{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 1345,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03727865796831314,
      "grad_norm": 142.35989379882812,
      "learning_rate": 6.666666666666667e-07,
      "loss": 2.7091,
      "step": 10
    },
    {
      "epoch": 0.07455731593662628,
      "grad_norm": 72.79329681396484,
      "learning_rate": 1.4074074074074075e-06,
      "loss": 2.1574,
      "step": 20
    },
    {
      "epoch": 0.11183597390493942,
      "grad_norm": 38.433494567871094,
      "learning_rate": 2.148148148148148e-06,
      "loss": 1.1004,
      "step": 30
    },
    {
      "epoch": 0.14911463187325255,
      "grad_norm": 16.615192413330078,
      "learning_rate": 2.888888888888889e-06,
      "loss": 0.665,
      "step": 40
    },
    {
      "epoch": 0.1863932898415657,
      "grad_norm": 23.41881561279297,
      "learning_rate": 3.6296296296296302e-06,
      "loss": 0.5175,
      "step": 50
    },
    {
      "epoch": 0.22367194780987884,
      "grad_norm": 27.603975296020508,
      "learning_rate": 4.370370370370371e-06,
      "loss": 0.4866,
      "step": 60
    },
    {
      "epoch": 0.26095060577819196,
      "grad_norm": 28.732698440551758,
      "learning_rate": 5.1111111111111115e-06,
      "loss": 0.5481,
      "step": 70
    },
    {
      "epoch": 0.2982292637465051,
      "grad_norm": 15.001837730407715,
      "learning_rate": 5.8518518518518515e-06,
      "loss": 0.4032,
      "step": 80
    },
    {
      "epoch": 0.33550792171481825,
      "grad_norm": 21.928102493286133,
      "learning_rate": 6.592592592592592e-06,
      "loss": 0.4906,
      "step": 90
    },
    {
      "epoch": 0.3727865796831314,
      "grad_norm": 18.61629295349121,
      "learning_rate": 7.333333333333333e-06,
      "loss": 0.4936,
      "step": 100
    },
    {
      "epoch": 0.41006523765144454,
      "grad_norm": 21.106975555419922,
      "learning_rate": 8.074074074074075e-06,
      "loss": 0.3966,
      "step": 110
    },
    {
      "epoch": 0.4473438956197577,
      "grad_norm": 16.85749626159668,
      "learning_rate": 8.814814814814817e-06,
      "loss": 0.4041,
      "step": 120
    },
    {
      "epoch": 0.4846225535880708,
      "grad_norm": 17.09644317626953,
      "learning_rate": 9.555555555555556e-06,
      "loss": 0.4176,
      "step": 130
    },
    {
      "epoch": 0.5219012115563839,
      "grad_norm": 15.114425659179688,
      "learning_rate": 9.999730359519366e-06,
      "loss": 0.4001,
      "step": 140
    },
    {
      "epoch": 0.5591798695246971,
      "grad_norm": 17.893978118896484,
      "learning_rate": 9.99669723809545e-06,
      "loss": 0.4301,
      "step": 150
    },
    {
      "epoch": 0.5964585274930102,
      "grad_norm": 18.640336990356445,
      "learning_rate": 9.99029599599737e-06,
      "loss": 0.4511,
      "step": 160
    },
    {
      "epoch": 0.6337371854613234,
      "grad_norm": 14.082732200622559,
      "learning_rate": 9.980530948106503e-06,
      "loss": 0.5251,
      "step": 170
    },
    {
      "epoch": 0.6710158434296365,
      "grad_norm": 13.453099250793457,
      "learning_rate": 9.96740867674275e-06,
      "loss": 0.4482,
      "step": 180
    },
    {
      "epoch": 0.7082945013979497,
      "grad_norm": 13.484246253967285,
      "learning_rate": 9.950938027227609e-06,
      "loss": 0.4946,
      "step": 190
    },
    {
      "epoch": 0.7455731593662628,
      "grad_norm": 17.218990325927734,
      "learning_rate": 9.931130101921795e-06,
      "loss": 0.4883,
      "step": 200
    },
    {
      "epoch": 0.782851817334576,
      "grad_norm": 14.79829216003418,
      "learning_rate": 9.9079982527415e-06,
      "loss": 0.3728,
      "step": 210
    },
    {
      "epoch": 0.8201304753028891,
      "grad_norm": 15.066132545471191,
      "learning_rate": 9.881558072158253e-06,
      "loss": 0.3713,
      "step": 220
    },
    {
      "epoch": 0.8574091332712023,
      "grad_norm": 14.823089599609375,
      "learning_rate": 9.851827382688535e-06,
      "loss": 0.534,
      "step": 230
    },
    {
      "epoch": 0.8946877912395154,
      "grad_norm": 12.547226905822754,
      "learning_rate": 9.818826224880149e-06,
      "loss": 0.3804,
      "step": 240
    },
    {
      "epoch": 0.9319664492078286,
      "grad_norm": 20.883176803588867,
      "learning_rate": 9.782576843803503e-06,
      "loss": 0.4312,
      "step": 250
    },
    {
      "epoch": 0.9692451071761417,
      "grad_norm": 11.794534683227539,
      "learning_rate": 9.743103674056901e-06,
      "loss": 0.4405,
      "step": 260
    },
    {
      "epoch": 1.0037278657968314,
      "grad_norm": 16.345706939697266,
      "learning_rate": 9.700433323295908e-06,
      "loss": 0.4058,
      "step": 270
    },
    {
      "epoch": 1.0410065237651445,
      "grad_norm": 9.839822769165039,
      "learning_rate": 9.65459455429796e-06,
      "loss": 0.2184,
      "step": 280
    },
    {
      "epoch": 1.0782851817334576,
      "grad_norm": 10.497920989990234,
      "learning_rate": 9.60561826557425e-06,
      "loss": 0.1515,
      "step": 290
    },
    {
      "epoch": 1.1155638397017706,
      "grad_norm": 11.623759269714355,
      "learning_rate": 9.553537470541993e-06,
      "loss": 0.2554,
      "step": 300
    },
    {
      "epoch": 1.152842497670084,
      "grad_norm": 6.473010540008545,
      "learning_rate": 9.498387275271074e-06,
      "loss": 0.1673,
      "step": 310
    },
    {
      "epoch": 1.190121155638397,
      "grad_norm": 16.041898727416992,
      "learning_rate": 9.44020485482015e-06,
      "loss": 0.1857,
      "step": 320
    },
    {
      "epoch": 1.2273998136067101,
      "grad_norm": 17.941225051879883,
      "learning_rate": 9.37902942817805e-06,
      "loss": 0.234,
      "step": 330
    },
    {
      "epoch": 1.2646784715750232,
      "grad_norm": 18.714780807495117,
      "learning_rate": 9.314902231827478e-06,
      "loss": 0.177,
      "step": 340
    },
    {
      "epoch": 1.3019571295433363,
      "grad_norm": 9.163804054260254,
      "learning_rate": 9.247866491948752e-06,
      "loss": 0.2347,
      "step": 350
    },
    {
      "epoch": 1.3392357875116496,
      "grad_norm": 10.53104019165039,
      "learning_rate": 9.177967395282359e-06,
      "loss": 0.1857,
      "step": 360
    },
    {
      "epoch": 1.3765144454799627,
      "grad_norm": 13.568281173706055,
      "learning_rate": 9.105252058669958e-06,
      "loss": 0.2465,
      "step": 370
    },
    {
      "epoch": 1.4137931034482758,
      "grad_norm": 15.50161361694336,
      "learning_rate": 9.029769497294358e-06,
      "loss": 0.2228,
      "step": 380
    },
    {
      "epoch": 1.4510717614165891,
      "grad_norm": 13.434879302978516,
      "learning_rate": 8.951570591639888e-06,
      "loss": 0.201,
      "step": 390
    },
    {
      "epoch": 1.4883504193849022,
      "grad_norm": 8.03503704071045,
      "learning_rate": 8.870708053195414e-06,
      "loss": 0.1881,
      "step": 400
    },
    {
      "epoch": 1.5256290773532153,
      "grad_norm": 6.359331130981445,
      "learning_rate": 8.787236388923138e-06,
      "loss": 0.2014,
      "step": 410
    },
    {
      "epoch": 1.5629077353215284,
      "grad_norm": 10.878474235534668,
      "learning_rate": 8.701211864517126e-06,
      "loss": 0.2524,
      "step": 420
    },
    {
      "epoch": 1.6001863932898415,
      "grad_norm": 12.528131484985352,
      "learning_rate": 8.612692466476327e-06,
      "loss": 0.2235,
      "step": 430
    },
    {
      "epoch": 1.6374650512581548,
      "grad_norm": 8.81213092803955,
      "learning_rate": 8.521737863017644e-06,
      "loss": 0.2091,
      "step": 440
    },
    {
      "epoch": 1.6747437092264679,
      "grad_norm": 12.80145263671875,
      "learning_rate": 8.428409363855423e-06,
      "loss": 0.2504,
      "step": 450
    },
    {
      "epoch": 1.712022367194781,
      "grad_norm": 10.278597831726074,
      "learning_rate": 8.332769878874433e-06,
      "loss": 0.2653,
      "step": 460
    },
    {
      "epoch": 1.7493010251630943,
      "grad_norm": 6.535762786865234,
      "learning_rate": 8.234883875724269e-06,
      "loss": 0.1698,
      "step": 470
    },
    {
      "epoch": 1.7865796831314071,
      "grad_norm": 9.170923233032227,
      "learning_rate": 8.134817336363648e-06,
      "loss": 0.2188,
      "step": 480
    },
    {
      "epoch": 1.8238583410997204,
      "grad_norm": 17.866756439208984,
      "learning_rate": 8.032637712584017e-06,
      "loss": 0.2237,
      "step": 490
    },
    {
      "epoch": 1.8611369990680335,
      "grad_norm": 11.461530685424805,
      "learning_rate": 7.928413880542349e-06,
      "loss": 0.2007,
      "step": 500
    },
    {
      "epoch": 1.8984156570363466,
      "grad_norm": 11.155534744262695,
      "learning_rate": 7.822216094333847e-06,
      "loss": 0.1605,
      "step": 510
    },
    {
      "epoch": 1.93569431500466,
      "grad_norm": 14.84400463104248,
      "learning_rate": 7.714115938635792e-06,
      "loss": 0.2295,
      "step": 520
    },
    {
      "epoch": 1.972972972972973,
      "grad_norm": 16.546606063842773,
      "learning_rate": 7.6041862804545155e-06,
      "loss": 0.2054,
      "step": 530
    },
    {
      "epoch": 2.0074557315936628,
      "grad_norm": 9.678475379943848,
      "learning_rate": 7.4925012200079795e-06,
      "loss": 0.158,
      "step": 540
    },
    {
      "epoch": 2.0447343895619756,
      "grad_norm": 3.3748250007629395,
      "learning_rate": 7.379136040777084e-06,
      "loss": 0.0512,
      "step": 550
    },
    {
      "epoch": 2.082013047530289,
      "grad_norm": 13.04529857635498,
      "learning_rate": 7.264167158759375e-06,
      "loss": 0.072,
      "step": 560
    },
    {
      "epoch": 2.1192917054986022,
      "grad_norm": 2.89333176612854,
      "learning_rate": 7.147672070959367e-06,
      "loss": 0.0567,
      "step": 570
    },
    {
      "epoch": 2.156570363466915,
      "grad_norm": 21.26679039001465,
      "learning_rate": 7.029729303150179e-06,
      "loss": 0.0862,
      "step": 580
    },
    {
      "epoch": 2.1938490214352284,
      "grad_norm": 1.28345787525177,
      "learning_rate": 6.910418356941712e-06,
      "loss": 0.0639,
      "step": 590
    },
    {
      "epoch": 2.2311276794035413,
      "grad_norm": 9.119125366210938,
      "learning_rate": 6.7898196561910524e-06,
      "loss": 0.0794,
      "step": 600
    },
    {
      "epoch": 2.2684063373718546,
      "grad_norm": 227.37673950195312,
      "learning_rate": 6.66801449279121e-06,
      "loss": 0.2074,
      "step": 610
    },
    {
      "epoch": 2.305684995340168,
      "grad_norm": 12.226691246032715,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.0744,
      "step": 620
    },
    {
      "epoch": 2.3429636533084808,
      "grad_norm": 4.885424613952637,
      "learning_rate": 6.42111395646918e-06,
      "loss": 0.1148,
      "step": 630
    },
    {
      "epoch": 2.380242311276794,
      "grad_norm": 6.862027645111084,
      "learning_rate": 6.296185011641634e-06,
      "loss": 0.0983,
      "step": 640
    },
    {
      "epoch": 2.417520969245107,
      "grad_norm": 3.4426121711730957,
      "learning_rate": 6.17038234817011e-06,
      "loss": 0.0577,
      "step": 650
    },
    {
      "epoch": 2.4547996272134203,
      "grad_norm": 3.5107228755950928,
      "learning_rate": 6.043790765779604e-06,
      "loss": 0.1013,
      "step": 660
    },
    {
      "epoch": 2.4920782851817336,
      "grad_norm": 2.4559249877929688,
      "learning_rate": 5.916495595981218e-06,
      "loss": 0.0717,
      "step": 670
    },
    {
      "epoch": 2.5293569431500464,
      "grad_norm": 8.287261009216309,
      "learning_rate": 5.788582644552782e-06,
      "loss": 0.1242,
      "step": 680
    },
    {
      "epoch": 2.5666356011183598,
      "grad_norm": 5.529751777648926,
      "learning_rate": 5.660138133699825e-06,
      "loss": 0.0685,
      "step": 690
    },
    {
      "epoch": 2.6039142590866726,
      "grad_norm": 1.219261884689331,
      "learning_rate": 5.531248643935803e-06,
      "loss": 0.0405,
      "step": 700
    },
    {
      "epoch": 2.641192917054986,
      "grad_norm": 11.259958267211914,
      "learning_rate": 5.402001055720845e-06,
      "loss": 0.0563,
      "step": 710
    },
    {
      "epoch": 2.6784715750232992,
      "grad_norm": 11.031204223632812,
      "learning_rate": 5.2724824908982815e-06,
      "loss": 0.054,
      "step": 720
    },
    {
      "epoch": 2.7157502329916126,
      "grad_norm": 0.6370890736579895,
      "learning_rate": 5.142780253968481e-06,
      "loss": 0.0663,
      "step": 730
    },
    {
      "epoch": 2.7530288909599254,
      "grad_norm": 13.135482788085938,
      "learning_rate": 5.01298177323956e-06,
      "loss": 0.1242,
      "step": 740
    },
    {
      "epoch": 2.7903075489282387,
      "grad_norm": 5.429863452911377,
      "learning_rate": 4.883174541894633e-06,
      "loss": 0.0546,
      "step": 750
    },
    {
      "epoch": 2.8275862068965516,
      "grad_norm": 10.01339340209961,
      "learning_rate": 4.753446059015334e-06,
      "loss": 0.0625,
      "step": 760
    },
    {
      "epoch": 2.864864864864865,
      "grad_norm": 13.857867240905762,
      "learning_rate": 4.623883770601369e-06,
      "loss": 0.0972,
      "step": 770
    },
    {
      "epoch": 2.9021435228331782,
      "grad_norm": 8.516348838806152,
      "learning_rate": 4.494575010625845e-06,
      "loss": 0.0501,
      "step": 780
    },
    {
      "epoch": 2.939422180801491,
      "grad_norm": 8.758301734924316,
      "learning_rate": 4.3656069421661025e-06,
      "loss": 0.1,
      "step": 790
    },
    {
      "epoch": 2.9767008387698044,
      "grad_norm": 5.620622634887695,
      "learning_rate": 4.2370664986497605e-06,
      "loss": 0.0458,
      "step": 800
    },
    {
      "epoch": 3.011183597390494,
      "grad_norm": 1.509472370147705,
      "learning_rate": 4.109040325255554e-06,
      "loss": 0.0617,
      "step": 810
    },
    {
      "epoch": 3.048462255358807,
      "grad_norm": 6.817937850952148,
      "learning_rate": 3.981614720508469e-06,
      "loss": 0.0125,
      "step": 820
    },
    {
      "epoch": 3.08574091332712,
      "grad_norm": 7.452628135681152,
      "learning_rate": 3.854875578108544e-06,
      "loss": 0.0185,
      "step": 830
    },
    {
      "epoch": 3.1230195712954334,
      "grad_norm": 3.0916385650634766,
      "learning_rate": 3.7289083290325668e-06,
      "loss": 0.0384,
      "step": 840
    },
    {
      "epoch": 3.1602982292637467,
      "grad_norm": 2.345118761062622,
      "learning_rate": 3.6037978839476513e-06,
      "loss": 0.0063,
      "step": 850
    },
    {
      "epoch": 3.1975768872320596,
      "grad_norm": 3.9550981521606445,
      "learning_rate": 3.4796285759755767e-06,
      "loss": 0.0132,
      "step": 860
    },
    {
      "epoch": 3.234855545200373,
      "grad_norm": 8.894234657287598,
      "learning_rate": 3.356484103846389e-06,
      "loss": 0.0232,
      "step": 870
    },
    {
      "epoch": 3.2721342031686858,
      "grad_norm": 1.8498859405517578,
      "learning_rate": 3.234447475479668e-06,
      "loss": 0.036,
      "step": 880
    },
    {
      "epoch": 3.309412861136999,
      "grad_norm": 0.3681803047657013,
      "learning_rate": 3.113600952031415e-06,
      "loss": 0.006,
      "step": 890
    },
    {
      "epoch": 3.3466915191053124,
      "grad_norm": 5.196049690246582,
      "learning_rate": 2.994025992444334e-06,
      "loss": 0.0207,
      "step": 900
    },
    {
      "epoch": 3.3839701770736252,
      "grad_norm": 23.478836059570312,
      "learning_rate": 2.8758031985388515e-06,
      "loss": 0.0527,
      "step": 910
    },
    {
      "epoch": 3.4212488350419386,
      "grad_norm": 8.055231094360352,
      "learning_rate": 2.7590122606818892e-06,
      "loss": 0.0244,
      "step": 920
    },
    {
      "epoch": 3.4585274930102514,
      "grad_norm": 6.720674991607666,
      "learning_rate": 2.6437319040700347e-06,
      "loss": 0.0218,
      "step": 930
    },
    {
      "epoch": 3.4958061509785647,
      "grad_norm": 10.457154273986816,
      "learning_rate": 2.5300398356632883e-06,
      "loss": 0.0196,
      "step": 940
    },
    {
      "epoch": 3.533084808946878,
      "grad_norm": 6.857097625732422,
      "learning_rate": 2.418012691805191e-06,
      "loss": 0.0195,
      "step": 950
    },
    {
      "epoch": 3.570363466915191,
      "grad_norm": 11.561606407165527,
      "learning_rate": 2.307725986564598e-06,
      "loss": 0.0298,
      "step": 960
    },
    {
      "epoch": 3.607642124883504,
      "grad_norm": 9.121880531311035,
      "learning_rate": 2.1992540608339624e-06,
      "loss": 0.0274,
      "step": 970
    },
    {
      "epoch": 3.644920782851817,
      "grad_norm": 8.414935111999512,
      "learning_rate": 2.0926700322184197e-06,
      "loss": 0.0277,
      "step": 980
    },
    {
      "epoch": 3.6821994408201304,
      "grad_norm": 16.52871322631836,
      "learning_rate": 1.9880457457494358e-06,
      "loss": 0.0285,
      "step": 990
    },
    {
      "epoch": 3.7194780987884437,
      "grad_norm": 6.088344097137451,
      "learning_rate": 1.8854517254562853e-06,
      "loss": 0.0091,
      "step": 1000
    },
    {
      "epoch": 3.756756756756757,
      "grad_norm": 2.94254469871521,
      "learning_rate": 1.784957126827948e-06,
      "loss": 0.017,
      "step": 1010
    },
    {
      "epoch": 3.79403541472507,
      "grad_norm": 8.226703643798828,
      "learning_rate": 1.6866296901975177e-06,
      "loss": 0.0254,
      "step": 1020
    },
    {
      "epoch": 3.831314072693383,
      "grad_norm": 0.32786825299263,
      "learning_rate": 1.5905356950805184e-06,
      "loss": 0.0094,
      "step": 1030
    },
    {
      "epoch": 3.868592730661696,
      "grad_norm": 0.34963831305503845,
      "learning_rate": 1.4967399154979074e-06,
      "loss": 0.0117,
      "step": 1040
    },
    {
      "epoch": 3.9058713886300094,
      "grad_norm": 3.5587544441223145,
      "learning_rate": 1.4053055763138884e-06,
      "loss": 0.0121,
      "step": 1050
    },
    {
      "epoch": 3.9431500465983227,
      "grad_norm": 1.3276591300964355,
      "learning_rate": 1.3162943106179748e-06,
      "loss": 0.0181,
      "step": 1060
    },
    {
      "epoch": 3.9804287045666356,
      "grad_norm": 1.2272201776504517,
      "learning_rate": 1.2297661181800036e-06,
      "loss": 0.0342,
      "step": 1070
    },
    {
      "epoch": 4.0149114631873255,
      "grad_norm": 0.08773171901702881,
      "learning_rate": 1.145779325006125e-06,
      "loss": 0.0027,
      "step": 1080
    },
    {
      "epoch": 4.052190121155639,
      "grad_norm": 0.04279715195298195,
      "learning_rate": 1.0643905440230318e-06,
      "loss": 0.0023,
      "step": 1090
    },
    {
      "epoch": 4.089468779123951,
      "grad_norm": 0.6202743053436279,
      "learning_rate": 9.85654636916914e-07,
      "loss": 0.0036,
      "step": 1100
    },
    {
      "epoch": 4.1267474370922645,
      "grad_norm": 0.3144533038139343,
      "learning_rate": 9.096246771528838e-07,
      "loss": 0.0062,
      "step": 1110
    },
    {
      "epoch": 4.164026095060578,
      "grad_norm": 0.170346200466156,
      "learning_rate": 8.363519141997645e-07,
      "loss": 0.0073,
      "step": 1120
    },
    {
      "epoch": 4.201304753028891,
      "grad_norm": 0.10405213385820389,
      "learning_rate": 7.658857389844038e-07,
      "loss": 0.0014,
      "step": 1130
    },
    {
      "epoch": 4.2385834109972045,
      "grad_norm": 1.366269588470459,
      "learning_rate": 6.98273650598752e-07,
      "loss": 0.0064,
      "step": 1140
    },
    {
      "epoch": 4.275862068965517,
      "grad_norm": 0.09792446345090866,
      "learning_rate": 6.33561224282187e-07,
      "loss": 0.0016,
      "step": 1150
    },
    {
      "epoch": 4.31314072693383,
      "grad_norm": 7.333868503570557,
      "learning_rate": 5.717920807006333e-07,
      "loss": 0.0034,
      "step": 1160
    },
    {
      "epoch": 4.3504193849021435,
      "grad_norm": 0.0547269769012928,
      "learning_rate": 5.130078565432089e-07,
      "loss": 0.0009,
      "step": 1170
    },
    {
      "epoch": 4.387698042870457,
      "grad_norm": 0.13646848499774933,
      "learning_rate": 4.5724817645620813e-07,
      "loss": 0.003,
      "step": 1180
    },
    {
      "epoch": 4.42497670083877,
      "grad_norm": 0.04732973873615265,
      "learning_rate": 4.0455062633333986e-07,
      "loss": 0.0215,
      "step": 1190
    },
    {
      "epoch": 4.462255358807083,
      "grad_norm": 3.078120231628418,
      "learning_rate": 3.549507279802278e-07,
      "loss": 0.0021,
      "step": 1200
    },
    {
      "epoch": 4.499534016775396,
      "grad_norm": 5.298852920532227,
      "learning_rate": 3.084819151702456e-07,
      "loss": 0.0023,
      "step": 1210
    },
    {
      "epoch": 4.536812674743709,
      "grad_norm": 0.1440030038356781,
      "learning_rate": 2.651755111078358e-07,
      "loss": 0.0119,
      "step": 1220
    },
    {
      "epoch": 4.5740913327120225,
      "grad_norm": 0.2678074836730957,
      "learning_rate": 2.250607073144978e-07,
      "loss": 0.001,
      "step": 1230
    },
    {
      "epoch": 4.611369990680336,
      "grad_norm": 0.10277917236089706,
      "learning_rate": 1.8816454395167328e-07,
      "loss": 0.0008,
      "step": 1240
    },
    {
      "epoch": 4.648648648648649,
      "grad_norm": 0.5624837875366211,
      "learning_rate": 1.5451189159380266e-07,
      "loss": 0.0009,
      "step": 1250
    },
    {
      "epoch": 4.6859273066169616,
      "grad_norm": 2.3667943477630615,
      "learning_rate": 1.2412543446383373e-07,
      "loss": 0.0048,
      "step": 1260
    },
    {
      "epoch": 4.723205964585275,
      "grad_norm": 2.504687786102295,
      "learning_rate": 9.702565514247919e-08,
      "loss": 0.0043,
      "step": 1270
    },
    {
      "epoch": 4.760484622553588,
      "grad_norm": 0.17930255830287933,
      "learning_rate": 7.32308207615351e-08,
      "loss": 0.005,
      "step": 1280
    },
    {
      "epoch": 4.7977632805219015,
      "grad_norm": 0.11154995858669281,
      "learning_rate": 5.275697069056895e-08,
      "loss": 0.0047,
      "step": 1290
    },
    {
      "epoch": 4.835041938490214,
      "grad_norm": 0.13471058011054993,
      "learning_rate": 3.561790572526924e-08,
      "loss": 0.0072,
      "step": 1300
    },
    {
      "epoch": 4.872320596458527,
      "grad_norm": 0.05573806166648865,
      "learning_rate": 2.1825178784753985e-08,
      "loss": 0.0041,
      "step": 1310
    },
    {
      "epoch": 4.9095992544268405,
      "grad_norm": 0.27573004364967346,
      "learning_rate": 1.1388087124099557e-08,
      "loss": 0.0034,
      "step": 1320
    },
    {
      "epoch": 4.946877912395154,
      "grad_norm": 5.542865753173828,
      "learning_rate": 4.313666067345401e-09,
      "loss": 0.0039,
      "step": 1330
    },
    {
      "epoch": 4.984156570363467,
      "grad_norm": 0.2552098035812378,
      "learning_rate": 6.066842651997195e-10,
      "loss": 0.0017,
      "step": 1340
    }
  ],
  "logging_steps": 10,
  "max_steps": 1345,
  "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": 6536924658401280.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}