{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 1326,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0015082956259426848,
      "grad_norm": 29.815479278564453,
      "learning_rate": 2.2388059701492537e-07,
      "loss": 0.9120703339576721,
      "step": 1,
      "token_acc": 0.8818734015345269
    },
    {
      "epoch": 0.015082956259426848,
      "grad_norm": 12.396344184875488,
      "learning_rate": 2.2388059701492535e-06,
      "loss": 0.7334412468804253,
      "step": 10,
      "token_acc": 0.8877795797013311
    },
    {
      "epoch": 0.030165912518853696,
      "grad_norm": 2.1095762252807617,
      "learning_rate": 4.477611940298507e-06,
      "loss": 0.22399492263793946,
      "step": 20,
      "token_acc": 0.9162146601618133
    },
    {
      "epoch": 0.04524886877828054,
      "grad_norm": 1.1608413457870483,
      "learning_rate": 6.716417910447762e-06,
      "loss": 0.16365647315979004,
      "step": 30,
      "token_acc": 0.9324771509612354
    },
    {
      "epoch": 0.06033182503770739,
      "grad_norm": 0.7096361517906189,
      "learning_rate": 8.955223880597014e-06,
      "loss": 0.1502757787704468,
      "step": 40,
      "token_acc": 0.9384272679242049
    },
    {
      "epoch": 0.07541478129713423,
      "grad_norm": 0.8093337416648865,
      "learning_rate": 1.119402985074627e-05,
      "loss": 0.1409849166870117,
      "step": 50,
      "token_acc": 0.943997583773129
    },
    {
      "epoch": 0.09049773755656108,
      "grad_norm": 0.8115042448043823,
      "learning_rate": 1.3432835820895523e-05,
      "loss": 0.13948071002960205,
      "step": 60,
      "token_acc": 0.9437285085323135
    },
    {
      "epoch": 0.10558069381598793,
      "grad_norm": 0.9792377948760986,
      "learning_rate": 1.4999789854529206e-05,
      "loss": 0.13599691390991211,
      "step": 70,
      "token_acc": 0.9456004846281993
    },
    {
      "epoch": 0.12066365007541478,
      "grad_norm": 0.9598162174224854,
      "learning_rate": 1.4996054262643387e-05,
      "loss": 0.13396866321563722,
      "step": 80,
      "token_acc": 0.9470223855769079
    },
    {
      "epoch": 0.13574660633484162,
      "grad_norm": 0.8766047358512878,
      "learning_rate": 1.498765144861652e-05,
      "loss": 0.13383805751800537,
      "step": 90,
      "token_acc": 0.9467140586100363
    },
    {
      "epoch": 0.15082956259426847,
      "grad_norm": 0.6658370494842529,
      "learning_rate": 1.4974586644241e-05,
      "loss": 0.1294718623161316,
      "step": 100,
      "token_acc": 0.9488610561781293
    },
    {
      "epoch": 0.16591251885369532,
      "grad_norm": 0.7449970245361328,
      "learning_rate": 1.4956867983975587e-05,
      "loss": 0.1265101194381714,
      "step": 110,
      "token_acc": 0.9505024772205117
    },
    {
      "epoch": 0.18099547511312217,
      "grad_norm": 0.9009830355644226,
      "learning_rate": 1.493450649988069e-05,
      "loss": 0.12676594257354737,
      "step": 120,
      "token_acc": 0.9508201932858596
    },
    {
      "epoch": 0.19607843137254902,
      "grad_norm": 0.6432599425315857,
      "learning_rate": 1.4907516114749562e-05,
      "loss": 0.1273956298828125,
      "step": 130,
      "token_acc": 0.9494921556877026
    },
    {
      "epoch": 0.21116138763197587,
      "grad_norm": 0.698387861251831,
      "learning_rate": 1.487591363343962e-05,
      "loss": 0.12959996461868287,
      "step": 140,
      "token_acc": 0.9489460940865406
    },
    {
      "epoch": 0.22624434389140272,
      "grad_norm": 0.7407552599906921,
      "learning_rate": 1.4839718732409347e-05,
      "loss": 0.12531228065490724,
      "step": 150,
      "token_acc": 0.9505425260504718
    },
    {
      "epoch": 0.24132730015082957,
      "grad_norm": 0.6289209127426147,
      "learning_rate": 1.4798953947467258e-05,
      "loss": 0.1264174222946167,
      "step": 160,
      "token_acc": 0.9510897650241663
    },
    {
      "epoch": 0.2564102564102564,
      "grad_norm": 0.5078371167182922,
      "learning_rate": 1.4753644659740567e-05,
      "loss": 0.12367002964019776,
      "step": 170,
      "token_acc": 0.9527126937910639
    },
    {
      "epoch": 0.27149321266968324,
      "grad_norm": 0.6037768721580505,
      "learning_rate": 1.4703819079872297e-05,
      "loss": 0.12230026721954346,
      "step": 180,
      "token_acc": 0.9514374689714311
    },
    {
      "epoch": 0.2865761689291101,
      "grad_norm": 1.126369595527649,
      "learning_rate": 1.4649508230456655e-05,
      "loss": 0.12020505666732788,
      "step": 190,
      "token_acc": 0.9517103550845882
    },
    {
      "epoch": 0.30165912518853694,
      "grad_norm": 0.706250786781311,
      "learning_rate": 1.459074592672363e-05,
      "loss": 0.12032895088195801,
      "step": 200,
      "token_acc": 0.9516314228303512
    },
    {
      "epoch": 0.3167420814479638,
      "grad_norm": 17.569059371948242,
      "learning_rate": 1.4527568755484822e-05,
      "loss": 0.12428458929061889,
      "step": 210,
      "token_acc": 0.9539214311314728
    },
    {
      "epoch": 0.33182503770739064,
      "grad_norm": 0.7467140555381775,
      "learning_rate": 1.4460016052353615e-05,
      "loss": 0.11901211738586426,
      "step": 220,
      "token_acc": 0.9542842215256009
    },
    {
      "epoch": 0.3469079939668175,
      "grad_norm": 0.8222630620002747,
      "learning_rate": 1.43881298772539e-05,
      "loss": 0.11689404249191285,
      "step": 230,
      "token_acc": 0.9536877379054559
    },
    {
      "epoch": 0.36199095022624433,
      "grad_norm": 0.5848267674446106,
      "learning_rate": 1.4311954988232549e-05,
      "loss": 0.1140936017036438,
      "step": 240,
      "token_acc": 0.9551059271601999
    },
    {
      "epoch": 0.3770739064856712,
      "grad_norm": 0.4587470591068268,
      "learning_rate": 1.4231538813591998e-05,
      "loss": 0.11686046123504638,
      "step": 250,
      "token_acc": 0.9549980476376415
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 0.8135445713996887,
      "learning_rate": 1.4146931422360251e-05,
      "loss": 0.1165614128112793,
      "step": 260,
      "token_acc": 0.9541351915539069
    },
    {
      "epoch": 0.4072398190045249,
      "grad_norm": 0.6098634004592896,
      "learning_rate": 1.4058185493116703e-05,
      "loss": 0.11796810626983642,
      "step": 270,
      "token_acc": 0.9544430810332241
    },
    {
      "epoch": 0.42232277526395173,
      "grad_norm": 0.6988653540611267,
      "learning_rate": 1.396535628119321e-05,
      "loss": 0.11589565277099609,
      "step": 280,
      "token_acc": 0.9535664210494914
    },
    {
      "epoch": 0.4374057315233786,
      "grad_norm": 0.5857917070388794,
      "learning_rate": 1.3868501584270802e-05,
      "loss": 0.1144603967666626,
      "step": 290,
      "token_acc": 0.9557559372521086
    },
    {
      "epoch": 0.45248868778280543,
      "grad_norm": 0.5581201910972595,
      "learning_rate": 1.3767681706393457e-05,
      "loss": 0.11729376316070557,
      "step": 300,
      "token_acc": 0.9539887259097483
    },
    {
      "epoch": 0.4675716440422323,
      "grad_norm": 0.634107232093811,
      "learning_rate": 1.3662959420421375e-05,
      "loss": 0.1152184247970581,
      "step": 310,
      "token_acc": 0.9554923666443647
    },
    {
      "epoch": 0.48265460030165913,
      "grad_norm": 0.7922980785369873,
      "learning_rate": 1.35543999289471e-05,
      "loss": 0.11476943492889405,
      "step": 320,
      "token_acc": 0.955084578023417
    },
    {
      "epoch": 0.497737556561086,
      "grad_norm": 0.49707281589508057,
      "learning_rate": 1.3442070823698822e-05,
      "loss": 0.11693618297576905,
      "step": 330,
      "token_acc": 0.954469228143597
    },
    {
      "epoch": 0.5128205128205128,
      "grad_norm": 0.5565932989120483,
      "learning_rate": 1.332604204345616e-05,
      "loss": 0.11409028768539428,
      "step": 340,
      "token_acc": 0.9557340869758506
    },
    {
      "epoch": 0.5279034690799397,
      "grad_norm": 0.7205259799957275,
      "learning_rate": 1.3206385830504614e-05,
      "loss": 0.11583917140960694,
      "step": 350,
      "token_acc": 0.9547551099112996
    },
    {
      "epoch": 0.5429864253393665,
      "grad_norm": 0.6330074071884155,
      "learning_rate": 1.3083176685655786e-05,
      "loss": 0.11165814399719239,
      "step": 360,
      "token_acc": 0.9570610839166415
    },
    {
      "epoch": 0.5580693815987934,
      "grad_norm": 0.5063838958740234,
      "learning_rate": 1.295649132186141e-05,
      "loss": 0.11352043151855469,
      "step": 370,
      "token_acc": 0.9562457422431412
    },
    {
      "epoch": 0.5731523378582202,
      "grad_norm": 0.4470171332359314,
      "learning_rate": 1.2826408616450027e-05,
      "loss": 0.11494969129562378,
      "step": 380,
      "token_acc": 0.9556565656565656
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.5665348768234253,
      "learning_rate": 1.2693009562016084e-05,
      "loss": 0.11389807462692261,
      "step": 390,
      "token_acc": 0.9554707753947679
    },
    {
      "epoch": 0.6033182503770739,
      "grad_norm": 0.5819860696792603,
      "learning_rate": 1.2556377215992019e-05,
      "loss": 0.11242160797119141,
      "step": 400,
      "token_acc": 0.9572872256434549
    },
    {
      "epoch": 0.6184012066365008,
      "grad_norm": 0.6842740178108215,
      "learning_rate": 1.2416596648934716e-05,
      "loss": 0.11078722476959228,
      "step": 410,
      "token_acc": 0.9570846029075144
    },
    {
      "epoch": 0.6334841628959276,
      "grad_norm": 0.6340636014938354,
      "learning_rate": 1.2273754891558542e-05,
      "loss": 0.11188424825668335,
      "step": 420,
      "token_acc": 0.9564591028214992
    },
    {
      "epoch": 0.6485671191553545,
      "grad_norm": 0.7150080800056458,
      "learning_rate": 1.212794088054797e-05,
      "loss": 0.11114382743835449,
      "step": 430,
      "token_acc": 0.9571518106914105
    },
    {
      "epoch": 0.6636500754147813,
      "grad_norm": 0.6401362419128418,
      "learning_rate": 1.1979245403183451e-05,
      "loss": 0.10850663185119629,
      "step": 440,
      "token_acc": 0.9578280044101434
    },
    {
      "epoch": 0.6787330316742082,
      "grad_norm": 0.3908592164516449,
      "learning_rate": 1.182776104081511e-05,
      "loss": 0.10802350044250489,
      "step": 450,
      "token_acc": 0.9573075771188979
    },
    {
      "epoch": 0.693815987933635,
      "grad_norm": 0.5129520893096924,
      "learning_rate": 1.167358211121938e-05,
      "loss": 0.10646429061889648,
      "step": 460,
      "token_acc": 0.9588181522395922
    },
    {
      "epoch": 0.7088989441930619,
      "grad_norm": 0.5238049626350403,
      "learning_rate": 1.1516804609874507e-05,
      "loss": 0.11202960014343262,
      "step": 470,
      "token_acc": 0.9568884020347834
    },
    {
      "epoch": 0.7239819004524887,
      "grad_norm": 0.6929338574409485,
      "learning_rate": 1.1357526150191471e-05,
      "loss": 0.10924453735351562,
      "step": 480,
      "token_acc": 0.9574212846838145
    },
    {
      "epoch": 0.7390648567119156,
      "grad_norm": 0.47621390223503113,
      "learning_rate": 1.1195845902737558e-05,
      "loss": 0.11051700115203858,
      "step": 490,
      "token_acc": 0.9569450689157971
    },
    {
      "epoch": 0.7541478129713424,
      "grad_norm": 0.5051868557929993,
      "learning_rate": 1.1031864533490397e-05,
      "loss": 0.10972610712051392,
      "step": 500,
      "token_acc": 0.957463275464368
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 0.5780250430107117,
      "learning_rate": 1.0865684141160912e-05,
      "loss": 0.10704238414764404,
      "step": 510,
      "token_acc": 0.9587042036251446
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 0.5643448233604431,
      "learning_rate": 1.0697408193624248e-05,
      "loss": 0.10957627296447754,
      "step": 520,
      "token_acc": 0.9574098818903685
    },
    {
      "epoch": 0.799396681749623,
      "grad_norm": 0.5077806115150452,
      "learning_rate": 1.0527141463498186e-05,
      "loss": 0.10917415618896484,
      "step": 530,
      "token_acc": 0.9577429299816209
    },
    {
      "epoch": 0.8144796380090498,
      "grad_norm": 0.3003750145435333,
      "learning_rate": 1.0354989962909223e-05,
      "loss": 0.1055147409439087,
      "step": 540,
      "token_acc": 0.9590716835170364
    },
    {
      "epoch": 0.8295625942684767,
      "grad_norm": 0.3721616566181183,
      "learning_rate": 1.0181060877486896e-05,
      "loss": 0.1066395878791809,
      "step": 550,
      "token_acc": 0.9577666855964372
    },
    {
      "epoch": 0.8446455505279035,
      "grad_norm": 0.6667331457138062,
      "learning_rate": 1.0005462499627443e-05,
      "loss": 0.10366250276565551,
      "step": 560,
      "token_acc": 0.9599628167911544
    },
    {
      "epoch": 0.8597285067873304,
      "grad_norm": 0.5545596480369568,
      "learning_rate": 9.82830416106839e-06,
      "loss": 0.10734113454818725,
      "step": 570,
      "token_acc": 0.9578964143426295
    },
    {
      "epoch": 0.8748114630467572,
      "grad_norm": 0.43999919295310974,
      "learning_rate": 9.649696164815983e-06,
      "loss": 0.1027806282043457,
      "step": 580,
      "token_acc": 0.9604619208620876
    },
    {
      "epoch": 0.889894419306184,
      "grad_norm": 0.4688698649406433,
      "learning_rate": 9.469749716467922e-06,
      "loss": 0.10514171123504638,
      "step": 590,
      "token_acc": 0.9594212797665249
    },
    {
      "epoch": 0.9049773755656109,
      "grad_norm": 0.40809348225593567,
      "learning_rate": 9.288576854974107e-06,
      "loss": 0.10727782249450683,
      "step": 600,
      "token_acc": 0.9590070037556371
    },
    {
      "epoch": 0.9200603318250377,
      "grad_norm": 0.49828872084617615,
      "learning_rate": 9.106290382878496e-06,
      "loss": 0.10498321056365967,
      "step": 610,
      "token_acc": 0.9591861672878352
    },
    {
      "epoch": 0.9351432880844646,
      "grad_norm": 0.4879675805568695,
      "learning_rate": 8.923003796085571e-06,
      "loss": 0.10296416282653809,
      "step": 620,
      "token_acc": 0.9602477477477478
    },
    {
      "epoch": 0.9502262443438914,
      "grad_norm": 0.3828055262565613,
      "learning_rate": 8.738831213195076e-06,
      "loss": 0.1069521188735962,
      "step": 630,
      "token_acc": 0.9581954843577689
    },
    {
      "epoch": 0.9653092006033183,
      "grad_norm": 0.6066528558731079,
      "learning_rate": 8.553887304449051e-06,
      "loss": 0.10204867124557496,
      "step": 640,
      "token_acc": 0.960210117690554
    },
    {
      "epoch": 0.9803921568627451,
      "grad_norm": 0.5788474082946777,
      "learning_rate": 8.368287220335405e-06,
      "loss": 0.10363715887069702,
      "step": 650,
      "token_acc": 0.959661719127198
    },
    {
      "epoch": 0.995475113122172,
      "grad_norm": 0.45377716422080994,
      "learning_rate": 8.18214651989248e-06,
      "loss": 0.10440789461135865,
      "step": 660,
      "token_acc": 0.9594376692706295
    },
    {
      "epoch": 1.0105580693815988,
      "grad_norm": 0.39804208278656006,
      "learning_rate": 7.99558109875924e-06,
      "loss": 0.09972409009933472,
      "step": 670,
      "token_acc": 0.96149672939207
    },
    {
      "epoch": 1.0256410256410255,
      "grad_norm": 0.4112270176410675,
      "learning_rate": 7.808707117015871e-06,
      "loss": 0.09616597890853881,
      "step": 680,
      "token_acc": 0.9632930129633229
    },
    {
      "epoch": 1.0407239819004526,
      "grad_norm": 0.8032577633857727,
      "learning_rate": 7.621640926859767e-06,
      "loss": 0.09788082242012024,
      "step": 690,
      "token_acc": 0.9619020874286249
    },
    {
      "epoch": 1.0558069381598794,
      "grad_norm": 0.5775718092918396,
      "learning_rate": 7.434499000161847e-06,
      "loss": 0.09358711242675781,
      "step": 700,
      "token_acc": 0.9633524141363863
    },
    {
      "epoch": 1.0708898944193062,
      "grad_norm": 0.5401419401168823,
      "learning_rate": 7.247397855948417e-06,
      "loss": 0.09408066868782043,
      "step": 710,
      "token_acc": 0.9634460570946094
    },
    {
      "epoch": 1.085972850678733,
      "grad_norm": 0.3692776560783386,
      "learning_rate": 7.060453987853635e-06,
      "loss": 0.09346169233322144,
      "step": 720,
      "token_acc": 0.9638752317240268
    },
    {
      "epoch": 1.10105580693816,
      "grad_norm": 0.4132152199745178,
      "learning_rate": 6.873783791587794e-06,
      "loss": 0.09388680458068847,
      "step": 730,
      "token_acc": 0.9637559772180428
    },
    {
      "epoch": 1.1161387631975868,
      "grad_norm": 0.4544770419597626,
      "learning_rate": 6.6875034924665825e-06,
      "loss": 0.09155543446540833,
      "step": 740,
      "token_acc": 0.9647887323943662
    },
    {
      "epoch": 1.1312217194570136,
      "grad_norm": 0.4763709008693695,
      "learning_rate": 6.501729073046421e-06,
      "loss": 0.09151909351348878,
      "step": 750,
      "token_acc": 0.9641338049016095
    },
    {
      "epoch": 1.1463046757164403,
      "grad_norm": 0.3512129783630371,
      "learning_rate": 6.316576200910982e-06,
      "loss": 0.08871339559555054,
      "step": 760,
      "token_acc": 0.9654217212723278
    },
    {
      "epoch": 1.1613876319758674,
      "grad_norm": 0.4989905059337616,
      "learning_rate": 6.1321601566537614e-06,
      "loss": 0.09526662230491638,
      "step": 770,
      "token_acc": 0.9631938857602574
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 0.6301065683364868,
      "learning_rate": 5.94859576210165e-06,
      "loss": 0.09378224611282349,
      "step": 780,
      "token_acc": 0.9638894054925299
    },
    {
      "epoch": 1.191553544494721,
      "grad_norm": 0.5111472606658936,
      "learning_rate": 5.765997308824127e-06,
      "loss": 0.09456103444099426,
      "step": 790,
      "token_acc": 0.9630646797222439
    },
    {
      "epoch": 1.2066365007541477,
      "grad_norm": 0.4381602704524994,
      "learning_rate": 5.584478486972594e-06,
      "loss": 0.09292151927947997,
      "step": 800,
      "token_acc": 0.9635370603112539
    },
    {
      "epoch": 1.2217194570135748,
      "grad_norm": 0.46196287870407104,
      "learning_rate": 5.404152314494185e-06,
      "loss": 0.09235544800758362,
      "step": 810,
      "token_acc": 0.9644699668345412
    },
    {
      "epoch": 1.2368024132730016,
      "grad_norm": 0.4581727683544159,
      "learning_rate": 5.225131066764096e-06,
      "loss": 0.0894450843334198,
      "step": 820,
      "token_acc": 0.9649870350432923
    },
    {
      "epoch": 1.2518853695324283,
      "grad_norm": 0.4270052909851074,
      "learning_rate": 5.04752620668025e-06,
      "loss": 0.0947738230228424,
      "step": 830,
      "token_acc": 0.9627476129400029
    },
    {
      "epoch": 1.2669683257918551,
      "grad_norm": 0.45762163400650024,
      "learning_rate": 4.871448315263849e-06,
      "loss": 0.09065256118774415,
      "step": 840,
      "token_acc": 0.9639841068656632
    },
    {
      "epoch": 1.282051282051282,
      "grad_norm": 0.46436837315559387,
      "learning_rate": 4.697007022808989e-06,
      "loss": 0.08982598781585693,
      "step": 850,
      "token_acc": 0.9650109525782962
    },
    {
      "epoch": 1.297134238310709,
      "grad_norm": 0.6739473342895508,
      "learning_rate": 4.52431094062422e-06,
      "loss": 0.09014194011688233,
      "step": 860,
      "token_acc": 0.9642136708704825
    },
    {
      "epoch": 1.3122171945701357,
      "grad_norm": 0.45681968331336975,
      "learning_rate": 4.353467593408566e-06,
      "loss": 0.09165002107620239,
      "step": 870,
      "token_acc": 0.964421518513901
    },
    {
      "epoch": 1.3273001508295625,
      "grad_norm": 0.42422786355018616,
      "learning_rate": 4.1845833523040775e-06,
      "loss": 0.09196000099182129,
      "step": 880,
      "token_acc": 0.9632488905325444
    },
    {
      "epoch": 1.3423831070889896,
      "grad_norm": 0.41552069783210754,
      "learning_rate": 4.0177633686666225e-06,
      "loss": 0.0871020257472992,
      "step": 890,
      "token_acc": 0.9665725819685582
    },
    {
      "epoch": 1.3574660633484164,
      "grad_norm": 0.5335479378700256,
      "learning_rate": 3.853111508596134e-06,
      "loss": 0.0898630142211914,
      "step": 900,
      "token_acc": 0.9652458695864699
    },
    {
      "epoch": 1.3725490196078431,
      "grad_norm": 0.44376763701438904,
      "learning_rate": 3.6907302882671214e-06,
      "loss": 0.09012840986251831,
      "step": 910,
      "token_acc": 0.9647732077067102
    },
    {
      "epoch": 1.38763197586727,
      "grad_norm": 0.5066686272621155,
      "learning_rate": 3.5307208100996376e-06,
      "loss": 0.09017069339752197,
      "step": 920,
      "token_acc": 0.964361793105593
    },
    {
      "epoch": 1.4027149321266967,
      "grad_norm": 0.3902303874492645,
      "learning_rate": 3.373182699810508e-06,
      "loss": 0.09142625331878662,
      "step": 930,
      "token_acc": 0.9647421163923876
    },
    {
      "epoch": 1.4177978883861238,
      "grad_norm": 0.41397470235824585,
      "learning_rate": 3.2182140443839834e-06,
      "loss": 0.09144071340560914,
      "step": 940,
      "token_acc": 0.9640816822657104
    },
    {
      "epoch": 1.4328808446455505,
      "grad_norm": 0.5350260734558105,
      "learning_rate": 3.0659113310004437e-06,
      "loss": 0.09406846761703491,
      "step": 950,
      "token_acc": 0.9636620346288177
    },
    {
      "epoch": 1.4479638009049773,
      "grad_norm": 0.3802923858165741,
      "learning_rate": 2.9163693869611806e-06,
      "loss": 0.08901124000549317,
      "step": 960,
      "token_acc": 0.9656810008299266
    },
    {
      "epoch": 1.4630467571644044,
      "grad_norm": 0.35188332200050354,
      "learning_rate": 2.7696813206466494e-06,
      "loss": 0.08718435764312744,
      "step": 970,
      "token_acc": 0.966524814193239
    },
    {
      "epoch": 1.4781297134238311,
      "grad_norm": 0.4129849672317505,
      "learning_rate": 2.6259384635449823e-06,
      "loss": 0.08675562143325806,
      "step": 980,
      "token_acc": 0.9662713856203166
    },
    {
      "epoch": 1.493212669683258,
      "grad_norm": 0.45535895228385925,
      "learning_rate": 2.48523031338683e-06,
      "loss": 0.09237306118011475,
      "step": 990,
      "token_acc": 0.9645306006091614
    },
    {
      "epoch": 1.5082956259426847,
      "grad_norm": 0.4584350287914276,
      "learning_rate": 2.347644478421916e-06,
      "loss": 0.09014174342155457,
      "step": 1000,
      "token_acc": 0.9647957202640322
    },
    {
      "epoch": 1.5233785822021115,
      "grad_norm": 0.4994698762893677,
      "learning_rate": 2.2132666228720783e-06,
      "loss": 0.09264277219772339,
      "step": 1010,
      "token_acc": 0.9640724946695096
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 0.39258939027786255,
      "learning_rate": 2.0821804135946685e-06,
      "loss": 0.0883907675743103,
      "step": 1020,
      "token_acc": 0.9648448140492557
    },
    {
      "epoch": 1.5535444947209653,
      "grad_norm": 0.3761826753616333,
      "learning_rate": 1.954467467989575e-06,
      "loss": 0.08799158334732056,
      "step": 1030,
      "token_acc": 0.9661682044755542
    },
    {
      "epoch": 1.5686274509803921,
      "grad_norm": 0.4628184139728546,
      "learning_rate": 1.8302073031822736e-06,
      "loss": 0.08756601214408874,
      "step": 1040,
      "token_acc": 0.9655115037286905
    },
    {
      "epoch": 1.5837104072398192,
      "grad_norm": 0.5046730041503906,
      "learning_rate": 1.7094772865145908e-06,
      "loss": 0.08718575239181518,
      "step": 1050,
      "token_acc": 0.966780388576554
    },
    {
      "epoch": 1.598793363499246,
      "grad_norm": 0.45518726110458374,
      "learning_rate": 1.5923525873739312e-06,
      "loss": 0.08757206201553344,
      "step": 1060,
      "token_acc": 0.9656194052315541
    },
    {
      "epoch": 1.6138763197586727,
      "grad_norm": 0.3528100252151489,
      "learning_rate": 1.4789061303910304e-06,
      "loss": 0.09095115661621093,
      "step": 1070,
      "token_acc": 0.9643856022374901
    },
    {
      "epoch": 1.6289592760180995,
      "grad_norm": 0.3653790056705475,
      "learning_rate": 1.3692085500353444e-06,
      "loss": 0.09028034210205078,
      "step": 1080,
      "token_acc": 0.9654089878495437
    },
    {
      "epoch": 1.6440422322775263,
      "grad_norm": 0.3657921254634857,
      "learning_rate": 1.263328146636342e-06,
      "loss": 0.08679715394973755,
      "step": 1090,
      "token_acc": 0.9662785326303381
    },
    {
      "epoch": 1.6591251885369531,
      "grad_norm": 0.3641863465309143,
      "learning_rate": 1.1613308438580924e-06,
      "loss": 0.08758020401000977,
      "step": 1100,
      "token_acc": 0.9654987184799791
    },
    {
      "epoch": 1.6742081447963801,
      "grad_norm": 0.4320993721485138,
      "learning_rate": 1.063280147653618e-06,
      "loss": 0.08432935476303101,
      "step": 1110,
      "token_acc": 0.9675015490982618
    },
    {
      "epoch": 1.689291101055807,
      "grad_norm": 0.34107470512390137,
      "learning_rate": 9.692371067245726e-07,
      "loss": 0.09100235104560853,
      "step": 1120,
      "token_acc": 0.964681418950374
    },
    {
      "epoch": 1.704374057315234,
      "grad_norm": 0.32863467931747437,
      "learning_rate": 8.792602745108685e-07,
      "loss": 0.08523865342140198,
      "step": 1130,
      "token_acc": 0.9660080165341016
    },
    {
      "epoch": 1.7194570135746607,
      "grad_norm": 0.44436898827552795,
      "learning_rate": 7.93405672733902e-07,
      "loss": 0.09061864614486695,
      "step": 1140,
      "token_acc": 0.9646512324867011
    },
    {
      "epoch": 1.7345399698340875,
      "grad_norm": 0.365710586309433,
      "learning_rate": 7.117267565160989e-07,
      "loss": 0.0857430338859558,
      "step": 1150,
      "token_acc": 0.9665197002508296
    },
    {
      "epoch": 1.7496229260935143,
      "grad_norm": 0.4832117557525635,
      "learning_rate": 6.342743810984836e-07,
      "loss": 0.08890118002891541,
      "step": 1160,
      "token_acc": 0.9660351545073312
    },
    {
      "epoch": 1.7647058823529411,
      "grad_norm": 0.3681408762931824,
      "learning_rate": 5.610967701769814e-07,
      "loss": 0.08722085952758789,
      "step": 1170,
      "token_acc": 0.9660664603457438
    },
    {
      "epoch": 1.779788838612368,
      "grad_norm": 0.4544508457183838,
      "learning_rate": 4.922394858772067e-07,
      "loss": 0.08912365436553955,
      "step": 1180,
      "token_acc": 0.9653174628232248
    },
    {
      "epoch": 1.7948717948717947,
      "grad_norm": 0.4088412821292877,
      "learning_rate": 4.2774540038639286e-07,
      "loss": 0.08246009349822998,
      "step": 1190,
      "token_acc": 0.9678997594096655
    },
    {
      "epoch": 1.8099547511312217,
      "grad_norm": 0.4180281162261963,
      "learning_rate": 3.676546692601465e-07,
      "loss": 0.08530747294425964,
      "step": 1200,
      "token_acc": 0.966511905135364
    },
    {
      "epoch": 1.8250377073906485,
      "grad_norm": 0.3279511332511902,
      "learning_rate": 3.1200470642063716e-07,
      "loss": 0.08611878156661987,
      "step": 1210,
      "token_acc": 0.9662316426015056
    },
    {
      "epoch": 1.8401206636500755,
      "grad_norm": 0.4073133170604706,
      "learning_rate": 2.6083016086180715e-07,
      "loss": 0.08319737315177918,
      "step": 1220,
      "token_acc": 0.9677166096159282
    },
    {
      "epoch": 1.8552036199095023,
      "grad_norm": 0.3561563789844513,
      "learning_rate": 2.14162895076071e-07,
      "loss": 0.0903146207332611,
      "step": 1230,
      "token_acc": 0.9648374348841386
    },
    {
      "epoch": 1.8702865761689291,
      "grad_norm": 0.42073923349380493,
      "learning_rate": 1.7203196521597354e-07,
      "loss": 0.08666027188301087,
      "step": 1240,
      "token_acc": 0.9662166414904331
    },
    {
      "epoch": 1.885369532428356,
      "grad_norm": 0.3711903691291809,
      "learning_rate": 1.3446360300313375e-07,
      "loss": 0.0845598578453064,
      "step": 1250,
      "token_acc": 0.9672128552204544
    },
    {
      "epoch": 1.9004524886877827,
      "grad_norm": 0.4648366868495941,
      "learning_rate": 1.0148119939575207e-07,
      "loss": 0.08463501334190368,
      "step": 1260,
      "token_acc": 0.9671166666666666
    },
    {
      "epoch": 1.9155354449472095,
      "grad_norm": 0.34604692459106445,
      "learning_rate": 7.310529002484079e-08,
      "loss": 0.0892053484916687,
      "step": 1270,
      "token_acc": 0.9649981072187764
    },
    {
      "epoch": 1.9306184012066365,
      "grad_norm": 0.38359424471855164,
      "learning_rate": 4.9353542408249465e-08,
      "loss": 0.08901722431182861,
      "step": 1280,
      "token_acc": 0.9650786774148629
    },
    {
      "epoch": 1.9457013574660633,
      "grad_norm": 0.42977797985076904,
      "learning_rate": 3.024074495044826e-08,
      "loss": 0.08712854385375976,
      "step": 1290,
      "token_acc": 0.9666139518538043
    },
    {
      "epoch": 1.9607843137254903,
      "grad_norm": 0.3914060592651367,
      "learning_rate": 1.5778797734913307e-08,
      "loss": 0.08645588755607606,
      "step": 1300,
      "token_acc": 0.96584090325116
    },
    {
      "epoch": 1.9758672699849171,
      "grad_norm": 0.3837912380695343,
      "learning_rate": 5.976705114852654e-09,
      "loss": 0.08520195484161378,
      "step": 1310,
      "token_acc": 0.966562089829369
    },
    {
      "epoch": 1.990950226244344,
      "grad_norm": 0.38043421506881714,
      "learning_rate": 8.40570106880878e-10,
      "loss": 0.08780958652496337,
      "step": 1320,
      "token_acc": 0.9663121369842348
    }
  ],
  "logging_steps": 10,
  "max_steps": 1326,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 1.5775587374832026e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}