{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 200.0,
  "global_step": 1641,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006093845216331506,
      "grad_norm": 5.139511599456721,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.97658371925354,
      "step": 1,
      "token_acc": 0.752757254524039
    },
    {
      "epoch": 0.006093845216331505,
      "grad_norm": 4.02627916602829,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.979477776421441,
      "step": 10,
      "token_acc": 0.7513394629500156
    },
    {
      "epoch": 0.01218769043266301,
      "grad_norm": 2.133243665962445,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.8085936546325684,
      "step": 20,
      "token_acc": 0.7841330046869243
    },
    {
      "epoch": 0.018281535648994516,
      "grad_norm": 1.6130698467539089,
      "learning_rate": 6e-06,
      "loss": 0.6815763473510742,
      "step": 30,
      "token_acc": 0.8088598674958689
    },
    {
      "epoch": 0.02437538086532602,
      "grad_norm": 1.5958993846193437,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.6228148937225342,
      "step": 40,
      "token_acc": 0.820709892041175
    },
    {
      "epoch": 0.030469226081657527,
      "grad_norm": 1.5835525047156958,
      "learning_rate": 1e-05,
      "loss": 0.5660243034362793,
      "step": 50,
      "token_acc": 0.8334044587872124
    },
    {
      "epoch": 0.03656307129798903,
      "grad_norm": 1.486912724454441,
      "learning_rate": 9.999025267866269e-06,
      "loss": 0.5425637722015381,
      "step": 60,
      "token_acc": 0.8377751665687463
    },
    {
      "epoch": 0.042656916514320534,
      "grad_norm": 1.5337888751937572,
      "learning_rate": 9.996101451506166e-06,
      "loss": 0.5277560710906982,
      "step": 70,
      "token_acc": 0.8436990187965481
    },
    {
      "epoch": 0.04875076173065204,
      "grad_norm": 1.3940886051095893,
      "learning_rate": 9.991229690894796e-06,
      "loss": 0.5164490699768066,
      "step": 80,
      "token_acc": 0.843745632858913
    },
    {
      "epoch": 0.054844606946983544,
      "grad_norm": 1.732808941800786,
      "learning_rate": 9.984411885496807e-06,
      "loss": 0.5111546516418457,
      "step": 90,
      "token_acc": 0.8453230842547292
    },
    {
      "epoch": 0.06093845216331505,
      "grad_norm": 1.502010177161109,
      "learning_rate": 9.975650693525798e-06,
      "loss": 0.5041120052337646,
      "step": 100,
      "token_acc": 0.8452364415692656
    },
    {
      "epoch": 0.06703229737964655,
      "grad_norm": 1.4601020474033113,
      "learning_rate": 9.964949530907907e-06,
      "loss": 0.5016684532165527,
      "step": 110,
      "token_acc": 0.8476517417815049
    },
    {
      "epoch": 0.07312614259597806,
      "grad_norm": 1.2952511202685608,
      "learning_rate": 9.952312569949963e-06,
      "loss": 0.481311559677124,
      "step": 120,
      "token_acc": 0.8533387907153323
    },
    {
      "epoch": 0.07921998781230957,
      "grad_norm": 1.4971652792627679,
      "learning_rate": 9.937744737712734e-06,
      "loss": 0.4774615287780762,
      "step": 130,
      "token_acc": 0.8534425009304056
    },
    {
      "epoch": 0.08531383302864107,
      "grad_norm": 1.3779208298780503,
      "learning_rate": 9.921251714089898e-06,
      "loss": 0.4781217575073242,
      "step": 140,
      "token_acc": 0.8540719832383131
    },
    {
      "epoch": 0.09140767824497258,
      "grad_norm": 1.2833076700573953,
      "learning_rate": 9.9028399295935e-06,
      "loss": 0.470335865020752,
      "step": 150,
      "token_acc": 0.855729364137813
    },
    {
      "epoch": 0.09750152346130408,
      "grad_norm": 1.275667403101274,
      "learning_rate": 9.882516562846735e-06,
      "loss": 0.4639917850494385,
      "step": 160,
      "token_acc": 0.856720295350119
    },
    {
      "epoch": 0.1035953686776356,
      "grad_norm": 1.3612747480512406,
      "learning_rate": 9.860289537785058e-06,
      "loss": 0.46750926971435547,
      "step": 170,
      "token_acc": 0.8558511446900368
    },
    {
      "epoch": 0.10968921389396709,
      "grad_norm": 1.415269189793894,
      "learning_rate": 9.83616752056669e-06,
      "loss": 0.4647522926330566,
      "step": 180,
      "token_acc": 0.8561545157582173
    },
    {
      "epoch": 0.1157830591102986,
      "grad_norm": 1.1892931159809241,
      "learning_rate": 9.810159916193763e-06,
      "loss": 0.45995321273803713,
      "step": 190,
      "token_acc": 0.8573448602405447
    },
    {
      "epoch": 0.1218769043266301,
      "grad_norm": 1.3064527572301536,
      "learning_rate": 9.782276864845351e-06,
      "loss": 0.4638189792633057,
      "step": 200,
      "token_acc": 0.8556243509610675
    },
    {
      "epoch": 0.12797074954296161,
      "grad_norm": 1.314500952457575,
      "learning_rate": 9.752529237923914e-06,
      "loss": 0.4490074634552002,
      "step": 210,
      "token_acc": 0.8597137978120026
    },
    {
      "epoch": 0.1340645947592931,
      "grad_norm": 1.1373343380715917,
      "learning_rate": 9.720928633816596e-06,
      "loss": 0.4501980781555176,
      "step": 220,
      "token_acc": 0.859648604947868
    },
    {
      "epoch": 0.14015843997562463,
      "grad_norm": 1.230125123804479,
      "learning_rate": 9.687487373373103e-06,
      "loss": 0.44935040473937987,
      "step": 230,
      "token_acc": 0.8588833095343918
    },
    {
      "epoch": 0.14625228519195613,
      "grad_norm": 1.6077603667743245,
      "learning_rate": 9.652218495101894e-06,
      "loss": 0.44729223251342776,
      "step": 240,
      "token_acc": 0.85995841942315
    },
    {
      "epoch": 0.15234613040828762,
      "grad_norm": 1.2688011877880512,
      "learning_rate": 9.61513575008656e-06,
      "loss": 0.43803844451904295,
      "step": 250,
      "token_acc": 0.8627407782309685
    },
    {
      "epoch": 0.15843997562461914,
      "grad_norm": 1.178112411608835,
      "learning_rate": 9.576253596624367e-06,
      "loss": 0.43675899505615234,
      "step": 260,
      "token_acc": 0.8637918234089942
    },
    {
      "epoch": 0.16453382084095064,
      "grad_norm": 1.294840110909924,
      "learning_rate": 9.53558719458908e-06,
      "loss": 0.4456604480743408,
      "step": 270,
      "token_acc": 0.8607995996575384
    },
    {
      "epoch": 0.17062766605728213,
      "grad_norm": 1.2275948762976965,
      "learning_rate": 9.49315239952023e-06,
      "loss": 0.44009056091308596,
      "step": 280,
      "token_acc": 0.8624380989923378
    },
    {
      "epoch": 0.17672151127361366,
      "grad_norm": 1.1835184130640346,
      "learning_rate": 9.448965756441154e-06,
      "loss": 0.43228535652160643,
      "step": 290,
      "token_acc": 0.8642425086011517
    },
    {
      "epoch": 0.18281535648994515,
      "grad_norm": 1.0506109771841785,
      "learning_rate": 9.403044493408205e-06,
      "loss": 0.4331789970397949,
      "step": 300,
      "token_acc": 0.8650773124725752
    },
    {
      "epoch": 0.18890920170627665,
      "grad_norm": 1.2538833747077607,
      "learning_rate": 9.355406514793667e-06,
      "loss": 0.44378862380981443,
      "step": 310,
      "token_acc": 0.8612388746191983
    },
    {
      "epoch": 0.19500304692260817,
      "grad_norm": 1.1649745576637627,
      "learning_rate": 9.306070394304955e-06,
      "loss": 0.4216612339019775,
      "step": 320,
      "token_acc": 0.8665872154728236
    },
    {
      "epoch": 0.20109689213893966,
      "grad_norm": 1.1640287810040342,
      "learning_rate": 9.255055367742868e-06,
      "loss": 0.43276224136352537,
      "step": 330,
      "token_acc": 0.864183550146075
    },
    {
      "epoch": 0.2071907373552712,
      "grad_norm": 1.1205258611684763,
      "learning_rate": 9.202381325501683e-06,
      "loss": 0.42910175323486327,
      "step": 340,
      "token_acc": 0.8651370039640893
    },
    {
      "epoch": 0.21328458257160268,
      "grad_norm": 1.1044569461174318,
      "learning_rate": 9.148068804814032e-06,
      "loss": 0.425107479095459,
      "step": 350,
      "token_acc": 0.8655852823220787
    },
    {
      "epoch": 0.21937842778793418,
      "grad_norm": 1.256804160100686,
      "learning_rate": 9.092138981743588e-06,
      "loss": 0.4197092533111572,
      "step": 360,
      "token_acc": 0.8678029564108461
    },
    {
      "epoch": 0.2254722730042657,
      "grad_norm": 1.219596805077906,
      "learning_rate": 9.034613662928665e-06,
      "loss": 0.4218160629272461,
      "step": 370,
      "token_acc": 0.8669598748703018
    },
    {
      "epoch": 0.2315661182205972,
      "grad_norm": 1.0651000148469036,
      "learning_rate": 8.975515277079961e-06,
      "loss": 0.4222999095916748,
      "step": 380,
      "token_acc": 0.8668562219942147
    },
    {
      "epoch": 0.2376599634369287,
      "grad_norm": 1.1478789808745513,
      "learning_rate": 8.91486686623577e-06,
      "loss": 0.41972966194152833,
      "step": 390,
      "token_acc": 0.8667315262188772
    },
    {
      "epoch": 0.2437538086532602,
      "grad_norm": 1.0060533858058822,
      "learning_rate": 8.85269207677806e-06,
      "loss": 0.4143358707427979,
      "step": 400,
      "token_acc": 0.8689943563130941
    },
    {
      "epoch": 0.2498476538695917,
      "grad_norm": 1.2219261737292129,
      "learning_rate": 8.789015150212907e-06,
      "loss": 0.41486186981201173,
      "step": 410,
      "token_acc": 0.867653374528066
    },
    {
      "epoch": 0.25594149908592323,
      "grad_norm": 1.2842286146778168,
      "learning_rate": 8.72386091371891e-06,
      "loss": 0.4264723777770996,
      "step": 420,
      "token_acc": 0.865345114787771
    },
    {
      "epoch": 0.2620353443022547,
      "grad_norm": 1.0550559155752623,
      "learning_rate": 8.657254770467252e-06,
      "loss": 0.40860881805419924,
      "step": 430,
      "token_acc": 0.8694270527928576
    },
    {
      "epoch": 0.2681291895185862,
      "grad_norm": 1.1246909396790437,
      "learning_rate": 8.58922268971719e-06,
      "loss": 0.4148720264434814,
      "step": 440,
      "token_acc": 0.86849521403236
    },
    {
      "epoch": 0.2742230347349177,
      "grad_norm": 1.1877909840033853,
      "learning_rate": 8.51979119669081e-06,
      "loss": 0.4155715465545654,
      "step": 450,
      "token_acc": 0.8686248236499153
    },
    {
      "epoch": 0.28031687995124926,
      "grad_norm": 1.03946599413896,
      "learning_rate": 8.448987362231054e-06,
      "loss": 0.4156056880950928,
      "step": 460,
      "token_acc": 0.8682606492506055
    },
    {
      "epoch": 0.28641072516758076,
      "grad_norm": 1.1045440790462375,
      "learning_rate": 8.376838792246978e-06,
      "loss": 0.41259098052978516,
      "step": 470,
      "token_acc": 0.868615067345492
    },
    {
      "epoch": 0.29250457038391225,
      "grad_norm": 1.1044055109636997,
      "learning_rate": 8.303373616950408e-06,
      "loss": 0.41626744270324706,
      "step": 480,
      "token_acc": 0.867445116993405
    },
    {
      "epoch": 0.29859841560024375,
      "grad_norm": 1.0612884186160958,
      "learning_rate": 8.228620479888172e-06,
      "loss": 0.4087618350982666,
      "step": 490,
      "token_acc": 0.869433255622514
    },
    {
      "epoch": 0.30469226081657524,
      "grad_norm": 1.079879116921211,
      "learning_rate": 8.152608526774188e-06,
      "loss": 0.40863656997680664,
      "step": 500,
      "token_acc": 0.8705444341829626
    },
    {
      "epoch": 0.31078610603290674,
      "grad_norm": 1.0470334273877924,
      "learning_rate": 8.075367394125755e-06,
      "loss": 0.41130657196044923,
      "step": 510,
      "token_acc": 0.8699947913802195
    },
    {
      "epoch": 0.3168799512492383,
      "grad_norm": 1.2778777056879977,
      "learning_rate": 7.996927197708486e-06,
      "loss": 0.4074504852294922,
      "step": 520,
      "token_acc": 0.8711178129454153
    },
    {
      "epoch": 0.3229737964655698,
      "grad_norm": 1.133795250933889,
      "learning_rate": 7.917318520794395e-06,
      "loss": 0.4040180206298828,
      "step": 530,
      "token_acc": 0.8719991647774729
    },
    {
      "epoch": 0.3290676416819013,
      "grad_norm": 1.1320221274981666,
      "learning_rate": 7.836572402237683e-06,
      "loss": 0.4074112892150879,
      "step": 540,
      "token_acc": 0.8696679374619692
    },
    {
      "epoch": 0.3351614868982328,
      "grad_norm": 1.0153565229717176,
      "learning_rate": 7.754720324372924e-06,
      "loss": 0.4030743598937988,
      "step": 550,
      "token_acc": 0.8720831783254012
    },
    {
      "epoch": 0.34125533211456427,
      "grad_norm": 1.0985579621580885,
      "learning_rate": 7.67179420074032e-06,
      "loss": 0.3988363742828369,
      "step": 560,
      "token_acc": 0.8726780258889484
    },
    {
      "epoch": 0.3473491773308958,
      "grad_norm": 1.0584699143582574,
      "learning_rate": 7.587826363642845e-06,
      "loss": 0.4028042793273926,
      "step": 570,
      "token_acc": 0.8709437860238254
    },
    {
      "epoch": 0.3534430225472273,
      "grad_norm": 1.1632651891282637,
      "learning_rate": 7.502849551540106e-06,
      "loss": 0.3974143028259277,
      "step": 580,
      "token_acc": 0.8732772418431721
    },
    {
      "epoch": 0.3595368677635588,
      "grad_norm": 0.9585380945132779,
      "learning_rate": 7.4168968962838524e-06,
      "loss": 0.40021185874938964,
      "step": 590,
      "token_acc": 0.8715715660830257
    },
    {
      "epoch": 0.3656307129798903,
      "grad_norm": 0.939779800665415,
      "learning_rate": 7.330001910200111e-06,
      "loss": 0.39843976497650146,
      "step": 600,
      "token_acc": 0.8733910783350537
    },
    {
      "epoch": 0.3717245581962218,
      "grad_norm": 0.9815164073943617,
      "learning_rate": 7.242198473022958e-06,
      "loss": 0.3972899913787842,
      "step": 610,
      "token_acc": 0.8731910420095998
    },
    {
      "epoch": 0.3778184034125533,
      "grad_norm": 1.0569386302509218,
      "learning_rate": 7.15352081868506e-06,
      "loss": 0.4026960372924805,
      "step": 620,
      "token_acc": 0.8716591305210795
    },
    {
      "epoch": 0.38391224862888484,
      "grad_norm": 1.0897077358900225,
      "learning_rate": 7.0640035219701085e-06,
      "loss": 0.39238433837890624,
      "step": 630,
      "token_acc": 0.8741110700683207
    },
    {
      "epoch": 0.39000609384521634,
      "grad_norm": 1.0094259905078886,
      "learning_rate": 6.973681485032359e-06,
      "loss": 0.3934662342071533,
      "step": 640,
      "token_acc": 0.874180305698641
    },
    {
      "epoch": 0.39609993906154783,
      "grad_norm": 0.9880095870102604,
      "learning_rate": 6.8825899237885215e-06,
      "loss": 0.3929059743881226,
      "step": 650,
      "token_acc": 0.873847849697677
    },
    {
      "epoch": 0.40219378427787933,
      "grad_norm": 0.9583618057687778,
      "learning_rate": 6.7907643541873446e-06,
      "loss": 0.38638834953308104,
      "step": 660,
      "token_acc": 0.8764517709444076
    },
    {
      "epoch": 0.4082876294942108,
      "grad_norm": 1.1091462631909463,
      "learning_rate": 6.698240578362179e-06,
      "loss": 0.3935162782669067,
      "step": 670,
      "token_acc": 0.8743182876186542
    },
    {
      "epoch": 0.4143814747105424,
      "grad_norm": 0.959273015275344,
      "learning_rate": 6.6050546706719984e-06,
      "loss": 0.38172011375427245,
      "step": 680,
      "token_acc": 0.8772576395099669
    },
    {
      "epoch": 0.42047531992687387,
      "grad_norm": 1.0010757728338364,
      "learning_rate": 6.511242963636257e-06,
      "loss": 0.3927836179733276,
      "step": 690,
      "token_acc": 0.8740263817041508
    },
    {
      "epoch": 0.42656916514320536,
      "grad_norm": 1.045230237684538,
      "learning_rate": 6.416842033769106e-06,
      "loss": 0.38949809074401853,
      "step": 700,
      "token_acc": 0.8748742675586352
    },
    {
      "epoch": 0.43266301035953686,
      "grad_norm": 0.9849032327305663,
      "learning_rate": 6.321888687318457e-06,
      "loss": 0.39299988746643066,
      "step": 710,
      "token_acc": 0.8744398373706392
    },
    {
      "epoch": 0.43875685557586835,
      "grad_norm": 0.9773426657855283,
      "learning_rate": 6.2264199459155105e-06,
      "loss": 0.38987624645233154,
      "step": 720,
      "token_acc": 0.8749521585172907
    },
    {
      "epoch": 0.4448507007921999,
      "grad_norm": 1.037517468712357,
      "learning_rate": 6.130473032140272e-06,
      "loss": 0.38550682067871095,
      "step": 730,
      "token_acc": 0.8752092114104209
    },
    {
      "epoch": 0.4509445460085314,
      "grad_norm": 1.0310013780608072,
      "learning_rate": 6.0340853550087345e-06,
      "loss": 0.378936243057251,
      "step": 740,
      "token_acc": 0.878043851367452
    },
    {
      "epoch": 0.4570383912248629,
      "grad_norm": 0.8055934899750623,
      "learning_rate": 5.937294495387377e-06,
      "loss": 0.38777313232421873,
      "step": 750,
      "token_acc": 0.8762303990063655
    },
    {
      "epoch": 0.4631322364411944,
      "grad_norm": 1.0076731680308868,
      "learning_rate": 5.840138191340651e-06,
      "loss": 0.3867051601409912,
      "step": 760,
      "token_acc": 0.875447200037364
    },
    {
      "epoch": 0.4692260816575259,
      "grad_norm": 0.9392775195574543,
      "learning_rate": 5.7426543234171736e-06,
      "loss": 0.3799318552017212,
      "step": 770,
      "token_acc": 0.8780739671196323
    },
    {
      "epoch": 0.4753199268738574,
      "grad_norm": 0.9059297874010275,
      "learning_rate": 5.644880899880382e-06,
      "loss": 0.38845138549804686,
      "step": 780,
      "token_acc": 0.8756513846485855
    },
    {
      "epoch": 0.48141377209018893,
      "grad_norm": 1.0364591251718924,
      "learning_rate": 5.546856041889374e-06,
      "loss": 0.384658670425415,
      "step": 790,
      "token_acc": 0.8760285406658391
    },
    {
      "epoch": 0.4875076173065204,
      "grad_norm": 0.9573686942596932,
      "learning_rate": 5.448617968635741e-06,
      "loss": 0.3791942596435547,
      "step": 800,
      "token_acc": 0.8779162415307187
    },
    {
      "epoch": 0.4936014625228519,
      "grad_norm": 0.9636242802763855,
      "learning_rate": 5.35020498244219e-06,
      "loss": 0.37176291942596434,
      "step": 810,
      "token_acc": 0.8793090876456928
    },
    {
      "epoch": 0.4996953077391834,
      "grad_norm": 1.037660587481492,
      "learning_rate": 5.251655453828728e-06,
      "loss": 0.37394251823425295,
      "step": 820,
      "token_acc": 0.8786210190654307
    },
    {
      "epoch": 0.505789152955515,
      "grad_norm": 1.0719330406024963,
      "learning_rate": 5.153007806552275e-06,
      "loss": 0.3745760679244995,
      "step": 830,
      "token_acc": 0.8784241641412887
    },
    {
      "epoch": 0.5118829981718465,
      "grad_norm": 0.8899515496236061,
      "learning_rate": 5.054300502625517e-06,
      "loss": 0.3706503868103027,
      "step": 840,
      "token_acc": 0.8798184912767585
    },
    {
      "epoch": 0.517976843388178,
      "grad_norm": 0.9136772226114551,
      "learning_rate": 4.9555720273208475e-06,
      "loss": 0.3767611742019653,
      "step": 850,
      "token_acc": 0.8780427238279765
    },
    {
      "epoch": 0.5240706886045094,
      "grad_norm": 0.9760538746168989,
      "learning_rate": 4.856860874165218e-06,
      "loss": 0.37979438304901125,
      "step": 860,
      "token_acc": 0.8784071947906439
    },
    {
      "epoch": 0.5301645338208409,
      "grad_norm": 0.9424993647974058,
      "learning_rate": 4.758205529931808e-06,
      "loss": 0.3839302062988281,
      "step": 870,
      "token_acc": 0.8770481761661205
    },
    {
      "epoch": 0.5362583790371724,
      "grad_norm": 1.0293112779306877,
      "learning_rate": 4.659644459634293e-06,
      "loss": 0.3767723321914673,
      "step": 880,
      "token_acc": 0.8782181679486365
    },
    {
      "epoch": 0.5423522242535039,
      "grad_norm": 1.0743397927299763,
      "learning_rate": 4.56121609152961e-06,
      "loss": 0.3791919946670532,
      "step": 890,
      "token_acc": 0.8769342677312787
    },
    {
      "epoch": 0.5484460694698354,
      "grad_norm": 0.8651643017417293,
      "learning_rate": 4.462958802135069e-06,
      "loss": 0.36331801414489745,
      "step": 900,
      "token_acc": 0.8819762679763837
    },
    {
      "epoch": 0.5545399146861669,
      "grad_norm": 0.9197439306994798,
      "learning_rate": 4.364910901265607e-06,
      "loss": 0.3720353603363037,
      "step": 910,
      "token_acc": 0.8795370329732339
    },
    {
      "epoch": 0.5606337599024985,
      "grad_norm": 0.9973864478854872,
      "learning_rate": 4.2671106170970734e-06,
      "loss": 0.37818198204040526,
      "step": 920,
      "token_acc": 0.8787091854009224
    },
    {
      "epoch": 0.56672760511883,
      "grad_norm": 0.9979320322546561,
      "learning_rate": 4.169596081261332e-06,
      "loss": 0.368232798576355,
      "step": 930,
      "token_acc": 0.8808049967885766
    },
    {
      "epoch": 0.5728214503351615,
      "grad_norm": 0.9817455772913783,
      "learning_rate": 4.072405313979021e-06,
      "loss": 0.37091827392578125,
      "step": 940,
      "token_acc": 0.8796466097957818
    },
    {
      "epoch": 0.578915295551493,
      "grad_norm": 1.0935297334377472,
      "learning_rate": 3.975576209235726e-06,
      "loss": 0.3674028396606445,
      "step": 950,
      "token_acc": 0.8807917695163083
    },
    {
      "epoch": 0.5850091407678245,
      "grad_norm": 0.9835469765967159,
      "learning_rate": 3.879146520007399e-06,
      "loss": 0.3728478908538818,
      "step": 960,
      "token_acc": 0.8795413152600885
    },
    {
      "epoch": 0.591102985984156,
      "grad_norm": 0.9625183356689964,
      "learning_rate": 3.7831538435407344e-06,
      "loss": 0.37494525909423826,
      "step": 970,
      "token_acc": 0.8792245580635571
    },
    {
      "epoch": 0.5971968312004875,
      "grad_norm": 0.9012795424730173,
      "learning_rate": 3.687635606694271e-06,
      "loss": 0.3702352046966553,
      "step": 980,
      "token_acc": 0.8801223453080008
    },
    {
      "epoch": 0.603290676416819,
      "grad_norm": 0.9782757486531443,
      "learning_rate": 3.592629051345936e-06,
      "loss": 0.3673159837722778,
      "step": 990,
      "token_acc": 0.8810825035648933
    },
    {
      "epoch": 0.6093845216331505,
      "grad_norm": 1.0059100640922563,
      "learning_rate": 3.4981712198726956e-06,
      "loss": 0.3642214059829712,
      "step": 1000,
      "token_acc": 0.8818312088488447
    },
    {
      "epoch": 0.615478366849482,
      "grad_norm": 0.9395189399708234,
      "learning_rate": 3.4042989407079986e-06,
      "loss": 0.3784639358520508,
      "step": 1010,
      "token_acc": 0.8780194366406157
    },
    {
      "epoch": 0.6215722120658135,
      "grad_norm": 1.0425592930772825,
      "learning_rate": 3.311048813982627e-06,
      "loss": 0.36695384979248047,
      "step": 1020,
      "token_acc": 0.8809777292779815
    },
    {
      "epoch": 0.6276660572821451,
      "grad_norm": 0.9146308056797927,
      "learning_rate": 3.218457197254583e-06,
      "loss": 0.36698212623596194,
      "step": 1030,
      "token_acc": 0.8810339710207495
    },
    {
      "epoch": 0.6337599024984766,
      "grad_norm": 0.976263078958663,
      "learning_rate": 3.1265601913335196e-06,
      "loss": 0.365465784072876,
      "step": 1040,
      "token_acc": 0.8814162812670944
    },
    {
      "epoch": 0.6398537477148081,
      "grad_norm": 1.0567379406046713,
      "learning_rate": 3.035393626205306e-06,
      "loss": 0.3610874891281128,
      "step": 1050,
      "token_acc": 0.8824792140002385
    },
    {
      "epoch": 0.6459475929311396,
      "grad_norm": 1.0205537815943757,
      "learning_rate": 2.944993047062161e-06,
      "loss": 0.35759830474853516,
      "step": 1060,
      "token_acc": 0.8834624031976018
    },
    {
      "epoch": 0.6520414381474711,
      "grad_norm": 1.0280714401242652,
      "learning_rate": 2.8553937004438425e-06,
      "loss": 0.3574142217636108,
      "step": 1070,
      "token_acc": 0.884169503378651
    },
    {
      "epoch": 0.6581352833638026,
      "grad_norm": 1.0187298702407688,
      "learning_rate": 2.766630520495277e-06,
      "loss": 0.36029987335205077,
      "step": 1080,
      "token_acc": 0.8823869756562952
    },
    {
      "epoch": 0.664229128580134,
      "grad_norm": 0.9191494153561297,
      "learning_rate": 2.67873811534598e-06,
      "loss": 0.35897092819213866,
      "step": 1090,
      "token_acc": 0.8827260508533868
    },
    {
      "epoch": 0.6703229737964655,
      "grad_norm": 0.9492740813391064,
      "learning_rate": 2.591750753616596e-06,
      "loss": 0.36168532371520995,
      "step": 1100,
      "token_acc": 0.8825941425209475
    },
    {
      "epoch": 0.676416819012797,
      "grad_norm": 0.9644543574186545,
      "learning_rate": 2.505702351057804e-06,
      "loss": 0.3665107488632202,
      "step": 1110,
      "token_acc": 0.8816928952036972
    },
    {
      "epoch": 0.6825106642291285,
      "grad_norm": 0.9521683470371731,
      "learning_rate": 2.4206264573268174e-06,
      "loss": 0.35790448188781737,
      "step": 1120,
      "token_acc": 0.8832886728694526
    },
    {
      "epoch": 0.68860450944546,
      "grad_norm": 1.0783164983743936,
      "learning_rate": 2.336556242906608e-06,
      "loss": 0.3561516284942627,
      "step": 1130,
      "token_acc": 0.8839432945670233
    },
    {
      "epoch": 0.6946983546617916,
      "grad_norm": 0.9994299291097577,
      "learning_rate": 2.2535244861729707e-06,
      "loss": 0.3557067632675171,
      "step": 1140,
      "token_acc": 0.8837923958883728
    },
    {
      "epoch": 0.7007921998781231,
      "grad_norm": 1.039214819811771,
      "learning_rate": 2.1715635606144653e-06,
      "loss": 0.3563429832458496,
      "step": 1150,
      "token_acc": 0.8836427544336156
    },
    {
      "epoch": 0.7068860450944546,
      "grad_norm": 0.8549094000634878,
      "learning_rate": 2.0907054222102367e-06,
      "loss": 0.35337374210357664,
      "step": 1160,
      "token_acc": 0.8852147256677358
    },
    {
      "epoch": 0.7129798903107861,
      "grad_norm": 0.894156191232295,
      "learning_rate": 2.0109815969705922e-06,
      "loss": 0.359290337562561,
      "step": 1170,
      "token_acc": 0.8828725266946272
    },
    {
      "epoch": 0.7190737355271176,
      "grad_norm": 0.8673526133846996,
      "learning_rate": 1.9324231686452478e-06,
      "loss": 0.35991313457489016,
      "step": 1180,
      "token_acc": 0.8837700799671174
    },
    {
      "epoch": 0.7251675807434491,
      "grad_norm": 0.9356232121590031,
      "learning_rate": 1.8550607666039877e-06,
      "loss": 0.3538203716278076,
      "step": 1190,
      "token_acc": 0.8850202284200351
    },
    {
      "epoch": 0.7312614259597806,
      "grad_norm": 1.0163312252270116,
      "learning_rate": 1.7789245538944971e-06,
      "loss": 0.3607466459274292,
      "step": 1200,
      "token_acc": 0.8824661130842316
    },
    {
      "epoch": 0.7373552711761121,
      "grad_norm": 0.8390316456040804,
      "learning_rate": 1.7040442154820036e-06,
      "loss": 0.35505869388580324,
      "step": 1210,
      "token_acc": 0.8845901901507859
    },
    {
      "epoch": 0.7434491163924436,
      "grad_norm": 0.921086850463397,
      "learning_rate": 1.6304489466753237e-06,
      "loss": 0.35682291984558107,
      "step": 1220,
      "token_acc": 0.884017590582417
    },
    {
      "epoch": 0.7495429616087751,
      "grad_norm": 0.8352814372993298,
      "learning_rate": 1.5581674417438143e-06,
      "loss": 0.3599454164505005,
      "step": 1230,
      "token_acc": 0.8830610223076613
    },
    {
      "epoch": 0.7556368068251066,
      "grad_norm": 0.9561368940432438,
      "learning_rate": 1.4872278827296855e-06,
      "loss": 0.3544511079788208,
      "step": 1240,
      "token_acc": 0.884971241183666
    },
    {
      "epoch": 0.7617306520414382,
      "grad_norm": 0.9963256225377098,
      "learning_rate": 1.417657928460029e-06,
      "loss": 0.35143122673034666,
      "step": 1250,
      "token_acc": 0.8854597977852672
    },
    {
      "epoch": 0.7678244972577697,
      "grad_norm": 1.0464860200353496,
      "learning_rate": 1.349484703762834e-06,
      "loss": 0.3545159101486206,
      "step": 1260,
      "token_acc": 0.8848001191868091
    },
    {
      "epoch": 0.7739183424741012,
      "grad_norm": 0.9553675018651967,
      "learning_rate": 1.2827347888912057e-06,
      "loss": 0.3540821552276611,
      "step": 1270,
      "token_acc": 0.8845431750704823
    },
    {
      "epoch": 0.7800121876904327,
      "grad_norm": 0.9171124221466627,
      "learning_rate": 1.2174342091599277e-06,
      "loss": 0.3459270477294922,
      "step": 1280,
      "token_acc": 0.8876378370255273
    },
    {
      "epoch": 0.7861060329067642,
      "grad_norm": 0.9897434740336704,
      "learning_rate": 1.1536084247983626e-06,
      "loss": 0.3577150821685791,
      "step": 1290,
      "token_acc": 0.8842498302783435
    },
    {
      "epoch": 0.7921998781230957,
      "grad_norm": 0.88979092902762,
      "learning_rate": 1.0912823210237033e-06,
      "loss": 0.350811505317688,
      "step": 1300,
      "token_acc": 0.8856008373344852
    },
    {
      "epoch": 0.7982937233394272,
      "grad_norm": 0.9287859784083828,
      "learning_rate": 1.0304801983383989e-06,
      "loss": 0.3551754951477051,
      "step": 1310,
      "token_acc": 0.8848410538592661
    },
    {
      "epoch": 0.8043875685557587,
      "grad_norm": 0.8802985747226686,
      "learning_rate": 9.712257630555589e-07,
      "loss": 0.35124433040618896,
      "step": 1320,
      "token_acc": 0.8857088187898194
    },
    {
      "epoch": 0.8104814137720902,
      "grad_norm": 0.9867993671885138,
      "learning_rate": 9.135421180560394e-07,
      "loss": 0.3533953666687012,
      "step": 1330,
      "token_acc": 0.8847630099080603
    },
    {
      "epoch": 0.8165752589884216,
      "grad_norm": 0.926446790364043,
      "learning_rate": 8.574517537807897e-07,
      "loss": 0.345960807800293,
      "step": 1340,
      "token_acc": 0.8876687663254338
    },
    {
      "epoch": 0.8226691042047533,
      "grad_norm": 0.9083144656784397,
      "learning_rate": 8.029765394619899e-07,
      "loss": 0.35233092308044434,
      "step": 1350,
      "token_acc": 0.8852270821778219
    },
    {
      "epoch": 0.8287629494210847,
      "grad_norm": 0.8865701179019319,
      "learning_rate": 7.501377145963939e-07,
      "loss": 0.35347394943237304,
      "step": 1360,
      "token_acc": 0.8848507491917527
    },
    {
      "epoch": 0.8348567946374162,
      "grad_norm": 0.8797844443235806,
      "learning_rate": 6.98955880664205e-07,
      "loss": 0.35233142375946047,
      "step": 1370,
      "token_acc": 0.8857494626572902
    },
    {
      "epoch": 0.8409506398537477,
      "grad_norm": 0.985930499180468,
      "learning_rate": 6.494509930967019e-07,
      "loss": 0.3484508991241455,
      "step": 1380,
      "token_acc": 0.8862226663569039
    },
    {
      "epoch": 0.8470444850700792,
      "grad_norm": 0.8385926015490823,
      "learning_rate": 6.016423534957616e-07,
      "loss": 0.34513344764709475,
      "step": 1390,
      "token_acc": 0.88766630420385
    },
    {
      "epoch": 0.8531383302864107,
      "grad_norm": 0.9469060182104153,
      "learning_rate": 5.555486021082979e-07,
      "loss": 0.3453853130340576,
      "step": 1400,
      "token_acc": 0.8872980190401473
    },
    {
      "epoch": 0.8592321755027422,
      "grad_norm": 1.0619116209691746,
      "learning_rate": 5.111877105585672e-07,
      "loss": 0.35715694427490235,
      "step": 1410,
      "token_acc": 0.8840584828365589
    },
    {
      "epoch": 0.8653260207190737,
      "grad_norm": 0.9990471419637711,
      "learning_rate": 4.6857697484116006e-07,
      "loss": 0.34844698905944826,
      "step": 1420,
      "token_acc": 0.8861595746957418
    },
    {
      "epoch": 0.8714198659354052,
      "grad_norm": 0.8653174829680845,
      "learning_rate": 4.277330085774156e-07,
      "loss": 0.34473817348480223,
      "step": 1430,
      "token_acc": 0.8869124712097335
    },
    {
      "epoch": 0.8775137111517367,
      "grad_norm": 0.978323586867761,
      "learning_rate": 3.886717365378867e-07,
      "loss": 0.3523882865905762,
      "step": 1440,
      "token_acc": 0.8849034480348013
    },
    {
      "epoch": 0.8836075563680682,
      "grad_norm": 1.0140389777919647,
      "learning_rate": 3.5140838843339073e-07,
      "loss": 0.3476292848587036,
      "step": 1450,
      "token_acc": 0.8866329934005767
    },
    {
      "epoch": 0.8897014015843998,
      "grad_norm": 1.0064657138214737,
      "learning_rate": 3.159574929770515e-07,
      "loss": 0.35365211963653564,
      "step": 1460,
      "token_acc": 0.8852465385385505
    },
    {
      "epoch": 0.8957952468007313,
      "grad_norm": 0.9324871915195588,
      "learning_rate": 2.8233287221965555e-07,
      "loss": 0.3441819190979004,
      "step": 1470,
      "token_acc": 0.8871095878318941
    },
    {
      "epoch": 0.9018890920170628,
      "grad_norm": 0.9055988245681192,
      "learning_rate": 2.5054763616053967e-07,
      "loss": 0.34738845825195314,
      "step": 1480,
      "token_acc": 0.8870410481583068
    },
    {
      "epoch": 0.9079829372333943,
      "grad_norm": 0.8845337059700371,
      "learning_rate": 2.2061417763608818e-07,
      "loss": 0.3496507167816162,
      "step": 1490,
      "token_acc": 0.8858089991712572
    },
    {
      "epoch": 0.9140767824497258,
      "grad_norm": 0.8884170981985747,
      "learning_rate": 1.9254416748786086e-07,
      "loss": 0.34417023658752444,
      "step": 1500,
      "token_acc": 0.8876897324425693
    },
    {
      "epoch": 0.9201706276660573,
      "grad_norm": 0.991921395955364,
      "learning_rate": 1.6634855001221195e-07,
      "loss": 0.3475677490234375,
      "step": 1510,
      "token_acc": 0.8866243585461391
    },
    {
      "epoch": 0.9262644728823888,
      "grad_norm": 0.8822494961130495,
      "learning_rate": 1.4203753869318882e-07,
      "loss": 0.35834810733795164,
      "step": 1520,
      "token_acc": 0.8836910930175179
    },
    {
      "epoch": 0.9323583180987203,
      "grad_norm": 1.0007870631310825,
      "learning_rate": 1.196206122203647e-07,
      "loss": 0.3498887777328491,
      "step": 1530,
      "token_acc": 0.8859136668935295
    },
    {
      "epoch": 0.9384521633150518,
      "grad_norm": 0.9115641715955437,
      "learning_rate": 9.910651079316824e-08,
      "loss": 0.3380606651306152,
      "step": 1540,
      "token_acc": 0.8888504997761748
    },
    {
      "epoch": 0.9445460085313833,
      "grad_norm": 0.9336474945041258,
      "learning_rate": 8.050323271314331e-08,
      "loss": 0.34683611392974856,
      "step": 1550,
      "token_acc": 0.8867626671565236
    },
    {
      "epoch": 0.9506398537477148,
      "grad_norm": 0.9153041920993996,
      "learning_rate": 6.381803126546405e-08,
      "loss": 0.3438985824584961,
      "step": 1560,
      "token_acc": 0.8876278171714178
    },
    {
      "epoch": 0.9567336989640464,
      "grad_norm": 0.8723491469657118,
      "learning_rate": 4.9057411890933714e-08,
      "loss": 0.35089046955108644,
      "step": 1570,
      "token_acc": 0.8854520115332541
    },
    {
      "epoch": 0.9628275441803779,
      "grad_norm": 0.8955372459045878,
      "learning_rate": 3.622712964956032e-08,
      "loss": 0.34657576084136965,
      "step": 1580,
      "token_acc": 0.8870879211520062
    },
    {
      "epoch": 0.9689213893967094,
      "grad_norm": 0.9416702515233623,
      "learning_rate": 2.5332186976697037e-08,
      "loss": 0.35133283138275145,
      "step": 1590,
      "token_acc": 0.8860156117328086
    },
    {
      "epoch": 0.9750152346130408,
      "grad_norm": 0.8711895076149625,
      "learning_rate": 1.637683173263238e-08,
      "loss": 0.35227146148681643,
      "step": 1600,
      "token_acc": 0.8855705009128142
    },
    {
      "epoch": 0.9811090798293723,
      "grad_norm": 0.983729311544991,
      "learning_rate": 9.364555546375054e-09,
      "loss": 0.34629082679748535,
      "step": 1610,
      "token_acc": 0.8869587094319709
    },
    {
      "epoch": 0.9872029250457038,
      "grad_norm": 0.94198406227018,
      "learning_rate": 4.2980924542984634e-09,
      "loss": 0.3403524875640869,
      "step": 1620,
      "token_acc": 0.8887918722020187
    },
    {
      "epoch": 0.9932967702620353,
      "grad_norm": 0.8500481071531769,
      "learning_rate": 1.179417834153429e-09,
      "loss": 0.3546321868896484,
      "step": 1630,
      "token_acc": 0.8848728077900511
    },
    {
      "epoch": 0.9993906154783668,
      "grad_norm": 0.9946340081463461,
      "learning_rate": 9.74763488759134e-12,
      "loss": 0.35070624351501467,
      "step": 1640,
      "token_acc": 0.8863686895606487
    }
  ],
  "logging_steps": 10,
  "max_steps": 1641,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2214001985716224.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}