{
  "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": 102.95948028564453,
      "learning_rate": 2.2388059701492537e-07,
      "loss": 0.9106237292289734,
      "step": 1,
      "token_acc": 0.8820332480818415
    },
    {
      "epoch": 0.015082956259426848,
      "grad_norm": 86.42596435546875,
      "learning_rate": 2.2388059701492535e-06,
      "loss": 0.7358827590942383,
      "step": 10,
      "token_acc": 0.8881174403675924
    },
    {
      "epoch": 0.030165912518853696,
      "grad_norm": 8.159674644470215,
      "learning_rate": 4.477611940298507e-06,
      "loss": 0.22912912368774413,
      "step": 20,
      "token_acc": 0.9142753249901518
    },
    {
      "epoch": 0.04524886877828054,
      "grad_norm": 5.822969913482666,
      "learning_rate": 6.716417910447762e-06,
      "loss": 0.16711633205413817,
      "step": 30,
      "token_acc": 0.931531673495115
    },
    {
      "epoch": 0.06033182503770739,
      "grad_norm": 3.3466780185699463,
      "learning_rate": 8.955223880597014e-06,
      "loss": 0.18157639503479003,
      "step": 40,
      "token_acc": 0.9353571989547486
    },
    {
      "epoch": 0.07541478129713423,
      "grad_norm": 11.269857406616211,
      "learning_rate": 1.119402985074627e-05,
      "loss": 0.1465624213218689,
      "step": 50,
      "token_acc": 0.9408342341196668
    },
    {
      "epoch": 0.09049773755656108,
      "grad_norm": 4.60915470123291,
      "learning_rate": 1.3432835820895523e-05,
      "loss": 0.14280350208282472,
      "step": 60,
      "token_acc": 0.942812610470161
    },
    {
      "epoch": 0.10558069381598793,
      "grad_norm": 5.3014421463012695,
      "learning_rate": 1.4999789854529206e-05,
      "loss": 0.14221720695495604,
      "step": 70,
      "token_acc": 0.9433893684688778
    },
    {
      "epoch": 0.12066365007541478,
      "grad_norm": 2.8947246074676514,
      "learning_rate": 1.4996054262643387e-05,
      "loss": 0.140693998336792,
      "step": 80,
      "token_acc": 0.9434419606786965
    },
    {
      "epoch": 0.13574660633484162,
      "grad_norm": 27.04346466064453,
      "learning_rate": 1.498765144861652e-05,
      "loss": 0.14304617643356324,
      "step": 90,
      "token_acc": 0.9435046042172831
    },
    {
      "epoch": 0.15082956259426847,
      "grad_norm": 2.744962215423584,
      "learning_rate": 1.4974586644241e-05,
      "loss": 0.1404696822166443,
      "step": 100,
      "token_acc": 0.944554310407969
    },
    {
      "epoch": 0.16591251885369532,
      "grad_norm": 2.598541259765625,
      "learning_rate": 1.4956867983975587e-05,
      "loss": 0.13427730798721313,
      "step": 110,
      "token_acc": 0.9465170435897036
    },
    {
      "epoch": 0.18099547511312217,
      "grad_norm": 2.129733085632324,
      "learning_rate": 1.493450649988069e-05,
      "loss": 0.13516327142715454,
      "step": 120,
      "token_acc": 0.9463536368260427
    },
    {
      "epoch": 0.19607843137254902,
      "grad_norm": 1.705326795578003,
      "learning_rate": 1.4907516114749562e-05,
      "loss": 0.13605421781539917,
      "step": 130,
      "token_acc": 0.9467877502727454
    },
    {
      "epoch": 0.21116138763197587,
      "grad_norm": 1.479612112045288,
      "learning_rate": 1.487591363343962e-05,
      "loss": 0.1342160224914551,
      "step": 140,
      "token_acc": 0.9476240697553512
    },
    {
      "epoch": 0.22624434389140272,
      "grad_norm": 1.3483514785766602,
      "learning_rate": 1.4839718732409347e-05,
      "loss": 0.13110628128051757,
      "step": 150,
      "token_acc": 0.9483754956505702
    },
    {
      "epoch": 0.24132730015082957,
      "grad_norm": 1.838326334953308,
      "learning_rate": 1.4798953947467258e-05,
      "loss": 0.13084776401519777,
      "step": 160,
      "token_acc": 0.9485363406997599
    },
    {
      "epoch": 0.2564102564102564,
      "grad_norm": 1.149587631225586,
      "learning_rate": 1.4753644659740567e-05,
      "loss": 0.1279315710067749,
      "step": 170,
      "token_acc": 0.9509291300172259
    },
    {
      "epoch": 0.27149321266968324,
      "grad_norm": 1.747382402420044,
      "learning_rate": 1.4703819079872297e-05,
      "loss": 0.12598202228546143,
      "step": 180,
      "token_acc": 0.9506852612417446
    },
    {
      "epoch": 0.2865761689291101,
      "grad_norm": 1.8707523345947266,
      "learning_rate": 1.4649508230456655e-05,
      "loss": 0.12346329689025878,
      "step": 190,
      "token_acc": 0.9505019520356943
    },
    {
      "epoch": 0.30165912518853694,
      "grad_norm": 0.9066284894943237,
      "learning_rate": 1.459074592672363e-05,
      "loss": 0.12236473560333253,
      "step": 200,
      "token_acc": 0.9514878958951296
    },
    {
      "epoch": 0.3167420814479638,
      "grad_norm": 1.7287721633911133,
      "learning_rate": 1.4527568755484822e-05,
      "loss": 0.11908667087554932,
      "step": 210,
      "token_acc": 0.9531890358873669
    },
    {
      "epoch": 0.33182503770739064,
      "grad_norm": 0.9215026497840881,
      "learning_rate": 1.4460016052353615e-05,
      "loss": 0.12045620679855347,
      "step": 220,
      "token_acc": 0.9537771221341201
    },
    {
      "epoch": 0.3469079939668175,
      "grad_norm": 0.9963810443878174,
      "learning_rate": 1.43881298772539e-05,
      "loss": 0.12107706069946289,
      "step": 230,
      "token_acc": 0.952223704330286
    },
    {
      "epoch": 0.36199095022624433,
      "grad_norm": 3.457219362258911,
      "learning_rate": 1.4311954988232549e-05,
      "loss": 0.11647148132324218,
      "step": 240,
      "token_acc": 0.9550424502102673
    },
    {
      "epoch": 0.3770739064856712,
      "grad_norm": 0.9486528038978577,
      "learning_rate": 1.4231538813591998e-05,
      "loss": 0.12029435634613037,
      "step": 250,
      "token_acc": 0.9532897305739946
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 0.8173623085021973,
      "learning_rate": 1.4146931422360251e-05,
      "loss": 0.11927635669708252,
      "step": 260,
      "token_acc": 0.9536245813747428
    },
    {
      "epoch": 0.4072398190045249,
      "grad_norm": 0.7389430999755859,
      "learning_rate": 1.4058185493116703e-05,
      "loss": 0.11926217079162597,
      "step": 270,
      "token_acc": 0.954318820770103
    },
    {
      "epoch": 0.42232277526395173,
      "grad_norm": 0.9154466390609741,
      "learning_rate": 1.396535628119321e-05,
      "loss": 0.11332788467407226,
      "step": 280,
      "token_acc": 0.9558336565138271
    },
    {
      "epoch": 0.4374057315233786,
      "grad_norm": 1.3126405477523804,
      "learning_rate": 1.3868501584270802e-05,
      "loss": 0.11628217697143554,
      "step": 290,
      "token_acc": 0.9555292936815011
    },
    {
      "epoch": 0.45248868778280543,
      "grad_norm": 0.8111864328384399,
      "learning_rate": 1.3767681706393457e-05,
      "loss": 0.11811476945877075,
      "step": 300,
      "token_acc": 0.9534279727296875
    },
    {
      "epoch": 0.4675716440422323,
      "grad_norm": 0.8737362027168274,
      "learning_rate": 1.3662959420421375e-05,
      "loss": 0.11337666511535645,
      "step": 310,
      "token_acc": 0.9552794842162885
    },
    {
      "epoch": 0.48265460030165913,
      "grad_norm": 1.1029027700424194,
      "learning_rate": 1.35543999289471e-05,
      "loss": 0.11433804035186768,
      "step": 320,
      "token_acc": 0.9556174728604272
    },
    {
      "epoch": 0.497737556561086,
      "grad_norm": 0.6877908110618591,
      "learning_rate": 1.3442070823698822e-05,
      "loss": 0.11494088172912598,
      "step": 330,
      "token_acc": 0.9556713117551905
    },
    {
      "epoch": 0.5128205128205128,
      "grad_norm": 1.1351238489151,
      "learning_rate": 1.332604204345616e-05,
      "loss": 0.11259114742279053,
      "step": 340,
      "token_acc": 0.9564803200863725
    },
    {
      "epoch": 0.5279034690799397,
      "grad_norm": 1.1910680532455444,
      "learning_rate": 1.3206385830504614e-05,
      "loss": 0.1151119589805603,
      "step": 350,
      "token_acc": 0.9553413035094486
    },
    {
      "epoch": 0.5429864253393665,
      "grad_norm": 0.99884432554245,
      "learning_rate": 1.3083176685655786e-05,
      "loss": 0.10849617719650269,
      "step": 360,
      "token_acc": 0.9579699284085654
    },
    {
      "epoch": 0.5580693815987934,
      "grad_norm": 0.6889309883117676,
      "learning_rate": 1.295649132186141e-05,
      "loss": 0.11037075519561768,
      "step": 370,
      "token_acc": 0.9574533969158358
    },
    {
      "epoch": 0.5731523378582202,
      "grad_norm": 0.6279752254486084,
      "learning_rate": 1.2826408616450027e-05,
      "loss": 0.112148118019104,
      "step": 380,
      "token_acc": 0.957041847041847
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.7004132270812988,
      "learning_rate": 1.2693009562016084e-05,
      "loss": 0.1127288818359375,
      "step": 390,
      "token_acc": 0.956207282583078
    },
    {
      "epoch": 0.6033182503770739,
      "grad_norm": 0.7210747599601746,
      "learning_rate": 1.2556377215992019e-05,
      "loss": 0.11080955266952515,
      "step": 400,
      "token_acc": 0.9577135638717827
    },
    {
      "epoch": 0.6184012066365008,
      "grad_norm": 0.8611845374107361,
      "learning_rate": 1.2416596648934716e-05,
      "loss": 0.10928909778594971,
      "step": 410,
      "token_acc": 0.9575848530325769
    },
    {
      "epoch": 0.6334841628959276,
      "grad_norm": 0.7873753309249878,
      "learning_rate": 1.2273754891558542e-05,
      "loss": 0.11083755493164063,
      "step": 420,
      "token_acc": 0.9573786571441708
    },
    {
      "epoch": 0.6485671191553545,
      "grad_norm": 0.8866596221923828,
      "learning_rate": 1.212794088054797e-05,
      "loss": 0.11016602516174316,
      "step": 430,
      "token_acc": 0.9571059510524788
    },
    {
      "epoch": 0.6636500754147813,
      "grad_norm": 0.5996086001396179,
      "learning_rate": 1.1979245403183451e-05,
      "loss": 0.10762960910797119,
      "step": 440,
      "token_acc": 0.9582261423496263
    },
    {
      "epoch": 0.6787330316742082,
      "grad_norm": 0.8173403739929199,
      "learning_rate": 1.182776104081511e-05,
      "loss": 0.1057842493057251,
      "step": 450,
      "token_acc": 0.9589248277927523
    },
    {
      "epoch": 0.693815987933635,
      "grad_norm": 0.8952581882476807,
      "learning_rate": 1.167358211121938e-05,
      "loss": 0.10421741008758545,
      "step": 460,
      "token_acc": 0.959809739243605
    },
    {
      "epoch": 0.7088989441930619,
      "grad_norm": 0.8190169334411621,
      "learning_rate": 1.1516804609874507e-05,
      "loss": 0.10857417583465576,
      "step": 470,
      "token_acc": 0.9578637775543585
    },
    {
      "epoch": 0.7239819004524887,
      "grad_norm": 0.8312342762947083,
      "learning_rate": 1.1357526150191471e-05,
      "loss": 0.10505218505859375,
      "step": 480,
      "token_acc": 0.958963653076168
    },
    {
      "epoch": 0.7390648567119156,
      "grad_norm": 0.6930398941040039,
      "learning_rate": 1.1195845902737558e-05,
      "loss": 0.10784991979598998,
      "step": 490,
      "token_acc": 0.9578753158963705
    },
    {
      "epoch": 0.7541478129713424,
      "grad_norm": 0.7167505025863647,
      "learning_rate": 1.1031864533490397e-05,
      "loss": 0.10691661834716797,
      "step": 500,
      "token_acc": 0.9590415199708632
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 0.9776201844215393,
      "learning_rate": 1.0865684141160912e-05,
      "loss": 0.10396910905838012,
      "step": 510,
      "token_acc": 0.9596143463170074
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 0.6151344180107117,
      "learning_rate": 1.0697408193624248e-05,
      "loss": 0.10538628101348876,
      "step": 520,
      "token_acc": 0.9587614615490946
    },
    {
      "epoch": 0.799396681749623,
      "grad_norm": 0.6072873473167419,
      "learning_rate": 1.0527141463498186e-05,
      "loss": 0.10715539455413818,
      "step": 530,
      "token_acc": 0.9581875852255884
    },
    {
      "epoch": 0.8144796380090498,
      "grad_norm": 0.6179074048995972,
      "learning_rate": 1.0354989962909223e-05,
      "loss": 0.10194606781005859,
      "step": 540,
      "token_acc": 0.9605654176952467
    },
    {
      "epoch": 0.8295625942684767,
      "grad_norm": 0.6735125184059143,
      "learning_rate": 1.0181060877486896e-05,
      "loss": 0.10430541038513183,
      "step": 550,
      "token_acc": 0.9597842006157157
    },
    {
      "epoch": 0.8446455505279035,
      "grad_norm": 0.9804912209510803,
      "learning_rate": 1.0005462499627443e-05,
      "loss": 0.09949544668197632,
      "step": 560,
      "token_acc": 0.9620176783326266
    },
    {
      "epoch": 0.8597285067873304,
      "grad_norm": 0.7045010328292847,
      "learning_rate": 9.82830416106839e-06,
      "loss": 0.10367201566696167,
      "step": 570,
      "token_acc": 0.96001593625498
    },
    {
      "epoch": 0.8748114630467572,
      "grad_norm": 0.7430729866027832,
      "learning_rate": 9.649696164815983e-06,
      "loss": 0.09992576837539673,
      "step": 580,
      "token_acc": 0.9614963024717089
    },
    {
      "epoch": 0.889894419306184,
      "grad_norm": 0.6927557587623596,
      "learning_rate": 9.469749716467922e-06,
      "loss": 0.10162270069122314,
      "step": 590,
      "token_acc": 0.9600887950572822
    },
    {
      "epoch": 0.9049773755656109,
      "grad_norm": 0.7048885226249695,
      "learning_rate": 9.288576854974107e-06,
      "loss": 0.1029390573501587,
      "step": 600,
      "token_acc": 0.9599495381581428
    },
    {
      "epoch": 0.9200603318250377,
      "grad_norm": 0.7786235809326172,
      "learning_rate": 9.106290382878496e-06,
      "loss": 0.10250804424285889,
      "step": 610,
      "token_acc": 0.9600262723773446
    },
    {
      "epoch": 0.9351432880844646,
      "grad_norm": 0.6432821154594421,
      "learning_rate": 8.923003796085571e-06,
      "loss": 0.09980993270874024,
      "step": 620,
      "token_acc": 0.9616151866151866
    },
    {
      "epoch": 0.9502262443438914,
      "grad_norm": 0.4739595353603363,
      "learning_rate": 8.738831213195076e-06,
      "loss": 0.10440936088562011,
      "step": 630,
      "token_acc": 0.9596175053688548
    },
    {
      "epoch": 0.9653092006033183,
      "grad_norm": 0.5999048352241516,
      "learning_rate": 8.553887304449051e-06,
      "loss": 0.09771686196327209,
      "step": 640,
      "token_acc": 0.9621664920820753
    },
    {
      "epoch": 0.9803921568627451,
      "grad_norm": 0.6257745027542114,
      "learning_rate": 8.368287220335405e-06,
      "loss": 0.09986623525619506,
      "step": 650,
      "token_acc": 0.9612191817091594
    },
    {
      "epoch": 0.995475113122172,
      "grad_norm": 0.6971683502197266,
      "learning_rate": 8.18214651989248e-06,
      "loss": 0.10077552795410157,
      "step": 660,
      "token_acc": 0.9610657983311677
    },
    {
      "epoch": 1.0105580693815988,
      "grad_norm": 0.5287856459617615,
      "learning_rate": 7.99558109875924e-06,
      "loss": 0.09711869359016419,
      "step": 670,
      "token_acc": 0.9632017856589268
    },
    {
      "epoch": 1.0256410256410255,
      "grad_norm": 0.5113791227340698,
      "learning_rate": 7.808707117015871e-06,
      "loss": 0.09611727595329285,
      "step": 680,
      "token_acc": 0.9628295031473811
    },
    {
      "epoch": 1.0407239819004526,
      "grad_norm": 0.9405093193054199,
      "learning_rate": 7.621640926859767e-06,
      "loss": 0.09766757488250732,
      "step": 690,
      "token_acc": 0.961746076320634
    },
    {
      "epoch": 1.0558069381598794,
      "grad_norm": 0.7931028604507446,
      "learning_rate": 7.434499000161847e-06,
      "loss": 0.09249839186668396,
      "step": 700,
      "token_acc": 0.9642857142857143
    },
    {
      "epoch": 1.0708898944193062,
      "grad_norm": 0.5711884498596191,
      "learning_rate": 7.247397855948417e-06,
      "loss": 0.09391047954559326,
      "step": 710,
      "token_acc": 0.9633587119502715
    },
    {
      "epoch": 1.085972850678733,
      "grad_norm": 0.5200867056846619,
      "learning_rate": 7.060453987853635e-06,
      "loss": 0.09218854904174804,
      "step": 720,
      "token_acc": 0.9643104698960264
    },
    {
      "epoch": 1.10105580693816,
      "grad_norm": 0.5937605500221252,
      "learning_rate": 6.873783791587794e-06,
      "loss": 0.09332010746002198,
      "step": 730,
      "token_acc": 0.963314348369019
    },
    {
      "epoch": 1.1161387631975868,
      "grad_norm": 0.5634381771087646,
      "learning_rate": 6.6875034924665825e-06,
      "loss": 0.09174969792366028,
      "step": 740,
      "token_acc": 0.9643075846382644
    },
    {
      "epoch": 1.1312217194570136,
      "grad_norm": 0.575001060962677,
      "learning_rate": 6.501729073046421e-06,
      "loss": 0.09115147590637207,
      "step": 750,
      "token_acc": 0.96492840783385
    },
    {
      "epoch": 1.1463046757164403,
      "grad_norm": 0.5640040040016174,
      "learning_rate": 6.316576200910982e-06,
      "loss": 0.08881928324699402,
      "step": 760,
      "token_acc": 0.9655490814149712
    },
    {
      "epoch": 1.1613876319758674,
      "grad_norm": 0.6095544099807739,
      "learning_rate": 6.1321601566537614e-06,
      "loss": 0.09392951130867004,
      "step": 770,
      "token_acc": 0.9636105045397081
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 0.6617202758789062,
      "learning_rate": 5.94859576210165e-06,
      "loss": 0.09372889995574951,
      "step": 780,
      "token_acc": 0.9637654206186845
    },
    {
      "epoch": 1.191553544494721,
      "grad_norm": 1.2394925355911255,
      "learning_rate": 5.765997308824127e-06,
      "loss": 0.09473921060562134,
      "step": 790,
      "token_acc": 0.9626901771085278
    },
    {
      "epoch": 1.2066365007541477,
      "grad_norm": 0.5932070016860962,
      "learning_rate": 5.584478486972594e-06,
      "loss": 0.09192266464233398,
      "step": 800,
      "token_acc": 0.9635680926003507
    },
    {
      "epoch": 1.2217194570135748,
      "grad_norm": 0.9952248930931091,
      "learning_rate": 5.404152314494185e-06,
      "loss": 0.09385214447975158,
      "step": 810,
      "token_acc": 0.9637176022601646
    },
    {
      "epoch": 1.2368024132730016,
      "grad_norm": 0.6110525727272034,
      "learning_rate": 5.225131066764096e-06,
      "loss": 0.08856210112571716,
      "step": 820,
      "token_acc": 0.9655632553414105
    },
    {
      "epoch": 1.2518853695324283,
      "grad_norm": 0.6847214698791504,
      "learning_rate": 5.04752620668025e-06,
      "loss": 0.09417049884796143,
      "step": 830,
      "token_acc": 0.9632891549095055
    },
    {
      "epoch": 1.2669683257918551,
      "grad_norm": 0.6929960250854492,
      "learning_rate": 4.871448315263849e-06,
      "loss": 0.09022506475448608,
      "step": 840,
      "token_acc": 0.9648385764536664
    },
    {
      "epoch": 1.282051282051282,
      "grad_norm": 0.5659875869750977,
      "learning_rate": 4.697007022808989e-06,
      "loss": 0.08967013955116272,
      "step": 850,
      "token_acc": 0.9652773666449589
    },
    {
      "epoch": 1.297134238310709,
      "grad_norm": 0.819989025592804,
      "learning_rate": 4.52431094062422e-06,
      "loss": 0.09020506143569947,
      "step": 860,
      "token_acc": 0.964875810350934
    },
    {
      "epoch": 1.3122171945701357,
      "grad_norm": 0.6961911916732788,
      "learning_rate": 4.353467593408566e-06,
      "loss": 0.09065240621566772,
      "step": 870,
      "token_acc": 0.964779952624361
    },
    {
      "epoch": 1.3273001508295625,
      "grad_norm": 0.711963951587677,
      "learning_rate": 4.1845833523040775e-06,
      "loss": 0.09238415956497192,
      "step": 880,
      "token_acc": 0.9640809911242604
    },
    {
      "epoch": 1.3423831070889896,
      "grad_norm": 0.5482915639877319,
      "learning_rate": 4.0177633686666225e-06,
      "loss": 0.08626269102096558,
      "step": 890,
      "token_acc": 0.9664413009944534
    },
    {
      "epoch": 1.3574660633484164,
      "grad_norm": 0.6547002792358398,
      "learning_rate": 3.853111508596134e-06,
      "loss": 0.08824948668479919,
      "step": 900,
      "token_acc": 0.9652458695864699
    },
    {
      "epoch": 1.3725490196078431,
      "grad_norm": 0.7393091917037964,
      "learning_rate": 3.6907302882671214e-06,
      "loss": 0.0905347228050232,
      "step": 910,
      "token_acc": 0.965180890257897
    },
    {
      "epoch": 1.38763197586727,
      "grad_norm": 0.7055743932723999,
      "learning_rate": 3.5307208100996376e-06,
      "loss": 0.08957746028900146,
      "step": 920,
      "token_acc": 0.965201436723524
    },
    {
      "epoch": 1.4027149321266967,
      "grad_norm": 0.5165733695030212,
      "learning_rate": 3.373182699810508e-06,
      "loss": 0.0900674045085907,
      "step": 930,
      "token_acc": 0.9648800220649076
    },
    {
      "epoch": 1.4177978883861238,
      "grad_norm": 0.6470308303833008,
      "learning_rate": 3.2182140443839834e-06,
      "loss": 0.09126996994018555,
      "step": 940,
      "token_acc": 0.9645374984806127
    },
    {
      "epoch": 1.4328808446455505,
      "grad_norm": 0.7675370573997498,
      "learning_rate": 3.0659113310004437e-06,
      "loss": 0.09289331436157226,
      "step": 950,
      "token_acc": 0.9638552426246563
    },
    {
      "epoch": 1.4479638009049773,
      "grad_norm": 0.5420223474502563,
      "learning_rate": 2.9163693869611806e-06,
      "loss": 0.08824996948242188,
      "step": 960,
      "token_acc": 0.9658039529093536
    },
    {
      "epoch": 1.4630467571644044,
      "grad_norm": 0.49305713176727295,
      "learning_rate": 2.7696813206466494e-06,
      "loss": 0.0872989296913147,
      "step": 970,
      "token_acc": 0.966210141452889
    },
    {
      "epoch": 1.4781297134238311,
      "grad_norm": 0.5599616169929504,
      "learning_rate": 2.6259384635449823e-06,
      "loss": 0.08567045331001281,
      "step": 980,
      "token_acc": 0.9661950765380095
    },
    {
      "epoch": 1.493212669683258,
      "grad_norm": 0.5709689259529114,
      "learning_rate": 2.48523031338683e-06,
      "loss": 0.09245221614837647,
      "step": 990,
      "token_acc": 0.9648156761541808
    },
    {
      "epoch": 1.5082956259426847,
      "grad_norm": 0.6077988147735596,
      "learning_rate": 2.347644478421916e-06,
      "loss": 0.09017724990844726,
      "step": 1000,
      "token_acc": 0.9644074377669443
    },
    {
      "epoch": 1.5233785822021115,
      "grad_norm": 0.5771783590316772,
      "learning_rate": 2.2132666228720783e-06,
      "loss": 0.09128254652023315,
      "step": 1010,
      "token_acc": 0.964498933901919
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 0.5295813083648682,
      "learning_rate": 2.0821804135946685e-06,
      "loss": 0.0867614507675171,
      "step": 1020,
      "token_acc": 0.9657581438574565
    },
    {
      "epoch": 1.5535444947209653,
      "grad_norm": 0.6218671202659607,
      "learning_rate": 1.954467467989575e-06,
      "loss": 0.08816643357276917,
      "step": 1030,
      "token_acc": 0.9664409193521507
    },
    {
      "epoch": 1.5686274509803921,
      "grad_norm": 0.5615459084510803,
      "learning_rate": 1.8302073031822736e-06,
      "loss": 0.08652478456497192,
      "step": 1040,
      "token_acc": 0.9657081486635708
    },
    {
      "epoch": 1.5837104072398192,
      "grad_norm": 0.6350222229957581,
      "learning_rate": 1.7094772865145908e-06,
      "loss": 0.08563481569290161,
      "step": 1050,
      "token_acc": 0.9670686976438743
    },
    {
      "epoch": 1.598793363499246,
      "grad_norm": 0.611201286315918,
      "learning_rate": 1.5923525873739312e-06,
      "loss": 0.08522043824195862,
      "step": 1060,
      "token_acc": 0.9665700207551056
    },
    {
      "epoch": 1.6138763197586727,
      "grad_norm": 0.5586323142051697,
      "learning_rate": 1.4789061303910304e-06,
      "loss": 0.09052041172981262,
      "step": 1070,
      "token_acc": 0.9644768042804159
    },
    {
      "epoch": 1.6289592760180995,
      "grad_norm": 0.6737985014915466,
      "learning_rate": 1.3692085500353444e-06,
      "loss": 0.08938039541244507,
      "step": 1080,
      "token_acc": 0.965438055926981
    },
    {
      "epoch": 1.6440422322775263,
      "grad_norm": 0.5793459415435791,
      "learning_rate": 1.263328146636342e-06,
      "loss": 0.08627030849456788,
      "step": 1090,
      "token_acc": 0.9662307459500789
    },
    {
      "epoch": 1.6591251885369531,
      "grad_norm": 0.553979754447937,
      "learning_rate": 1.1613308438580924e-06,
      "loss": 0.08699219822883605,
      "step": 1100,
      "token_acc": 0.9655294135702994
    },
    {
      "epoch": 1.6742081447963801,
      "grad_norm": 0.6702016592025757,
      "learning_rate": 1.063280147653618e-06,
      "loss": 0.08392992615699768,
      "step": 1110,
      "token_acc": 0.9671428105534358
    },
    {
      "epoch": 1.689291101055807,
      "grad_norm": 0.5012071132659912,
      "learning_rate": 9.692371067245726e-07,
      "loss": 0.09083313941955566,
      "step": 1120,
      "token_acc": 0.9637969026411347
    },
    {
      "epoch": 1.704374057315234,
      "grad_norm": 0.5176061987876892,
      "learning_rate": 8.792602745108685e-07,
      "loss": 0.0841116487979889,
      "step": 1130,
      "token_acc": 0.9662585332247761
    },
    {
      "epoch": 1.7194570135746607,
      "grad_norm": 0.7949459552764893,
      "learning_rate": 7.93405672733902e-07,
      "loss": 0.08959006071090699,
      "step": 1140,
      "token_acc": 0.9651756949127145
    },
    {
      "epoch": 1.7345399698340875,
      "grad_norm": 0.4745083749294281,
      "learning_rate": 7.117267565160989e-07,
      "loss": 0.08536668419837952,
      "step": 1150,
      "token_acc": 0.966504120772119
    },
    {
      "epoch": 1.7496229260935143,
      "grad_norm": 0.6467885375022888,
      "learning_rate": 6.342743810984836e-07,
      "loss": 0.08896956443786622,
      "step": 1160,
      "token_acc": 0.9651875204473128
    },
    {
      "epoch": 1.7647058823529411,
      "grad_norm": 0.6844164729118347,
      "learning_rate": 5.610967701769814e-07,
      "loss": 0.08681308031082154,
      "step": 1170,
      "token_acc": 0.966941761230423
    },
    {
      "epoch": 1.779788838612368,
      "grad_norm": 0.6476861238479614,
      "learning_rate": 4.922394858772067e-07,
      "loss": 0.087920880317688,
      "step": 1180,
      "token_acc": 0.9660436333785875
    },
    {
      "epoch": 1.7948717948717947,
      "grad_norm": 0.5709762573242188,
      "learning_rate": 4.2774540038639286e-07,
      "loss": 0.0817031741142273,
      "step": 1190,
      "token_acc": 0.9678674654050475
    },
    {
      "epoch": 1.8099547511312217,
      "grad_norm": 0.5973171591758728,
      "learning_rate": 3.676546692601465e-07,
      "loss": 0.08370765447616577,
      "step": 1200,
      "token_acc": 0.9672491137810961
    },
    {
      "epoch": 1.8250377073906485,
      "grad_norm": 0.775956392288208,
      "learning_rate": 3.1200470642063716e-07,
      "loss": 0.0849980354309082,
      "step": 1210,
      "token_acc": 0.9670338146365544
    },
    {
      "epoch": 1.8401206636500755,
      "grad_norm": 0.5492123365402222,
      "learning_rate": 2.6083016086180715e-07,
      "loss": 0.08112341165542603,
      "step": 1220,
      "token_acc": 0.9692397073140093
    },
    {
      "epoch": 1.8552036199095023,
      "grad_norm": 0.5747104287147522,
      "learning_rate": 2.14162895076071e-07,
      "loss": 0.0888913631439209,
      "step": 1230,
      "token_acc": 0.964867373211185
    },
    {
      "epoch": 1.8702865761689291,
      "grad_norm": 0.5583189725875854,
      "learning_rate": 1.7203196521597354e-07,
      "loss": 0.08525767922401428,
      "step": 1240,
      "token_acc": 0.9667044310171199
    },
    {
      "epoch": 1.885369532428356,
      "grad_norm": 0.5149325728416443,
      "learning_rate": 1.3446360300313375e-07,
      "loss": 0.08332073092460632,
      "step": 1250,
      "token_acc": 0.9677986226549513
    },
    {
      "epoch": 1.9004524886877827,
      "grad_norm": 0.6395967602729797,
      "learning_rate": 1.0148119939575207e-07,
      "loss": 0.0832521915435791,
      "step": 1260,
      "token_acc": 0.9674833333333334
    },
    {
      "epoch": 1.9155354449472095,
      "grad_norm": 0.5390537977218628,
      "learning_rate": 7.310529002484079e-08,
      "loss": 0.0882462739944458,
      "step": 1270,
      "token_acc": 0.9657406598526542
    },
    {
      "epoch": 1.9306184012066365,
      "grad_norm": 0.5799115896224976,
      "learning_rate": 4.9353542408249465e-08,
      "loss": 0.08766170740127563,
      "step": 1280,
      "token_acc": 0.9653571815382711
    },
    {
      "epoch": 1.9457013574660633,
      "grad_norm": 0.5990166664123535,
      "learning_rate": 3.024074495044826e-08,
      "loss": 0.0859516441822052,
      "step": 1290,
      "token_acc": 0.9666938227823837
    },
    {
      "epoch": 1.9607843137254903,
      "grad_norm": 1.4371193647384644,
      "learning_rate": 1.5778797734913307e-08,
      "loss": 0.08548210859298706,
      "step": 1300,
      "token_acc": 0.9662036547210592
    },
    {
      "epoch": 1.9758672699849171,
      "grad_norm": 0.5660617351531982,
      "learning_rate": 5.976705114852654e-09,
      "loss": 0.08374635577201843,
      "step": 1310,
      "token_acc": 0.9671063162275506
    },
    {
      "epoch": 1.990950226244344,
      "grad_norm": 0.6013316512107849,
      "learning_rate": 8.40570106880878e-10,
      "loss": 0.08703867197036744,
      "step": 1320,
      "token_acc": 0.9658897186392095
    }
  ],
  "logging_steps": 10,
  "max_steps": 1326,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.5869318005205762e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}