{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 1473,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00679001867255135,
      "grad_norm": 0.3864942491054535,
      "learning_rate": 6.081081081081082e-06,
      "loss": 1.15,
      "step": 10
    },
    {
      "epoch": 0.0135800373451027,
      "grad_norm": 0.4518967866897583,
      "learning_rate": 1.2837837837837838e-05,
      "loss": 1.1633,
      "step": 20
    },
    {
      "epoch": 0.02037005601765405,
      "grad_norm": 0.36683163046836853,
      "learning_rate": 1.9594594594594595e-05,
      "loss": 1.1004,
      "step": 30
    },
    {
      "epoch": 0.0271600746902054,
      "grad_norm": 0.2317497730255127,
      "learning_rate": 2.635135135135135e-05,
      "loss": 1.014,
      "step": 40
    },
    {
      "epoch": 0.03395009336275675,
      "grad_norm": 0.2007765769958496,
      "learning_rate": 3.310810810810811e-05,
      "loss": 0.9387,
      "step": 50
    },
    {
      "epoch": 0.0407401120353081,
      "grad_norm": 0.19630813598632812,
      "learning_rate": 3.986486486486487e-05,
      "loss": 0.8715,
      "step": 60
    },
    {
      "epoch": 0.04753013070785945,
      "grad_norm": 0.20267479121685028,
      "learning_rate": 4.662162162162162e-05,
      "loss": 0.8385,
      "step": 70
    },
    {
      "epoch": 0.0543201493804108,
      "grad_norm": 0.21056681871414185,
      "learning_rate": 5.337837837837838e-05,
      "loss": 0.7904,
      "step": 80
    },
    {
      "epoch": 0.06111016805296215,
      "grad_norm": 0.181947723031044,
      "learning_rate": 6.013513513513514e-05,
      "loss": 0.7601,
      "step": 90
    },
    {
      "epoch": 0.0679001867255135,
      "grad_norm": 0.1922345757484436,
      "learning_rate": 6.68918918918919e-05,
      "loss": 0.7673,
      "step": 100
    },
    {
      "epoch": 0.07469020539806484,
      "grad_norm": 0.18776988983154297,
      "learning_rate": 7.364864864864865e-05,
      "loss": 0.7375,
      "step": 110
    },
    {
      "epoch": 0.0814802240706162,
      "grad_norm": 0.2282622903585434,
      "learning_rate": 8.040540540540541e-05,
      "loss": 0.7381,
      "step": 120
    },
    {
      "epoch": 0.08827024274316754,
      "grad_norm": 0.2738005220890045,
      "learning_rate": 8.716216216216216e-05,
      "loss": 0.7237,
      "step": 130
    },
    {
      "epoch": 0.0950602614157189,
      "grad_norm": 0.21742618083953857,
      "learning_rate": 9.391891891891892e-05,
      "loss": 0.7023,
      "step": 140
    },
    {
      "epoch": 0.10185028008827024,
      "grad_norm": 0.25216951966285706,
      "learning_rate": 9.999985945746134e-05,
      "loss": 0.7101,
      "step": 150
    },
    {
      "epoch": 0.1086402987608216,
      "grad_norm": 0.25243276357650757,
      "learning_rate": 9.998299530880787e-05,
      "loss": 0.687,
      "step": 160
    },
    {
      "epoch": 0.11543031743337294,
      "grad_norm": 0.3072853684425354,
      "learning_rate": 9.993803351513094e-05,
      "loss": 0.7097,
      "step": 170
    },
    {
      "epoch": 0.1222203361059243,
      "grad_norm": 0.2733334004878998,
      "learning_rate": 9.986499935143679e-05,
      "loss": 0.6722,
      "step": 180
    },
    {
      "epoch": 0.12901035477847564,
      "grad_norm": 0.2821713387966156,
      "learning_rate": 9.976393387344834e-05,
      "loss": 0.6935,
      "step": 190
    },
    {
      "epoch": 0.135800373451027,
      "grad_norm": 0.28884661197662354,
      "learning_rate": 9.963489389452596e-05,
      "loss": 0.6964,
      "step": 200
    },
    {
      "epoch": 0.14259039212357835,
      "grad_norm": 0.25266194343566895,
      "learning_rate": 9.947795195373016e-05,
      "loss": 0.6926,
      "step": 210
    },
    {
      "epoch": 0.14938041079612968,
      "grad_norm": 0.3153521418571472,
      "learning_rate": 9.92931962750442e-05,
      "loss": 0.6727,
      "step": 220
    },
    {
      "epoch": 0.15617042946868104,
      "grad_norm": 0.29505372047424316,
      "learning_rate": 9.908073071777954e-05,
      "loss": 0.6814,
      "step": 230
    },
    {
      "epoch": 0.1629604481412324,
      "grad_norm": 0.2789740562438965,
      "learning_rate": 9.884067471819184e-05,
      "loss": 0.6817,
      "step": 240
    },
    {
      "epoch": 0.16975046681378375,
      "grad_norm": 0.2661297917366028,
      "learning_rate": 9.857316322234067e-05,
      "loss": 0.6795,
      "step": 250
    },
    {
      "epoch": 0.17654048548633508,
      "grad_norm": 0.2866443395614624,
      "learning_rate": 9.827834661023034e-05,
      "loss": 0.6623,
      "step": 260
    },
    {
      "epoch": 0.18333050415888644,
      "grad_norm": 0.2814137637615204,
      "learning_rate": 9.795639061127468e-05,
      "loss": 0.6606,
      "step": 270
    },
    {
      "epoch": 0.1901205228314378,
      "grad_norm": 0.30371809005737305,
      "learning_rate": 9.760747621113325e-05,
      "loss": 0.6679,
      "step": 280
    },
    {
      "epoch": 0.19691054150398912,
      "grad_norm": 0.311916321516037,
      "learning_rate": 9.723179954997125e-05,
      "loss": 0.6817,
      "step": 290
    },
    {
      "epoch": 0.20370056017654048,
      "grad_norm": 0.2893620431423187,
      "learning_rate": 9.682957181220062e-05,
      "loss": 0.6678,
      "step": 300
    },
    {
      "epoch": 0.21049057884909184,
      "grad_norm": 0.3415584862232208,
      "learning_rate": 9.640101910776381e-05,
      "loss": 0.6512,
      "step": 310
    },
    {
      "epoch": 0.2172805975216432,
      "grad_norm": 0.2780122756958008,
      "learning_rate": 9.594638234502753e-05,
      "loss": 0.6454,
      "step": 320
    },
    {
      "epoch": 0.22407061619419452,
      "grad_norm": 0.28491947054862976,
      "learning_rate": 9.546591709535751e-05,
      "loss": 0.6597,
      "step": 330
    },
    {
      "epoch": 0.23086063486674588,
      "grad_norm": 0.28796347975730896,
      "learning_rate": 9.495989344945056e-05,
      "loss": 0.6515,
      "step": 340
    },
    {
      "epoch": 0.23765065353929724,
      "grad_norm": 0.2798394560813904,
      "learning_rate": 9.442859586550468e-05,
      "loss": 0.654,
      "step": 350
    },
    {
      "epoch": 0.2444406722118486,
      "grad_norm": 0.31528720259666443,
      "learning_rate": 9.387232300931255e-05,
      "loss": 0.6477,
      "step": 360
    },
    {
      "epoch": 0.2512306908843999,
      "grad_norm": 0.329709529876709,
      "learning_rate": 9.32913875863684e-05,
      "loss": 0.6576,
      "step": 370
    },
    {
      "epoch": 0.2580207095569513,
      "grad_norm": 0.31941738724708557,
      "learning_rate": 9.268611616608236e-05,
      "loss": 0.6561,
      "step": 380
    },
    {
      "epoch": 0.26481072822950263,
      "grad_norm": 0.30430927872657776,
      "learning_rate": 9.205684899820147e-05,
      "loss": 0.6453,
      "step": 390
    },
    {
      "epoch": 0.271600746902054,
      "grad_norm": 0.31718093156814575,
      "learning_rate": 9.140393982154026e-05,
      "loss": 0.6434,
      "step": 400
    },
    {
      "epoch": 0.27839076557460535,
      "grad_norm": 0.30542615056037903,
      "learning_rate": 9.072775566512863e-05,
      "loss": 0.6355,
      "step": 410
    },
    {
      "epoch": 0.2851807842471567,
      "grad_norm": 0.3371242880821228,
      "learning_rate": 9.002867664188858e-05,
      "loss": 0.6565,
      "step": 420
    },
    {
      "epoch": 0.291970802919708,
      "grad_norm": 0.30639946460723877,
      "learning_rate": 8.930709573495602e-05,
      "loss": 0.6525,
      "step": 430
    },
    {
      "epoch": 0.29876082159225936,
      "grad_norm": 0.29678982496261597,
      "learning_rate": 8.856341857676758e-05,
      "loss": 0.6435,
      "step": 440
    },
    {
      "epoch": 0.3055508402648107,
      "grad_norm": 0.295890748500824,
      "learning_rate": 8.779806322103682e-05,
      "loss": 0.6458,
      "step": 450
    },
    {
      "epoch": 0.3123408589373621,
      "grad_norm": 0.39472854137420654,
      "learning_rate": 8.701145990774775e-05,
      "loss": 0.6582,
      "step": 460
    },
    {
      "epoch": 0.31913087760991343,
      "grad_norm": 0.3048432469367981,
      "learning_rate": 8.620405082129798e-05,
      "loss": 0.6482,
      "step": 470
    },
    {
      "epoch": 0.3259208962824648,
      "grad_norm": 0.30375346541404724,
      "learning_rate": 8.537628984192736e-05,
      "loss": 0.6473,
      "step": 480
    },
    {
      "epoch": 0.33271091495501615,
      "grad_norm": 0.3189581036567688,
      "learning_rate": 8.452864229057191e-05,
      "loss": 0.6423,
      "step": 490
    },
    {
      "epoch": 0.3395009336275675,
      "grad_norm": 0.30964285135269165,
      "learning_rate": 8.366158466728644e-05,
      "loss": 0.6239,
      "step": 500
    },
    {
      "epoch": 0.3462909523001188,
      "grad_norm": 0.30495893955230713,
      "learning_rate": 8.277560438338278e-05,
      "loss": 0.6307,
      "step": 510
    },
    {
      "epoch": 0.35308097097267016,
      "grad_norm": 0.3121732175350189,
      "learning_rate": 8.18711994874345e-05,
      "loss": 0.6363,
      "step": 520
    },
    {
      "epoch": 0.3598709896452215,
      "grad_norm": 0.32421112060546875,
      "learning_rate": 8.094887838530174e-05,
      "loss": 0.636,
      "step": 530
    },
    {
      "epoch": 0.3666610083177729,
      "grad_norm": 0.3044915497303009,
      "learning_rate": 8.000915955433396e-05,
      "loss": 0.6276,
      "step": 540
    },
    {
      "epoch": 0.37345102699032423,
      "grad_norm": 0.3056237995624542,
      "learning_rate": 7.905257125191094e-05,
      "loss": 0.627,
      "step": 550
    },
    {
      "epoch": 0.3802410456628756,
      "grad_norm": 0.29212453961372375,
      "learning_rate": 7.807965121848597e-05,
      "loss": 0.6295,
      "step": 560
    },
    {
      "epoch": 0.38703106433542694,
      "grad_norm": 0.2936370074748993,
      "learning_rate": 7.70909463752983e-05,
      "loss": 0.6449,
      "step": 570
    },
    {
      "epoch": 0.39382108300797825,
      "grad_norm": 0.3245689868927002,
      "learning_rate": 7.608701251692457e-05,
      "loss": 0.6506,
      "step": 580
    },
    {
      "epoch": 0.4006111016805296,
      "grad_norm": 0.33466264605522156,
      "learning_rate": 7.506841399884217e-05,
      "loss": 0.6217,
      "step": 590
    },
    {
      "epoch": 0.40740112035308096,
      "grad_norm": 0.3045268952846527,
      "learning_rate": 7.403572342018021e-05,
      "loss": 0.6284,
      "step": 600
    },
    {
      "epoch": 0.4141911390256323,
      "grad_norm": 0.31157055497169495,
      "learning_rate": 7.29895213018362e-05,
      "loss": 0.6459,
      "step": 610
    },
    {
      "epoch": 0.4209811576981837,
      "grad_norm": 0.30077850818634033,
      "learning_rate": 7.193039576013976e-05,
      "loss": 0.638,
      "step": 620
    },
    {
      "epoch": 0.42777117637073503,
      "grad_norm": 0.31292667984962463,
      "learning_rate": 7.085894217624645e-05,
      "loss": 0.6411,
      "step": 630
    },
    {
      "epoch": 0.4345611950432864,
      "grad_norm": 0.30501753091812134,
      "learning_rate": 6.977576286144784e-05,
      "loss": 0.6177,
      "step": 640
    },
    {
      "epoch": 0.44135121371583774,
      "grad_norm": 0.32817915081977844,
      "learning_rate": 6.868146671858567e-05,
      "loss": 0.624,
      "step": 650
    },
    {
      "epoch": 0.44814123238838904,
      "grad_norm": 0.3311234414577484,
      "learning_rate": 6.757666889976085e-05,
      "loss": 0.6319,
      "step": 660
    },
    {
      "epoch": 0.4549312510609404,
      "grad_norm": 0.3192739486694336,
      "learning_rate": 6.646199046052921e-05,
      "loss": 0.6193,
      "step": 670
    },
    {
      "epoch": 0.46172126973349176,
      "grad_norm": 0.29953694343566895,
      "learning_rate": 6.533805801077899e-05,
      "loss": 0.615,
      "step": 680
    },
    {
      "epoch": 0.4685112884060431,
      "grad_norm": 0.32995176315307617,
      "learning_rate": 6.420550336248558e-05,
      "loss": 0.6357,
      "step": 690
    },
    {
      "epoch": 0.47530130707859447,
      "grad_norm": 0.29778003692626953,
      "learning_rate": 6.306496317454227e-05,
      "loss": 0.6164,
      "step": 700
    },
    {
      "epoch": 0.4820913257511458,
      "grad_norm": 0.3111869990825653,
      "learning_rate": 6.191707859486622e-05,
      "loss": 0.6133,
      "step": 710
    },
    {
      "epoch": 0.4888813444236972,
      "grad_norm": 0.32019391655921936,
      "learning_rate": 6.076249489998097e-05,
      "loss": 0.6243,
      "step": 720
    },
    {
      "epoch": 0.49567136309624854,
      "grad_norm": 0.323253870010376,
      "learning_rate": 5.9601861132278e-05,
      "loss": 0.6253,
      "step": 730
    },
    {
      "epoch": 0.5024613817687998,
      "grad_norm": 0.3075142800807953,
      "learning_rate": 5.8435829735161416e-05,
      "loss": 0.6129,
      "step": 740
    },
    {
      "epoch": 0.5092514004413512,
      "grad_norm": 0.34016624093055725,
      "learning_rate": 5.7265056186280754e-05,
      "loss": 0.6381,
      "step": 750
    },
    {
      "epoch": 0.5160414191139026,
      "grad_norm": 0.34157323837280273,
      "learning_rate": 5.6090198629058134e-05,
      "loss": 0.6211,
      "step": 760
    },
    {
      "epoch": 0.5228314377864539,
      "grad_norm": 0.29887712001800537,
      "learning_rate": 5.491191750271677e-05,
      "loss": 0.6132,
      "step": 770
    },
    {
      "epoch": 0.5296214564590053,
      "grad_norm": 0.32949894666671753,
      "learning_rate": 5.373087517101899e-05,
      "loss": 0.6296,
      "step": 780
    },
    {
      "epoch": 0.5364114751315566,
      "grad_norm": 0.33202993869781494,
      "learning_rate": 5.254773554992242e-05,
      "loss": 0.6195,
      "step": 790
    },
    {
      "epoch": 0.543201493804108,
      "grad_norm": 0.3175908029079437,
      "learning_rate": 5.136316373436343e-05,
      "loss": 0.6203,
      "step": 800
    },
    {
      "epoch": 0.5499915124766593,
      "grad_norm": 0.33497434854507446,
      "learning_rate": 5.0177825624378114e-05,
      "loss": 0.6257,
      "step": 810
    },
    {
      "epoch": 0.5567815311492107,
      "grad_norm": 0.3095554709434509,
      "learning_rate": 4.8992387550770373e-05,
      "loss": 0.6174,
      "step": 820
    },
    {
      "epoch": 0.563571549821762,
      "grad_norm": 0.3134996294975281,
      "learning_rate": 4.780751590053813e-05,
      "loss": 0.6041,
      "step": 830
    },
    {
      "epoch": 0.5703615684943134,
      "grad_norm": 0.34364205598831177,
      "learning_rate": 4.662387674226771e-05,
      "loss": 0.6102,
      "step": 840
    },
    {
      "epoch": 0.5771515871668647,
      "grad_norm": 0.3402939438819885,
      "learning_rate": 4.544213545170741e-05,
      "loss": 0.6238,
      "step": 850
    },
    {
      "epoch": 0.583941605839416,
      "grad_norm": 0.3603513240814209,
      "learning_rate": 4.4262956337730385e-05,
      "loss": 0.6338,
      "step": 860
    },
    {
      "epoch": 0.5907316245119674,
      "grad_norm": 0.3301768898963928,
      "learning_rate": 4.3087002268897434e-05,
      "loss": 0.6249,
      "step": 870
    },
    {
      "epoch": 0.5975216431845187,
      "grad_norm": 0.32261279225349426,
      "learning_rate": 4.191493430082929e-05,
      "loss": 0.6174,
      "step": 880
    },
    {
      "epoch": 0.6043116618570701,
      "grad_norm": 0.3522423505783081,
      "learning_rate": 4.074741130459813e-05,
      "loss": 0.6227,
      "step": 890
    },
    {
      "epoch": 0.6111016805296214,
      "grad_norm": 0.3201356828212738,
      "learning_rate": 3.9585089596347094e-05,
      "loss": 0.6335,
      "step": 900
    },
    {
      "epoch": 0.6178916992021728,
      "grad_norm": 0.3597421944141388,
      "learning_rate": 3.842862256834613e-05,
      "loss": 0.6325,
      "step": 910
    },
    {
      "epoch": 0.6246817178747242,
      "grad_norm": 0.30802375078201294,
      "learning_rate": 3.727866032169127e-05,
      "loss": 0.6161,
      "step": 920
    },
    {
      "epoch": 0.6314717365472755,
      "grad_norm": 0.3393535912036896,
      "learning_rate": 3.6135849300854225e-05,
      "loss": 0.6192,
      "step": 930
    },
    {
      "epoch": 0.6382617552198269,
      "grad_norm": 0.3308396339416504,
      "learning_rate": 3.500083193028741e-05,
      "loss": 0.6138,
      "step": 940
    },
    {
      "epoch": 0.6450517738923782,
      "grad_norm": 0.33988067507743835,
      "learning_rate": 3.387424625328892e-05,
      "loss": 0.6113,
      "step": 950
    },
    {
      "epoch": 0.6518417925649296,
      "grad_norm": 0.3523455262184143,
      "learning_rate": 3.275672557333015e-05,
      "loss": 0.6162,
      "step": 960
    },
    {
      "epoch": 0.6586318112374809,
      "grad_norm": 0.3228614628314972,
      "learning_rate": 3.164889809804808e-05,
      "loss": 0.6062,
      "step": 970
    },
    {
      "epoch": 0.6654218299100323,
      "grad_norm": 0.311490923166275,
      "learning_rate": 3.055138658610202e-05,
      "loss": 0.607,
      "step": 980
    },
    {
      "epoch": 0.6722118485825836,
      "grad_norm": 0.33996137976646423,
      "learning_rate": 2.946480799709358e-05,
      "loss": 0.6144,
      "step": 990
    },
    {
      "epoch": 0.679001867255135,
      "grad_norm": 0.3363693952560425,
      "learning_rate": 2.83897731447464e-05,
      "loss": 0.6031,
      "step": 1000
    },
    {
      "epoch": 0.6857918859276863,
      "grad_norm": 0.32421648502349854,
      "learning_rate": 2.7326886353540916e-05,
      "loss": 0.6148,
      "step": 1010
    },
    {
      "epoch": 0.6925819046002376,
      "grad_norm": 0.3457362949848175,
      "learning_rate": 2.627674511899678e-05,
      "loss": 0.6169,
      "step": 1020
    },
    {
      "epoch": 0.699371923272789,
      "grad_norm": 0.3354722261428833,
      "learning_rate": 2.5239939771794563e-05,
      "loss": 0.6158,
      "step": 1030
    },
    {
      "epoch": 0.7061619419453403,
      "grad_norm": 0.4128516614437103,
      "learning_rate": 2.421705314592452e-05,
      "loss": 0.6138,
      "step": 1040
    },
    {
      "epoch": 0.7129519606178917,
      "grad_norm": 0.3348320722579956,
      "learning_rate": 2.3208660251050158e-05,
      "loss": 0.6089,
      "step": 1050
    },
    {
      "epoch": 0.719741979290443,
      "grad_norm": 0.36343589425086975,
      "learning_rate": 2.2215327949269886e-05,
      "loss": 0.6155,
      "step": 1060
    },
    {
      "epoch": 0.7265319979629944,
      "grad_norm": 0.3468243479728699,
      "learning_rate": 2.123761463645878e-05,
      "loss": 0.6282,
      "step": 1070
    },
    {
      "epoch": 0.7333220166355457,
      "grad_norm": 0.37657150626182556,
      "learning_rate": 2.0276069928369746e-05,
      "loss": 0.6094,
      "step": 1080
    },
    {
      "epoch": 0.7401120353080971,
      "grad_norm": 0.3502194583415985,
      "learning_rate": 1.933123435167018e-05,
      "loss": 0.6191,
      "step": 1090
    },
    {
      "epoch": 0.7469020539806485,
      "grad_norm": 0.3248715102672577,
      "learning_rate": 1.840363904008819e-05,
      "loss": 0.6099,
      "step": 1100
    },
    {
      "epoch": 0.7536920726531998,
      "grad_norm": 0.33478298783302307,
      "learning_rate": 1.7493805435838874e-05,
      "loss": 0.6126,
      "step": 1110
    },
    {
      "epoch": 0.7604820913257512,
      "grad_norm": 0.3416365385055542,
      "learning_rate": 1.660224499649861e-05,
      "loss": 0.6202,
      "step": 1120
    },
    {
      "epoch": 0.7672721099983025,
      "grad_norm": 0.37717097997665405,
      "learning_rate": 1.5729458907492327e-05,
      "loss": 0.6179,
      "step": 1130
    },
    {
      "epoch": 0.7740621286708539,
      "grad_norm": 0.33141157031059265,
      "learning_rate": 1.4875937800354988e-05,
      "loss": 0.6101,
      "step": 1140
    },
    {
      "epoch": 0.7808521473434052,
      "grad_norm": 0.3471899926662445,
      "learning_rate": 1.404216147692598e-05,
      "loss": 0.6176,
      "step": 1150
    },
    {
      "epoch": 0.7876421660159565,
      "grad_norm": 0.349560409784317,
      "learning_rate": 1.3228598639631418e-05,
      "loss": 0.6166,
      "step": 1160
    },
    {
      "epoch": 0.7944321846885078,
      "grad_norm": 0.3692922592163086,
      "learning_rate": 1.2435706628005767e-05,
      "loss": 0.6067,
      "step": 1170
    },
    {
      "epoch": 0.8012222033610592,
      "grad_norm": 0.3713966906070709,
      "learning_rate": 1.1663931161601188e-05,
      "loss": 0.6105,
      "step": 1180
    },
    {
      "epoch": 0.8080122220336106,
      "grad_norm": 0.3440631031990051,
      "learning_rate": 1.0913706089428932e-05,
      "loss": 0.6,
      "step": 1190
    },
    {
      "epoch": 0.8148022407061619,
      "grad_norm": 0.35651665925979614,
      "learning_rate": 1.0185453146073604e-05,
      "loss": 0.6151,
      "step": 1200
    },
    {
      "epoch": 0.8215922593787133,
      "grad_norm": 0.4277671277523041,
      "learning_rate": 9.479581714617668e-06,
      "loss": 0.6116,
      "step": 1210
    },
    {
      "epoch": 0.8283822780512646,
      "grad_norm": 0.3536016345024109,
      "learning_rate": 8.796488596509133e-06,
      "loss": 0.6179,
      "step": 1220
    },
    {
      "epoch": 0.835172296723816,
      "grad_norm": 0.3368649184703827,
      "learning_rate": 8.136557788501903e-06,
      "loss": 0.6168,
      "step": 1230
    },
    {
      "epoch": 0.8419623153963673,
      "grad_norm": 0.3577621281147003,
      "learning_rate": 7.500160266794371e-06,
      "loss": 0.6163,
      "step": 1240
    },
    {
      "epoch": 0.8487523340689187,
      "grad_norm": 0.34164294600486755,
      "learning_rate": 6.887653778487307e-06,
      "loss": 0.6151,
      "step": 1250
    },
    {
      "epoch": 0.8555423527414701,
      "grad_norm": 0.3569304049015045,
      "learning_rate": 6.2993826404783965e-06,
      "loss": 0.6141,
      "step": 1260
    },
    {
      "epoch": 0.8623323714140214,
      "grad_norm": 0.37374386191368103,
      "learning_rate": 5.735677545906626e-06,
      "loss": 0.6161,
      "step": 1270
    },
    {
      "epoch": 0.8691223900865728,
      "grad_norm": 0.3596641719341278,
      "learning_rate": 5.196855378254989e-06,
      "loss": 0.6075,
      "step": 1280
    },
    {
      "epoch": 0.8759124087591241,
      "grad_norm": 0.35499128699302673,
      "learning_rate": 4.683219033216402e-06,
      "loss": 0.6001,
      "step": 1290
    },
    {
      "epoch": 0.8827024274316755,
      "grad_norm": 0.32738348841667175,
      "learning_rate": 4.1950572484226345e-06,
      "loss": 0.6094,
      "step": 1300
    },
    {
      "epoch": 0.8894924461042268,
      "grad_norm": 0.3689960539340973,
      "learning_rate": 3.7326444411321547e-06,
      "loss": 0.6119,
      "step": 1310
    },
    {
      "epoch": 0.8962824647767781,
      "grad_norm": 0.3303182125091553,
      "learning_rate": 3.2962405539681217e-06,
      "loss": 0.6084,
      "step": 1320
    },
    {
      "epoch": 0.9030724834493294,
      "grad_norm": 0.3562391698360443,
      "learning_rate": 2.8860909087931543e-06,
      "loss": 0.6103,
      "step": 1330
    },
    {
      "epoch": 0.9098625021218808,
      "grad_norm": 0.3637941777706146,
      "learning_rate": 2.5024260688030987e-06,
      "loss": 0.5962,
      "step": 1340
    },
    {
      "epoch": 0.9166525207944322,
      "grad_norm": 0.3294752240180969,
      "learning_rate": 2.145461708917312e-06,
      "loss": 0.6043,
      "step": 1350
    },
    {
      "epoch": 0.9234425394669835,
      "grad_norm": 0.35764122009277344,
      "learning_rate": 1.8153984945382452e-06,
      "loss": 0.6038,
      "step": 1360
    },
    {
      "epoch": 0.9302325581395349,
      "grad_norm": 0.35339877009391785,
      "learning_rate": 1.512421968748623e-06,
      "loss": 0.5948,
      "step": 1370
    },
    {
      "epoch": 0.9370225768120862,
      "grad_norm": 0.3505209982395172,
      "learning_rate": 1.2367024480094691e-06,
      "loss": 0.62,
      "step": 1380
    },
    {
      "epoch": 0.9438125954846376,
      "grad_norm": 0.3490634560585022,
      "learning_rate": 9.883949264176962e-07,
      "loss": 0.5942,
      "step": 1390
    },
    {
      "epoch": 0.9506026141571889,
      "grad_norm": 0.4094849228858948,
      "learning_rate": 7.676389885771518e-07,
      "loss": 0.6072,
      "step": 1400
    },
    {
      "epoch": 0.9573926328297403,
      "grad_norm": 0.3647181987762451,
      "learning_rate": 5.745587311318968e-07,
      "loss": 0.603,
      "step": 1410
    },
    {
      "epoch": 0.9641826515022917,
      "grad_norm": 0.3373996615409851,
      "learning_rate": 4.0926269300603503e-07,
      "loss": 0.6036,
      "step": 1420
    },
    {
      "epoch": 0.970972670174843,
      "grad_norm": 0.3408346176147461,
      "learning_rate": 2.71843794389226e-07,
      "loss": 0.6101,
      "step": 1430
    },
    {
      "epoch": 0.9777626888473944,
      "grad_norm": 0.361774742603302,
      "learning_rate": 1.623792845020955e-07,
      "loss": 0.5909,
      "step": 1440
    },
    {
      "epoch": 0.9845527075199457,
      "grad_norm": 0.3575396239757538,
      "learning_rate": 8.093069817109445e-08,
      "loss": 0.5922,
      "step": 1450
    },
    {
      "epoch": 0.9913427261924971,
      "grad_norm": 0.3574063181877136,
      "learning_rate": 2.7543821237030475e-08,
      "loss": 0.6109,
      "step": 1460
    },
    {
      "epoch": 0.9981327448650483,
      "grad_norm": 0.33741241693496704,
      "learning_rate": 2.2486648168396075e-09,
      "loss": 0.5931,
      "step": 1470
    }
  ],
  "logging_steps": 10,
  "max_steps": 1473,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "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": 2.1892094394273628e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}