{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.6618863761720905,
  "eval_steps": 300.0,
  "global_step": 600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0011031439602868175,
      "grad_norm": 80.42967325823248,
      "learning_rate": 2.3e-06,
      "loss": 1.6655548810958862,
      "step": 1,
      "token_acc": 0.6208588957055214
    },
    {
      "epoch": 0.005515719801434087,
      "grad_norm": 12.936447236966638,
      "learning_rate": 2.3e-06,
      "loss": 1.6539018154144287,
      "step": 5,
      "token_acc": 0.6644312612844984
    },
    {
      "epoch": 0.011031439602868174,
      "grad_norm": 7.485145348211075,
      "learning_rate": 2.3e-06,
      "loss": 1.1399970054626465,
      "step": 10,
      "token_acc": 0.716547901821061
    },
    {
      "epoch": 0.01654715940430226,
      "grad_norm": 7.10761756326815,
      "learning_rate": 2.3e-06,
      "loss": 1.1828346252441406,
      "step": 15,
      "token_acc": 0.6928796755295178
    },
    {
      "epoch": 0.02206287920573635,
      "grad_norm": 5.659684098393193,
      "learning_rate": 2.3e-06,
      "loss": 1.0604573249816895,
      "step": 20,
      "token_acc": 0.7371653156472611
    },
    {
      "epoch": 0.027578599007170437,
      "grad_norm": 6.432203040875114,
      "learning_rate": 2.3e-06,
      "loss": 1.0374173164367675,
      "step": 25,
      "token_acc": 0.7369089984748348
    },
    {
      "epoch": 0.03309431880860452,
      "grad_norm": 6.0637637354614125,
      "learning_rate": 2.3e-06,
      "loss": 0.9596109390258789,
      "step": 30,
      "token_acc": 0.7449127906976745
    },
    {
      "epoch": 0.03861003861003861,
      "grad_norm": 5.9792048911606335,
      "learning_rate": 2.3e-06,
      "loss": 0.9912214279174805,
      "step": 35,
      "token_acc": 0.7413617886178862
    },
    {
      "epoch": 0.0441257584114727,
      "grad_norm": 6.031366828095079,
      "learning_rate": 2.3e-06,
      "loss": 1.0473726272583008,
      "step": 40,
      "token_acc": 0.727144535840188
    },
    {
      "epoch": 0.049641478212906785,
      "grad_norm": 6.186527950833231,
      "learning_rate": 2.3e-06,
      "loss": 0.9992570877075195,
      "step": 45,
      "token_acc": 0.7360217714002969
    },
    {
      "epoch": 0.05515719801434087,
      "grad_norm": 5.923085345386395,
      "learning_rate": 2.3e-06,
      "loss": 1.002680206298828,
      "step": 50,
      "token_acc": 0.7391618497109826
    },
    {
      "epoch": 0.06067291781577496,
      "grad_norm": 4.9260097226563255,
      "learning_rate": 2.3e-06,
      "loss": 0.8863202095031738,
      "step": 55,
      "token_acc": 0.7527737578388809
    },
    {
      "epoch": 0.06618863761720904,
      "grad_norm": 5.560254131314214,
      "learning_rate": 2.3e-06,
      "loss": 0.9248697280883789,
      "step": 60,
      "token_acc": 0.7579394848712178
    },
    {
      "epoch": 0.07170435741864313,
      "grad_norm": 5.180260599645047,
      "learning_rate": 2.3e-06,
      "loss": 1.004658317565918,
      "step": 65,
      "token_acc": 0.7360208062418726
    },
    {
      "epoch": 0.07722007722007722,
      "grad_norm": 6.299333554712354,
      "learning_rate": 2.3e-06,
      "loss": 0.9439926147460938,
      "step": 70,
      "token_acc": 0.751812046988253
    },
    {
      "epoch": 0.0827357970215113,
      "grad_norm": 5.622631000909693,
      "learning_rate": 2.3e-06,
      "loss": 0.8676441192626954,
      "step": 75,
      "token_acc": 0.760814889336016
    },
    {
      "epoch": 0.0882515168229454,
      "grad_norm": 5.613397002114218,
      "learning_rate": 2.3e-06,
      "loss": 0.8931824684143066,
      "step": 80,
      "token_acc": 0.7616038882138517
    },
    {
      "epoch": 0.09376723662437948,
      "grad_norm": 5.897499862699952,
      "learning_rate": 2.3e-06,
      "loss": 1.0003661155700683,
      "step": 85,
      "token_acc": 0.7361477572559367
    },
    {
      "epoch": 0.09928295642581357,
      "grad_norm": 5.838989842949948,
      "learning_rate": 2.3e-06,
      "loss": 0.8919829368591309,
      "step": 90,
      "token_acc": 0.7569311663479924
    },
    {
      "epoch": 0.10479867622724766,
      "grad_norm": 5.732772113211347,
      "learning_rate": 2.3e-06,
      "loss": 0.9857375144958496,
      "step": 95,
      "token_acc": 0.7319040543409042
    },
    {
      "epoch": 0.11031439602868175,
      "grad_norm": 5.887155914575355,
      "learning_rate": 2.3e-06,
      "loss": 0.9643180847167969,
      "step": 100,
      "token_acc": 0.7468728678644531
    },
    {
      "epoch": 0.11583011583011583,
      "grad_norm": 5.409153898947996,
      "learning_rate": 2.3e-06,
      "loss": 0.9775169372558594,
      "step": 105,
      "token_acc": 0.7443522360534809
    },
    {
      "epoch": 0.12134583563154992,
      "grad_norm": 5.738347427048291,
      "learning_rate": 2.3e-06,
      "loss": 1.0265106201171874,
      "step": 110,
      "token_acc": 0.7265641025641025
    },
    {
      "epoch": 0.126861555432984,
      "grad_norm": 5.38681474950728,
      "learning_rate": 2.3e-06,
      "loss": 0.9541057586669922,
      "step": 115,
      "token_acc": 0.747302805082714
    },
    {
      "epoch": 0.13237727523441808,
      "grad_norm": 5.856303394018206,
      "learning_rate": 2.3e-06,
      "loss": 0.8228609085083007,
      "step": 120,
      "token_acc": 0.7779299014238773
    },
    {
      "epoch": 0.13789299503585217,
      "grad_norm": 6.568516319919617,
      "learning_rate": 2.3e-06,
      "loss": 0.9721288681030273,
      "step": 125,
      "token_acc": 0.7395016151361329
    },
    {
      "epoch": 0.14340871483728626,
      "grad_norm": 5.54045872938136,
      "learning_rate": 2.3e-06,
      "loss": 0.9882600784301758,
      "step": 130,
      "token_acc": 0.7350409836065573
    },
    {
      "epoch": 0.14892443463872035,
      "grad_norm": 5.612320181227532,
      "learning_rate": 2.3e-06,
      "loss": 0.9342703819274902,
      "step": 135,
      "token_acc": 0.7464104263309035
    },
    {
      "epoch": 0.15444015444015444,
      "grad_norm": 5.925987985108847,
      "learning_rate": 2.3e-06,
      "loss": 0.9266027450561524,
      "step": 140,
      "token_acc": 0.7487753673897831
    },
    {
      "epoch": 0.15995587424158852,
      "grad_norm": 5.59653045493395,
      "learning_rate": 2.3e-06,
      "loss": 0.8937458038330078,
      "step": 145,
      "token_acc": 0.7624970664163342
    },
    {
      "epoch": 0.1654715940430226,
      "grad_norm": 5.374039667260155,
      "learning_rate": 2.3e-06,
      "loss": 0.8537099838256836,
      "step": 150,
      "token_acc": 0.7640309304065852
    },
    {
      "epoch": 0.1709873138444567,
      "grad_norm": 5.376937826461383,
      "learning_rate": 2.3e-06,
      "loss": 0.8238465309143066,
      "step": 155,
      "token_acc": 0.7761599210266535
    },
    {
      "epoch": 0.1765030336458908,
      "grad_norm": 6.0245180163205285,
      "learning_rate": 2.3e-06,
      "loss": 0.8189101219177246,
      "step": 160,
      "token_acc": 0.7774302840761012
    },
    {
      "epoch": 0.18201875344732488,
      "grad_norm": 5.571460118689288,
      "learning_rate": 2.3e-06,
      "loss": 0.8788368225097656,
      "step": 165,
      "token_acc": 0.760459995106435
    },
    {
      "epoch": 0.18753447324875896,
      "grad_norm": 5.088793082374291,
      "learning_rate": 2.3e-06,
      "loss": 0.8897226333618165,
      "step": 170,
      "token_acc": 0.7589285714285714
    },
    {
      "epoch": 0.19305019305019305,
      "grad_norm": 5.578076711526945,
      "learning_rate": 2.3e-06,
      "loss": 0.8561611175537109,
      "step": 175,
      "token_acc": 0.7607285429141717
    },
    {
      "epoch": 0.19856591285162714,
      "grad_norm": 5.81713562547314,
      "learning_rate": 2.3e-06,
      "loss": 0.8790461540222168,
      "step": 180,
      "token_acc": 0.753576372865713
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 5.254765297480428,
      "learning_rate": 2.3e-06,
      "loss": 0.7894742965698243,
      "step": 185,
      "token_acc": 0.7727930535455861
    },
    {
      "epoch": 0.20959735245449532,
      "grad_norm": 5.076375360468776,
      "learning_rate": 2.3e-06,
      "loss": 0.8446966171264648,
      "step": 190,
      "token_acc": 0.7672064777327935
    },
    {
      "epoch": 0.2151130722559294,
      "grad_norm": 5.763028979469674,
      "learning_rate": 2.3e-06,
      "loss": 0.961794662475586,
      "step": 195,
      "token_acc": 0.7398701589433624
    },
    {
      "epoch": 0.2206287920573635,
      "grad_norm": 5.351427081900536,
      "learning_rate": 2.3e-06,
      "loss": 0.9339935302734375,
      "step": 200,
      "token_acc": 0.7431607506217499
    },
    {
      "epoch": 0.22614451185879758,
      "grad_norm": 5.713651299205875,
      "learning_rate": 2.3e-06,
      "loss": 0.918415641784668,
      "step": 205,
      "token_acc": 0.7481903926299627
    },
    {
      "epoch": 0.23166023166023167,
      "grad_norm": 5.8891737450130535,
      "learning_rate": 2.3e-06,
      "loss": 0.9010303497314454,
      "step": 210,
      "token_acc": 0.7571428571428571
    },
    {
      "epoch": 0.23717595146166576,
      "grad_norm": 5.4902683255452915,
      "learning_rate": 2.3e-06,
      "loss": 0.8169702529907227,
      "step": 215,
      "token_acc": 0.7799597180261832
    },
    {
      "epoch": 0.24269167126309985,
      "grad_norm": 5.777757326601303,
      "learning_rate": 2.3e-06,
      "loss": 0.9132566452026367,
      "step": 220,
      "token_acc": 0.7451829400303096
    },
    {
      "epoch": 0.24820739106453393,
      "grad_norm": 5.440176086672954,
      "learning_rate": 2.3e-06,
      "loss": 0.9655065536499023,
      "step": 225,
      "token_acc": 0.7356965174129353
    },
    {
      "epoch": 0.253723110865968,
      "grad_norm": 5.65109550423067,
      "learning_rate": 2.3e-06,
      "loss": 0.9338722229003906,
      "step": 230,
      "token_acc": 0.740830755634114
    },
    {
      "epoch": 0.2592388306674021,
      "grad_norm": 5.558767495868624,
      "learning_rate": 2.3e-06,
      "loss": 0.9302779197692871,
      "step": 235,
      "token_acc": 0.7472228044435129
    },
    {
      "epoch": 0.26475455046883617,
      "grad_norm": 5.0652164247691,
      "learning_rate": 2.3e-06,
      "loss": 0.8661215782165528,
      "step": 240,
      "token_acc": 0.7648968235566891
    },
    {
      "epoch": 0.2702702702702703,
      "grad_norm": 5.301144006085399,
      "learning_rate": 2.3e-06,
      "loss": 0.9218810081481934,
      "step": 245,
      "token_acc": 0.7493386243386243
    },
    {
      "epoch": 0.27578599007170435,
      "grad_norm": 5.6495062752144625,
      "learning_rate": 2.3e-06,
      "loss": 0.7810210227966309,
      "step": 250,
      "token_acc": 0.7860394537177542
    },
    {
      "epoch": 0.28130170987313846,
      "grad_norm": 5.522127957882286,
      "learning_rate": 2.3e-06,
      "loss": 0.8038553237915039,
      "step": 255,
      "token_acc": 0.7727514635444386
    },
    {
      "epoch": 0.2868174296745725,
      "grad_norm": 5.2610053652109405,
      "learning_rate": 2.3e-06,
      "loss": 0.8923781394958497,
      "step": 260,
      "token_acc": 0.7542778918548939
    },
    {
      "epoch": 0.29233314947600664,
      "grad_norm": 5.613909953834712,
      "learning_rate": 2.3e-06,
      "loss": 0.8282554626464844,
      "step": 265,
      "token_acc": 0.7708383377372088
    },
    {
      "epoch": 0.2978488692774407,
      "grad_norm": 4.960174308621677,
      "learning_rate": 2.3e-06,
      "loss": 0.939533805847168,
      "step": 270,
      "token_acc": 0.745945945945946
    },
    {
      "epoch": 0.3033645890788748,
      "grad_norm": 5.338393166552948,
      "learning_rate": 2.3e-06,
      "loss": 0.7566696166992187,
      "step": 275,
      "token_acc": 0.7783882783882784
    },
    {
      "epoch": 0.3088803088803089,
      "grad_norm": 5.85539301408898,
      "learning_rate": 2.3e-06,
      "loss": 0.7774906158447266,
      "step": 280,
      "token_acc": 0.778774673160728
    },
    {
      "epoch": 0.314396028681743,
      "grad_norm": 5.5554378762103305,
      "learning_rate": 2.3e-06,
      "loss": 0.8302905082702636,
      "step": 285,
      "token_acc": 0.7736331119059785
    },
    {
      "epoch": 0.31991174848317705,
      "grad_norm": 5.64607545742485,
      "learning_rate": 2.3e-06,
      "loss": 0.8949955940246582,
      "step": 290,
      "token_acc": 0.7522200563136235
    },
    {
      "epoch": 0.32542746828461117,
      "grad_norm": 5.545223764081705,
      "learning_rate": 2.3e-06,
      "loss": 0.9496533393859863,
      "step": 295,
      "token_acc": 0.739816799844085
    },
    {
      "epoch": 0.3309431880860452,
      "grad_norm": 5.102348588101583,
      "learning_rate": 2.3e-06,
      "loss": 0.8521288871765137,
      "step": 300,
      "token_acc": 0.7637759710930443
    },
    {
      "epoch": 0.33645890788747934,
      "grad_norm": 5.799379698214005,
      "learning_rate": 2.3e-06,
      "loss": 0.8561582565307617,
      "step": 305,
      "token_acc": 0.7558797909407665
    },
    {
      "epoch": 0.3419746276889134,
      "grad_norm": 5.245842611420519,
      "learning_rate": 2.3e-06,
      "loss": 0.855079460144043,
      "step": 310,
      "token_acc": 0.7621547591874002
    },
    {
      "epoch": 0.3474903474903475,
      "grad_norm": 5.339700948065717,
      "learning_rate": 2.3e-06,
      "loss": 0.8615094184875488,
      "step": 315,
      "token_acc": 0.7575689616505943
    },
    {
      "epoch": 0.3530060672917816,
      "grad_norm": 5.608312178321082,
      "learning_rate": 2.3e-06,
      "loss": 0.8664688110351563,
      "step": 320,
      "token_acc": 0.7657678244972578
    },
    {
      "epoch": 0.35852178709321564,
      "grad_norm": 5.41193963454708,
      "learning_rate": 2.3e-06,
      "loss": 0.83935546875,
      "step": 325,
      "token_acc": 0.7757989753598439
    },
    {
      "epoch": 0.36403750689464975,
      "grad_norm": 5.220504527588121,
      "learning_rate": 2.3e-06,
      "loss": 0.8449893951416015,
      "step": 330,
      "token_acc": 0.772705078125
    },
    {
      "epoch": 0.3695532266960838,
      "grad_norm": 5.4190449827537135,
      "learning_rate": 2.3e-06,
      "loss": 0.8168981552124024,
      "step": 335,
      "token_acc": 0.7692491630798661
    },
    {
      "epoch": 0.37506894649751793,
      "grad_norm": 5.590435008250206,
      "learning_rate": 2.3e-06,
      "loss": 0.8648731231689453,
      "step": 340,
      "token_acc": 0.7575426322693485
    },
    {
      "epoch": 0.380584666298952,
      "grad_norm": 5.174005161957214,
      "learning_rate": 2.3e-06,
      "loss": 0.9896360397338867,
      "step": 345,
      "token_acc": 0.7302983932670237
    },
    {
      "epoch": 0.3861003861003861,
      "grad_norm": 5.511715259575185,
      "learning_rate": 2.3e-06,
      "loss": 0.8663737297058105,
      "step": 350,
      "token_acc": 0.7572226268511775
    },
    {
      "epoch": 0.39161610590182017,
      "grad_norm": 5.295113853870287,
      "learning_rate": 2.3e-06,
      "loss": 0.8019681930541992,
      "step": 355,
      "token_acc": 0.7747044917257683
    },
    {
      "epoch": 0.3971318257032543,
      "grad_norm": 5.2190425564256415,
      "learning_rate": 2.3e-06,
      "loss": 0.8304034233093261,
      "step": 360,
      "token_acc": 0.7567389875082183
    },
    {
      "epoch": 0.40264754550468834,
      "grad_norm": 5.594923827868745,
      "learning_rate": 2.3e-06,
      "loss": 0.8835367202758789,
      "step": 365,
      "token_acc": 0.7492839832562238
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 5.399822330777811,
      "learning_rate": 2.3e-06,
      "loss": 0.8703033447265625,
      "step": 370,
      "token_acc": 0.7619047619047619
    },
    {
      "epoch": 0.4136789851075565,
      "grad_norm": 5.718517969581787,
      "learning_rate": 2.3e-06,
      "loss": 0.7746540069580078,
      "step": 375,
      "token_acc": 0.7788509162951956
    },
    {
      "epoch": 0.41919470490899063,
      "grad_norm": 5.358207310874379,
      "learning_rate": 2.3e-06,
      "loss": 0.7805415630340576,
      "step": 380,
      "token_acc": 0.7823351786612984
    },
    {
      "epoch": 0.4247104247104247,
      "grad_norm": 5.240031526062139,
      "learning_rate": 2.3e-06,
      "loss": 0.892514419555664,
      "step": 385,
      "token_acc": 0.7574438845625286
    },
    {
      "epoch": 0.4302261445118588,
      "grad_norm": 5.090203604118294,
      "learning_rate": 2.3e-06,
      "loss": 0.7661482810974121,
      "step": 390,
      "token_acc": 0.7836170737725142
    },
    {
      "epoch": 0.43574186431329287,
      "grad_norm": 5.270675577305129,
      "learning_rate": 2.3e-06,
      "loss": 0.7520760059356689,
      "step": 395,
      "token_acc": 0.7893167028199566
    },
    {
      "epoch": 0.441257584114727,
      "grad_norm": 5.4625302980431165,
      "learning_rate": 2.3e-06,
      "loss": 0.8551373481750488,
      "step": 400,
      "token_acc": 0.7630287450338864
    },
    {
      "epoch": 0.44677330391616105,
      "grad_norm": 5.249069454600804,
      "learning_rate": 2.3e-06,
      "loss": 0.910746955871582,
      "step": 405,
      "token_acc": 0.7516659262549977
    },
    {
      "epoch": 0.45228902371759516,
      "grad_norm": 5.236854652942137,
      "learning_rate": 2.3e-06,
      "loss": 0.8199291229248047,
      "step": 410,
      "token_acc": 0.7682183767974653
    },
    {
      "epoch": 0.4578047435190292,
      "grad_norm": 6.027185729170618,
      "learning_rate": 2.3e-06,
      "loss": 0.8450562477111816,
      "step": 415,
      "token_acc": 0.7607934655775963
    },
    {
      "epoch": 0.46332046332046334,
      "grad_norm": 4.870989267188852,
      "learning_rate": 2.3e-06,
      "loss": 0.9044759750366211,
      "step": 420,
      "token_acc": 0.7498462169366413
    },
    {
      "epoch": 0.4688361831218974,
      "grad_norm": 5.291951695042281,
      "learning_rate": 2.3e-06,
      "loss": 0.8467486381530762,
      "step": 425,
      "token_acc": 0.7576541612764123
    },
    {
      "epoch": 0.4743519029233315,
      "grad_norm": 5.62593903563829,
      "learning_rate": 2.3e-06,
      "loss": 0.8833629608154296,
      "step": 430,
      "token_acc": 0.7509285051067781
    },
    {
      "epoch": 0.4798676227247656,
      "grad_norm": 5.128799172526207,
      "learning_rate": 2.3e-06,
      "loss": 0.8988676071166992,
      "step": 435,
      "token_acc": 0.7490849938999593
    },
    {
      "epoch": 0.4853833425261997,
      "grad_norm": 5.239699648307152,
      "learning_rate": 2.3e-06,
      "loss": 0.7166192054748535,
      "step": 440,
      "token_acc": 0.7969401947148818
    },
    {
      "epoch": 0.49089906232763375,
      "grad_norm": 5.4681083200632905,
      "learning_rate": 2.3e-06,
      "loss": 0.9467007637023925,
      "step": 445,
      "token_acc": 0.7396415494314897
    },
    {
      "epoch": 0.49641478212906787,
      "grad_norm": 4.685989886939226,
      "learning_rate": 2.3e-06,
      "loss": 0.7727291584014893,
      "step": 450,
      "token_acc": 0.7793472966390648
    },
    {
      "epoch": 0.5019305019305019,
      "grad_norm": 5.538988406196133,
      "learning_rate": 2.3e-06,
      "loss": 0.8206657409667969,
      "step": 455,
      "token_acc": 0.7731393129770993
    },
    {
      "epoch": 0.507446221731936,
      "grad_norm": 4.63939429197835,
      "learning_rate": 2.3e-06,
      "loss": 0.8562620162963868,
      "step": 460,
      "token_acc": 0.7662946428571429
    },
    {
      "epoch": 0.5129619415333702,
      "grad_norm": 5.1240560151191445,
      "learning_rate": 2.3e-06,
      "loss": 0.7872621536254882,
      "step": 465,
      "token_acc": 0.7787717601547389
    },
    {
      "epoch": 0.5184776613348042,
      "grad_norm": 5.189600840792027,
      "learning_rate": 2.3e-06,
      "loss": 0.8223024368286133,
      "step": 470,
      "token_acc": 0.7725631768953068
    },
    {
      "epoch": 0.5239933811362383,
      "grad_norm": 4.625349854858132,
      "learning_rate": 2.3e-06,
      "loss": 0.7645779609680176,
      "step": 475,
      "token_acc": 0.7779833487511564
    },
    {
      "epoch": 0.5295091009376723,
      "grad_norm": 5.392461676265602,
      "learning_rate": 2.3e-06,
      "loss": 0.863577938079834,
      "step": 480,
      "token_acc": 0.7583280288563548
    },
    {
      "epoch": 0.5350248207391064,
      "grad_norm": 5.809387834675999,
      "learning_rate": 2.3e-06,
      "loss": 0.7966668128967285,
      "step": 485,
      "token_acc": 0.7756534365924492
    },
    {
      "epoch": 0.5405405405405406,
      "grad_norm": 5.304450067691333,
      "learning_rate": 2.3e-06,
      "loss": 0.9444230079650879,
      "step": 490,
      "token_acc": 0.7450345423143351
    },
    {
      "epoch": 0.5460562603419746,
      "grad_norm": 5.583804788469441,
      "learning_rate": 2.3e-06,
      "loss": 0.9105037689208985,
      "step": 495,
      "token_acc": 0.7439771335238873
    },
    {
      "epoch": 0.5515719801434087,
      "grad_norm": 6.054938244093431,
      "learning_rate": 2.3e-06,
      "loss": 0.9297395706176758,
      "step": 500,
      "token_acc": 0.7482532288799492
    },
    {
      "epoch": 0.5570876999448428,
      "grad_norm": 5.391503075184599,
      "learning_rate": 2.3e-06,
      "loss": 0.846977424621582,
      "step": 505,
      "token_acc": 0.7518848526387937
    },
    {
      "epoch": 0.5626034197462769,
      "grad_norm": 5.030658472043065,
      "learning_rate": 2.3e-06,
      "loss": 0.8991237640380859,
      "step": 510,
      "token_acc": 0.7539121114683816
    },
    {
      "epoch": 0.568119139547711,
      "grad_norm": 5.104958695838692,
      "learning_rate": 2.3e-06,
      "loss": 0.8426759719848633,
      "step": 515,
      "token_acc": 0.7644656228727025
    },
    {
      "epoch": 0.573634859349145,
      "grad_norm": 6.009882223493915,
      "learning_rate": 2.3e-06,
      "loss": 0.7240866661071778,
      "step": 520,
      "token_acc": 0.7948649371489703
    },
    {
      "epoch": 0.5791505791505791,
      "grad_norm": 5.637457445639113,
      "learning_rate": 2.3e-06,
      "loss": 0.8636373519897461,
      "step": 525,
      "token_acc": 0.7649465787679018
    },
    {
      "epoch": 0.5846662989520133,
      "grad_norm": 5.3912830874814555,
      "learning_rate": 2.3e-06,
      "loss": 0.8545609474182129,
      "step": 530,
      "token_acc": 0.7600548446069469
    },
    {
      "epoch": 0.5901820187534473,
      "grad_norm": 5.223533669476021,
      "learning_rate": 2.3e-06,
      "loss": 0.7941525936126709,
      "step": 535,
      "token_acc": 0.7761698022190062
    },
    {
      "epoch": 0.5956977385548814,
      "grad_norm": 5.115121860566828,
      "learning_rate": 2.3e-06,
      "loss": 0.8650318145751953,
      "step": 540,
      "token_acc": 0.7646005817856344
    },
    {
      "epoch": 0.6012134583563155,
      "grad_norm": 4.715835836641467,
      "learning_rate": 2.3e-06,
      "loss": 0.8254315376281738,
      "step": 545,
      "token_acc": 0.7698187377156585
    },
    {
      "epoch": 0.6067291781577496,
      "grad_norm": 5.200096789536537,
      "learning_rate": 2.3e-06,
      "loss": 0.7865041255950928,
      "step": 550,
      "token_acc": 0.7788415509813308
    },
    {
      "epoch": 0.6122448979591837,
      "grad_norm": 5.62598549880936,
      "learning_rate": 2.3e-06,
      "loss": 0.9002869606018067,
      "step": 555,
      "token_acc": 0.7615146147032772
    },
    {
      "epoch": 0.6177606177606177,
      "grad_norm": 6.376332687516541,
      "learning_rate": 2.3e-06,
      "loss": 0.8603617668151855,
      "step": 560,
      "token_acc": 0.758905299739357
    },
    {
      "epoch": 0.6232763375620518,
      "grad_norm": 5.370356585251875,
      "learning_rate": 2.3e-06,
      "loss": 0.8701972961425781,
      "step": 565,
      "token_acc": 0.7542935653766888
    },
    {
      "epoch": 0.628792057363486,
      "grad_norm": 4.884591590707582,
      "learning_rate": 2.3e-06,
      "loss": 0.840721321105957,
      "step": 570,
      "token_acc": 0.7689443714410863
    },
    {
      "epoch": 0.63430777716492,
      "grad_norm": 5.138799986301503,
      "learning_rate": 2.3e-06,
      "loss": 0.7376739501953125,
      "step": 575,
      "token_acc": 0.7904834996162702
    },
    {
      "epoch": 0.6398234969663541,
      "grad_norm": 5.3450150515914885,
      "learning_rate": 2.3e-06,
      "loss": 0.755347728729248,
      "step": 580,
      "token_acc": 0.7790931989924433
    },
    {
      "epoch": 0.6453392167677882,
      "grad_norm": 4.895436117128604,
      "learning_rate": 2.3e-06,
      "loss": 0.8265247344970703,
      "step": 585,
      "token_acc": 0.7729456991830851
    },
    {
      "epoch": 0.6508549365692223,
      "grad_norm": 5.143509534036041,
      "learning_rate": 2.3e-06,
      "loss": 0.8215790748596191,
      "step": 590,
      "token_acc": 0.7608391608391608
    },
    {
      "epoch": 0.6563706563706564,
      "grad_norm": 5.082286518166302,
      "learning_rate": 2.3e-06,
      "loss": 0.8152825355529785,
      "step": 595,
      "token_acc": 0.7685413399058929
    },
    {
      "epoch": 0.6618863761720905,
      "grad_norm": 5.326424315651092,
      "learning_rate": 2.3e-06,
      "loss": 0.8500395774841308,
      "step": 600,
      "token_acc": 0.7704613957801993
    }
  ],
  "logging_steps": 5,
  "max_steps": 907,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 300,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1514130748997632.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}