975,
      "grad_norm": 0.2727565064513333,
      "learning_rate": 7.119462931476056e-06,
      "loss": 0.6802,
      "num_tokens": 8511385097.0,
      "step": 2786
    },
    {
      "epoch": 4.080600842645173,
      "grad_norm": 0.25546163663204835,
      "learning_rate": 7.109863044892539e-06,
      "loss": 0.6651,
      "num_tokens": 8514332038.0,
      "step": 2787
    },
    {
      "epoch": 4.082066312511449,
      "grad_norm": 0.2925241338954218,
      "learning_rate": 7.100276555623402e-06,
      "loss": 0.6779,
      "num_tokens": 8517413361.0,
      "step": 2788
    },
    {
      "epoch": 4.083531782377725,
      "grad_norm": 0.2691651755761821,
      "learning_rate": 7.090703472294047e-06,
      "loss": 0.6654,
      "num_tokens": 8520505156.0,
      "step": 2789
    },
    {
      "epoch": 4.084997252244,
      "grad_norm": 0.28101539473689946,
      "learning_rate": 7.081143803517795e-06,
      "loss": 0.6715,
      "num_tokens": 8523606749.0,
      "step": 2790
    },
    {
      "epoch": 4.0864627221102765,
      "grad_norm": 0.2551357394655886,
      "learning_rate": 7.071597557895903e-06,
      "loss": 0.6778,
      "num_tokens": 8526591258.0,
      "step": 2791
    },
    {
      "epoch": 4.087928191976553,
      "grad_norm": 0.2877777014996474,
      "learning_rate": 7.062064744017549e-06,
      "loss": 0.672,
      "num_tokens": 8529705104.0,
      "step": 2792
    },
    {
      "epoch": 4.089393661842828,
      "grad_norm": 0.28732767977322776,
      "learning_rate": 7.052545370459829e-06,
      "loss": 0.6709,
      "num_tokens": 8532854574.0,
      "step": 2793
    },
    {
      "epoch": 4.090859131709104,
      "grad_norm": 0.2562060466659093,
      "learning_rate": 7.043039445787738e-06,
      "loss": 0.6689,
      "num_tokens": 8535883563.0,
      "step": 2794
    },
    {
      "epoch": 4.09232460157538,
      "grad_norm": 0.27597843091009455,
      "learning_rate": 7.0335469785541845e-06,
      "loss": 0.6681,
      "num_tokens": 8538956887.0,
      "step": 2795
    },
    {
      "epoch": 4.093790071441656,
      "grad_norm": 0.2520921805910092,
      "learning_rate": 7.0240679772999644e-06,
      "loss": 0.6668,
      "num_tokens": 8542050016.0,
      "step": 2796
    },
    {
      "epoch": 4.095255541307932,
      "grad_norm": 0.2688655414760752,
      "learning_rate": 7.014602450553738e-06,
      "loss": 0.6677,
      "num_tokens": 8545002503.0,
      "step": 2797
    },
    {
      "epoch": 4.096721011174208,
      "grad_norm": 0.2525052421507422,
      "learning_rate": 7.005150406832072e-06,
      "loss": 0.6625,
      "num_tokens": 8548222591.0,
      "step": 2798
    },
    {
      "epoch": 4.098186481040484,
      "grad_norm": 0.27306484967325173,
      "learning_rate": 6.995711854639384e-06,
      "loss": 0.674,
      "num_tokens": 8551180054.0,
      "step": 2799
    },
    {
      "epoch": 4.099651950906759,
      "grad_norm": 0.2573022679759736,
      "learning_rate": 6.986286802467955e-06,
      "loss": 0.6532,
      "num_tokens": 8554225194.0,
      "step": 2800
    },
    {
      "epoch": 4.101117420773035,
      "grad_norm": 0.24682498994688504,
      "learning_rate": 6.9768752587979236e-06,
      "loss": 0.6608,
      "num_tokens": 8557174742.0,
      "step": 2801
    },
    {
      "epoch": 4.102582890639312,
      "grad_norm": 0.25770724829122393,
      "learning_rate": 6.967477232097271e-06,
      "loss": 0.6638,
      "num_tokens": 8560076199.0,
      "step": 2802
    },
    {
      "epoch": 4.104048360505587,
      "grad_norm": 0.2766893510700863,
      "learning_rate": 6.9580927308218126e-06,
      "loss": 0.6577,
      "num_tokens": 8563106130.0,
      "step": 2803
    },
    {
      "epoch": 4.105513830371863,
      "grad_norm": 0.2668644049813433,
      "learning_rate": 6.948721763415209e-06,
      "loss": 0.6637,
      "num_tokens": 8566291199.0,
      "step": 2804
    },
    {
      "epoch": 4.106979300238139,
      "grad_norm": 0.24544302409094326,
      "learning_rate": 6.9393643383089206e-06,
      "loss": 0.6773,
      "num_tokens": 8569385724.0,
      "step": 2805
    },
    {
      "epoch": 4.1084447701044144,
      "grad_norm": 0.25711325814094943,
      "learning_rate": 6.930020463922248e-06,
      "loss": 0.6445,
      "num_tokens": 8572235763.0,
      "step": 2806
    },
    {
      "epoch": 4.109910239970691,
      "grad_norm": 0.2745625023578714,
      "learning_rate": 6.920690148662285e-06,
      "loss": 0.6659,
      "num_tokens": 8575382442.0,
      "step": 2807
    },
    {
      "epoch": 4.111375709836967,
      "grad_norm": 0.28046579867527816,
      "learning_rate": 6.9113734009239246e-06,
      "loss": 0.6521,
      "num_tokens": 8578473083.0,
      "step": 2808
    },
    {
      "epoch": 4.112841179703242,
      "grad_norm": 0.27516697154458775,
      "learning_rate": 6.902070229089861e-06,
      "loss": 0.6823,
      "num_tokens": 8581550469.0,
      "step": 2809
    },
    {
      "epoch": 4.114306649569518,
      "grad_norm": 0.28370926594353296,
      "learning_rate": 6.892780641530565e-06,
      "loss": 0.6581,
      "num_tokens": 8584598882.0,
      "step": 2810
    },
    {
      "epoch": 4.115772119435794,
      "grad_norm": 0.2935151083423503,
      "learning_rate": 6.88350464660429e-06,
      "loss": 0.6895,
      "num_tokens": 8587514006.0,
      "step": 2811
    },
    {
      "epoch": 4.11723758930207,
      "grad_norm": 0.27530642435058994,
      "learning_rate": 6.874242252657062e-06,
      "loss": 0.6741,
      "num_tokens": 8590242141.0,
      "step": 2812
    },
    {
      "epoch": 4.118703059168346,
      "grad_norm": 0.2804765048147015,
      "learning_rate": 6.864993468022666e-06,
      "loss": 0.6473,
      "num_tokens": 8593551261.0,
      "step": 2813
    },
    {
      "epoch": 4.120168529034622,
      "grad_norm": 0.2819955620599147,
      "learning_rate": 6.855758301022641e-06,
      "loss": 0.6588,
      "num_tokens": 8596582826.0,
      "step": 2814
    },
    {
      "epoch": 4.121633998900897,
      "grad_norm": 0.25105784008232,
      "learning_rate": 6.8465367599662745e-06,
      "loss": 0.6601,
      "num_tokens": 8599685975.0,
      "step": 2815
    },
    {
      "epoch": 4.123099468767173,
      "grad_norm": 0.26715117170841746,
      "learning_rate": 6.837328853150594e-06,
      "loss": 0.6747,
      "num_tokens": 8602655170.0,
      "step": 2816
    },
    {
      "epoch": 4.1245649386334495,
      "grad_norm": 0.2908835889941841,
      "learning_rate": 6.828134588860358e-06,
      "loss": 0.6684,
      "num_tokens": 8605736562.0,
      "step": 2817
    },
    {
      "epoch": 4.126030408499725,
      "grad_norm": 0.26599414503431307,
      "learning_rate": 6.818953975368061e-06,
      "loss": 0.6757,
      "num_tokens": 8608775792.0,
      "step": 2818
    },
    {
      "epoch": 4.127495878366001,
      "grad_norm": 0.27629417984564264,
      "learning_rate": 6.809787020933896e-06,
      "loss": 0.6834,
      "num_tokens": 8611967538.0,
      "step": 2819
    },
    {
      "epoch": 4.128961348232277,
      "grad_norm": 0.2509651557802986,
      "learning_rate": 6.800633733805786e-06,
      "loss": 0.6945,
      "num_tokens": 8614899672.0,
      "step": 2820
    },
    {
      "epoch": 4.130426818098553,
      "grad_norm": 0.29996320575949453,
      "learning_rate": 6.791494122219343e-06,
      "loss": 0.6672,
      "num_tokens": 8618060382.0,
      "step": 2821
    },
    {
      "epoch": 4.1318922879648285,
      "grad_norm": 0.256748611839633,
      "learning_rate": 6.782368194397881e-06,
      "loss": 0.6623,
      "num_tokens": 8621093792.0,
      "step": 2822
    },
    {
      "epoch": 4.133357757831105,
      "grad_norm": 0.2624601956029337,
      "learning_rate": 6.773255958552401e-06,
      "loss": 0.6722,
      "num_tokens": 8624242371.0,
      "step": 2823
    },
    {
      "epoch": 4.134823227697381,
      "grad_norm": 0.2581501662833647,
      "learning_rate": 6.7641574228815854e-06,
      "loss": 0.6801,
      "num_tokens": 8627004038.0,
      "step": 2824
    },
    {
      "epoch": 4.136288697563656,
      "grad_norm": 0.2549748239321414,
      "learning_rate": 6.755072595571781e-06,
      "loss": 0.6774,
      "num_tokens": 8630118940.0,
      "step": 2825
    },
    {
      "epoch": 4.137754167429932,
      "grad_norm": 0.2601687660352944,
      "learning_rate": 6.74600148479702e-06,
      "loss": 0.6697,
      "num_tokens": 8633338909.0,
      "step": 2826
    },
    {
      "epoch": 4.1392196372962085,
      "grad_norm": 0.25277188567077785,
      "learning_rate": 6.736944098718978e-06,
      "loss": 0.6749,
      "num_tokens": 8636207710.0,
      "step": 2827
    },
    {
      "epoch": 4.140685107162484,
      "grad_norm": 0.2607393759310363,
      "learning_rate": 6.727900445486983e-06,
      "loss": 0.6769,
      "num_tokens": 8639401699.0,
      "step": 2828
    },
    {
      "epoch": 4.14215057702876,
      "grad_norm": 0.24285268823983108,
      "learning_rate": 6.718870533238011e-06,
      "loss": 0.6766,
      "num_tokens": 8642491244.0,
      "step": 2829
    },
    {
      "epoch": 4.143616046895036,
      "grad_norm": 0.262592554246135,
      "learning_rate": 6.709854370096672e-06,
      "loss": 0.6607,
      "num_tokens": 8645718000.0,
      "step": 2830
    },
    {
      "epoch": 4.145081516761311,
      "grad_norm": 0.25758937452229047,
      "learning_rate": 6.700851964175206e-06,
      "loss": 0.67,
      "num_tokens": 8648763538.0,
      "step": 2831
    },
    {
      "epoch": 4.1465469866275875,
      "grad_norm": 0.24844567767016623,
      "learning_rate": 6.691863323573482e-06,
      "loss": 0.6618,
      "num_tokens": 8651611884.0,
      "step": 2832
    },
    {
      "epoch": 4.148012456493864,
      "grad_norm": 0.26467954844147473,
      "learning_rate": 6.682888456378966e-06,
      "loss": 0.6547,
      "num_tokens": 8654621547.0,
      "step": 2833
    },
    {
      "epoch": 4.149477926360139,
      "grad_norm": 0.2741363742067731,
      "learning_rate": 6.673927370666753e-06,
      "loss": 0.6553,
      "num_tokens": 8657819976.0,
      "step": 2834
    },
    {
      "epoch": 4.150943396226415,
      "grad_norm": 0.24974125668223987,
      "learning_rate": 6.664980074499524e-06,
      "loss": 0.6659,
      "num_tokens": 8660939684.0,
      "step": 2835
    },
    {
      "epoch": 4.152408866092691,
      "grad_norm": 0.2615721039527689,
      "learning_rate": 6.656046575927557e-06,
      "loss": 0.6565,
      "num_tokens": 8663893867.0,
      "step": 2836
    },
    {
      "epoch": 4.1538743359589665,
      "grad_norm": 0.27134340688799297,
      "learning_rate": 6.647126882988715e-06,
      "loss": 0.6632,
      "num_tokens": 8667133857.0,
      "step": 2837
    },
    {
      "epoch": 4.155339805825243,
      "grad_norm": 0.2591575113788516,
      "learning_rate": 6.638221003708441e-06,
      "loss": 0.6774,
      "num_tokens": 8670073016.0,
      "step": 2838
    },
    {
      "epoch": 4.156805275691519,
      "grad_norm": 0.2598364447796385,
      "learning_rate": 6.6293289460997466e-06,
      "loss": 0.6545,
      "num_tokens": 8673070846.0,
      "step": 2839
    },
    {
      "epoch": 4.158270745557794,
      "grad_norm": 0.26041529014438575,
      "learning_rate": 6.620450718163207e-06,
      "loss": 0.6548,
      "num_tokens": 8676158455.0,
      "step": 2840
    },
    {
      "epoch": 4.15973621542407,
      "grad_norm": 0.23040494689385213,
      "learning_rate": 6.611586327886965e-06,
      "loss": 0.6503,
      "num_tokens": 8679130398.0,
      "step": 2841
    },
    {
      "epoch": 4.161201685290346,
      "grad_norm": 0.2619688483576692,
      "learning_rate": 6.6027357832466935e-06,
      "loss": 0.6638,
      "num_tokens": 8682027738.0,
      "step": 2842
    },
    {
      "epoch": 4.162667155156622,
      "grad_norm": 0.28026662814851494,
      "learning_rate": 6.593899092205627e-06,
      "loss": 0.6805,
      "num_tokens": 8685076316.0,
      "step": 2843
    },
    {
      "epoch": 4.164132625022898,
      "grad_norm": 0.2330718353126558,
      "learning_rate": 6.585076262714523e-06,
      "loss": 0.6484,
      "num_tokens": 8688143963.0,
      "step": 2844
    },
    {
      "epoch": 4.165598094889174,
      "grad_norm": 0.27017623742026037,
      "learning_rate": 6.576267302711674e-06,
      "loss": 0.6536,
      "num_tokens": 8691420125.0,
      "step": 2845
    },
    {
      "epoch": 4.16706356475545,
      "grad_norm": 0.23910696414953064,
      "learning_rate": 6.567472220122888e-06,
      "loss": 0.6639,
      "num_tokens": 8694567921.0,
      "step": 2846
    },
    {
      "epoch": 4.168529034621725,
      "grad_norm": 0.2404539323189428,
      "learning_rate": 6.5586910228614895e-06,
      "loss": 0.6605,
      "num_tokens": 8697466357.0,
      "step": 2847
    },
    {
      "epoch": 4.169994504488002,
      "grad_norm": 0.26139040310734346,
      "learning_rate": 6.549923718828306e-06,
      "loss": 0.6366,
      "num_tokens": 8700709023.0,
      "step": 2848
    },
    {
      "epoch": 4.171459974354278,
      "grad_norm": 0.2483274716151525,
      "learning_rate": 6.541170315911685e-06,
      "loss": 0.6591,
      "num_tokens": 8703447761.0,
      "step": 2849
    },
    {
      "epoch": 4.172925444220553,
      "grad_norm": 0.25282619636532216,
      "learning_rate": 6.532430821987426e-06,
      "loss": 0.6573,
      "num_tokens": 8706505608.0,
      "step": 2850
    },
    {
      "epoch": 4.174390914086829,
      "grad_norm": 0.2580020252215847,
      "learning_rate": 6.523705244918856e-06,
      "loss": 0.658,
      "num_tokens": 8709830150.0,
      "step": 2851
    },
    {
      "epoch": 4.175856383953105,
      "grad_norm": 0.2828820598737101,
      "learning_rate": 6.514993592556757e-06,
      "loss": 0.6582,
      "num_tokens": 8712884175.0,
      "step": 2852
    },
    {
      "epoch": 4.177321853819381,
      "grad_norm": 0.2502952083921611,
      "learning_rate": 6.506295872739388e-06,
      "loss": 0.6598,
      "num_tokens": 8715834191.0,
      "step": 2853
    },
    {
      "epoch": 4.178787323685657,
      "grad_norm": 0.24524123441380385,
      "learning_rate": 6.497612093292466e-06,
      "loss": 0.6504,
      "num_tokens": 8718560140.0,
      "step": 2854
    },
    {
      "epoch": 4.180252793551933,
      "grad_norm": 0.2611638560558994,
      "learning_rate": 6.488942262029185e-06,
      "loss": 0.6584,
      "num_tokens": 8721477799.0,
      "step": 2855
    },
    {
      "epoch": 4.181718263418208,
      "grad_norm": 0.24563653514466768,
      "learning_rate": 6.480286386750158e-06,
      "loss": 0.6677,
      "num_tokens": 8724416503.0,
      "step": 2856
    },
    {
      "epoch": 4.183183733284484,
      "grad_norm": 0.2580087824639078,
      "learning_rate": 6.471644475243472e-06,
      "loss": 0.6572,
      "num_tokens": 8727425012.0,
      "step": 2857
    },
    {
      "epoch": 4.1846492031507605,
      "grad_norm": 0.2637983930041308,
      "learning_rate": 6.463016535284633e-06,
      "loss": 0.6717,
      "num_tokens": 8730651387.0,
      "step": 2858
    },
    {
      "epoch": 4.186114673017036,
      "grad_norm": 0.27755898259284695,
      "learning_rate": 6.454402574636577e-06,
      "loss": 0.6617,
      "num_tokens": 8733772734.0,
      "step": 2859
    },
    {
      "epoch": 4.187580142883312,
      "grad_norm": 0.25581723319365895,
      "learning_rate": 6.44580260104967e-06,
      "loss": 0.6547,
      "num_tokens": 8736713599.0,
      "step": 2860
    },
    {
      "epoch": 4.189045612749588,
      "grad_norm": 0.27881942024675727,
      "learning_rate": 6.4372166222616835e-06,
      "loss": 0.6658,
      "num_tokens": 8739663552.0,
      "step": 2861
    },
    {
      "epoch": 4.190511082615863,
      "grad_norm": 0.2496936762919609,
      "learning_rate": 6.4286446459978e-06,
      "loss": 0.6581,
      "num_tokens": 8742915201.0,
      "step": 2862
    },
    {
      "epoch": 4.1919765524821395,
      "grad_norm": 0.23853767658519257,
      "learning_rate": 6.42008667997062e-06,
      "loss": 0.659,
      "num_tokens": 8745808807.0,
      "step": 2863
    },
    {
      "epoch": 4.193442022348416,
      "grad_norm": 0.2713529171051726,
      "learning_rate": 6.411542731880104e-06,
      "loss": 0.6427,
      "num_tokens": 8748824116.0,
      "step": 2864
    },
    {
      "epoch": 4.194907492214691,
      "grad_norm": 0.23794906023817944,
      "learning_rate": 6.403012809413634e-06,
      "loss": 0.6707,
      "num_tokens": 8751978825.0,
      "step": 2865
    },
    {
      "epoch": 4.196372962080967,
      "grad_norm": 0.23760338127296238,
      "learning_rate": 6.394496920245954e-06,
      "loss": 0.67,
      "num_tokens": 8754980546.0,
      "step": 2866
    },
    {
      "epoch": 4.197838431947243,
      "grad_norm": 0.264154055797665,
      "learning_rate": 6.385995072039185e-06,
      "loss": 0.6527,
      "num_tokens": 8758237295.0,
      "step": 2867
    },
    {
      "epoch": 4.1993039018135185,
      "grad_norm": 0.25264813842148964,
      "learning_rate": 6.377507272442817e-06,
      "loss": 0.6666,
      "num_tokens": 8761494596.0,
      "step": 2868
    },
    {
      "epoch": 4.200769371679795,
      "grad_norm": 0.24218431788640982,
      "learning_rate": 6.3690335290936966e-06,
      "loss": 0.6654,
      "num_tokens": 8764659144.0,
      "step": 2869
    },
    {
      "epoch": 4.202234841546071,
      "grad_norm": 0.2531054971525206,
      "learning_rate": 6.360573849616022e-06,
      "loss": 0.6709,
      "num_tokens": 8767829885.0,
      "step": 2870
    },
    {
      "epoch": 4.203700311412347,
      "grad_norm": 0.24521970143812893,
      "learning_rate": 6.3521282416213495e-06,
      "loss": 0.6725,
      "num_tokens": 8770847273.0,
      "step": 2871
    },
    {
      "epoch": 4.205165781278622,
      "grad_norm": 0.2802457881309137,
      "learning_rate": 6.34369671270856e-06,
      "loss": 0.6565,
      "num_tokens": 8773900011.0,
      "step": 2872
    },
    {
      "epoch": 4.206631251144898,
      "grad_norm": 0.24324892766984155,
      "learning_rate": 6.3352792704638764e-06,
      "loss": 0.6477,
      "num_tokens": 8776905823.0,
      "step": 2873
    },
    {
      "epoch": 4.208096721011175,
      "grad_norm": 0.2496964377387776,
      "learning_rate": 6.32687592246084e-06,
      "loss": 0.6791,
      "num_tokens": 8780039376.0,
      "step": 2874
    },
    {
      "epoch": 4.20956219087745,
      "grad_norm": 0.2487931872826802,
      "learning_rate": 6.318486676260316e-06,
      "loss": 0.6551,
      "num_tokens": 8783231643.0,
      "step": 2875
    },
    {
      "epoch": 4.211027660743726,
      "grad_norm": 0.2721115568072086,
      "learning_rate": 6.310111539410476e-06,
      "loss": 0.664,
      "num_tokens": 8786122538.0,
      "step": 2876
    },
    {
      "epoch": 4.212493130610002,
      "grad_norm": 0.2417328134338784,
      "learning_rate": 6.301750519446814e-06,
      "loss": 0.6642,
      "num_tokens": 8789043097.0,
      "step": 2877
    },
    {
      "epoch": 4.2139586004762775,
      "grad_norm": 0.26092837863741114,
      "learning_rate": 6.293403623892093e-06,
      "loss": 0.6494,
      "num_tokens": 8791887076.0,
      "step": 2878
    },
    {
      "epoch": 4.215424070342554,
      "grad_norm": 0.2506319313665538,
      "learning_rate": 6.285070860256389e-06,
      "loss": 0.6612,
      "num_tokens": 8794950138.0,
      "step": 2879
    },
    {
      "epoch": 4.21688954020883,
      "grad_norm": 0.2458467697589475,
      "learning_rate": 6.2767522360370684e-06,
      "loss": 0.687,
      "num_tokens": 8797936881.0,
      "step": 2880
    },
    {
      "epoch": 4.218355010075105,
      "grad_norm": 0.25915370976686813,
      "learning_rate": 6.268447758718745e-06,
      "loss": 0.6591,
      "num_tokens": 8800988506.0,
      "step": 2881
    },
    {
      "epoch": 4.219820479941381,
      "grad_norm": 0.24791290494883722,
      "learning_rate": 6.260157435773341e-06,
      "loss": 0.6694,
      "num_tokens": 8804042327.0,
      "step": 2882
    },
    {
      "epoch": 4.221285949807657,
      "grad_norm": 0.25292771854886675,
      "learning_rate": 6.251881274660021e-06,
      "loss": 0.6827,
      "num_tokens": 8806999940.0,
      "step": 2883
    },
    {
      "epoch": 4.222751419673933,
      "grad_norm": 0.27677073018613024,
      "learning_rate": 6.243619282825213e-06,
      "loss": 0.6655,
      "num_tokens": 8809887968.0,
      "step": 2884
    },
    {
      "epoch": 4.224216889540209,
      "grad_norm": 0.25024258490495654,
      "learning_rate": 6.23537146770259e-06,
      "loss": 0.6723,
      "num_tokens": 8812734312.0,
      "step": 2885
    },
    {
      "epoch": 4.225682359406485,
      "grad_norm": 0.2808465310623127,
      "learning_rate": 6.2271378367130904e-06,
      "loss": 0.6591,
      "num_tokens": 8815611205.0,
      "step": 2886
    },
    {
      "epoch": 4.22714782927276,
      "grad_norm": 0.27012340916364197,
      "learning_rate": 6.2189183972648625e-06,
      "loss": 0.6689,
      "num_tokens": 8818558300.0,
      "step": 2887
    },
    {
      "epoch": 4.228613299139036,
      "grad_norm": 0.2857031130655367,
      "learning_rate": 6.210713156753304e-06,
      "loss": 0.677,
      "num_tokens": 8821721587.0,
      "step": 2888
    },
    {
      "epoch": 4.2300787690053125,
      "grad_norm": 0.2617659988500074,
      "learning_rate": 6.202522122561035e-06,
      "loss": 0.6761,
      "num_tokens": 8824729628.0,
      "step": 2889
    },
    {
      "epoch": 4.231544238871588,
      "grad_norm": 0.25616308577420405,
      "learning_rate": 6.194345302057889e-06,
      "loss": 0.6842,
      "num_tokens": 8827667180.0,
      "step": 2890
    },
    {
      "epoch": 4.233009708737864,
      "grad_norm": 0.2762435884193145,
      "learning_rate": 6.186182702600909e-06,
      "loss": 0.6622,
      "num_tokens": 8830766132.0,
      "step": 2891
    },
    {
      "epoch": 4.23447517860414,
      "grad_norm": 0.2904026436138465,
      "learning_rate": 6.178034331534354e-06,
      "loss": 0.653,
      "num_tokens": 8833841554.0,
      "step": 2892
    },
    {
      "epoch": 4.235940648470415,
      "grad_norm": 0.2582396038046331,
      "learning_rate": 6.1699001961896645e-06,
      "loss": 0.6848,
      "num_tokens": 8836907717.0,
      "step": 2893
    },
    {
      "epoch": 4.237406118336692,
      "grad_norm": 0.2587652405353099,
      "learning_rate": 6.161780303885493e-06,
      "loss": 0.6766,
      "num_tokens": 8840181205.0,
      "step": 2894
    },
    {
      "epoch": 4.238871588202968,
      "grad_norm": 0.27095115657505847,
      "learning_rate": 6.153674661927653e-06,
      "loss": 0.6636,
      "num_tokens": 8843717963.0,
      "step": 2895
    },
    {
      "epoch": 4.240337058069244,
      "grad_norm": 0.2555779841673245,
      "learning_rate": 6.1455832776091584e-06,
      "loss": 0.6735,
      "num_tokens": 8846821275.0,
      "step": 2896
    },
    {
      "epoch": 4.241802527935519,
      "grad_norm": 0.26732524935651025,
      "learning_rate": 6.137506158210182e-06,
      "loss": 0.6672,
      "num_tokens": 8849770596.0,
      "step": 2897
    },
    {
      "epoch": 4.243267997801795,
      "grad_norm": 0.26390330405484164,
      "learning_rate": 6.1294433109980664e-06,
      "loss": 0.6583,
      "num_tokens": 8852800120.0,
      "step": 2898
    },
    {
      "epoch": 4.2447334676680715,
      "grad_norm": 0.24757050148622006,
      "learning_rate": 6.121394743227304e-06,
      "loss": 0.6731,
      "num_tokens": 8855748216.0,
      "step": 2899
    },
    {
      "epoch": 4.246198937534347,
      "grad_norm": 0.2636225204941994,
      "learning_rate": 6.113360462139561e-06,
      "loss": 0.6425,
      "num_tokens": 8858782991.0,
      "step": 2900
    },
    {
      "epoch": 4.247664407400623,
      "grad_norm": 0.2586106043606745,
      "learning_rate": 6.105340474963623e-06,
      "loss": 0.6713,
      "num_tokens": 8861839646.0,
      "step": 2901
    },
    {
      "epoch": 4.249129877266899,
      "grad_norm": 0.2513157899039302,
      "learning_rate": 6.097334788915435e-06,
      "loss": 0.6469,
      "num_tokens": 8865049976.0,
      "step": 2902
    },
    {
      "epoch": 4.250595347133174,
      "grad_norm": 0.2779915978737713,
      "learning_rate": 6.089343411198061e-06,
      "loss": 0.6644,
      "num_tokens": 8868380907.0,
      "step": 2903
    },
    {
      "epoch": 4.2520608169994505,
      "grad_norm": 0.275583662234401,
      "learning_rate": 6.0813663490017e-06,
      "loss": 0.6834,
      "num_tokens": 8871348100.0,
      "step": 2904
    },
    {
      "epoch": 4.253526286865727,
      "grad_norm": 0.24855474410357584,
      "learning_rate": 6.073403609503665e-06,
      "loss": 0.6528,
      "num_tokens": 8874711875.0,
      "step": 2905
    },
    {
      "epoch": 4.254991756732002,
      "grad_norm": 0.2704408871162648,
      "learning_rate": 6.065455199868389e-06,
      "loss": 0.6655,
      "num_tokens": 8877844541.0,
      "step": 2906
    },
    {
      "epoch": 4.256457226598278,
      "grad_norm": 0.24880828697263843,
      "learning_rate": 6.057521127247399e-06,
      "loss": 0.659,
      "num_tokens": 8880770597.0,
      "step": 2907
    },
    {
      "epoch": 4.257922696464554,
      "grad_norm": 0.25273243633482995,
      "learning_rate": 6.049601398779343e-06,
      "loss": 0.66,
      "num_tokens": 8883876628.0,
      "step": 2908
    },
    {
      "epoch": 4.2593881663308295,
      "grad_norm": 0.2547349628226953,
      "learning_rate": 6.041696021589938e-06,
      "loss": 0.6593,
      "num_tokens": 8886981639.0,
      "step": 2909
    },
    {
      "epoch": 4.260853636197106,
      "grad_norm": 0.24118291803712041,
      "learning_rate": 6.03380500279201e-06,
      "loss": 0.647,
      "num_tokens": 8890165868.0,
      "step": 2910
    },
    {
      "epoch": 4.262319106063382,
      "grad_norm": 0.26183331853578995,
      "learning_rate": 6.025928349485457e-06,
      "loss": 0.6803,
      "num_tokens": 8893127415.0,
      "step": 2911
    },
    {
      "epoch": 4.263784575929657,
      "grad_norm": 0.2507686391805737,
      "learning_rate": 6.018066068757247e-06,
      "loss": 0.6772,
      "num_tokens": 8896204145.0,
      "step": 2912
    },
    {
      "epoch": 4.265250045795933,
      "grad_norm": 0.25131707587143337,
      "learning_rate": 6.0102181676814245e-06,
      "loss": 0.6629,
      "num_tokens": 8899336560.0,
      "step": 2913
    },
    {
      "epoch": 4.266715515662209,
      "grad_norm": 0.271370526408866,
      "learning_rate": 6.002384653319098e-06,
      "loss": 0.6804,
      "num_tokens": 8902180196.0,
      "step": 2914
    },
    {
      "epoch": 4.268180985528485,
      "grad_norm": 0.2529260157711417,
      "learning_rate": 5.994565532718414e-06,
      "loss": 0.6565,
      "num_tokens": 8905191814.0,
      "step": 2915
    },
    {
      "epoch": 4.269646455394761,
      "grad_norm": 0.27827917896034154,
      "learning_rate": 5.986760812914593e-06,
      "loss": 0.677,
      "num_tokens": 8908129168.0,
      "step": 2916
    },
    {
      "epoch": 4.271111925261037,
      "grad_norm": 0.256103388878135,
      "learning_rate": 5.978970500929881e-06,
      "loss": 0.6797,
      "num_tokens": 8911150269.0,
      "step": 2917
    },
    {
      "epoch": 4.272577395127312,
      "grad_norm": 0.2571496760861382,
      "learning_rate": 5.971194603773567e-06,
      "loss": 0.6589,
      "num_tokens": 8914092763.0,
      "step": 2918
    },
    {
      "epoch": 4.274042864993588,
      "grad_norm": 0.25676599110027704,
      "learning_rate": 5.963433128441971e-06,
      "loss": 0.6895,
      "num_tokens": 8916937345.0,
      "step": 2919
    },
    {
      "epoch": 4.275508334859865,
      "grad_norm": 0.2874288739296436,
      "learning_rate": 5.955686081918433e-06,
      "loss": 0.674,
      "num_tokens": 8919928325.0,
      "step": 2920
    },
    {
      "epoch": 4.27697380472614,
      "grad_norm": 0.2557608302007733,
      "learning_rate": 5.947953471173313e-06,
      "loss": 0.6718,
      "num_tokens": 8923009217.0,
      "step": 2921
    },
    {
      "epoch": 4.278439274592416,
      "grad_norm": 0.25476046662613056,
      "learning_rate": 5.940235303163986e-06,
      "loss": 0.6567,
      "num_tokens": 8925982174.0,
      "step": 2922
    },
    {
      "epoch": 4.279904744458692,
      "grad_norm": 0.2418554541072236,
      "learning_rate": 5.932531584834824e-06,
      "loss": 0.6674,
      "num_tokens": 8928994112.0,
      "step": 2923
    },
    {
      "epoch": 4.281370214324968,
      "grad_norm": 0.2636409820372933,
      "learning_rate": 5.924842323117204e-06,
      "loss": 0.6656,
      "num_tokens": 8932099767.0,
      "step": 2924
    },
    {
      "epoch": 4.282835684191244,
      "grad_norm": 0.23415339047416747,
      "learning_rate": 5.917167524929503e-06,
      "loss": 0.647,
      "num_tokens": 8935453991.0,
      "step": 2925
    },
    {
      "epoch": 4.28430115405752,
      "grad_norm": 0.24510598513910214,
      "learning_rate": 5.909507197177059e-06,
      "loss": 0.6724,
      "num_tokens": 8938770001.0,
      "step": 2926
    },
    {
      "epoch": 4.285766623923796,
      "grad_norm": 0.261555535345284,
      "learning_rate": 5.901861346752225e-06,
      "loss": 0.6615,
      "num_tokens": 8941796843.0,
      "step": 2927
    },
    {
      "epoch": 4.287232093790071,
      "grad_norm": 0.26597895667242,
      "learning_rate": 5.894229980534301e-06,
      "loss": 0.6534,
      "num_tokens": 8944849932.0,
      "step": 2928
    },
    {
      "epoch": 4.288697563656347,
      "grad_norm": 0.2628441241582817,
      "learning_rate": 5.886613105389568e-06,
      "loss": 0.6676,
      "num_tokens": 8947656855.0,
      "step": 2929
    },
    {
      "epoch": 4.2901630335226235,
      "grad_norm": 0.23753247674510014,
      "learning_rate": 5.87901072817126e-06,
      "loss": 0.6581,
      "num_tokens": 8950790029.0,
      "step": 2930
    },
    {
      "epoch": 4.291628503388899,
      "grad_norm": 0.2563824854021634,
      "learning_rate": 5.871422855719584e-06,
      "loss": 0.6488,
      "num_tokens": 8953868789.0,
      "step": 2931
    },
    {
      "epoch": 4.293093973255175,
      "grad_norm": 0.2718253643156808,
      "learning_rate": 5.863849494861668e-06,
      "loss": 0.6627,
      "num_tokens": 8957115650.0,
      "step": 2932
    },
    {
      "epoch": 4.294559443121451,
      "grad_norm": 0.26680257667825785,
      "learning_rate": 5.856290652411613e-06,
      "loss": 0.6684,
      "num_tokens": 8960117245.0,
      "step": 2933
    },
    {
      "epoch": 4.296024912987726,
      "grad_norm": 0.24551325964835713,
      "learning_rate": 5.848746335170436e-06,
      "loss": 0.659,
      "num_tokens": 8963456827.0,
      "step": 2934
    },
    {
      "epoch": 4.2974903828540025,
      "grad_norm": 0.26038924072771213,
      "learning_rate": 5.8412165499260966e-06,
      "loss": 0.6457,
      "num_tokens": 8966604911.0,
      "step": 2935
    },
    {
      "epoch": 4.298955852720279,
      "grad_norm": 0.2573026694235447,
      "learning_rate": 5.8337013034534725e-06,
      "loss": 0.6648,
      "num_tokens": 8969538753.0,
      "step": 2936
    },
    {
      "epoch": 4.300421322586554,
      "grad_norm": 0.24785788184841212,
      "learning_rate": 5.826200602514365e-06,
      "loss": 0.6645,
      "num_tokens": 8972722707.0,
      "step": 2937
    },
    {
      "epoch": 4.30188679245283,
      "grad_norm": 0.2298122002317854,
      "learning_rate": 5.818714453857482e-06,
      "loss": 0.6783,
      "num_tokens": 8975702522.0,
      "step": 2938
    },
    {
      "epoch": 4.303352262319106,
      "grad_norm": 0.24378118601884363,
      "learning_rate": 5.811242864218455e-06,
      "loss": 0.6688,
      "num_tokens": 8978819858.0,
      "step": 2939
    },
    {
      "epoch": 4.3048177321853816,
      "grad_norm": 0.25781975641764454,
      "learning_rate": 5.803785840319784e-06,
      "loss": 0.6656,
      "num_tokens": 8981930141.0,
      "step": 2940
    },
    {
      "epoch": 4.306283202051658,
      "grad_norm": 0.23521527800797498,
      "learning_rate": 5.7963433888709e-06,
      "loss": 0.6677,
      "num_tokens": 8984918108.0,
      "step": 2941
    },
    {
      "epoch": 4.307748671917934,
      "grad_norm": 0.24151862352338208,
      "learning_rate": 5.7889155165681e-06,
      "loss": 0.6952,
      "num_tokens": 8987883045.0,
      "step": 2942
    },
    {
      "epoch": 4.30921414178421,
      "grad_norm": 0.26479391523318335,
      "learning_rate": 5.7815022300945664e-06,
      "loss": 0.6368,
      "num_tokens": 8990861932.0,
      "step": 2943
    },
    {
      "epoch": 4.310679611650485,
      "grad_norm": 0.24856290082720506,
      "learning_rate": 5.774103536120359e-06,
      "loss": 0.6624,
      "num_tokens": 8993873835.0,
      "step": 2944
    },
    {
      "epoch": 4.3121450815167615,
      "grad_norm": 0.2650284973756836,
      "learning_rate": 5.7667194413024195e-06,
      "loss": 0.6623,
      "num_tokens": 8996861868.0,
      "step": 2945
    },
    {
      "epoch": 4.313610551383038,
      "grad_norm": 0.25650339318741877,
      "learning_rate": 5.759349952284532e-06,
      "loss": 0.6574,
      "num_tokens": 8999905691.0,
      "step": 2946
    },
    {
      "epoch": 4.315076021249313,
      "grad_norm": 0.24879595612203237,
      "learning_rate": 5.7519950756973616e-06,
      "loss": 0.6619,
      "num_tokens": 9002912553.0,
      "step": 2947
    },
    {
      "epoch": 4.316541491115589,
      "grad_norm": 0.2685118006379776,
      "learning_rate": 5.744654818158412e-06,
      "loss": 0.6799,
      "num_tokens": 9005807209.0,
      "step": 2948
    },
    {
      "epoch": 4.318006960981865,
      "grad_norm": 0.23683261390694316,
      "learning_rate": 5.7373291862720355e-06,
      "loss": 0.6786,
      "num_tokens": 9008950336.0,
      "step": 2949
    },
    {
      "epoch": 4.3194724308481405,
      "grad_norm": 0.2513178957167594,
      "learning_rate": 5.730018186629431e-06,
      "loss": 0.6576,
      "num_tokens": 9012097105.0,
      "step": 2950
    },
    {
      "epoch": 4.320937900714417,
      "grad_norm": 0.2648886196152374,
      "learning_rate": 5.722721825808626e-06,
      "loss": 0.656,
      "num_tokens": 9015199091.0,
      "step": 2951
    },
    {
      "epoch": 4.322403370580693,
      "grad_norm": 0.24830275113741962,
      "learning_rate": 5.715440110374474e-06,
      "loss": 0.655,
      "num_tokens": 9018232510.0,
      "step": 2952
    },
    {
      "epoch": 4.323868840446968,
      "grad_norm": 0.2577958467251042,
      "learning_rate": 5.708173046878668e-06,
      "loss": 0.6763,
      "num_tokens": 9021460544.0,
      "step": 2953
    },
    {
      "epoch": 4.325334310313244,
      "grad_norm": 0.2410976995603928,
      "learning_rate": 5.700920641859692e-06,
      "loss": 0.6641,
      "num_tokens": 9024328676.0,
      "step": 2954
    },
    {
      "epoch": 4.32679978017952,
      "grad_norm": 0.2573195185907293,
      "learning_rate": 5.693682901842867e-06,
      "loss": 0.6627,
      "num_tokens": 9027538989.0,
      "step": 2955
    },
    {
      "epoch": 4.328265250045796,
      "grad_norm": 0.24827090498728469,
      "learning_rate": 5.686459833340302e-06,
      "loss": 0.6674,
      "num_tokens": 9030531867.0,
      "step": 2956
    },
    {
      "epoch": 4.329730719912072,
      "grad_norm": 0.2540006400350655,
      "learning_rate": 5.679251442850911e-06,
      "loss": 0.6818,
      "num_tokens": 9033739154.0,
      "step": 2957
    },
    {
      "epoch": 4.331196189778348,
      "grad_norm": 0.24283018345125743,
      "learning_rate": 5.672057736860401e-06,
      "loss": 0.6749,
      "num_tokens": 9036634402.0,
      "step": 2958
    },
    {
      "epoch": 4.332661659644623,
      "grad_norm": 0.25243851792249855,
      "learning_rate": 5.664878721841276e-06,
      "loss": 0.6447,
      "num_tokens": 9039650200.0,
      "step": 2959
    },
    {
      "epoch": 4.334127129510899,
      "grad_norm": 0.25317559043121124,
      "learning_rate": 5.657714404252797e-06,
      "loss": 0.6983,
      "num_tokens": 9042587643.0,
      "step": 2960
    },
    {
      "epoch": 4.335592599377176,
      "grad_norm": 0.27123755428168456,
      "learning_rate": 5.650564790541032e-06,
      "loss": 0.6526,
      "num_tokens": 9045569332.0,
      "step": 2961
    },
    {
      "epoch": 4.337058069243451,
      "grad_norm": 0.24613108412945423,
      "learning_rate": 5.643429887138801e-06,
      "loss": 0.6811,
      "num_tokens": 9048627409.0,
      "step": 2962
    },
    {
      "epoch": 4.338523539109727,
      "grad_norm": 0.2526093710011062,
      "learning_rate": 5.636309700465684e-06,
      "loss": 0.6483,
      "num_tokens": 9051586770.0,
      "step": 2963
    },
    {
      "epoch": 4.339989008976003,
      "grad_norm": 0.27446649767461523,
      "learning_rate": 5.6292042369280355e-06,
      "loss": 0.6543,
      "num_tokens": 9054476839.0,
      "step": 2964
    },
    {
      "epoch": 4.341454478842278,
      "grad_norm": 0.2533795337780764,
      "learning_rate": 5.622113502918954e-06,
      "loss": 0.6539,
      "num_tokens": 9057624054.0,
      "step": 2965
    },
    {
      "epoch": 4.342919948708555,
      "grad_norm": 0.2558906120866623,
      "learning_rate": 5.615037504818282e-06,
      "loss": 0.6695,
      "num_tokens": 9060600046.0,
      "step": 2966
    },
    {
      "epoch": 4.344385418574831,
      "grad_norm": 0.2441813103136406,
      "learning_rate": 5.607976248992615e-06,
      "loss": 0.6573,
      "num_tokens": 9063526586.0,
      "step": 2967
    },
    {
      "epoch": 4.345850888441106,
      "grad_norm": 0.26140083132104486,
      "learning_rate": 5.600929741795271e-06,
      "loss": 0.6681,
      "num_tokens": 9066662266.0,
      "step": 2968
    },
    {
      "epoch": 4.347316358307382,
      "grad_norm": 0.23781288769666115,
      "learning_rate": 5.5938979895663035e-06,
      "loss": 0.6759,
      "num_tokens": 9069972720.0,
      "step": 2969
    },
    {
      "epoch": 4.348781828173658,
      "grad_norm": 0.2739594547858151,
      "learning_rate": 5.5868809986325e-06,
      "loss": 0.6716,
      "num_tokens": 9072920280.0,
      "step": 2970
    },
    {
      "epoch": 4.350247298039934,
      "grad_norm": 0.2663442627153427,
      "learning_rate": 5.579878775307344e-06,
      "loss": 0.6446,
      "num_tokens": 9076009077.0,
      "step": 2971
    },
    {
      "epoch": 4.35171276790621,
      "grad_norm": 0.25357554724588605,
      "learning_rate": 5.572891325891057e-06,
      "loss": 0.6681,
      "num_tokens": 9079004417.0,
      "step": 2972
    },
    {
      "epoch": 4.353178237772486,
      "grad_norm": 0.2860076905235898,
      "learning_rate": 5.565918656670548e-06,
      "loss": 0.6504,
      "num_tokens": 9082244154.0,
      "step": 2973
    },
    {
      "epoch": 4.354643707638762,
      "grad_norm": 0.244471391375885,
      "learning_rate": 5.558960773919441e-06,
      "loss": 0.6533,
      "num_tokens": 9085322485.0,
      "step": 2974
    },
    {
      "epoch": 4.356109177505037,
      "grad_norm": 0.24498590224750422,
      "learning_rate": 5.552017683898044e-06,
      "loss": 0.666,
      "num_tokens": 9088414128.0,
      "step": 2975
    },
    {
      "epoch": 4.3575746473713135,
      "grad_norm": 0.2742922929432383,
      "learning_rate": 5.545089392853375e-06,
      "loss": 0.6757,
      "num_tokens": 9091527741.0,
      "step": 2976
    },
    {
      "epoch": 4.35904011723759,
      "grad_norm": 0.25982029122973377,
      "learning_rate": 5.538175907019105e-06,
      "loss": 0.6609,
      "num_tokens": 9094643094.0,
      "step": 2977
    },
    {
      "epoch": 4.360505587103865,
      "grad_norm": 0.23976058825854804,
      "learning_rate": 5.53127723261562e-06,
      "loss": 0.6665,
      "num_tokens": 9097677935.0,
      "step": 2978
    },
    {
      "epoch": 4.361971056970141,
      "grad_norm": 0.24473297605558886,
      "learning_rate": 5.524393375849952e-06,
      "loss": 0.6625,
      "num_tokens": 9100990404.0,
      "step": 2979
    },
    {
      "epoch": 4.363436526836417,
      "grad_norm": 0.23032098211621221,
      "learning_rate": 5.517524342915818e-06,
      "loss": 0.6803,
      "num_tokens": 9104124426.0,
      "step": 2980
    },
    {
      "epoch": 4.3649019967026925,
      "grad_norm": 0.23313893483942788,
      "learning_rate": 5.510670139993584e-06,
      "loss": 0.6775,
      "num_tokens": 9107003233.0,
      "step": 2981
    },
    {
      "epoch": 4.366367466568969,
      "grad_norm": 0.24991895956648105,
      "learning_rate": 5.503830773250285e-06,
      "loss": 0.6695,
      "num_tokens": 9110038268.0,
      "step": 2982
    },
    {
      "epoch": 4.367832936435245,
      "grad_norm": 0.24693325419310477,
      "learning_rate": 5.497006248839593e-06,
      "loss": 0.6778,
      "num_tokens": 9113097541.0,
      "step": 2983
    },
    {
      "epoch": 4.36929840630152,
      "grad_norm": 0.24577445698076777,
      "learning_rate": 5.490196572901851e-06,
      "loss": 0.6685,
      "num_tokens": 9116114831.0,
      "step": 2984
    },
    {
      "epoch": 4.370763876167796,
      "grad_norm": 0.26108320791907447,
      "learning_rate": 5.4834017515640045e-06,
      "loss": 0.6839,
      "num_tokens": 9119248578.0,
      "step": 2985
    },
    {
      "epoch": 4.372229346034072,
      "grad_norm": 0.2373901653532449,
      "learning_rate": 5.476621790939669e-06,
      "loss": 0.6573,
      "num_tokens": 9122182871.0,
      "step": 2986
    },
    {
      "epoch": 4.373694815900348,
      "grad_norm": 0.25092928070548365,
      "learning_rate": 5.469856697129069e-06,
      "loss": 0.6604,
      "num_tokens": 9125281733.0,
      "step": 2987
    },
    {
      "epoch": 4.375160285766624,
      "grad_norm": 0.2615713411172448,
      "learning_rate": 5.463106476219062e-06,
      "loss": 0.6673,
      "num_tokens": 9128129068.0,
      "step": 2988
    },
    {
      "epoch": 4.3766257556329,
      "grad_norm": 0.26742743717009315,
      "learning_rate": 5.456371134283115e-06,
      "loss": 0.6619,
      "num_tokens": 9131171155.0,
      "step": 2989
    },
    {
      "epoch": 4.378091225499175,
      "grad_norm": 0.28309691375388185,
      "learning_rate": 5.449650677381321e-06,
      "loss": 0.6684,
      "num_tokens": 9134325295.0,
      "step": 2990
    },
    {
      "epoch": 4.3795566953654514,
      "grad_norm": 0.23548124503230008,
      "learning_rate": 5.442945111560363e-06,
      "loss": 0.6669,
      "num_tokens": 9137373978.0,
      "step": 2991
    },
    {
      "epoch": 4.381022165231728,
      "grad_norm": 0.2525629540082861,
      "learning_rate": 5.4362544428535395e-06,
      "loss": 0.6514,
      "num_tokens": 9140732009.0,
      "step": 2992
    },
    {
      "epoch": 4.382487635098004,
      "grad_norm": 0.24729094648386607,
      "learning_rate": 5.4295786772807416e-06,
      "loss": 0.6615,
      "num_tokens": 9143876369.0,
      "step": 2993
    },
    {
      "epoch": 4.383953104964279,
      "grad_norm": 0.25021136302442404,
      "learning_rate": 5.422917820848449e-06,
      "loss": 0.6482,
      "num_tokens": 9147034724.0,
      "step": 2994
    },
    {
      "epoch": 4.385418574830555,
      "grad_norm": 0.2729141084807848,
      "learning_rate": 5.416271879549731e-06,
      "loss": 0.6736,
      "num_tokens": 9149948555.0,
      "step": 2995
    },
    {
      "epoch": 4.386884044696831,
      "grad_norm": 0.27813257687185355,
      "learning_rate": 5.409640859364232e-06,
      "loss": 0.6833,
      "num_tokens": 9152851620.0,
      "step": 2996
    },
    {
      "epoch": 4.388349514563107,
      "grad_norm": 0.24031644443037073,
      "learning_rate": 5.403024766258169e-06,
      "loss": 0.6525,
      "num_tokens": 9155798891.0,
      "step": 2997
    },
    {
      "epoch": 4.389814984429383,
      "grad_norm": 0.2631805434843908,
      "learning_rate": 5.396423606184349e-06,
      "loss": 0.6707,
      "num_tokens": 9158861765.0,
      "step": 2998
    },
    {
      "epoch": 4.391280454295659,
      "grad_norm": 0.2930265187067786,
      "learning_rate": 5.389837385082108e-06,
      "loss": 0.657,
      "num_tokens": 9162057423.0,
      "step": 2999
    },
    {
      "epoch": 4.392745924161934,
      "grad_norm": 0.25188901190683516,
      "learning_rate": 5.383266108877379e-06,
      "loss": 0.6694,
      "num_tokens": 9165191595.0,
      "step": 3000
    },
    {
      "epoch": 4.39421139402821,
      "grad_norm": 0.2958681206092923,
      "learning_rate": 5.376709783482616e-06,
      "loss": 0.6656,
      "num_tokens": 9168313516.0,
      "step": 3001
    },
    {
      "epoch": 4.3956768638944865,
      "grad_norm": 0.2771785114395411,
      "learning_rate": 5.370168414796839e-06,
      "loss": 0.6564,
      "num_tokens": 9171131040.0,
      "step": 3002
    },
    {
      "epoch": 4.397142333760762,
      "grad_norm": 0.26956782026445514,
      "learning_rate": 5.3636420087056095e-06,
      "loss": 0.6644,
      "num_tokens": 9174159896.0,
      "step": 3003
    },
    {
      "epoch": 4.398607803627038,
      "grad_norm": 0.25843294104764697,
      "learning_rate": 5.3571305710810195e-06,
      "loss": 0.6704,
      "num_tokens": 9177117611.0,
      "step": 3004
    },
    {
      "epoch": 4.400073273493314,
      "grad_norm": 0.27172850750197225,
      "learning_rate": 5.350634107781699e-06,
      "loss": 0.6666,
      "num_tokens": 9179910355.0,
      "step": 3005
    },
    {
      "epoch": 4.401538743359589,
      "grad_norm": 0.27383478236259534,
      "learning_rate": 5.344152624652801e-06,
      "loss": 0.6803,
      "num_tokens": 9182927015.0,
      "step": 3006
    },
    {
      "epoch": 4.4030042132258655,
      "grad_norm": 0.2647415176351087,
      "learning_rate": 5.337686127526009e-06,
      "loss": 0.6595,
      "num_tokens": 9186168990.0,
      "step": 3007
    },
    {
      "epoch": 4.404469683092142,
      "grad_norm": 0.2715294900144207,
      "learning_rate": 5.3312346222195055e-06,
      "loss": 0.6633,
      "num_tokens": 9189138932.0,
      "step": 3008
    },
    {
      "epoch": 4.405935152958417,
      "grad_norm": 0.23933675334435248,
      "learning_rate": 5.324798114538006e-06,
      "loss": 0.6815,
      "num_tokens": 9192107120.0,
      "step": 3009
    },
    {
      "epoch": 4.407400622824693,
      "grad_norm": 0.2534735141952917,
      "learning_rate": 5.318376610272715e-06,
      "loss": 0.6749,
      "num_tokens": 9194938935.0,
      "step": 3010
    },
    {
      "epoch": 4.408866092690969,
      "grad_norm": 0.26419113552766715,
      "learning_rate": 5.311970115201345e-06,
      "loss": 0.666,
      "num_tokens": 9197794276.0,
      "step": 3011
    },
    {
      "epoch": 4.410331562557245,
      "grad_norm": 0.2561183335323131,
      "learning_rate": 5.305578635088101e-06,
      "loss": 0.6536,
      "num_tokens": 9200669509.0,
      "step": 3012
    },
    {
      "epoch": 4.411797032423521,
      "grad_norm": 0.26324245265136,
      "learning_rate": 5.299202175683684e-06,
      "loss": 0.6779,
      "num_tokens": 9203601947.0,
      "step": 3013
    },
    {
      "epoch": 4.413262502289797,
      "grad_norm": 0.25112355329905733,
      "learning_rate": 5.29284074272527e-06,
      "loss": 0.6559,
      "num_tokens": 9206491246.0,
      "step": 3014
    },
    {
      "epoch": 4.414727972156072,
      "grad_norm": 0.2548767866261667,
      "learning_rate": 5.286494341936532e-06,
      "loss": 0.6701,
      "num_tokens": 9209451921.0,
      "step": 3015
    },
    {
      "epoch": 4.416193442022348,
      "grad_norm": 0.24301911547249022,
      "learning_rate": 5.280162979027594e-06,
      "loss": 0.6591,
      "num_tokens": 9212603036.0,
      "step": 3016
    },
    {
      "epoch": 4.4176589118886245,
      "grad_norm": 0.2510433669552997,
      "learning_rate": 5.2738466596950746e-06,
      "loss": 0.6601,
      "num_tokens": 9215474892.0,
      "step": 3017
    },
    {
      "epoch": 4.4191243817549,
      "grad_norm": 0.24716221088257032,
      "learning_rate": 5.26754538962204e-06,
      "loss": 0.6609,
      "num_tokens": 9218450657.0,
      "step": 3018
    },
    {
      "epoch": 4.420589851621176,
      "grad_norm": 0.24420092126624518,
      "learning_rate": 5.261259174478022e-06,
      "loss": 0.6607,
      "num_tokens": 9221475868.0,
      "step": 3019
    },
    {
      "epoch": 4.422055321487452,
      "grad_norm": 0.24019242512303168,
      "learning_rate": 5.254988019919004e-06,
      "loss": 0.6747,
      "num_tokens": 9224630529.0,
      "step": 3020
    },
    {
      "epoch": 4.423520791353727,
      "grad_norm": 0.24335358092685966,
      "learning_rate": 5.248731931587428e-06,
      "loss": 0.6461,
      "num_tokens": 9227722974.0,
      "step": 3021
    },
    {
      "epoch": 4.4249862612200035,
      "grad_norm": 0.23644840592204705,
      "learning_rate": 5.242490915112163e-06,
      "loss": 0.6592,
      "num_tokens": 9230740438.0,
      "step": 3022
    },
    {
      "epoch": 4.42645173108628,
      "grad_norm": 0.24539493526892406,
      "learning_rate": 5.236264976108537e-06,
      "loss": 0.6646,
      "num_tokens": 9233723359.0,
      "step": 3023
    },
    {
      "epoch": 4.427917200952556,
      "grad_norm": 0.24802410516707743,
      "learning_rate": 5.2300541201782985e-06,
      "loss": 0.6511,
      "num_tokens": 9236712034.0,
      "step": 3024
    },
    {
      "epoch": 4.429382670818831,
      "grad_norm": 0.23896057537849677,
      "learning_rate": 5.2238583529096285e-06,
      "loss": 0.6683,
      "num_tokens": 9239765774.0,
      "step": 3025
    },
    {
      "epoch": 4.430848140685107,
      "grad_norm": 0.2590115412654243,
      "learning_rate": 5.217677679877135e-06,
      "loss": 0.6654,
      "num_tokens": 9242854259.0,
      "step": 3026
    },
    {
      "epoch": 4.432313610551383,
      "grad_norm": 0.26691164948157764,
      "learning_rate": 5.211512106641838e-06,
      "loss": 0.663,
      "num_tokens": 9245876296.0,
      "step": 3027
    },
    {
      "epoch": 4.433779080417659,
      "grad_norm": 0.24520438050268956,
      "learning_rate": 5.205361638751179e-06,
      "loss": 0.6532,
      "num_tokens": 9248905143.0,
      "step": 3028
    },
    {
      "epoch": 4.435244550283935,
      "grad_norm": 0.24319863078662501,
      "learning_rate": 5.19922628173901e-06,
      "loss": 0.6697,
      "num_tokens": 9252154869.0,
      "step": 3029
    },
    {
      "epoch": 4.436710020150211,
      "grad_norm": 0.26749816316740893,
      "learning_rate": 5.193106041125575e-06,
      "loss": 0.6832,
      "num_tokens": 9255207219.0,
      "step": 3030
    },
    {
      "epoch": 4.438175490016486,
      "grad_norm": 0.24125730426563372,
      "learning_rate": 5.187000922417531e-06,
      "loss": 0.6707,
      "num_tokens": 9258411751.0,
      "step": 3031
    },
    {
      "epoch": 4.439640959882762,
      "grad_norm": 0.2560535423299746,
      "learning_rate": 5.180910931107922e-06,
      "loss": 0.6737,
      "num_tokens": 9261258869.0,
      "step": 3032
    },
    {
      "epoch": 4.441106429749039,
      "grad_norm": 0.2568454262845663,
      "learning_rate": 5.174836072676182e-06,
      "loss": 0.6601,
      "num_tokens": 9264370602.0,
      "step": 3033
    },
    {
      "epoch": 4.442571899615314,
      "grad_norm": 0.23836527014953598,
      "learning_rate": 5.168776352588123e-06,
      "loss": 0.6547,
      "num_tokens": 9267436557.0,
      "step": 3034
    },
    {
      "epoch": 4.44403736948159,
      "grad_norm": 0.25561704449376954,
      "learning_rate": 5.162731776295958e-06,
      "loss": 0.6753,
      "num_tokens": 9270551120.0,
      "step": 3035
    },
    {
      "epoch": 4.445502839347866,
      "grad_norm": 0.25393343699997695,
      "learning_rate": 5.156702349238244e-06,
      "loss": 0.6702,
      "num_tokens": 9273512269.0,
      "step": 3036
    },
    {
      "epoch": 4.446968309214141,
      "grad_norm": 0.25010216771630867,
      "learning_rate": 5.150688076839934e-06,
      "loss": 0.6652,
      "num_tokens": 9276414489.0,
      "step": 3037
    },
    {
      "epoch": 4.448433779080418,
      "grad_norm": 0.2393702470898058,
      "learning_rate": 5.144688964512327e-06,
      "loss": 0.6598,
      "num_tokens": 9279350541.0,
      "step": 3038
    },
    {
      "epoch": 4.449899248946694,
      "grad_norm": 0.25037438553208136,
      "learning_rate": 5.1387050176530975e-06,
      "loss": 0.6665,
      "num_tokens": 9282277929.0,
      "step": 3039
    },
    {
      "epoch": 4.451364718812969,
      "grad_norm": 0.2511182281352977,
      "learning_rate": 5.132736241646259e-06,
      "loss": 0.6647,
      "num_tokens": 9285318838.0,
      "step": 3040
    },
    {
      "epoch": 4.452830188679245,
      "grad_norm": 0.23912898207650501,
      "learning_rate": 5.126782641862189e-06,
      "loss": 0.6697,
      "num_tokens": 9288501630.0,
      "step": 3041
    },
    {
      "epoch": 4.454295658545521,
      "grad_norm": 0.24645585039622456,
      "learning_rate": 5.120844223657598e-06,
      "loss": 0.6612,
      "num_tokens": 9291785281.0,
      "step": 3042
    },
    {
      "epoch": 4.4557611284117975,
      "grad_norm": 0.26553474098149116,
      "learning_rate": 5.114920992375553e-06,
      "loss": 0.668,
      "num_tokens": 9294763946.0,
      "step": 3043
    },
    {
      "epoch": 4.457226598278073,
      "grad_norm": 0.2406365541096892,
      "learning_rate": 5.109012953345438e-06,
      "loss": 0.6668,
      "num_tokens": 9297871294.0,
      "step": 3044
    },
    {
      "epoch": 4.458692068144349,
      "grad_norm": 0.25000972574977337,
      "learning_rate": 5.103120111882978e-06,
      "loss": 0.6851,
      "num_tokens": 9300854081.0,
      "step": 3045
    },
    {
      "epoch": 4.460157538010625,
      "grad_norm": 0.26001995647034953,
      "learning_rate": 5.097242473290231e-06,
      "loss": 0.6664,
      "num_tokens": 9303887882.0,
      "step": 3046
    },
    {
      "epoch": 4.4616230078769,
      "grad_norm": 0.23301341749670199,
      "learning_rate": 5.091380042855557e-06,
      "loss": 0.6524,
      "num_tokens": 9307084278.0,
      "step": 3047
    },
    {
      "epoch": 4.4630884777431765,
      "grad_norm": 0.24295331900565087,
      "learning_rate": 5.085532825853651e-06,
      "loss": 0.6602,
      "num_tokens": 9310135501.0,
      "step": 3048
    },
    {
      "epoch": 4.464553947609453,
      "grad_norm": 0.24379296827155128,
      "learning_rate": 5.079700827545512e-06,
      "loss": 0.6549,
      "num_tokens": 9313160885.0,
      "step": 3049
    },
    {
      "epoch": 4.466019417475728,
      "grad_norm": 0.22967579232457286,
      "learning_rate": 5.073884053178443e-06,
      "loss": 0.6658,
      "num_tokens": 9316091683.0,
      "step": 3050
    },
    {
      "epoch": 4.467484887342004,
      "grad_norm": 0.2567254338984546,
      "learning_rate": 5.068082507986054e-06,
      "loss": 0.667,
      "num_tokens": 9319084087.0,
      "step": 3051
    },
    {
      "epoch": 4.46895035720828,
      "grad_norm": 0.24916683409938942,
      "learning_rate": 5.06229619718826e-06,
      "loss": 0.6843,
      "num_tokens": 9321976414.0,
      "step": 3052
    },
    {
      "epoch": 4.4704158270745555,
      "grad_norm": 0.2474760451483358,
      "learning_rate": 5.056525125991247e-06,
      "loss": 0.6748,
      "num_tokens": 9324914518.0,
      "step": 3053
    },
    {
      "epoch": 4.471881296940832,
      "grad_norm": 0.2418707946050199,
      "learning_rate": 5.050769299587513e-06,
      "loss": 0.6462,
      "num_tokens": 9327972664.0,
      "step": 3054
    },
    {
      "epoch": 4.473346766807108,
      "grad_norm": 0.2501616817508447,
      "learning_rate": 5.045028723155824e-06,
      "loss": 0.6683,
      "num_tokens": 9330942140.0,
      "step": 3055
    },
    {
      "epoch": 4.474812236673383,
      "grad_norm": 0.25012670122543673,
      "learning_rate": 5.039303401861233e-06,
      "loss": 0.6312,
      "num_tokens": 9334136430.0,
      "step": 3056
    },
    {
      "epoch": 4.476277706539659,
      "grad_norm": 0.24434469311399956,
      "learning_rate": 5.033593340855064e-06,
      "loss": 0.6631,
      "num_tokens": 9337336690.0,
      "step": 3057
    },
    {
      "epoch": 4.477743176405935,
      "grad_norm": 0.2404181446960059,
      "learning_rate": 5.027898545274912e-06,
      "loss": 0.6853,
      "num_tokens": 9340374251.0,
      "step": 3058
    },
    {
      "epoch": 4.479208646272211,
      "grad_norm": 0.24067731470521994,
      "learning_rate": 5.0222190202446315e-06,
      "loss": 0.6399,
      "num_tokens": 9343573786.0,
      "step": 3059
    },
    {
      "epoch": 4.480674116138487,
      "grad_norm": 0.2358926521056054,
      "learning_rate": 5.0165547708743536e-06,
      "loss": 0.644,
      "num_tokens": 9346542321.0,
      "step": 3060
    },
    {
      "epoch": 4.482139586004763,
      "grad_norm": 0.23924674341849783,
      "learning_rate": 5.010905802260438e-06,
      "loss": 0.6444,
      "num_tokens": 9349748644.0,
      "step": 3061
    },
    {
      "epoch": 4.483605055871038,
      "grad_norm": 0.24750978685076047,
      "learning_rate": 5.005272119485527e-06,
      "loss": 0.6688,
      "num_tokens": 9352910658.0,
      "step": 3062
    },
    {
      "epoch": 4.4850705257373145,
      "grad_norm": 0.26378245773562387,
      "learning_rate": 4.999653727618482e-06,
      "loss": 0.6738,
      "num_tokens": 9356032399.0,
      "step": 3063
    },
    {
      "epoch": 4.486535995603591,
      "grad_norm": 0.2565058016236164,
      "learning_rate": 4.994050631714427e-06,
      "loss": 0.6737,
      "num_tokens": 9359047876.0,
      "step": 3064
    },
    {
      "epoch": 4.488001465469866,
      "grad_norm": 0.23923595273059367,
      "learning_rate": 4.9884628368147085e-06,
      "loss": 0.6579,
      "num_tokens": 9362175992.0,
      "step": 3065
    },
    {
      "epoch": 4.489466935336142,
      "grad_norm": 0.2475377416740997,
      "learning_rate": 4.98289034794692e-06,
      "loss": 0.6484,
      "num_tokens": 9365055642.0,
      "step": 3066
    },
    {
      "epoch": 4.490932405202418,
      "grad_norm": 0.2651839536777912,
      "learning_rate": 4.9773331701248675e-06,
      "loss": 0.6482,
      "num_tokens": 9368376213.0,
      "step": 3067
    },
    {
      "epoch": 4.4923978750686935,
      "grad_norm": 0.23336756506910022,
      "learning_rate": 4.971791308348594e-06,
      "loss": 0.6404,
      "num_tokens": 9371480182.0,
      "step": 3068
    },
    {
      "epoch": 4.49386334493497,
      "grad_norm": 0.24018715616980066,
      "learning_rate": 4.966264767604359e-06,
      "loss": 0.6532,
      "num_tokens": 9374593974.0,
      "step": 3069
    },
    {
      "epoch": 4.495328814801246,
      "grad_norm": 0.2628949187775641,
      "learning_rate": 4.960753552864632e-06,
      "loss": 0.6601,
      "num_tokens": 9377565471.0,
      "step": 3070
    },
    {
      "epoch": 4.496794284667521,
      "grad_norm": 0.2448462772266706,
      "learning_rate": 4.9552576690881e-06,
      "loss": 0.656,
      "num_tokens": 9380685284.0,
      "step": 3071
    },
    {
      "epoch": 4.498259754533797,
      "grad_norm": 0.2614733081968525,
      "learning_rate": 4.94977712121965e-06,
      "loss": 0.6616,
      "num_tokens": 9383789908.0,
      "step": 3072
    },
    {
      "epoch": 4.499725224400073,
      "grad_norm": 0.2685278770089857,
      "learning_rate": 4.94431191419037e-06,
      "loss": 0.6624,
      "num_tokens": 9386766155.0,
      "step": 3073
    },
    {
      "epoch": 4.501190694266349,
      "grad_norm": 0.26347025786882816,
      "learning_rate": 4.938862052917558e-06,
      "loss": 0.6619,
      "num_tokens": 9389813110.0,
      "step": 3074
    },
    {
      "epoch": 4.502656164132625,
      "grad_norm": 0.23478587322355565,
      "learning_rate": 4.933427542304685e-06,
      "loss": 0.6632,
      "num_tokens": 9392894911.0,
      "step": 3075
    },
    {
      "epoch": 4.504121633998901,
      "grad_norm": 0.2632934216447485,
      "learning_rate": 4.9280083872414265e-06,
      "loss": 0.6525,
      "num_tokens": 9395664696.0,
      "step": 3076
    },
    {
      "epoch": 4.505587103865177,
      "grad_norm": 0.26502632198467946,
      "learning_rate": 4.9226045926036345e-06,
      "loss": 0.6616,
      "num_tokens": 9398710816.0,
      "step": 3077
    },
    {
      "epoch": 4.507052573731452,
      "grad_norm": 0.27038893288904303,
      "learning_rate": 4.917216163253342e-06,
      "loss": 0.6604,
      "num_tokens": 9401922463.0,
      "step": 3078
    },
    {
      "epoch": 4.508518043597729,
      "grad_norm": 0.2492078967135333,
      "learning_rate": 4.911843104038753e-06,
      "loss": 0.6628,
      "num_tokens": 9405214006.0,
      "step": 3079
    },
    {
      "epoch": 4.509983513464005,
      "grad_norm": 0.2534780473072918,
      "learning_rate": 4.906485419794255e-06,
      "loss": 0.6522,
      "num_tokens": 9408511761.0,
      "step": 3080
    },
    {
      "epoch": 4.51144898333028,
      "grad_norm": 0.25325552585032607,
      "learning_rate": 4.901143115340383e-06,
      "loss": 0.6534,
      "num_tokens": 9411431706.0,
      "step": 3081
    },
    {
      "epoch": 4.512914453196556,
      "grad_norm": 0.2633102015608032,
      "learning_rate": 4.895816195483852e-06,
      "loss": 0.6496,
      "num_tokens": 9414867077.0,
      "step": 3082
    },
    {
      "epoch": 4.514379923062832,
      "grad_norm": 0.25229533764516526,
      "learning_rate": 4.8905046650175254e-06,
      "loss": 0.6524,
      "num_tokens": 9417960529.0,
      "step": 3083
    },
    {
      "epoch": 4.515845392929108,
      "grad_norm": 0.2521765058159813,
      "learning_rate": 4.885208528720422e-06,
      "loss": 0.6541,
      "num_tokens": 9420911832.0,
      "step": 3084
    },
    {
      "epoch": 4.517310862795384,
      "grad_norm": 0.25568851271747345,
      "learning_rate": 4.879927791357709e-06,
      "loss": 0.6656,
      "num_tokens": 9423841334.0,
      "step": 3085
    },
    {
      "epoch": 4.51877633266166,
      "grad_norm": 0.2536007775891062,
      "learning_rate": 4.874662457680698e-06,
      "loss": 0.6608,
      "num_tokens": 9426897131.0,
      "step": 3086
    },
    {
      "epoch": 4.520241802527935,
      "grad_norm": 0.2468046903956502,
      "learning_rate": 4.869412532426845e-06,
      "loss": 0.6558,
      "num_tokens": 9430209565.0,
      "step": 3087
    },
    {
      "epoch": 4.521707272394211,
      "grad_norm": 0.2516091020737002,
      "learning_rate": 4.864178020319738e-06,
      "loss": 0.6701,
      "num_tokens": 9433289190.0,
      "step": 3088
    },
    {
      "epoch": 4.5231727422604875,
      "grad_norm": 0.24168674288701958,
      "learning_rate": 4.8589589260691e-06,
      "loss": 0.6548,
      "num_tokens": 9436457441.0,
      "step": 3089
    },
    {
      "epoch": 4.524638212126763,
      "grad_norm": 0.24713022018059413,
      "learning_rate": 4.8537552543707765e-06,
      "loss": 0.6862,
      "num_tokens": 9439359876.0,
      "step": 3090
    },
    {
      "epoch": 4.526103681993039,
      "grad_norm": 0.24869570409656233,
      "learning_rate": 4.848567009906751e-06,
      "loss": 0.6771,
      "num_tokens": 9442317920.0,
      "step": 3091
    },
    {
      "epoch": 4.527569151859315,
      "grad_norm": 0.2652847871635881,
      "learning_rate": 4.843394197345106e-06,
      "loss": 0.6677,
      "num_tokens": 9445320252.0,
      "step": 3092
    },
    {
      "epoch": 4.529034621725591,
      "grad_norm": 0.2481197548929934,
      "learning_rate": 4.838236821340055e-06,
      "loss": 0.6602,
      "num_tokens": 9448510319.0,
      "step": 3093
    },
    {
      "epoch": 4.5305000915918665,
      "grad_norm": 0.24557960080565933,
      "learning_rate": 4.833094886531918e-06,
      "loss": 0.6672,
      "num_tokens": 9451771874.0,
      "step": 3094
    },
    {
      "epoch": 4.531965561458143,
      "grad_norm": 0.23236007777453788,
      "learning_rate": 4.827968397547117e-06,
      "loss": 0.6629,
      "num_tokens": 9454841248.0,
      "step": 3095
    },
    {
      "epoch": 4.533431031324419,
      "grad_norm": 0.22788506507911596,
      "learning_rate": 4.822857358998181e-06,
      "loss": 0.659,
      "num_tokens": 9458020466.0,
      "step": 3096
    },
    {
      "epoch": 4.534896501190694,
      "grad_norm": 0.24243085207560156,
      "learning_rate": 4.817761775483743e-06,
      "loss": 0.6582,
      "num_tokens": 9461068325.0,
      "step": 3097
    },
    {
      "epoch": 4.53636197105697,
      "grad_norm": 0.25270392872102915,
      "learning_rate": 4.812681651588517e-06,
      "loss": 0.6674,
      "num_tokens": 9464114018.0,
      "step": 3098
    },
    {
      "epoch": 4.537827440923246,
      "grad_norm": 0.24640723360501193,
      "learning_rate": 4.80761699188332e-06,
      "loss": 0.6703,
      "num_tokens": 9467002820.0,
      "step": 3099
    },
    {
      "epoch": 4.539292910789522,
      "grad_norm": 0.23476360464931922,
      "learning_rate": 4.8025678009250486e-06,
      "loss": 0.6669,
      "num_tokens": 9469844592.0,
      "step": 3100
    },
    {
      "epoch": 4.540758380655798,
      "grad_norm": 0.24447956596879658,
      "learning_rate": 4.797534083256684e-06,
      "loss": 0.6513,
      "num_tokens": 9472936593.0,
      "step": 3101
    },
    {
      "epoch": 4.542223850522074,
      "grad_norm": 0.2578389911532004,
      "learning_rate": 4.792515843407285e-06,
      "loss": 0.6748,
      "num_tokens": 9475959001.0,
      "step": 3102
    },
    {
      "epoch": 4.543689320388349,
      "grad_norm": 0.24163555739318807,
      "learning_rate": 4.787513085891982e-06,
      "loss": 0.6663,
      "num_tokens": 9479069095.0,
      "step": 3103
    },
    {
      "epoch": 4.545154790254625,
      "grad_norm": 0.2644419896641208,
      "learning_rate": 4.782525815211976e-06,
      "loss": 0.671,
      "num_tokens": 9482166921.0,
      "step": 3104
    },
    {
      "epoch": 4.546620260120902,
      "grad_norm": 0.24090272644347177,
      "learning_rate": 4.777554035854539e-06,
      "loss": 0.6583,
      "num_tokens": 9485629827.0,
      "step": 3105
    },
    {
      "epoch": 4.548085729987177,
      "grad_norm": 0.26245665493207015,
      "learning_rate": 4.772597752292996e-06,
      "loss": 0.6793,
      "num_tokens": 9488609550.0,
      "step": 3106
    },
    {
      "epoch": 4.549551199853453,
      "grad_norm": 0.2679507306789325,
      "learning_rate": 4.767656968986737e-06,
      "loss": 0.6747,
      "num_tokens": 9491645139.0,
      "step": 3107
    },
    {
      "epoch": 4.551016669719729,
      "grad_norm": 0.2461525236190638,
      "learning_rate": 4.762731690381207e-06,
      "loss": 0.6575,
      "num_tokens": 9494566343.0,
      "step": 3108
    },
    {
      "epoch": 4.5524821395860045,
      "grad_norm": 0.2595354303277881,
      "learning_rate": 4.757821920907888e-06,
      "loss": 0.6535,
      "num_tokens": 9497510935.0,
      "step": 3109
    },
    {
      "epoch": 4.553947609452281,
      "grad_norm": 0.27391688565560646,
      "learning_rate": 4.752927664984323e-06,
      "loss": 0.6564,
      "num_tokens": 9500450051.0,
      "step": 3110
    },
    {
      "epoch": 4.555413079318557,
      "grad_norm": 0.26349929539635497,
      "learning_rate": 4.748048927014089e-06,
      "loss": 0.6591,
      "num_tokens": 9503191478.0,
      "step": 3111
    },
    {
      "epoch": 4.556878549184832,
      "grad_norm": 0.24142314397658987,
      "learning_rate": 4.743185711386801e-06,
      "loss": 0.6783,
      "num_tokens": 9506258648.0,
      "step": 3112
    },
    {
      "epoch": 4.558344019051108,
      "grad_norm": 0.2950006362282494,
      "learning_rate": 4.7383380224781095e-06,
      "loss": 0.6708,
      "num_tokens": 9509195664.0,
      "step": 3113
    },
    {
      "epoch": 4.559809488917384,
      "grad_norm": 0.25056078725175035,
      "learning_rate": 4.733505864649698e-06,
      "loss": 0.6687,
      "num_tokens": 9512377555.0,
      "step": 3114
    },
    {
      "epoch": 4.56127495878366,
      "grad_norm": 0.26086783274426306,
      "learning_rate": 4.7286892422492656e-06,
      "loss": 0.6749,
      "num_tokens": 9515507625.0,
      "step": 3115
    },
    {
      "epoch": 4.562740428649936,
      "grad_norm": 0.2598962516385441,
      "learning_rate": 4.723888159610548e-06,
      "loss": 0.6688,
      "num_tokens": 9518696788.0,
      "step": 3116
    },
    {
      "epoch": 4.564205898516212,
      "grad_norm": 0.24064701676619346,
      "learning_rate": 4.7191026210532855e-06,
      "loss": 0.6458,
      "num_tokens": 9521805801.0,
      "step": 3117
    },
    {
      "epoch": 4.565671368382487,
      "grad_norm": 0.24684676347513262,
      "learning_rate": 4.714332630883239e-06,
      "loss": 0.6606,
      "num_tokens": 9524926683.0,
      "step": 3118
    },
    {
      "epoch": 4.567136838248763,
      "grad_norm": 0.24881284727513214,
      "learning_rate": 4.709578193392186e-06,
      "loss": 0.6418,
      "num_tokens": 9527959245.0,
      "step": 3119
    },
    {
      "epoch": 4.5686023081150395,
      "grad_norm": 0.26383064184416916,
      "learning_rate": 4.704839312857894e-06,
      "loss": 0.6641,
      "num_tokens": 9531076082.0,
      "step": 3120
    },
    {
      "epoch": 4.570067777981315,
      "grad_norm": 0.27875352524711977,
      "learning_rate": 4.7001159935441534e-06,
      "loss": 0.6534,
      "num_tokens": 9534184471.0,
      "step": 3121
    },
    {
      "epoch": 4.571533247847591,
      "grad_norm": 0.24557643302865986,
      "learning_rate": 4.6954082397007375e-06,
      "loss": 0.6748,
      "num_tokens": 9537157366.0,
      "step": 3122
    },
    {
      "epoch": 4.572998717713867,
      "grad_norm": 0.2286991700305166,
      "learning_rate": 4.690716055563422e-06,
      "loss": 0.6537,
      "num_tokens": 9540044608.0,
      "step": 3123
    },
    {
      "epoch": 4.574464187580142,
      "grad_norm": 0.2503774430311768,
      "learning_rate": 4.6860394453539695e-06,
      "loss": 0.6608,
      "num_tokens": 9542905355.0,
      "step": 3124
    },
    {
      "epoch": 4.5759296574464186,
      "grad_norm": 0.2325171985056932,
      "learning_rate": 4.6813784132801385e-06,
      "loss": 0.661,
      "num_tokens": 9546006814.0,
      "step": 3125
    },
    {
      "epoch": 4.577395127312695,
      "grad_norm": 0.23598743048757523,
      "learning_rate": 4.676732963535661e-06,
      "loss": 0.6727,
      "num_tokens": 9548916228.0,
      "step": 3126
    },
    {
      "epoch": 4.578860597178971,
      "grad_norm": 0.24626357597344972,
      "learning_rate": 4.672103100300254e-06,
      "loss": 0.6544,
      "num_tokens": 9551770976.0,
      "step": 3127
    },
    {
      "epoch": 4.580326067045246,
      "grad_norm": 0.24970856684041626,
      "learning_rate": 4.6674888277396165e-06,
      "loss": 0.6711,
      "num_tokens": 9555011020.0,
      "step": 3128
    },
    {
      "epoch": 4.581791536911522,
      "grad_norm": 0.24254050719442524,
      "learning_rate": 4.662890150005401e-06,
      "loss": 0.6658,
      "num_tokens": 9558195322.0,
      "step": 3129
    },
    {
      "epoch": 4.5832570067777985,
      "grad_norm": 0.24380745041135088,
      "learning_rate": 4.658307071235252e-06,
      "loss": 0.6813,
      "num_tokens": 9561278195.0,
      "step": 3130
    },
    {
      "epoch": 4.584722476644074,
      "grad_norm": 0.23652059789333404,
      "learning_rate": 4.653739595552763e-06,
      "loss": 0.6555,
      "num_tokens": 9564168330.0,
      "step": 3131
    },
    {
      "epoch": 4.58618794651035,
      "grad_norm": 0.2279606035662297,
      "learning_rate": 4.649187727067497e-06,
      "loss": 0.6692,
      "num_tokens": 9566963499.0,
      "step": 3132
    },
    {
      "epoch": 4.587653416376626,
      "grad_norm": 0.26163928787369495,
      "learning_rate": 4.644651469874969e-06,
      "loss": 0.6517,
      "num_tokens": 9570122172.0,
      "step": 3133
    },
    {
      "epoch": 4.589118886242901,
      "grad_norm": 0.24931470659508967,
      "learning_rate": 4.640130828056649e-06,
      "loss": 0.6644,
      "num_tokens": 9573113349.0,
      "step": 3134
    },
    {
      "epoch": 4.5905843561091775,
      "grad_norm": 0.22020684484307335,
      "learning_rate": 4.635625805679959e-06,
      "loss": 0.644,
      "num_tokens": 9576041319.0,
      "step": 3135
    },
    {
      "epoch": 4.592049825975454,
      "grad_norm": 0.23544641897564023,
      "learning_rate": 4.631136406798268e-06,
      "loss": 0.6524,
      "num_tokens": 9579296816.0,
      "step": 3136
    },
    {
      "epoch": 4.593515295841729,
      "grad_norm": 0.2408847685002871,
      "learning_rate": 4.6266626354508885e-06,
      "loss": 0.6819,
      "num_tokens": 9582150619.0,
      "step": 3137
    },
    {
      "epoch": 4.594980765708005,
      "grad_norm": 0.24266778695153038,
      "learning_rate": 4.622204495663065e-06,
      "loss": 0.6671,
      "num_tokens": 9585022466.0,
      "step": 3138
    },
    {
      "epoch": 4.596446235574281,
      "grad_norm": 0.2517241139222598,
      "learning_rate": 4.617761991445989e-06,
      "loss": 0.653,
      "num_tokens": 9587874297.0,
      "step": 3139
    },
    {
      "epoch": 4.5979117054405565,
      "grad_norm": 0.25976144370037896,
      "learning_rate": 4.613335126796773e-06,
      "loss": 0.6532,
      "num_tokens": 9591026873.0,
      "step": 3140
    },
    {
      "epoch": 4.599377175306833,
      "grad_norm": 0.23544791221315528,
      "learning_rate": 4.6089239056984615e-06,
      "loss": 0.6591,
      "num_tokens": 9594017898.0,
      "step": 3141
    },
    {
      "epoch": 4.600842645173109,
      "grad_norm": 0.25954948049907184,
      "learning_rate": 4.604528332120034e-06,
      "loss": 0.6619,
      "num_tokens": 9596938342.0,
      "step": 3142
    },
    {
      "epoch": 4.602308115039385,
      "grad_norm": 0.254850963966629,
      "learning_rate": 4.600148410016373e-06,
      "loss": 0.6728,
      "num_tokens": 9599989951.0,
      "step": 3143
    },
    {
      "epoch": 4.60377358490566,
      "grad_norm": 0.2413374671379886,
      "learning_rate": 4.595784143328291e-06,
      "loss": 0.6508,
      "num_tokens": 9603146795.0,
      "step": 3144
    },
    {
      "epoch": 4.605239054771936,
      "grad_norm": 0.26059191219883276,
      "learning_rate": 4.591435535982515e-06,
      "loss": 0.6567,
      "num_tokens": 9606213061.0,
      "step": 3145
    },
    {
      "epoch": 4.606704524638213,
      "grad_norm": 0.2497730198027195,
      "learning_rate": 4.587102591891675e-06,
      "loss": 0.6616,
      "num_tokens": 9609224122.0,
      "step": 3146
    },
    {
      "epoch": 4.608169994504488,
      "grad_norm": 0.2650162740434224,
      "learning_rate": 4.5827853149543125e-06,
      "loss": 0.6747,
      "num_tokens": 9612123028.0,
      "step": 3147
    },
    {
      "epoch": 4.609635464370764,
      "grad_norm": 0.26374805930768735,
      "learning_rate": 4.5784837090548746e-06,
      "loss": 0.6579,
      "num_tokens": 9615148939.0,
      "step": 3148
    },
    {
      "epoch": 4.61110093423704,
      "grad_norm": 0.23495707364358784,
      "learning_rate": 4.574197778063704e-06,
      "loss": 0.6436,
      "num_tokens": 9618503635.0,
      "step": 3149
    },
    {
      "epoch": 4.612566404103315,
      "grad_norm": 0.25226724596666,
      "learning_rate": 4.569927525837047e-06,
      "loss": 0.6582,
      "num_tokens": 9621480687.0,
      "step": 3150
    },
    {
      "epoch": 4.614031873969592,
      "grad_norm": 0.27335240446689496,
      "learning_rate": 4.56567295621703e-06,
      "loss": 0.6486,
      "num_tokens": 9624632016.0,
      "step": 3151
    },
    {
      "epoch": 4.615497343835868,
      "grad_norm": 0.23615235480397487,
      "learning_rate": 4.561434073031688e-06,
      "loss": 0.6645,
      "num_tokens": 9627687517.0,
      "step": 3152
    },
    {
      "epoch": 4.616962813702143,
      "grad_norm": 0.24285400431194604,
      "learning_rate": 4.557210880094921e-06,
      "loss": 0.6588,
      "num_tokens": 9630798861.0,
      "step": 3153
    },
    {
      "epoch": 4.618428283568419,
      "grad_norm": 0.2296383424533283,
      "learning_rate": 4.553003381206529e-06,
      "loss": 0.6588,
      "num_tokens": 9633633015.0,
      "step": 3154
    },
    {
      "epoch": 4.619893753434695,
      "grad_norm": 0.2691977765343534,
      "learning_rate": 4.548811580152182e-06,
      "loss": 0.6733,
      "num_tokens": 9636588431.0,
      "step": 3155
    },
    {
      "epoch": 4.621359223300971,
      "grad_norm": 0.23105229030853533,
      "learning_rate": 4.5446354807034325e-06,
      "loss": 0.6477,
      "num_tokens": 9639513640.0,
      "step": 3156
    },
    {
      "epoch": 4.622824693167247,
      "grad_norm": 0.2343453372552833,
      "learning_rate": 4.540475086617693e-06,
      "loss": 0.6516,
      "num_tokens": 9642651878.0,
      "step": 3157
    },
    {
      "epoch": 4.624290163033523,
      "grad_norm": 0.22182506465411253,
      "learning_rate": 4.536330401638264e-06,
      "loss": 0.6644,
      "num_tokens": 9645863142.0,
      "step": 3158
    },
    {
      "epoch": 4.625755632899798,
      "grad_norm": 0.23702082489281406,
      "learning_rate": 4.532201429494297e-06,
      "loss": 0.6603,
      "num_tokens": 9649034541.0,
      "step": 3159
    },
    {
      "epoch": 4.627221102766074,
      "grad_norm": 0.23992925925953218,
      "learning_rate": 4.528088173900811e-06,
      "loss": 0.6503,
      "num_tokens": 9652264814.0,
      "step": 3160
    },
    {
      "epoch": 4.6286865726323505,
      "grad_norm": 0.2657118247933047,
      "learning_rate": 4.523990638558684e-06,
      "loss": 0.6484,
      "num_tokens": 9655409200.0,
      "step": 3161
    },
    {
      "epoch": 4.630152042498626,
      "grad_norm": 0.23008309050235026,
      "learning_rate": 4.51990882715465e-06,
      "loss": 0.659,
      "num_tokens": 9658519407.0,
      "step": 3162
    },
    {
      "epoch": 4.631617512364902,
      "grad_norm": 0.2343287355289672,
      "learning_rate": 4.515842743361295e-06,
      "loss": 0.6671,
      "num_tokens": 9661643116.0,
      "step": 3163
    },
    {
      "epoch": 4.633082982231178,
      "grad_norm": 0.23154322628892454,
      "learning_rate": 4.511792390837054e-06,
      "loss": 0.6538,
      "num_tokens": 9664799914.0,
      "step": 3164
    },
    {
      "epoch": 4.634548452097453,
      "grad_norm": 0.2334435353828147,
      "learning_rate": 4.507757773226208e-06,
      "loss": 0.6651,
      "num_tokens": 9667615621.0,
      "step": 3165
    },
    {
      "epoch": 4.6360139219637295,
      "grad_norm": 0.27244751270346496,
      "learning_rate": 4.503738894158883e-06,
      "loss": 0.6557,
      "num_tokens": 9670738842.0,
      "step": 3166
    },
    {
      "epoch": 4.637479391830006,
      "grad_norm": 0.2628584793953414,
      "learning_rate": 4.499735757251039e-06,
      "loss": 0.6779,
      "num_tokens": 9673704428.0,
      "step": 3167
    },
    {
      "epoch": 4.638944861696281,
      "grad_norm": 0.24291804891678928,
      "learning_rate": 4.4957483661044785e-06,
      "loss": 0.6797,
      "num_tokens": 9676774032.0,
      "step": 3168
    },
    {
      "epoch": 4.640410331562557,
      "grad_norm": 0.2584464891252511,
      "learning_rate": 4.491776724306828e-06,
      "loss": 0.6715,
      "num_tokens": 9679559766.0,
      "step": 3169
    },
    {
      "epoch": 4.641875801428833,
      "grad_norm": 0.2541079840160581,
      "learning_rate": 4.4878208354315564e-06,
      "loss": 0.686,
      "num_tokens": 9682369239.0,
      "step": 3170
    },
    {
      "epoch": 4.6433412712951085,
      "grad_norm": 0.2767590535851318,
      "learning_rate": 4.4838807030379474e-06,
      "loss": 0.6422,
      "num_tokens": 9685524608.0,
      "step": 3171
    },
    {
      "epoch": 4.644806741161385,
      "grad_norm": 0.2491807577442647,
      "learning_rate": 4.479956330671111e-06,
      "loss": 0.6629,
      "num_tokens": 9688858592.0,
      "step": 3172
    },
    {
      "epoch": 4.646272211027661,
      "grad_norm": 0.24461944638180488,
      "learning_rate": 4.476047721861984e-06,
      "loss": 0.6621,
      "num_tokens": 9692024732.0,
      "step": 3173
    },
    {
      "epoch": 4.647737680893936,
      "grad_norm": 0.2553529752436928,
      "learning_rate": 4.4721548801273065e-06,
      "loss": 0.6698,
      "num_tokens": 9695101149.0,
      "step": 3174
    },
    {
      "epoch": 4.649203150760212,
      "grad_norm": 0.2426182586862576,
      "learning_rate": 4.46827780896965e-06,
      "loss": 0.6414,
      "num_tokens": 9698124728.0,
      "step": 3175
    },
    {
      "epoch": 4.6506686206264884,
      "grad_norm": 0.2492930330322641,
      "learning_rate": 4.464416511877378e-06,
      "loss": 0.6682,
      "num_tokens": 9701273187.0,
      "step": 3176
    },
    {
      "epoch": 4.652134090492765,
      "grad_norm": 0.23525775373213317,
      "learning_rate": 4.460570992324676e-06,
      "loss": 0.6572,
      "num_tokens": 9704196943.0,
      "step": 3177
    },
    {
      "epoch": 4.65359956035904,
      "grad_norm": 0.25483413796026966,
      "learning_rate": 4.456741253771526e-06,
      "loss": 0.6575,
      "num_tokens": 9707355315.0,
      "step": 3178
    },
    {
      "epoch": 4.655065030225316,
      "grad_norm": 0.22539062336536916,
      "learning_rate": 4.4529272996637115e-06,
      "loss": 0.6486,
      "num_tokens": 9710370667.0,
      "step": 3179
    },
    {
      "epoch": 4.656530500091592,
      "grad_norm": 0.24847912656574958,
      "learning_rate": 4.449129133432815e-06,
      "loss": 0.6617,
      "num_tokens": 9713465128.0,
      "step": 3180
    },
    {
      "epoch": 4.6579959699578675,
      "grad_norm": 0.24157126326010436,
      "learning_rate": 4.445346758496219e-06,
      "loss": 0.645,
      "num_tokens": 9716616870.0,
      "step": 3181
    },
    {
      "epoch": 4.659461439824144,
      "grad_norm": 0.2504792748548414,
      "learning_rate": 4.441580178257091e-06,
      "loss": 0.6547,
      "num_tokens": 9719653304.0,
      "step": 3182
    },
    {
      "epoch": 4.66092690969042,
      "grad_norm": 0.23925281359408035,
      "learning_rate": 4.43782939610439e-06,
      "loss": 0.6657,
      "num_tokens": 9722999263.0,
      "step": 3183
    },
    {
      "epoch": 4.662392379556695,
      "grad_norm": 0.25769197530750265,
      "learning_rate": 4.43409441541286e-06,
      "loss": 0.6696,
      "num_tokens": 9725959253.0,
      "step": 3184
    },
    {
      "epoch": 4.663857849422971,
      "grad_norm": 0.24250856154886063,
      "learning_rate": 4.430375239543027e-06,
      "loss": 0.6664,
      "num_tokens": 9728963406.0,
      "step": 3185
    },
    {
      "epoch": 4.665323319289247,
      "grad_norm": 0.24775972887256018,
      "learning_rate": 4.4266718718412e-06,
      "loss": 0.6824,
      "num_tokens": 9731846198.0,
      "step": 3186
    },
    {
      "epoch": 4.666788789155523,
      "grad_norm": 0.24830009204926612,
      "learning_rate": 4.422984315639464e-06,
      "loss": 0.6648,
      "num_tokens": 9734920651.0,
      "step": 3187
    },
    {
      "epoch": 4.668254259021799,
      "grad_norm": 0.2407462029748362,
      "learning_rate": 4.41931257425567e-06,
      "loss": 0.6887,
      "num_tokens": 9737901581.0,
      "step": 3188
    },
    {
      "epoch": 4.669719728888075,
      "grad_norm": 0.24666499385364385,
      "learning_rate": 4.415656650993452e-06,
      "loss": 0.6615,
      "num_tokens": 9741044442.0,
      "step": 3189
    },
    {
      "epoch": 4.67118519875435,
      "grad_norm": 0.24874117114856079,
      "learning_rate": 4.412016549142204e-06,
      "loss": 0.6537,
      "num_tokens": 9743947739.0,
      "step": 3190
    },
    {
      "epoch": 4.672650668620626,
      "grad_norm": 0.24672095360403817,
      "learning_rate": 4.4083922719770885e-06,
      "loss": 0.6708,
      "num_tokens": 9747002684.0,
      "step": 3191
    },
    {
      "epoch": 4.6741161384869025,
      "grad_norm": 0.24000872768165177,
      "learning_rate": 4.404783822759024e-06,
      "loss": 0.6602,
      "num_tokens": 9749998850.0,
      "step": 3192
    },
    {
      "epoch": 4.675581608353179,
      "grad_norm": 0.23546314955273942,
      "learning_rate": 4.401191204734695e-06,
      "loss": 0.6833,
      "num_tokens": 9753252556.0,
      "step": 3193
    },
    {
      "epoch": 4.677047078219454,
      "grad_norm": 0.262598571542368,
      "learning_rate": 4.397614421136535e-06,
      "loss": 0.6691,
      "num_tokens": 9756150874.0,
      "step": 3194
    },
    {
      "epoch": 4.67851254808573,
      "grad_norm": 0.24214230652999752,
      "learning_rate": 4.394053475182738e-06,
      "loss": 0.6582,
      "num_tokens": 9759145059.0,
      "step": 3195
    },
    {
      "epoch": 4.679978017952006,
      "grad_norm": 0.2498778674560754,
      "learning_rate": 4.390508370077242e-06,
      "loss": 0.6523,
      "num_tokens": 9762318321.0,
      "step": 3196
    },
    {
      "epoch": 4.681443487818282,
      "grad_norm": 0.23987640602371504,
      "learning_rate": 4.386979109009738e-06,
      "loss": 0.6452,
      "num_tokens": 9765366857.0,
      "step": 3197
    },
    {
      "epoch": 4.682908957684558,
      "grad_norm": 0.2490818408157706,
      "learning_rate": 4.383465695155656e-06,
      "loss": 0.6613,
      "num_tokens": 9768426145.0,
      "step": 3198
    },
    {
      "epoch": 4.684374427550834,
      "grad_norm": 0.23612215684525745,
      "learning_rate": 4.379968131676168e-06,
      "loss": 0.6623,
      "num_tokens": 9771593082.0,
      "step": 3199
    },
    {
      "epoch": 4.685839897417109,
      "grad_norm": 0.2352445108499555,
      "learning_rate": 4.376486421718188e-06,
      "loss": 0.664,
      "num_tokens": 9774735206.0,
      "step": 3200
    },
    {
      "epoch": 4.687305367283385,
      "grad_norm": 0.22456593319753507,
      "learning_rate": 4.373020568414368e-06,
      "loss": 0.647,
      "num_tokens": 9777850243.0,
      "step": 3201
    },
    {
      "epoch": 4.6887708371496615,
      "grad_norm": 0.227333265633815,
      "learning_rate": 4.369570574883083e-06,
      "loss": 0.6629,
      "num_tokens": 9780916718.0,
      "step": 3202
    },
    {
      "epoch": 4.690236307015937,
      "grad_norm": 0.23555955521248362,
      "learning_rate": 4.366136444228448e-06,
      "loss": 0.6641,
      "num_tokens": 9784002873.0,
      "step": 3203
    },
    {
      "epoch": 4.691701776882213,
      "grad_norm": 0.2513002389053168,
      "learning_rate": 4.362718179540304e-06,
      "loss": 0.6728,
      "num_tokens": 9787089548.0,
      "step": 3204
    },
    {
      "epoch": 4.693167246748489,
      "grad_norm": 0.2490840544173277,
      "learning_rate": 4.35931578389421e-06,
      "loss": 0.6634,
      "num_tokens": 9790040721.0,
      "step": 3205
    },
    {
      "epoch": 4.694632716614764,
      "grad_norm": 0.2580907252187034,
      "learning_rate": 4.355929260351456e-06,
      "loss": 0.6533,
      "num_tokens": 9793175364.0,
      "step": 3206
    },
    {
      "epoch": 4.6960981864810405,
      "grad_norm": 0.22117147293862097,
      "learning_rate": 4.352558611959046e-06,
      "loss": 0.6618,
      "num_tokens": 9796116806.0,
      "step": 3207
    },
    {
      "epoch": 4.697563656347317,
      "grad_norm": 0.2560623512179201,
      "learning_rate": 4.349203841749698e-06,
      "loss": 0.6698,
      "num_tokens": 9799122862.0,
      "step": 3208
    },
    {
      "epoch": 4.699029126213592,
      "grad_norm": 0.2438353414504536,
      "learning_rate": 4.345864952741853e-06,
      "loss": 0.6464,
      "num_tokens": 9802415982.0,
      "step": 3209
    },
    {
      "epoch": 4.700494596079868,
      "grad_norm": 0.2335379220954658,
      "learning_rate": 4.3425419479396545e-06,
      "loss": 0.6607,
      "num_tokens": 9805355812.0,
      "step": 3210
    },
    {
      "epoch": 4.701960065946144,
      "grad_norm": 0.24197016303883623,
      "learning_rate": 4.339234830332952e-06,
      "loss": 0.6748,
      "num_tokens": 9808451784.0,
      "step": 3211
    },
    {
      "epoch": 4.7034255358124195,
      "grad_norm": 0.2522544543083815,
      "learning_rate": 4.335943602897312e-06,
      "loss": 0.6624,
      "num_tokens": 9811598973.0,
      "step": 3212
    },
    {
      "epoch": 4.704891005678696,
      "grad_norm": 0.243330596236898,
      "learning_rate": 4.332668268593993e-06,
      "loss": 0.6761,
      "num_tokens": 9814707397.0,
      "step": 3213
    },
    {
      "epoch": 4.706356475544972,
      "grad_norm": 0.2282131411670486,
      "learning_rate": 4.329408830369962e-06,
      "loss": 0.6699,
      "num_tokens": 9817819783.0,
      "step": 3214
    },
    {
      "epoch": 4.707821945411247,
      "grad_norm": 0.2516872535273225,
      "learning_rate": 4.3261652911578765e-06,
      "loss": 0.6629,
      "num_tokens": 9820973680.0,
      "step": 3215
    },
    {
      "epoch": 4.709287415277523,
      "grad_norm": 0.24567433453193424,
      "learning_rate": 4.3229376538760945e-06,
      "loss": 0.657,
      "num_tokens": 9823770781.0,
      "step": 3216
    },
    {
      "epoch": 4.710752885143799,
      "grad_norm": 0.2380790398409783,
      "learning_rate": 4.31972592142866e-06,
      "loss": 0.6853,
      "num_tokens": 9826589526.0,
      "step": 3217
    },
    {
      "epoch": 4.712218355010075,
      "grad_norm": 0.22916306782888507,
      "learning_rate": 4.316530096705316e-06,
      "loss": 0.6571,
      "num_tokens": 9829657911.0,
      "step": 3218
    },
    {
      "epoch": 4.713683824876351,
      "grad_norm": 0.2539701942345983,
      "learning_rate": 4.313350182581481e-06,
      "loss": 0.6673,
      "num_tokens": 9832726758.0,
      "step": 3219
    },
    {
      "epoch": 4.715149294742627,
      "grad_norm": 0.23744850501956752,
      "learning_rate": 4.310186181918271e-06,
      "loss": 0.6615,
      "num_tokens": 9835777267.0,
      "step": 3220
    },
    {
      "epoch": 4.716614764608902,
      "grad_norm": 0.22385971952389067,
      "learning_rate": 4.307038097562472e-06,
      "loss": 0.6584,
      "num_tokens": 9838865584.0,
      "step": 3221
    },
    {
      "epoch": 4.718080234475178,
      "grad_norm": 0.253959929711729,
      "learning_rate": 4.303905932346557e-06,
      "loss": 0.6853,
      "num_tokens": 9841738476.0,
      "step": 3222
    },
    {
      "epoch": 4.719545704341455,
      "grad_norm": 0.2386376119433655,
      "learning_rate": 4.300789689088674e-06,
      "loss": 0.6739,
      "num_tokens": 9844811762.0,
      "step": 3223
    },
    {
      "epoch": 4.72101117420773,
      "grad_norm": 0.24065581363798533,
      "learning_rate": 4.297689370592644e-06,
      "loss": 0.6638,
      "num_tokens": 9847806862.0,
      "step": 3224
    },
    {
      "epoch": 4.722476644074006,
      "grad_norm": 0.26066726567547405,
      "learning_rate": 4.294604979647958e-06,
      "loss": 0.656,
      "num_tokens": 9850883640.0,
      "step": 3225
    },
    {
      "epoch": 4.723942113940282,
      "grad_norm": 0.248012163477068,
      "learning_rate": 4.2915365190297846e-06,
      "loss": 0.6345,
      "num_tokens": 9853979716.0,
      "step": 3226
    },
    {
      "epoch": 4.725407583806558,
      "grad_norm": 0.22996826068388163,
      "learning_rate": 4.28848399149895e-06,
      "loss": 0.6732,
      "num_tokens": 9856844009.0,
      "step": 3227
    },
    {
      "epoch": 4.726873053672834,
      "grad_norm": 0.26114838821045794,
      "learning_rate": 4.2854473998019456e-06,
      "loss": 0.6672,
      "num_tokens": 9859847945.0,
      "step": 3228
    },
    {
      "epoch": 4.72833852353911,
      "grad_norm": 0.25111034253763787,
      "learning_rate": 4.282426746670932e-06,
      "loss": 0.662,
      "num_tokens": 9862953698.0,
      "step": 3229
    },
    {
      "epoch": 4.729803993405386,
      "grad_norm": 0.23899361319790152,
      "learning_rate": 4.279422034823722e-06,
      "loss": 0.6707,
      "num_tokens": 9865924087.0,
      "step": 3230
    },
    {
      "epoch": 4.731269463271661,
      "grad_norm": 0.2537988783249213,
      "learning_rate": 4.276433266963787e-06,
      "loss": 0.6709,
      "num_tokens": 9869325655.0,
      "step": 3231
    },
    {
      "epoch": 4.732734933137937,
      "grad_norm": 0.25474148748692943,
      "learning_rate": 4.2734604457802565e-06,
      "loss": 0.66,
      "num_tokens": 9872281766.0,
      "step": 3232
    },
    {
      "epoch": 4.7342004030042135,
      "grad_norm": 0.2618608684616351,
      "learning_rate": 4.2705035739479054e-06,
      "loss": 0.6709,
      "num_tokens": 9875178863.0,
      "step": 3233
    },
    {
      "epoch": 4.735665872870489,
      "grad_norm": 0.24289791381796583,
      "learning_rate": 4.267562654127165e-06,
      "loss": 0.6612,
      "num_tokens": 9878259939.0,
      "step": 3234
    },
    {
      "epoch": 4.737131342736765,
      "grad_norm": 0.25050776014022175,
      "learning_rate": 4.2646376889641074e-06,
      "loss": 0.659,
      "num_tokens": 9881561522.0,
      "step": 3235
    },
    {
      "epoch": 4.738596812603041,
      "grad_norm": 0.2330725198540521,
      "learning_rate": 4.261728681090457e-06,
      "loss": 0.639,
      "num_tokens": 9884487591.0,
      "step": 3236
    },
    {
      "epoch": 4.740062282469316,
      "grad_norm": 0.2321698201883858,
      "learning_rate": 4.258835633123575e-06,
      "loss": 0.6593,
      "num_tokens": 9887475274.0,
      "step": 3237
    },
    {
      "epoch": 4.7415277523355925,
      "grad_norm": 0.258839333813259,
      "learning_rate": 4.255958547666466e-06,
      "loss": 0.6695,
      "num_tokens": 9890593188.0,
      "step": 3238
    },
    {
      "epoch": 4.742993222201869,
      "grad_norm": 0.2566708064822181,
      "learning_rate": 4.253097427307764e-06,
      "loss": 0.6574,
      "num_tokens": 9893588275.0,
      "step": 3239
    },
    {
      "epoch": 4.744458692068144,
      "grad_norm": 0.23361914107706985,
      "learning_rate": 4.250252274621756e-06,
      "loss": 0.6729,
      "num_tokens": 9896661742.0,
      "step": 3240
    },
    {
      "epoch": 4.74592416193442,
      "grad_norm": 0.21798728714874466,
      "learning_rate": 4.247423092168344e-06,
      "loss": 0.6469,
      "num_tokens": 9899981508.0,
      "step": 3241
    },
    {
      "epoch": 4.747389631800696,
      "grad_norm": 0.24417615223590142,
      "learning_rate": 4.244609882493074e-06,
      "loss": 0.6701,
      "num_tokens": 9903005838.0,
      "step": 3242
    },
    {
      "epoch": 4.7488551016669724,
      "grad_norm": 0.2550538565499943,
      "learning_rate": 4.241812648127109e-06,
      "loss": 0.7031,
      "num_tokens": 9905989448.0,
      "step": 3243
    },
    {
      "epoch": 4.750320571533248,
      "grad_norm": 0.2598865041747553,
      "learning_rate": 4.239031391587249e-06,
      "loss": 0.6628,
      "num_tokens": 9909184250.0,
      "step": 3244
    },
    {
      "epoch": 4.751786041399524,
      "grad_norm": 0.24269716607687597,
      "learning_rate": 4.236266115375911e-06,
      "loss": 0.657,
      "num_tokens": 9912216143.0,
      "step": 3245
    },
    {
      "epoch": 4.7532515112658,
      "grad_norm": 0.23367998877364038,
      "learning_rate": 4.2335168219811396e-06,
      "loss": 0.6606,
      "num_tokens": 9915129766.0,
      "step": 3246
    },
    {
      "epoch": 4.754716981132075,
      "grad_norm": 0.23238924686569898,
      "learning_rate": 4.230783513876591e-06,
      "loss": 0.6642,
      "num_tokens": 9918435461.0,
      "step": 3247
    },
    {
      "epoch": 4.7561824509983515,
      "grad_norm": 0.242077765294199,
      "learning_rate": 4.228066193521547e-06,
      "loss": 0.6616,
      "num_tokens": 9921380344.0,
      "step": 3248
    },
    {
      "epoch": 4.757647920864628,
      "grad_norm": 0.2267635806042762,
      "learning_rate": 4.2253648633609e-06,
      "loss": 0.6484,
      "num_tokens": 9924346977.0,
      "step": 3249
    },
    {
      "epoch": 4.759113390730903,
      "grad_norm": 0.23511091997274222,
      "learning_rate": 4.222679525825153e-06,
      "loss": 0.6717,
      "num_tokens": 9927343320.0,
      "step": 3250
    },
    {
      "epoch": 4.760578860597179,
      "grad_norm": 0.23299627854707206,
      "learning_rate": 4.220010183330428e-06,
      "loss": 0.6684,
      "num_tokens": 9930286988.0,
      "step": 3251
    },
    {
      "epoch": 4.762044330463455,
      "grad_norm": 0.24936479769332312,
      "learning_rate": 4.217356838278447e-06,
      "loss": 0.658,
      "num_tokens": 9933424168.0,
      "step": 3252
    },
    {
      "epoch": 4.7635098003297305,
      "grad_norm": 0.26399514450969874,
      "learning_rate": 4.214719493056541e-06,
      "loss": 0.6817,
      "num_tokens": 9936463892.0,
      "step": 3253
    },
    {
      "epoch": 4.764975270196007,
      "grad_norm": 0.2535653627566746,
      "learning_rate": 4.212098150037648e-06,
      "loss": 0.6461,
      "num_tokens": 9939448137.0,
      "step": 3254
    },
    {
      "epoch": 4.766440740062283,
      "grad_norm": 0.29373307329634873,
      "learning_rate": 4.209492811580309e-06,
      "loss": 0.6621,
      "num_tokens": 9942491238.0,
      "step": 3255
    },
    {
      "epoch": 4.767906209928558,
      "grad_norm": 0.2331822459096511,
      "learning_rate": 4.206903480028656e-06,
      "loss": 0.6648,
      "num_tokens": 9945687387.0,
      "step": 3256
    },
    {
      "epoch": 4.769371679794834,
      "grad_norm": 0.2691908336547314,
      "learning_rate": 4.20433015771243e-06,
      "loss": 0.6681,
      "num_tokens": 9948732309.0,
      "step": 3257
    },
    {
      "epoch": 4.77083714966111,
      "grad_norm": 0.26606557941007086,
      "learning_rate": 4.20177284694696e-06,
      "loss": 0.6517,
      "num_tokens": 9952204673.0,
      "step": 3258
    },
    {
      "epoch": 4.772302619527386,
      "grad_norm": 0.2522575303681543,
      "learning_rate": 4.199231550033172e-06,
      "loss": 0.6656,
      "num_tokens": 9955325545.0,
      "step": 3259
    },
    {
      "epoch": 4.773768089393662,
      "grad_norm": 0.2671076901908884,
      "learning_rate": 4.196706269257586e-06,
      "loss": 0.6842,
      "num_tokens": 9958199616.0,
      "step": 3260
    },
    {
      "epoch": 4.775233559259938,
      "grad_norm": 0.2940589621637163,
      "learning_rate": 4.194197006892304e-06,
      "loss": 0.6543,
      "num_tokens": 9961443942.0,
      "step": 3261
    },
    {
      "epoch": 4.776699029126213,
      "grad_norm": 0.24884280934453326,
      "learning_rate": 4.191703765195021e-06,
      "loss": 0.679,
      "num_tokens": 9964564645.0,
      "step": 3262
    },
    {
      "epoch": 4.778164498992489,
      "grad_norm": 0.25002426481833495,
      "learning_rate": 4.189226546409025e-06,
      "loss": 0.6662,
      "num_tokens": 9967512058.0,
      "step": 3263
    },
    {
      "epoch": 4.779629968858766,
      "grad_norm": 0.2696749269858804,
      "learning_rate": 4.186765352763166e-06,
      "loss": 0.664,
      "num_tokens": 9970667893.0,
      "step": 3264
    },
    {
      "epoch": 4.781095438725041,
      "grad_norm": 0.26987762053956615,
      "learning_rate": 4.184320186471899e-06,
      "loss": 0.6502,
      "num_tokens": 9973651724.0,
      "step": 3265
    },
    {
      "epoch": 4.782560908591317,
      "grad_norm": 0.23980771208494261,
      "learning_rate": 4.181891049735241e-06,
      "loss": 0.6474,
      "num_tokens": 9977106712.0,
      "step": 3266
    },
    {
      "epoch": 4.784026378457593,
      "grad_norm": 0.2705563622131156,
      "learning_rate": 4.1794779447388025e-06,
      "loss": 0.6399,
      "num_tokens": 9980350509.0,
      "step": 3267
    },
    {
      "epoch": 4.785491848323868,
      "grad_norm": 0.31286361667932133,
      "learning_rate": 4.177080873653751e-06,
      "loss": 0.6775,
      "num_tokens": 9983392498.0,
      "step": 3268
    },
    {
      "epoch": 4.786957318190145,
      "grad_norm": 0.24194233340772495,
      "learning_rate": 4.174699838636846e-06,
      "loss": 0.666,
      "num_tokens": 9986386239.0,
      "step": 3269
    },
    {
      "epoch": 4.788422788056421,
      "grad_norm": 0.24856860921812712,
      "learning_rate": 4.172334841830401e-06,
      "loss": 0.6568,
      "num_tokens": 9989337588.0,
      "step": 3270
    },
    {
      "epoch": 4.789888257922696,
      "grad_norm": 0.26337525176478194,
      "learning_rate": 4.169985885362317e-06,
      "loss": 0.6726,
      "num_tokens": 9992534969.0,
      "step": 3271
    },
    {
      "epoch": 4.791353727788972,
      "grad_norm": 0.269574565835146,
      "learning_rate": 4.1676529713460465e-06,
      "loss": 0.6814,
      "num_tokens": 9995569644.0,
      "step": 3272
    },
    {
      "epoch": 4.792819197655248,
      "grad_norm": 0.2330907210484345,
      "learning_rate": 4.165336101880621e-06,
      "loss": 0.6714,
      "num_tokens": 9998680944.0,
      "step": 3273
    },
    {
      "epoch": 4.794284667521524,
      "grad_norm": 0.2661359814443262,
      "learning_rate": 4.1630352790506275e-06,
      "loss": 0.6794,
      "num_tokens": 10001707598.0,
      "step": 3274
    },
    {
      "epoch": 4.7957501373878,
      "grad_norm": 0.2849957804870165,
      "learning_rate": 4.16075050492622e-06,
      "loss": 0.6632,
      "num_tokens": 10004781083.0,
      "step": 3275
    },
    {
      "epoch": 4.797215607254076,
      "grad_norm": 0.23669060225122585,
      "learning_rate": 4.158481781563107e-06,
      "loss": 0.6778,
      "num_tokens": 10007867448.0,
      "step": 3276
    },
    {
      "epoch": 4.798681077120352,
      "grad_norm": 0.23802948200338991,
      "learning_rate": 4.1562291110025635e-06,
      "loss": 0.6618,
      "num_tokens": 10010744972.0,
      "step": 3277
    },
    {
      "epoch": 4.800146546986627,
      "grad_norm": 0.25897445628236854,
      "learning_rate": 4.153992495271414e-06,
      "loss": 0.6648,
      "num_tokens": 10014041173.0,
      "step": 3278
    },
    {
      "epoch": 4.8016120168529035,
      "grad_norm": 0.25132548211075517,
      "learning_rate": 4.15177193638204e-06,
      "loss": 0.6444,
      "num_tokens": 10017096006.0,
      "step": 3279
    },
    {
      "epoch": 4.80307748671918,
      "grad_norm": 0.23136396369003998,
      "learning_rate": 4.149567436332381e-06,
      "loss": 0.6832,
      "num_tokens": 10020370573.0,
      "step": 3280
    },
    {
      "epoch": 4.804542956585455,
      "grad_norm": 0.22433115759099007,
      "learning_rate": 4.147378997105917e-06,
      "loss": 0.6537,
      "num_tokens": 10023307222.0,
      "step": 3281
    },
    {
      "epoch": 4.806008426451731,
      "grad_norm": 0.25336194850164406,
      "learning_rate": 4.14520662067169e-06,
      "loss": 0.6729,
      "num_tokens": 10026217663.0,
      "step": 3282
    },
    {
      "epoch": 4.807473896318007,
      "grad_norm": 0.25699339370988566,
      "learning_rate": 4.143050308984277e-06,
      "loss": 0.6496,
      "num_tokens": 10029394044.0,
      "step": 3283
    },
    {
      "epoch": 4.8089393661842825,
      "grad_norm": 0.23365787610299754,
      "learning_rate": 4.140910063983809e-06,
      "loss": 0.6681,
      "num_tokens": 10032436500.0,
      "step": 3284
    },
    {
      "epoch": 4.810404836050559,
      "grad_norm": 0.2319119913896515,
      "learning_rate": 4.138785887595959e-06,
      "loss": 0.657,
      "num_tokens": 10035585114.0,
      "step": 3285
    },
    {
      "epoch": 4.811870305916835,
      "grad_norm": 0.26755282059722246,
      "learning_rate": 4.136677781731945e-06,
      "loss": 0.669,
      "num_tokens": 10038692924.0,
      "step": 3286
    },
    {
      "epoch": 4.81333577578311,
      "grad_norm": 0.2546731434610133,
      "learning_rate": 4.134585748288522e-06,
      "loss": 0.6595,
      "num_tokens": 10041806551.0,
      "step": 3287
    },
    {
      "epoch": 4.814801245649386,
      "grad_norm": 0.24832616424684056,
      "learning_rate": 4.132509789147981e-06,
      "loss": 0.6632,
      "num_tokens": 10044783783.0,
      "step": 3288
    },
    {
      "epoch": 4.816266715515662,
      "grad_norm": 0.26345538123758616,
      "learning_rate": 4.13044990617816e-06,
      "loss": 0.6691,
      "num_tokens": 10047768359.0,
      "step": 3289
    },
    {
      "epoch": 4.817732185381938,
      "grad_norm": 0.24014538539821292,
      "learning_rate": 4.128406101232422e-06,
      "loss": 0.6575,
      "num_tokens": 10050773130.0,
      "step": 3290
    },
    {
      "epoch": 4.819197655248214,
      "grad_norm": 0.24555964909618153,
      "learning_rate": 4.126378376149673e-06,
      "loss": 0.6673,
      "num_tokens": 10054172675.0,
      "step": 3291
    },
    {
      "epoch": 4.82066312511449,
      "grad_norm": 0.25568723765575085,
      "learning_rate": 4.124366732754342e-06,
      "loss": 0.6485,
      "num_tokens": 10057095320.0,
      "step": 3292
    },
    {
      "epoch": 4.822128594980766,
      "grad_norm": 0.24059049295252455,
      "learning_rate": 4.122371172856397e-06,
      "loss": 0.6451,
      "num_tokens": 10060204040.0,
      "step": 3293
    },
    {
      "epoch": 4.8235940648470415,
      "grad_norm": 0.24503549214595582,
      "learning_rate": 4.12039169825133e-06,
      "loss": 0.6678,
      "num_tokens": 10063226174.0,
      "step": 3294
    },
    {
      "epoch": 4.825059534713318,
      "grad_norm": 0.2399578571878561,
      "learning_rate": 4.118428310720161e-06,
      "loss": 0.6563,
      "num_tokens": 10066457237.0,
      "step": 3295
    },
    {
      "epoch": 4.826525004579594,
      "grad_norm": 0.25626228724359756,
      "learning_rate": 4.116481012029441e-06,
      "loss": 0.675,
      "num_tokens": 10069541378.0,
      "step": 3296
    },
    {
      "epoch": 4.827990474445869,
      "grad_norm": 0.23492729610920066,
      "learning_rate": 4.1145498039312345e-06,
      "loss": 0.6803,
      "num_tokens": 10072602529.0,
      "step": 3297
    },
    {
      "epoch": 4.829455944312145,
      "grad_norm": 0.260984441026835,
      "learning_rate": 4.112634688163138e-06,
      "loss": 0.6446,
      "num_tokens": 10075734680.0,
      "step": 3298
    },
    {
      "epoch": 4.830921414178421,
      "grad_norm": 0.24818212062971204,
      "learning_rate": 4.110735666448265e-06,
      "loss": 0.6556,
      "num_tokens": 10078753094.0,
      "step": 3299
    },
    {
      "epoch": 4.832386884044697,
      "grad_norm": 0.2552422087123095,
      "learning_rate": 4.108852740495252e-06,
      "loss": 0.6802,
      "num_tokens": 10081764546.0,
      "step": 3300
    },
    {
      "epoch": 4.833852353910973,
      "grad_norm": 0.24745719766136143,
      "learning_rate": 4.106985911998247e-06,
      "loss": 0.6586,
      "num_tokens": 10084936144.0,
      "step": 3301
    },
    {
      "epoch": 4.835317823777249,
      "grad_norm": 0.24337760846373835,
      "learning_rate": 4.105135182636921e-06,
      "loss": 0.6811,
      "num_tokens": 10087830473.0,
      "step": 3302
    },
    {
      "epoch": 4.836783293643524,
      "grad_norm": 0.2512049470443849,
      "learning_rate": 4.103300554076455e-06,
      "loss": 0.6721,
      "num_tokens": 10090807432.0,
      "step": 3303
    },
    {
      "epoch": 4.8382487635098,
      "grad_norm": 0.2598909586128265,
      "learning_rate": 4.101482027967546e-06,
      "loss": 0.6694,
      "num_tokens": 10093693314.0,
      "step": 3304
    },
    {
      "epoch": 4.8397142333760765,
      "grad_norm": 0.2365767800164388,
      "learning_rate": 4.099679605946402e-06,
      "loss": 0.6466,
      "num_tokens": 10096860244.0,
      "step": 3305
    },
    {
      "epoch": 4.841179703242352,
      "grad_norm": 0.2344009449110385,
      "learning_rate": 4.097893289634743e-06,
      "loss": 0.6676,
      "num_tokens": 10099909930.0,
      "step": 3306
    },
    {
      "epoch": 4.842645173108628,
      "grad_norm": 0.2568835000319415,
      "learning_rate": 4.096123080639794e-06,
      "loss": 0.6666,
      "num_tokens": 10102919461.0,
      "step": 3307
    },
    {
      "epoch": 4.844110642974904,
      "grad_norm": 0.24003594968612912,
      "learning_rate": 4.094368980554294e-06,
      "loss": 0.6583,
      "num_tokens": 10105937192.0,
      "step": 3308
    },
    {
      "epoch": 4.845576112841179,
      "grad_norm": 0.24285154095192701,
      "learning_rate": 4.092630990956481e-06,
      "loss": 0.6734,
      "num_tokens": 10108988646.0,
      "step": 3309
    },
    {
      "epoch": 4.8470415827074556,
      "grad_norm": 0.23834506721390777,
      "learning_rate": 4.090909113410105e-06,
      "loss": 0.6787,
      "num_tokens": 10112025105.0,
      "step": 3310
    },
    {
      "epoch": 4.848507052573732,
      "grad_norm": 0.27767507346890546,
      "learning_rate": 4.089203349464411e-06,
      "loss": 0.6743,
      "num_tokens": 10115286424.0,
      "step": 3311
    },
    {
      "epoch": 4.849972522440007,
      "grad_norm": 0.232072509441997,
      "learning_rate": 4.087513700654153e-06,
      "loss": 0.6479,
      "num_tokens": 10118414853.0,
      "step": 3312
    },
    {
      "epoch": 4.851437992306283,
      "grad_norm": 0.2553948489461406,
      "learning_rate": 4.085840168499582e-06,
      "loss": 0.663,
      "num_tokens": 10121515221.0,
      "step": 3313
    },
    {
      "epoch": 4.852903462172559,
      "grad_norm": 0.24334159020693186,
      "learning_rate": 4.0841827545064475e-06,
      "loss": 0.668,
      "num_tokens": 10124236733.0,
      "step": 3314
    },
    {
      "epoch": 4.854368932038835,
      "grad_norm": 0.24204271783604037,
      "learning_rate": 4.082541460165999e-06,
      "loss": 0.6553,
      "num_tokens": 10127360795.0,
      "step": 3315
    },
    {
      "epoch": 4.855834401905111,
      "grad_norm": 0.23787273882824225,
      "learning_rate": 4.080916286954983e-06,
      "loss": 0.6431,
      "num_tokens": 10130508047.0,
      "step": 3316
    },
    {
      "epoch": 4.857299871771387,
      "grad_norm": 0.24359198731560341,
      "learning_rate": 4.079307236335639e-06,
      "loss": 0.6676,
      "num_tokens": 10133771785.0,
      "step": 3317
    },
    {
      "epoch": 4.858765341637662,
      "grad_norm": 0.24744561468587145,
      "learning_rate": 4.0777143097557025e-06,
      "loss": 0.6555,
      "num_tokens": 10136758561.0,
      "step": 3318
    },
    {
      "epoch": 4.860230811503938,
      "grad_norm": 0.2525095749707601,
      "learning_rate": 4.076137508648394e-06,
      "loss": 0.6765,
      "num_tokens": 10139934563.0,
      "step": 3319
    },
    {
      "epoch": 4.8616962813702145,
      "grad_norm": 0.22471336963876876,
      "learning_rate": 4.074576834432438e-06,
      "loss": 0.6567,
      "num_tokens": 10143138174.0,
      "step": 3320
    },
    {
      "epoch": 4.86316175123649,
      "grad_norm": 0.2593195577416882,
      "learning_rate": 4.073032288512037e-06,
      "loss": 0.6572,
      "num_tokens": 10146076023.0,
      "step": 3321
    },
    {
      "epoch": 4.864627221102766,
      "grad_norm": 0.24691720531379804,
      "learning_rate": 4.071503872276892e-06,
      "loss": 0.6727,
      "num_tokens": 10149043723.0,
      "step": 3322
    },
    {
      "epoch": 4.866092690969042,
      "grad_norm": 0.2353475597555255,
      "learning_rate": 4.069991587102184e-06,
      "loss": 0.6601,
      "num_tokens": 10152076693.0,
      "step": 3323
    },
    {
      "epoch": 4.867558160835317,
      "grad_norm": 0.27606765649864623,
      "learning_rate": 4.0684954343485806e-06,
      "loss": 0.6635,
      "num_tokens": 10155106221.0,
      "step": 3324
    },
    {
      "epoch": 4.8690236307015935,
      "grad_norm": 0.2725279679408378,
      "learning_rate": 4.06701541536224e-06,
      "loss": 0.6644,
      "num_tokens": 10158118374.0,
      "step": 3325
    },
    {
      "epoch": 4.87048910056787,
      "grad_norm": 0.259555626526974,
      "learning_rate": 4.065551531474801e-06,
      "loss": 0.662,
      "num_tokens": 10161233612.0,
      "step": 3326
    },
    {
      "epoch": 4.871954570434146,
      "grad_norm": 0.2401063679414391,
      "learning_rate": 4.064103784003381e-06,
      "loss": 0.6685,
      "num_tokens": 10164218144.0,
      "step": 3327
    },
    {
      "epoch": 4.873420040300421,
      "grad_norm": 0.24770852013881364,
      "learning_rate": 4.062672174250587e-06,
      "loss": 0.6727,
      "num_tokens": 10167201877.0,
      "step": 3328
    },
    {
      "epoch": 4.874885510166697,
      "grad_norm": 0.25899712969062416,
      "learning_rate": 4.061256703504498e-06,
      "loss": 0.6608,
      "num_tokens": 10170132568.0,
      "step": 3329
    },
    {
      "epoch": 4.876350980032973,
      "grad_norm": 0.24362490404046167,
      "learning_rate": 4.0598573730386765e-06,
      "loss": 0.6765,
      "num_tokens": 10173474458.0,
      "step": 3330
    },
    {
      "epoch": 4.877816449899249,
      "grad_norm": 0.25877921319908337,
      "learning_rate": 4.058474184112163e-06,
      "loss": 0.6741,
      "num_tokens": 10176764191.0,
      "step": 3331
    },
    {
      "epoch": 4.879281919765525,
      "grad_norm": 0.21890189252752154,
      "learning_rate": 4.057107137969473e-06,
      "loss": 0.6576,
      "num_tokens": 10179996567.0,
      "step": 3332
    },
    {
      "epoch": 4.880747389631801,
      "grad_norm": 0.23725423603431547,
      "learning_rate": 4.055756235840598e-06,
      "loss": 0.6741,
      "num_tokens": 10183090348.0,
      "step": 3333
    },
    {
      "epoch": 4.882212859498076,
      "grad_norm": 0.24763788137224907,
      "learning_rate": 4.054421478941005e-06,
      "loss": 0.6639,
      "num_tokens": 10186282764.0,
      "step": 3334
    },
    {
      "epoch": 4.883678329364352,
      "grad_norm": 0.23911064625330353,
      "learning_rate": 4.053102868471633e-06,
      "loss": 0.6587,
      "num_tokens": 10189444787.0,
      "step": 3335
    },
    {
      "epoch": 4.885143799230629,
      "grad_norm": 0.22712063537280114,
      "learning_rate": 4.051800405618894e-06,
      "loss": 0.6641,
      "num_tokens": 10192379450.0,
      "step": 3336
    },
    {
      "epoch": 4.886609269096904,
      "grad_norm": 0.2249114852172756,
      "learning_rate": 4.0505140915546744e-06,
      "loss": 0.6566,
      "num_tokens": 10195729246.0,
      "step": 3337
    },
    {
      "epoch": 4.88807473896318,
      "grad_norm": 0.2501779899369485,
      "learning_rate": 4.0492439274363235e-06,
      "loss": 0.6634,
      "num_tokens": 10198731706.0,
      "step": 3338
    },
    {
      "epoch": 4.889540208829456,
      "grad_norm": 0.2570072206332202,
      "learning_rate": 4.047989914406667e-06,
      "loss": 0.6671,
      "num_tokens": 10201638224.0,
      "step": 3339
    },
    {
      "epoch": 4.891005678695731,
      "grad_norm": 0.23983720583322143,
      "learning_rate": 4.046752053593994e-06,
      "loss": 0.6585,
      "num_tokens": 10204847586.0,
      "step": 3340
    },
    {
      "epoch": 4.892471148562008,
      "grad_norm": 0.24176996472566084,
      "learning_rate": 4.045530346112064e-06,
      "loss": 0.674,
      "num_tokens": 10207878070.0,
      "step": 3341
    },
    {
      "epoch": 4.893936618428284,
      "grad_norm": 0.2485242918178423,
      "learning_rate": 4.0443247930600995e-06,
      "loss": 0.6808,
      "num_tokens": 10210940245.0,
      "step": 3342
    },
    {
      "epoch": 4.89540208829456,
      "grad_norm": 0.22672509204028388,
      "learning_rate": 4.043135395522794e-06,
      "loss": 0.6444,
      "num_tokens": 10214085685.0,
      "step": 3343
    },
    {
      "epoch": 4.896867558160835,
      "grad_norm": 0.2540631253396701,
      "learning_rate": 4.041962154570294e-06,
      "loss": 0.6615,
      "num_tokens": 10217168715.0,
      "step": 3344
    },
    {
      "epoch": 4.898333028027111,
      "grad_norm": 0.23038009394376857,
      "learning_rate": 4.040805071258224e-06,
      "loss": 0.6561,
      "num_tokens": 10220309143.0,
      "step": 3345
    },
    {
      "epoch": 4.8997984978933875,
      "grad_norm": 0.23151992318663714,
      "learning_rate": 4.039664146627658e-06,
      "loss": 0.6484,
      "num_tokens": 10223425070.0,
      "step": 3346
    },
    {
      "epoch": 4.901263967759663,
      "grad_norm": 0.25249963455037666,
      "learning_rate": 4.038539381705138e-06,
      "loss": 0.6574,
      "num_tokens": 10226574466.0,
      "step": 3347
    },
    {
      "epoch": 4.902729437625939,
      "grad_norm": 0.24490407916529922,
      "learning_rate": 4.037430777502665e-06,
      "loss": 0.6525,
      "num_tokens": 10229521093.0,
      "step": 3348
    },
    {
      "epoch": 4.904194907492215,
      "grad_norm": 0.2312993830802392,
      "learning_rate": 4.036338335017697e-06,
      "loss": 0.6421,
      "num_tokens": 10232746150.0,
      "step": 3349
    },
    {
      "epoch": 4.90566037735849,
      "grad_norm": 0.24054832572824672,
      "learning_rate": 4.035262055233155e-06,
      "loss": 0.6808,
      "num_tokens": 10235799916.0,
      "step": 3350
    },
    {
      "epoch": 4.9071258472247665,
      "grad_norm": 0.24888798910212104,
      "learning_rate": 4.0342019391174155e-06,
      "loss": 0.6465,
      "num_tokens": 10238900773.0,
      "step": 3351
    },
    {
      "epoch": 4.908591317091043,
      "grad_norm": 0.21278156288114924,
      "learning_rate": 4.03315798762431e-06,
      "loss": 0.6621,
      "num_tokens": 10241899425.0,
      "step": 3352
    },
    {
      "epoch": 4.910056786957318,
      "grad_norm": 0.2450722550420986,
      "learning_rate": 4.032130201693129e-06,
      "loss": 0.6693,
      "num_tokens": 10245002655.0,
      "step": 3353
    },
    {
      "epoch": 4.911522256823594,
      "grad_norm": 0.23348002544815005,
      "learning_rate": 4.031118582248618e-06,
      "loss": 0.6618,
      "num_tokens": 10247980716.0,
      "step": 3354
    },
    {
      "epoch": 4.91298772668987,
      "grad_norm": 0.2620539362851663,
      "learning_rate": 4.030123130200975e-06,
      "loss": 0.6531,
      "num_tokens": 10251082567.0,
      "step": 3355
    },
    {
      "epoch": 4.9144531965561455,
      "grad_norm": 0.2397075984098191,
      "learning_rate": 4.029143846445851e-06,
      "loss": 0.6549,
      "num_tokens": 10254010801.0,
      "step": 3356
    },
    {
      "epoch": 4.915918666422422,
      "grad_norm": 0.23537613664779508,
      "learning_rate": 4.0281807318643526e-06,
      "loss": 0.6699,
      "num_tokens": 10257075458.0,
      "step": 3357
    },
    {
      "epoch": 4.917384136288698,
      "grad_norm": 0.2491329251044043,
      "learning_rate": 4.027233787323035e-06,
      "loss": 0.6715,
      "num_tokens": 10260128275.0,
      "step": 3358
    },
    {
      "epoch": 4.918849606154973,
      "grad_norm": 0.26525799562504165,
      "learning_rate": 4.026303013673908e-06,
      "loss": 0.6688,
      "num_tokens": 10263189137.0,
      "step": 3359
    },
    {
      "epoch": 4.920315076021249,
      "grad_norm": 0.27383866474262725,
      "learning_rate": 4.025388411754426e-06,
      "loss": 0.63,
      "num_tokens": 10266417442.0,
      "step": 3360
    },
    {
      "epoch": 4.9217805458875254,
      "grad_norm": 0.2794440005469038,
      "learning_rate": 4.024489982387502e-06,
      "loss": 0.6729,
      "num_tokens": 10269519743.0,
      "step": 3361
    },
    {
      "epoch": 4.923246015753801,
      "grad_norm": 0.2543682162679272,
      "learning_rate": 4.023607726381488e-06,
      "loss": 0.6871,
      "num_tokens": 10272295629.0,
      "step": 3362
    },
    {
      "epoch": 4.924711485620077,
      "grad_norm": 0.2638752217718156,
      "learning_rate": 4.022741644530191e-06,
      "loss": 0.6775,
      "num_tokens": 10275458029.0,
      "step": 3363
    },
    {
      "epoch": 4.926176955486353,
      "grad_norm": 0.24821375206446913,
      "learning_rate": 4.0218917376128635e-06,
      "loss": 0.6709,
      "num_tokens": 10278653826.0,
      "step": 3364
    },
    {
      "epoch": 4.927642425352628,
      "grad_norm": 0.24395056265839496,
      "learning_rate": 4.021058006394204e-06,
      "loss": 0.6818,
      "num_tokens": 10281817798.0,
      "step": 3365
    },
    {
      "epoch": 4.9291078952189045,
      "grad_norm": 0.22583025127807876,
      "learning_rate": 4.020240451624357e-06,
      "loss": 0.6501,
      "num_tokens": 10284930492.0,
      "step": 3366
    },
    {
      "epoch": 4.930573365085181,
      "grad_norm": 0.24937615285526646,
      "learning_rate": 4.019439074038912e-06,
      "loss": 0.6548,
      "num_tokens": 10288084622.0,
      "step": 3367
    },
    {
      "epoch": 4.932038834951456,
      "grad_norm": 0.24806917285337138,
      "learning_rate": 4.018653874358904e-06,
      "loss": 0.6703,
      "num_tokens": 10291226036.0,
      "step": 3368
    },
    {
      "epoch": 4.933504304817732,
      "grad_norm": 0.240888926967346,
      "learning_rate": 4.017884853290814e-06,
      "loss": 0.6559,
      "num_tokens": 10294172034.0,
      "step": 3369
    },
    {
      "epoch": 4.934969774684008,
      "grad_norm": 0.24705496568282145,
      "learning_rate": 4.01713201152656e-06,
      "loss": 0.6766,
      "num_tokens": 10297261754.0,
      "step": 3370
    },
    {
      "epoch": 4.9364352445502835,
      "grad_norm": 0.26204171156197903,
      "learning_rate": 4.016395349743512e-06,
      "loss": 0.6918,
      "num_tokens": 10300268809.0,
      "step": 3371
    },
    {
      "epoch": 4.93790071441656,
      "grad_norm": 0.25276420533569455,
      "learning_rate": 4.015674868604473e-06,
      "loss": 0.654,
      "num_tokens": 10303377808.0,
      "step": 3372
    },
    {
      "epoch": 4.939366184282836,
      "grad_norm": 0.24236960717055947,
      "learning_rate": 4.014970568757696e-06,
      "loss": 0.6726,
      "num_tokens": 10306272071.0,
      "step": 3373
    },
    {
      "epoch": 4.940831654149111,
      "grad_norm": 0.23758048217360095,
      "learning_rate": 4.0142824508368655e-06,
      "loss": 0.6681,
      "num_tokens": 10309415447.0,
      "step": 3374
    },
    {
      "epoch": 4.942297124015387,
      "grad_norm": 0.2554655586449771,
      "learning_rate": 4.0136105154611155e-06,
      "loss": 0.6728,
      "num_tokens": 10312503781.0,
      "step": 3375
    },
    {
      "epoch": 4.943762593881663,
      "grad_norm": 0.23146119287102876,
      "learning_rate": 4.012954763235016e-06,
      "loss": 0.6628,
      "num_tokens": 10315571007.0,
      "step": 3376
    },
    {
      "epoch": 4.9452280637479396,
      "grad_norm": 0.232604878560038,
      "learning_rate": 4.0123151947485725e-06,
      "loss": 0.6634,
      "num_tokens": 10318855057.0,
      "step": 3377
    },
    {
      "epoch": 4.946693533614215,
      "grad_norm": 0.22980342602434667,
      "learning_rate": 4.0116918105772375e-06,
      "loss": 0.6524,
      "num_tokens": 10322011598.0,
      "step": 3378
    },
    {
      "epoch": 4.948159003480491,
      "grad_norm": 0.22577381012648853,
      "learning_rate": 4.011084611281893e-06,
      "loss": 0.6556,
      "num_tokens": 10324931296.0,
      "step": 3379
    },
    {
      "epoch": 4.949624473346767,
      "grad_norm": 0.2199843360816917,
      "learning_rate": 4.010493597408868e-06,
      "loss": 0.6521,
      "num_tokens": 10328508880.0,
      "step": 3380
    },
    {
      "epoch": 4.951089943213042,
      "grad_norm": 0.2322028332239894,
      "learning_rate": 4.00991876948992e-06,
      "loss": 0.6451,
      "num_tokens": 10331620458.0,
      "step": 3381
    },
    {
      "epoch": 4.952555413079319,
      "grad_norm": 0.22451829435508014,
      "learning_rate": 4.009360128042249e-06,
      "loss": 0.6406,
      "num_tokens": 10334780085.0,
      "step": 3382
    },
    {
      "epoch": 4.954020882945595,
      "grad_norm": 0.22413339782695352,
      "learning_rate": 4.008817673568487e-06,
      "loss": 0.6777,
      "num_tokens": 10337749677.0,
      "step": 3383
    },
    {
      "epoch": 4.95548635281187,
      "grad_norm": 0.22848152385550569,
      "learning_rate": 4.008291406556708e-06,
      "loss": 0.6796,
      "num_tokens": 10340769683.0,
      "step": 3384
    },
    {
      "epoch": 4.956951822678146,
      "grad_norm": 0.22849980282508697,
      "learning_rate": 4.007781327480415e-06,
      "loss": 0.651,
      "num_tokens": 10343805466.0,
      "step": 3385
    },
    {
      "epoch": 4.958417292544422,
      "grad_norm": 0.2482882936373451,
      "learning_rate": 4.007287436798549e-06,
      "loss": 0.6551,
      "num_tokens": 10346722074.0,
      "step": 3386
    },
    {
      "epoch": 4.959882762410698,
      "grad_norm": 0.24365064933390798,
      "learning_rate": 4.0068097349554865e-06,
      "loss": 0.6872,
      "num_tokens": 10349658690.0,
      "step": 3387
    },
    {
      "epoch": 4.961348232276974,
      "grad_norm": 0.24685306430568574,
      "learning_rate": 4.006348222381034e-06,
      "loss": 0.6685,
      "num_tokens": 10352641667.0,
      "step": 3388
    },
    {
      "epoch": 4.96281370214325,
      "grad_norm": 0.24142202133098234,
      "learning_rate": 4.005902899490439e-06,
      "loss": 0.6687,
      "num_tokens": 10355791818.0,
      "step": 3389
    },
    {
      "epoch": 4.964279172009525,
      "grad_norm": 0.2492758251665183,
      "learning_rate": 4.005473766684375e-06,
      "loss": 0.6599,
      "num_tokens": 10358746371.0,
      "step": 3390
    },
    {
      "epoch": 4.965744641875801,
      "grad_norm": 0.25134199628739456,
      "learning_rate": 4.005060824348953e-06,
      "loss": 0.6724,
      "num_tokens": 10361658311.0,
      "step": 3391
    },
    {
      "epoch": 4.9672101117420775,
      "grad_norm": 0.22529792143437824,
      "learning_rate": 4.004664072855716e-06,
      "loss": 0.6582,
      "num_tokens": 10364628427.0,
      "step": 3392
    },
    {
      "epoch": 4.968675581608354,
      "grad_norm": 0.2531806118825763,
      "learning_rate": 4.0042835125616365e-06,
      "loss": 0.6645,
      "num_tokens": 10367891435.0,
      "step": 3393
    },
    {
      "epoch": 4.970141051474629,
      "grad_norm": 0.22867585661997294,
      "learning_rate": 4.003919143809124e-06,
      "loss": 0.6637,
      "num_tokens": 10371280949.0,
      "step": 3394
    },
    {
      "epoch": 4.971606521340905,
      "grad_norm": 0.23898093698479422,
      "learning_rate": 4.003570966926017e-06,
      "loss": 0.6502,
      "num_tokens": 10374431732.0,
      "step": 3395
    },
    {
      "epoch": 4.973071991207181,
      "grad_norm": 0.237246129076489,
      "learning_rate": 4.003238982225583e-06,
      "loss": 0.6543,
      "num_tokens": 10377365233.0,
      "step": 3396
    },
    {
      "epoch": 4.9745374610734565,
      "grad_norm": 0.24735010189655873,
      "learning_rate": 4.002923190006527e-06,
      "loss": 0.6514,
      "num_tokens": 10380460181.0,
      "step": 3397
    },
    {
      "epoch": 4.976002930939733,
      "grad_norm": 0.232742465942289,
      "learning_rate": 4.0026235905529785e-06,
      "loss": 0.6677,
      "num_tokens": 10383414225.0,
      "step": 3398
    },
    {
      "epoch": 4.977468400806009,
      "grad_norm": 0.231619986983961,
      "learning_rate": 4.002340184134502e-06,
      "loss": 0.6734,
      "num_tokens": 10386359973.0,
      "step": 3399
    },
    {
      "epoch": 4.978933870672284,
      "grad_norm": 0.2400915172151342,
      "learning_rate": 4.002072971006089e-06,
      "loss": 0.6763,
      "num_tokens": 10389357384.0,
      "step": 3400
    },
    {
      "epoch": 4.98039934053856,
      "grad_norm": 0.22642706479975402,
      "learning_rate": 4.001821951408163e-06,
      "loss": 0.6522,
      "num_tokens": 10392538884.0,
      "step": 3401
    },
    {
      "epoch": 4.981864810404836,
      "grad_norm": 0.25109936291397833,
      "learning_rate": 4.001587125566579e-06,
      "loss": 0.6767,
      "num_tokens": 10395221171.0,
      "step": 3402
    },
    {
      "epoch": 4.983330280271112,
      "grad_norm": 0.2621244881922184,
      "learning_rate": 4.001368493692621e-06,
      "loss": 0.6607,
      "num_tokens": 10398234249.0,
      "step": 3403
    },
    {
      "epoch": 4.984795750137388,
      "grad_norm": 0.2508753499169267,
      "learning_rate": 4.001166055983001e-06,
      "loss": 0.6481,
      "num_tokens": 10401430928.0,
      "step": 3404
    },
    {
      "epoch": 4.986261220003664,
      "grad_norm": 0.2426986135473989,
      "learning_rate": 4.000979812619857e-06,
      "loss": 0.6618,
      "num_tokens": 10404514616.0,
      "step": 3405
    },
    {
      "epoch": 4.987726689869939,
      "grad_norm": 0.23795954378718198,
      "learning_rate": 4.000809763770768e-06,
      "loss": 0.6779,
      "num_tokens": 10407453510.0,
      "step": 3406
    },
    {
      "epoch": 4.989192159736215,
      "grad_norm": 0.27657572023584226,
      "learning_rate": 4.000655909588727e-06,
      "loss": 0.6838,
      "num_tokens": 10410272540.0,
      "step": 3407
    },
    {
      "epoch": 4.990657629602492,
      "grad_norm": 0.2619521261201692,
      "learning_rate": 4.000518250212169e-06,
      "loss": 0.6728,
      "num_tokens": 10413277091.0,
      "step": 3408
    },
    {
      "epoch": 4.992123099468767,
      "grad_norm": 0.26956884101097617,
      "learning_rate": 4.000396785764949e-06,
      "loss": 0.679,
      "num_tokens": 10416149006.0,
      "step": 3409
    },
    {
      "epoch": 4.993588569335043,
      "grad_norm": 0.23539351351832585,
      "learning_rate": 4.000291516356355e-06,
      "loss": 0.6781,
      "num_tokens": 10419121779.0,
      "step": 3410
    },
    {
      "epoch": 4.995054039201319,
      "grad_norm": 0.23202792635745148,
      "learning_rate": 4.000202442081103e-06,
      "loss": 0.642,
      "num_tokens": 10422250208.0,
      "step": 3411
    },
    {
      "epoch": 4.9965195090675945,
      "grad_norm": 0.24622217854337639,
      "learning_rate": 4.000129563019336e-06,
      "loss": 0.6739,
      "num_tokens": 10425008750.0,
      "step": 3412
    },
    {
      "epoch": 4.997984978933871,
      "grad_norm": 0.26039551967189023,
      "learning_rate": 4.000072879236628e-06,
      "loss": 0.6734,
      "num_tokens": 10428097263.0,
      "step": 3413
    },
    {
      "epoch": 4.999450448800147,
      "grad_norm": 0.2442136306876934,
      "learning_rate": 4.000032390783978e-06,
      "loss": 0.657,
      "num_tokens": 10431157791.0,
      "step": 3414
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.2442136306876934,
      "learning_rate": 4.000008097697817e-06,
      "loss": 0.6751,
      "num_tokens": 10431803027.0,
      "step": 3415
    },
    {
      "epoch": 5.0,
      "step": 3415,
      "total_flos": 1.1159501200687104e+16,
      "train_loss": 0.7174374641820592,
      "train_runtime": 243574.8878,
      "train_samples_per_second": 7.171,
      "train_steps_per_second": 0.014
    }
  ],
  "logging_steps": 1,
  "max_steps": 3415,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1159501200687104e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}