{
  "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": 30.968734741210938,
      "learning_rate": 2.9850746268656716e-07,
      "loss": 0.8261417150497437,
      "step": 1,
      "token_acc": 0.8618925831202046
    },
    {
      "epoch": 0.015082956259426848,
      "grad_norm": 9.643964767456055,
      "learning_rate": 2.9850746268656716e-06,
      "loss": 0.6247194078233507,
      "step": 10,
      "token_acc": 0.8784715183458341
    },
    {
      "epoch": 0.030165912518853696,
      "grad_norm": 1.2653270959854126,
      "learning_rate": 5.970149253731343e-06,
      "loss": 0.21320788860321044,
      "step": 20,
      "token_acc": 0.9200478773370504
    },
    {
      "epoch": 0.04524886877828054,
      "grad_norm": 1.6828762292861938,
      "learning_rate": 8.955223880597016e-06,
      "loss": 0.1644657850265503,
      "step": 30,
      "token_acc": 0.9310746927198235
    },
    {
      "epoch": 0.06033182503770739,
      "grad_norm": 1.3016011714935303,
      "learning_rate": 1.1940298507462686e-05,
      "loss": 0.1581899881362915,
      "step": 40,
      "token_acc": 0.9363424769030858
    },
    {
      "epoch": 0.07541478129713423,
      "grad_norm": 1.772908329963684,
      "learning_rate": 1.492537313432836e-05,
      "loss": 0.14797013998031616,
      "step": 50,
      "token_acc": 0.9403096585489922
    },
    {
      "epoch": 0.09049773755656108,
      "grad_norm": 1.0607707500457764,
      "learning_rate": 1.791044776119403e-05,
      "loss": 0.14336094856262208,
      "step": 60,
      "token_acc": 0.94302149950188
    },
    {
      "epoch": 0.10558069381598793,
      "grad_norm": 0.8757421970367432,
      "learning_rate": 1.999971980603894e-05,
      "loss": 0.14081008434295655,
      "step": 70,
      "token_acc": 0.9429653187944873
    },
    {
      "epoch": 0.12066365007541478,
      "grad_norm": 0.9281443357467651,
      "learning_rate": 1.999473901685785e-05,
      "loss": 0.1406252861022949,
      "step": 80,
      "token_acc": 0.9438380253798261
    },
    {
      "epoch": 0.13574660633484162,
      "grad_norm": 0.8397683501243591,
      "learning_rate": 1.9983535264822027e-05,
      "loss": 0.13869004249572753,
      "step": 90,
      "token_acc": 0.9443744563424218
    },
    {
      "epoch": 0.15082956259426847,
      "grad_norm": 0.9427488446235657,
      "learning_rate": 1.9966115525654667e-05,
      "loss": 0.13446314334869386,
      "step": 100,
      "token_acc": 0.9466490881125027
    },
    {
      "epoch": 0.16591251885369532,
      "grad_norm": 0.8245936036109924,
      "learning_rate": 1.9942490645300782e-05,
      "loss": 0.13205068111419677,
      "step": 110,
      "token_acc": 0.9481581044965068
    },
    {
      "epoch": 0.18099547511312217,
      "grad_norm": 1.0142247676849365,
      "learning_rate": 1.9912675333174255e-05,
      "loss": 0.1309139370918274,
      "step": 120,
      "token_acc": 0.9484835961342828
    },
    {
      "epoch": 0.19607843137254902,
      "grad_norm": 1.0399980545043945,
      "learning_rate": 1.987668815299942e-05,
      "loss": 0.13243772983551025,
      "step": 130,
      "token_acc": 0.9475406813257733
    },
    {
      "epoch": 0.21116138763197587,
      "grad_norm": 0.825975775718689,
      "learning_rate": 1.9834551511252828e-05,
      "loss": 0.13408410549163818,
      "step": 140,
      "token_acc": 0.9473418398419512
    },
    {
      "epoch": 0.22624434389140272,
      "grad_norm": 0.7565873265266418,
      "learning_rate": 1.9786291643212466e-05,
      "loss": 0.12777891159057617,
      "step": 150,
      "token_acc": 0.949374481295915
    },
    {
      "epoch": 0.24132730015082957,
      "grad_norm": 0.7258040904998779,
      "learning_rate": 1.9731938596623012e-05,
      "loss": 0.12959227561950684,
      "step": 160,
      "token_acc": 0.9500562361309542
    },
    {
      "epoch": 0.2564102564102564,
      "grad_norm": 0.5646476149559021,
      "learning_rate": 1.9671526212987425e-05,
      "loss": 0.12599967718124389,
      "step": 170,
      "token_acc": 0.9519809753197457
    },
    {
      "epoch": 0.27149321266968324,
      "grad_norm": 0.871751606464386,
      "learning_rate": 1.9605092106496397e-05,
      "loss": 0.1252290725708008,
      "step": 180,
      "token_acc": 0.9504295106136511
    },
    {
      "epoch": 0.2865761689291101,
      "grad_norm": 0.9698847532272339,
      "learning_rate": 1.9532677640608874e-05,
      "loss": 0.12251553535461426,
      "step": 190,
      "token_acc": 0.9512455846811675
    },
    {
      "epoch": 0.30165912518853694,
      "grad_norm": 0.7135357856750488,
      "learning_rate": 1.9454327902298176e-05,
      "loss": 0.1227449893951416,
      "step": 200,
      "token_acc": 0.9516792651420917
    },
    {
      "epoch": 0.3167420814479638,
      "grad_norm": 1.152921199798584,
      "learning_rate": 1.9370091673979764e-05,
      "loss": 0.12003127336502076,
      "step": 210,
      "token_acc": 0.9536876879684603
    },
    {
      "epoch": 0.33182503770739064,
      "grad_norm": 0.7168576717376709,
      "learning_rate": 1.9280021403138154e-05,
      "loss": 0.13734824657440187,
      "step": 220,
      "token_acc": 0.9508574589710492
    },
    {
      "epoch": 0.3469079939668175,
      "grad_norm": 0.9696043729782104,
      "learning_rate": 1.9184173169671868e-05,
      "loss": 0.12161376476287841,
      "step": 230,
      "token_acc": 0.9523213065686307
    },
    {
      "epoch": 0.36199095022624433,
      "grad_norm": 0.6335675120353699,
      "learning_rate": 1.9082606650976735e-05,
      "loss": 0.1165058970451355,
      "step": 240,
      "token_acc": 0.9543600729984925
    },
    {
      "epoch": 0.3770739064856712,
      "grad_norm": 0.5023436546325684,
      "learning_rate": 1.8975385084789333e-05,
      "loss": 0.11858338117599487,
      "step": 250,
      "token_acc": 0.9543472601848236
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 0.8230804800987244,
      "learning_rate": 1.886257522981367e-05,
      "loss": 0.11891791820526124,
      "step": 260,
      "token_acc": 0.9533092044993767
    },
    {
      "epoch": 0.4072398190045249,
      "grad_norm": 0.7313021421432495,
      "learning_rate": 1.8744247324155604e-05,
      "loss": 0.11976690292358398,
      "step": 270,
      "token_acc": 0.9534955965269256
    },
    {
      "epoch": 0.42232277526395173,
      "grad_norm": 0.6101995706558228,
      "learning_rate": 1.8620475041590948e-05,
      "loss": 0.11677769422531128,
      "step": 280,
      "token_acc": 0.9540437337788252
    },
    {
      "epoch": 0.4374057315233786,
      "grad_norm": 0.6487584114074707,
      "learning_rate": 1.8491335445694404e-05,
      "loss": 0.11501857042312622,
      "step": 290,
      "token_acc": 0.95502744006087
    },
    {
      "epoch": 0.45248868778280543,
      "grad_norm": 0.6374943256378174,
      "learning_rate": 1.8356908941857942e-05,
      "loss": 0.12021658420562745,
      "step": 300,
      "token_acc": 0.9528524628869934
    },
    {
      "epoch": 0.4675716440422323,
      "grad_norm": 0.744682788848877,
      "learning_rate": 1.8217279227228503e-05,
      "loss": 0.1173781156539917,
      "step": 310,
      "token_acc": 0.9543975427285445
    },
    {
      "epoch": 0.48265460030165913,
      "grad_norm": 0.7849182486534119,
      "learning_rate": 1.8072533238596134e-05,
      "loss": 0.11671500205993653,
      "step": 320,
      "token_acc": 0.9545516831864066
    },
    {
      "epoch": 0.497737556561086,
      "grad_norm": 0.42268458008766174,
      "learning_rate": 1.7922761098265096e-05,
      "loss": 0.11766018867492675,
      "step": 330,
      "token_acc": 0.9539052876838372
    },
    {
      "epoch": 0.5128205128205128,
      "grad_norm": 0.6217177510261536,
      "learning_rate": 1.7768056057941546e-05,
      "loss": 0.11545138359069824,
      "step": 340,
      "token_acc": 0.9544162710572694
    },
    {
      "epoch": 0.5279034690799397,
      "grad_norm": 0.688858151435852,
      "learning_rate": 1.760851444067282e-05,
      "loss": 0.11715751886367798,
      "step": 350,
      "token_acc": 0.9544157346702661
    },
    {
      "epoch": 0.5429864253393665,
      "grad_norm": 0.570622444152832,
      "learning_rate": 1.7444235580874382e-05,
      "loss": 0.11272256374359131,
      "step": 360,
      "token_acc": 0.9569494714351771
    },
    {
      "epoch": 0.5580693815987934,
      "grad_norm": 0.43587830662727356,
      "learning_rate": 1.727532176248188e-05,
      "loss": 0.1147916078567505,
      "step": 370,
      "token_acc": 0.9553941908713693
    },
    {
      "epoch": 0.5731523378582202,
      "grad_norm": 0.39815858006477356,
      "learning_rate": 1.7101878155266702e-05,
      "loss": 0.11624550819396973,
      "step": 380,
      "token_acc": 0.9544733044733045
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.5202909708023071,
      "learning_rate": 1.692401274935478e-05,
      "loss": 0.11601772308349609,
      "step": 390,
      "token_acc": 0.9548668395003536
    },
    {
      "epoch": 0.6033182503770739,
      "grad_norm": 0.6442638039588928,
      "learning_rate": 1.674183628798936e-05,
      "loss": 0.1141278862953186,
      "step": 400,
      "token_acc": 0.9565292910153166
    },
    {
      "epoch": 0.6184012066365008,
      "grad_norm": 0.5705167651176453,
      "learning_rate": 1.655546219857962e-05,
      "loss": 0.11343185901641846,
      "step": 410,
      "token_acc": 0.955780920763412
    },
    {
      "epoch": 0.6334841628959276,
      "grad_norm": 0.7077851295471191,
      "learning_rate": 1.6365006522078057e-05,
      "loss": 0.11292238235473633,
      "step": 420,
      "token_acc": 0.956367147389232
    },
    {
      "epoch": 0.6485671191553545,
      "grad_norm": 0.7096773386001587,
      "learning_rate": 1.6170587840730627e-05,
      "loss": 0.11361818313598633,
      "step": 430,
      "token_acc": 0.9554244309583136
    },
    {
      "epoch": 0.6636500754147813,
      "grad_norm": 0.6647536754608154,
      "learning_rate": 1.59723272042446e-05,
      "loss": 0.1104124665260315,
      "step": 440,
      "token_acc": 0.9571542325125567
    },
    {
      "epoch": 0.6787330316742082,
      "grad_norm": 0.532390296459198,
      "learning_rate": 1.5770348054420147e-05,
      "loss": 0.10935736894607544,
      "step": 450,
      "token_acc": 0.9568433662773286
    },
    {
      "epoch": 0.693815987933635,
      "grad_norm": 0.6258285045623779,
      "learning_rate": 1.556477614829251e-05,
      "loss": 0.10925512313842774,
      "step": 460,
      "token_acc": 0.9575476813907008
    },
    {
      "epoch": 0.7088989441930619,
      "grad_norm": 0.6589264273643494,
      "learning_rate": 1.5355739479832675e-05,
      "loss": 0.11245393753051758,
      "step": 470,
      "token_acc": 0.9563331882774868
    },
    {
      "epoch": 0.7239819004524887,
      "grad_norm": 0.8117122650146484,
      "learning_rate": 1.5143368200255294e-05,
      "loss": 0.10972108840942382,
      "step": 480,
      "token_acc": 0.9572499104179975
    },
    {
      "epoch": 0.7390648567119156,
      "grad_norm": 0.6422977447509766,
      "learning_rate": 1.4927794536983411e-05,
      "loss": 0.1119101881980896,
      "step": 490,
      "token_acc": 0.9557822601900805
    },
    {
      "epoch": 0.7541478129713424,
      "grad_norm": 0.5782728791236877,
      "learning_rate": 1.4709152711320529e-05,
      "loss": 0.11008884906768798,
      "step": 500,
      "token_acc": 0.9574936263202622
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 0.6354721188545227,
      "learning_rate": 1.4487578854881217e-05,
      "loss": 0.10828690528869629,
      "step": 510,
      "token_acc": 0.9580254531430775
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 0.5012155771255493,
      "learning_rate": 1.4263210924832332e-05,
      "loss": 0.10970515012741089,
      "step": 520,
      "token_acc": 0.957624905926984
    },
    {
      "epoch": 0.799396681749623,
      "grad_norm": 0.5928862690925598,
      "learning_rate": 1.4036188617997581e-05,
      "loss": 0.1113357424736023,
      "step": 530,
      "token_acc": 0.9566905792375645
    },
    {
      "epoch": 0.8144796380090498,
      "grad_norm": 0.43102213740348816,
      "learning_rate": 1.3806653283878963e-05,
      "loss": 0.107137131690979,
      "step": 540,
      "token_acc": 0.9579867607981257
    },
    {
      "epoch": 0.8295625942684767,
      "grad_norm": 0.417781800031662,
      "learning_rate": 1.3574747836649197e-05,
      "loss": 0.10823215246200561,
      "step": 550,
      "token_acc": 0.9582748004901814
    },
    {
      "epoch": 0.8446455505279035,
      "grad_norm": 0.8108027577400208,
      "learning_rate": 1.3340616666169925e-05,
      "loss": 0.10481956005096435,
      "step": 560,
      "token_acc": 0.959131087119606
    },
    {
      "epoch": 0.8597285067873304,
      "grad_norm": 0.5911276340484619,
      "learning_rate": 1.3104405548091188e-05,
      "loss": 0.108212411403656,
      "step": 570,
      "token_acc": 0.9576892430278885
    },
    {
      "epoch": 0.8748114630467572,
      "grad_norm": 0.5372975468635559,
      "learning_rate": 1.2866261553087977e-05,
      "loss": 0.10410468578338623,
      "step": 580,
      "token_acc": 0.9592422768746237
    },
    {
      "epoch": 0.889894419306184,
      "grad_norm": 0.5496704578399658,
      "learning_rate": 1.2626332955290564e-05,
      "loss": 0.10734721422195434,
      "step": 590,
      "token_acc": 0.9583501505790307
    },
    {
      "epoch": 0.9049773755656109,
      "grad_norm": 0.4394945800304413,
      "learning_rate": 1.2384769139965476e-05,
      "loss": 0.10930960178375244,
      "step": 600,
      "token_acc": 0.9577309571799371
    },
    {
      "epoch": 0.9200603318250377,
      "grad_norm": 0.642622172832489,
      "learning_rate": 1.2141720510504662e-05,
      "loss": 0.10778934955596924,
      "step": 610,
      "token_acc": 0.9577198020406916
    },
    {
      "epoch": 0.9351432880844646,
      "grad_norm": 0.5836299657821655,
      "learning_rate": 1.1897338394780762e-05,
      "loss": 0.10450876951217651,
      "step": 620,
      "token_acc": 0.9592342342342343
    },
    {
      "epoch": 0.9502262443438914,
      "grad_norm": 0.37644606828689575,
      "learning_rate": 1.165177495092677e-05,
      "loss": 0.10912293195724487,
      "step": 630,
      "token_acc": 0.9573103488304603
    },
    {
      "epoch": 0.9653092006033183,
      "grad_norm": 0.5391575694084167,
      "learning_rate": 1.1405183072598735e-05,
      "loss": 0.10322842597961426,
      "step": 640,
      "token_acc": 0.9595477232115349
    },
    {
      "epoch": 0.9803921568627451,
      "grad_norm": 0.5731673240661621,
      "learning_rate": 1.115771629378054e-05,
      "loss": 0.10575449466705322,
      "step": 650,
      "token_acc": 0.9590387340944134
    },
    {
      "epoch": 0.995475113122172,
      "grad_norm": 0.4750414192676544,
      "learning_rate": 1.0909528693189974e-05,
      "loss": 0.10648643970489502,
      "step": 660,
      "token_acc": 0.9586862250888426
    },
    {
      "epoch": 1.0105580693815988,
      "grad_norm": 0.44234615564346313,
      "learning_rate": 1.0660774798345655e-05,
      "loss": 0.10033718347549439,
      "step": 670,
      "token_acc": 0.9616517345072387
    },
    {
      "epoch": 1.0256410256410255,
      "grad_norm": 0.4087628722190857,
      "learning_rate": 1.0411609489354495e-05,
      "loss": 0.09711532592773438,
      "step": 680,
      "token_acc": 0.9627995992882882
    },
    {
      "epoch": 1.0407239819004526,
      "grad_norm": 0.873450517654419,
      "learning_rate": 1.016218790247969e-05,
      "loss": 0.09997568130493165,
      "step": 690,
      "token_acc": 0.960326375237917
    },
    {
      "epoch": 1.0558069381598794,
      "grad_norm": 0.6006814241409302,
      "learning_rate": 9.91266533354913e-06,
      "loss": 0.09607951045036316,
      "step": 700,
      "token_acc": 0.962045793927327
    },
    {
      "epoch": 1.0708898944193062,
      "grad_norm": 0.5565671324729919,
      "learning_rate": 9.663197141264556e-06,
      "loss": 0.09508470296859742,
      "step": 710,
      "token_acc": 0.9628928711804696
    },
    {
      "epoch": 1.085972850678733,
      "grad_norm": 0.4465359151363373,
      "learning_rate": 9.413938650471515e-06,
      "loss": 0.09565139412879944,
      "step": 720,
      "token_acc": 0.9632787942290643
    },
    {
      "epoch": 1.10105580693816,
      "grad_norm": 0.5148073434829712,
      "learning_rate": 9.165045055450393e-06,
      "loss": 0.09551175832748413,
      "step": 730,
      "token_acc": 0.9626595193859836
    },
    {
      "epoch": 1.1161387631975868,
      "grad_norm": 0.526919960975647,
      "learning_rate": 8.916671323288777e-06,
      "loss": 0.09376778602600097,
      "step": 740,
      "token_acc": 0.9630099437202928
    },
    {
      "epoch": 1.1312217194570136,
      "grad_norm": 0.515141487121582,
      "learning_rate": 8.66897209739523e-06,
      "loss": 0.0931336224079132,
      "step": 750,
      "token_acc": 0.9638066154589221
    },
    {
      "epoch": 1.1463046757164403,
      "grad_norm": 0.40961524844169617,
      "learning_rate": 8.422101601214643e-06,
      "loss": 0.09019145965576172,
      "step": 760,
      "token_acc": 0.9654217212723278
    },
    {
      "epoch": 1.1613876319758674,
      "grad_norm": 0.498300701379776,
      "learning_rate": 8.176213542205016e-06,
      "loss": 0.09666601419448853,
      "step": 770,
      "token_acc": 0.9628490978048501
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 0.634801983833313,
      "learning_rate": 7.931461016135534e-06,
      "loss": 0.09509063959121704,
      "step": 780,
      "token_acc": 0.9631919905771495
    },
    {
      "epoch": 1.191553544494721,
      "grad_norm": 0.5501129627227783,
      "learning_rate": 7.687996411765503e-06,
      "loss": 0.09726064801216125,
      "step": 790,
      "token_acc": 0.9615822735429508
    },
    {
      "epoch": 1.2066365007541477,
      "grad_norm": 0.40932971239089966,
      "learning_rate": 7.445971315963459e-06,
      "loss": 0.09402073621749878,
      "step": 800,
      "token_acc": 0.9623268010364785
    },
    {
      "epoch": 1.2217194570135748,
      "grad_norm": 0.4873681962490082,
      "learning_rate": 7.205536419325581e-06,
      "loss": 0.0950639545917511,
      "step": 810,
      "token_acc": 0.9637483110183024
    },
    {
      "epoch": 1.2368024132730016,
      "grad_norm": 0.664931058883667,
      "learning_rate": 6.966841422352127e-06,
      "loss": 0.0921372652053833,
      "step": 820,
      "token_acc": 0.9640317224437807
    },
    {
      "epoch": 1.2518853695324283,
      "grad_norm": 0.43320387601852417,
      "learning_rate": 6.7300349422403344e-06,
      "loss": 0.09601188898086548,
      "step": 830,
      "token_acc": 0.9622773264928032
    },
    {
      "epoch": 1.2669683257918551,
      "grad_norm": 0.48374417424201965,
      "learning_rate": 6.495264420351799e-06,
      "loss": 0.09131983518600464,
      "step": 840,
      "token_acc": 0.9639983480254632
    },
    {
      "epoch": 1.282051282051282,
      "grad_norm": 0.43359142541885376,
      "learning_rate": 6.262676030411985e-06,
      "loss": 0.09142310619354248,
      "step": 850,
      "token_acc": 0.9637676869338702
    },
    {
      "epoch": 1.297134238310709,
      "grad_norm": 0.6854210495948792,
      "learning_rate": 6.0324145874989606e-06,
      "loss": 0.0924448013305664,
      "step": 860,
      "token_acc": 0.9636747201305801
    },
    {
      "epoch": 1.3122171945701357,
      "grad_norm": 0.4733792841434479,
      "learning_rate": 5.804623457878088e-06,
      "loss": 0.09350036382675171,
      "step": 870,
      "token_acc": 0.9638449071188131
    },
    {
      "epoch": 1.3273001508295625,
      "grad_norm": 0.4248761832714081,
      "learning_rate": 5.579444469738771e-06,
      "loss": 0.09371748566627502,
      "step": 880,
      "token_acc": 0.963402983234714
    },
    {
      "epoch": 1.3423831070889896,
      "grad_norm": 0.3817572593688965,
      "learning_rate": 5.357017824888831e-06,
      "loss": 0.08855417370796204,
      "step": 890,
      "token_acc": 0.965276182349273
    },
    {
      "epoch": 1.3574660633484164,
      "grad_norm": 0.5116860270500183,
      "learning_rate": 5.137482011461513e-06,
      "loss": 0.09090266823768615,
      "step": 900,
      "token_acc": 0.964496366539577
    },
    {
      "epoch": 1.3725490196078431,
      "grad_norm": 0.4829014539718628,
      "learning_rate": 4.920973717689495e-06,
      "loss": 0.0917883038520813,
      "step": 910,
      "token_acc": 0.9649694993054297
    },
    {
      "epoch": 1.38763197586727,
      "grad_norm": 0.5397462248802185,
      "learning_rate": 4.707627746799517e-06,
      "loss": 0.09063017964363099,
      "step": 920,
      "token_acc": 0.9644550868409186
    },
    {
      "epoch": 1.4027149321266967,
      "grad_norm": 0.446719229221344,
      "learning_rate": 4.497576933080677e-06,
      "loss": 0.09249653816223144,
      "step": 930,
      "token_acc": 0.963393705372192
    },
    {
      "epoch": 1.4177978883861238,
      "grad_norm": 0.38629722595214844,
      "learning_rate": 4.290952059178644e-06,
      "loss": 0.09290457963943481,
      "step": 940,
      "token_acc": 0.9633827640695272
    },
    {
      "epoch": 1.4328808446455505,
      "grad_norm": 0.5298126339912415,
      "learning_rate": 4.087881774667258e-06,
      "loss": 0.09557251930236817,
      "step": 950,
      "token_acc": 0.9631715835624582
    },
    {
      "epoch": 1.4479638009049773,
      "grad_norm": 0.3959716856479645,
      "learning_rate": 3.888492515948241e-06,
      "loss": 0.09034661054611207,
      "step": 960,
      "token_acc": 0.964851074293794
    },
    {
      "epoch": 1.4630467571644044,
      "grad_norm": 0.39855247735977173,
      "learning_rate": 3.692908427528866e-06,
      "loss": 0.08885170221328735,
      "step": 970,
      "token_acc": 0.9658355310477104
    },
    {
      "epoch": 1.4781297134238311,
      "grad_norm": 0.38074812293052673,
      "learning_rate": 3.5012512847266435e-06,
      "loss": 0.08835339546203613,
      "step": 980,
      "token_acc": 0.9653404148161714
    },
    {
      "epoch": 1.493212669683258,
      "grad_norm": 0.46875646710395813,
      "learning_rate": 3.313640417849107e-06,
      "loss": 0.09411121010780335,
      "step": 990,
      "token_acc": 0.963810409758586
    },
    {
      "epoch": 1.5082956259426847,
      "grad_norm": 0.540631890296936,
      "learning_rate": 3.130192637895888e-06,
      "loss": 0.09138602018356323,
      "step": 1000,
      "token_acc": 0.9636740152724449
    },
    {
      "epoch": 1.5233785822021115,
      "grad_norm": 0.5243298411369324,
      "learning_rate": 2.9510221638294378e-06,
      "loss": 0.09415959119796753,
      "step": 1010,
      "token_acc": 0.9632957660676211
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 0.4213015139102936,
      "learning_rate": 2.776240551459558e-06,
      "loss": 0.08891291618347168,
      "step": 1020,
      "token_acc": 0.9648928840391611
    },
    {
      "epoch": 1.5535444947209653,
      "grad_norm": 0.46009889245033264,
      "learning_rate": 2.6059566239861e-06,
      "loss": 0.08925861120223999,
      "step": 1030,
      "token_acc": 0.9647288759601836
    },
    {
      "epoch": 1.5686274509803921,
      "grad_norm": 0.48863378167152405,
      "learning_rate": 2.4402764042430317e-06,
      "loss": 0.08919482231140137,
      "step": 1040,
      "token_acc": 0.9651635934592869
    },
    {
      "epoch": 1.5837104072398192,
      "grad_norm": 0.535592257976532,
      "learning_rate": 2.2793030486861212e-06,
      "loss": 0.08794163465499878,
      "step": 1050,
      "token_acc": 0.9659154613745936
    },
    {
      "epoch": 1.598793363499246,
      "grad_norm": 0.49713340401649475,
      "learning_rate": 2.1231367831652415e-06,
      "loss": 0.08887355327606201,
      "step": 1060,
      "token_acc": 0.9647955384444762
    },
    {
      "epoch": 1.6138763197586727,
      "grad_norm": 0.3400653898715973,
      "learning_rate": 1.971874840521374e-06,
      "loss": 0.09162210822105407,
      "step": 1070,
      "token_acc": 0.96406639508725
    },
    {
      "epoch": 1.6289592760180995,
      "grad_norm": 0.44764232635498047,
      "learning_rate": 1.825611400047126e-06,
      "loss": 0.09081205129623413,
      "step": 1080,
      "token_acc": 0.9648421603395152
    },
    {
      "epoch": 1.6440422322775263,
      "grad_norm": 0.3386666178703308,
      "learning_rate": 1.684437528848456e-06,
      "loss": 0.08757848739624023,
      "step": 1090,
      "token_acc": 0.965227225664633
    },
    {
      "epoch": 1.6591251885369531,
      "grad_norm": 0.41909298300743103,
      "learning_rate": 1.5484411251441234e-06,
      "loss": 0.08880707025527954,
      "step": 1100,
      "token_acc": 0.9646852985864911
    },
    {
      "epoch": 1.6742081447963801,
      "grad_norm": 0.4352640211582184,
      "learning_rate": 1.4177068635381575e-06,
      "loss": 0.08535876274108886,
      "step": 1110,
      "token_acc": 0.9665068649512442
    },
    {
      "epoch": 1.689291101055807,
      "grad_norm": 0.3794000744819641,
      "learning_rate": 1.2923161422994302e-06,
      "loss": 0.09240926504135132,
      "step": 1120,
      "token_acc": 0.9636572421712548
    },
    {
      "epoch": 1.704374057315234,
      "grad_norm": 0.35188183188438416,
      "learning_rate": 1.1723470326811582e-06,
      "loss": 0.08627489805221558,
      "step": 1130,
      "token_acc": 0.9659923592409344
    },
    {
      "epoch": 1.7194570135746607,
      "grad_norm": 0.446426123380661,
      "learning_rate": 1.0578742303118695e-06,
      "loss": 0.09179290533065795,
      "step": 1140,
      "token_acc": 0.9638420618865663
    },
    {
      "epoch": 1.7345399698340875,
      "grad_norm": 0.3761591613292694,
      "learning_rate": 9.489690086881321e-07,
      "loss": 0.08692222833633423,
      "step": 1150,
      "token_acc": 0.9657407263152975
    },
    {
      "epoch": 1.7496229260935143,
      "grad_norm": 0.49052301049232483,
      "learning_rate": 8.456991747979782e-07,
      "loss": 0.089877450466156,
      "step": 1160,
      "token_acc": 0.9652767450852096
    },
    {
      "epoch": 1.7647058823529411,
      "grad_norm": 0.41365307569503784,
      "learning_rate": 7.48129026902642e-07,
      "loss": 0.08901469707489014,
      "step": 1170,
      "token_acc": 0.9654412454281159
    },
    {
      "epoch": 1.779788838612368,
      "grad_norm": 0.4520035982131958,
      "learning_rate": 6.563193145029423e-07,
      "loss": 0.09040895104408264,
      "step": 1180,
      "token_acc": 0.9646702238499668
    },
    {
      "epoch": 1.7948717948717947,
      "grad_norm": 0.4278804361820221,
      "learning_rate": 5.703272005151905e-07,
      "loss": 0.08340400457382202,
      "step": 1190,
      "token_acc": 0.9672215853126866
    },
    {
      "epoch": 1.8099547511312217,
      "grad_norm": 0.43313461542129517,
      "learning_rate": 4.902062256801954e-07,
      "loss": 0.08683132529258727,
      "step": 1200,
      "token_acc": 0.9662295699093391
    },
    {
      "epoch": 1.8250377073906485,
      "grad_norm": 0.32996904850006104,
      "learning_rate": 4.1600627522751624e-07,
      "loss": 0.08770017027854919,
      "step": 1210,
      "token_acc": 0.9661082315191905
    },
    {
      "epoch": 1.8401206636500755,
      "grad_norm": 0.4273052513599396,
      "learning_rate": 3.477735478157429e-07,
      "loss": 0.08336774706840515,
      "step": 1220,
      "token_acc": 0.9679050340734228
    },
    {
      "epoch": 1.8552036199095023,
      "grad_norm": 0.3404141068458557,
      "learning_rate": 2.855505267680947e-07,
      "loss": 0.09096280932426452,
      "step": 1230,
      "token_acc": 0.9640440692174121
    },
    {
      "epoch": 1.8702865761689291,
      "grad_norm": 0.4492412805557251,
      "learning_rate": 2.2937595362129805e-07,
      "loss": 0.08709906935691833,
      "step": 1240,
      "token_acc": 0.9661851711983888
    },
    {
      "epoch": 1.885369532428356,
      "grad_norm": 0.39498382806777954,
      "learning_rate": 1.7928480400417837e-07,
      "loss": 0.08509960174560546,
      "step": 1250,
      "token_acc": 0.9664054460539856
    },
    {
      "epoch": 1.9004524886877827,
      "grad_norm": 0.4895835518836975,
      "learning_rate": 1.3530826586100276e-07,
      "loss": 0.0855100154876709,
      "step": 1260,
      "token_acc": 0.9667
    },
    {
      "epoch": 1.9155354449472095,
      "grad_norm": 0.37063419818878174,
      "learning_rate": 9.747372003312106e-08,
      "loss": 0.08988954424858094,
      "step": 1270,
      "token_acc": 0.9651437057744388
    },
    {
      "epoch": 1.9306184012066365,
      "grad_norm": 0.37928536534309387,
      "learning_rate": 6.580472321099928e-08,
      "loss": 0.09078261852264405,
      "step": 1280,
      "token_acc": 0.9638408813108261
    },
    {
      "epoch": 1.9457013574660633,
      "grad_norm": 0.4147525429725647,
      "learning_rate": 4.0320993267264354e-08,
      "loss": 0.08869935870170594,
      "step": 1290,
      "token_acc": 0.9655277072251242
    },
    {
      "epoch": 1.9607843137254903,
      "grad_norm": 0.39682766795158386,
      "learning_rate": 2.1038396979884414e-08,
      "loss": 0.08754374980926513,
      "step": 1300,
      "token_acc": 0.9651909735342573
    },
    {
      "epoch": 1.9758672699849171,
      "grad_norm": 0.39889928698539734,
      "learning_rate": 7.968940153136872e-09,
      "loss": 0.0862058699131012,
      "step": 1310,
      "token_acc": 0.9660498767487274
    },
    {
      "epoch": 1.990950226244344,
      "grad_norm": 0.38509562611579895,
      "learning_rate": 1.1207601425078375e-09,
      "loss": 0.08868761658668518,
      "step": 1320,
      "token_acc": 0.9653918684468583
    }
  ],
  "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": 7.840189747375899e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}