{
  "best_metric": 0.5118223208478402,
  "best_model_checkpoint": "Mrohit01/1_M_cards-swinv2-base-patch4-window12to16-192to256-22kto1k-ft-finetuned-v3/checkpoint-16330",
  "epoch": 9.99540933435348,
  "eval_steps": 500,
  "global_step": 16330,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00612088752869166,
      "grad_norm": 12.164946556091309,
      "learning_rate": 3.061849357011635e-07,
      "loss": 2.3398,
      "step": 10
    },
    {
      "epoch": 0.01224177505738332,
      "grad_norm": 14.423832893371582,
      "learning_rate": 6.12369871402327e-07,
      "loss": 2.3002,
      "step": 20
    },
    {
      "epoch": 0.018362662586074982,
      "grad_norm": 10.38547134399414,
      "learning_rate": 9.185548071034905e-07,
      "loss": 2.2128,
      "step": 30
    },
    {
      "epoch": 0.02448355011476664,
      "grad_norm": 10.298561096191406,
      "learning_rate": 1.224739742804654e-06,
      "loss": 2.0935,
      "step": 40
    },
    {
      "epoch": 0.030604437643458302,
      "grad_norm": 8.211506843566895,
      "learning_rate": 1.5309246785058176e-06,
      "loss": 1.9766,
      "step": 50
    },
    {
      "epoch": 0.036725325172149964,
      "grad_norm": 50.141380310058594,
      "learning_rate": 1.837109614206981e-06,
      "loss": 1.861,
      "step": 60
    },
    {
      "epoch": 0.04284621270084162,
      "grad_norm": 7.632486343383789,
      "learning_rate": 2.1432945499081446e-06,
      "loss": 1.8087,
      "step": 70
    },
    {
      "epoch": 0.04896710022953328,
      "grad_norm": 79.66400909423828,
      "learning_rate": 2.449479485609308e-06,
      "loss": 1.7657,
      "step": 80
    },
    {
      "epoch": 0.05508798775822494,
      "grad_norm": 7.020776271820068,
      "learning_rate": 2.7556644213104716e-06,
      "loss": 1.7149,
      "step": 90
    },
    {
      "epoch": 0.061208875286916604,
      "grad_norm": 5.450502395629883,
      "learning_rate": 3.061849357011635e-06,
      "loss": 1.6573,
      "step": 100
    },
    {
      "epoch": 0.06732976281560826,
      "grad_norm": 5.259347438812256,
      "learning_rate": 3.368034292712799e-06,
      "loss": 1.6454,
      "step": 110
    },
    {
      "epoch": 0.07345065034429993,
      "grad_norm": 13.419878005981445,
      "learning_rate": 3.674219228413962e-06,
      "loss": 1.6188,
      "step": 120
    },
    {
      "epoch": 0.07957153787299158,
      "grad_norm": 12.368359565734863,
      "learning_rate": 3.980404164115126e-06,
      "loss": 1.6115,
      "step": 130
    },
    {
      "epoch": 0.08569242540168324,
      "grad_norm": 8.168076515197754,
      "learning_rate": 4.286589099816289e-06,
      "loss": 1.5702,
      "step": 140
    },
    {
      "epoch": 0.09181331293037491,
      "grad_norm": 9.173910140991211,
      "learning_rate": 4.592774035517453e-06,
      "loss": 1.5632,
      "step": 150
    },
    {
      "epoch": 0.09793420045906656,
      "grad_norm": 95.74823760986328,
      "learning_rate": 4.898958971218616e-06,
      "loss": 1.5163,
      "step": 160
    },
    {
      "epoch": 0.10405508798775823,
      "grad_norm": 7.6970744132995605,
      "learning_rate": 5.20514390691978e-06,
      "loss": 1.5255,
      "step": 170
    },
    {
      "epoch": 0.11017597551644988,
      "grad_norm": 8.821610450744629,
      "learning_rate": 5.511328842620943e-06,
      "loss": 1.5225,
      "step": 180
    },
    {
      "epoch": 0.11629686304514154,
      "grad_norm": 8.845519065856934,
      "learning_rate": 5.817513778322107e-06,
      "loss": 1.5166,
      "step": 190
    },
    {
      "epoch": 0.12241775057383321,
      "grad_norm": 5.295988082885742,
      "learning_rate": 6.12369871402327e-06,
      "loss": 1.5011,
      "step": 200
    },
    {
      "epoch": 0.12853863810252486,
      "grad_norm": 21.03306770324707,
      "learning_rate": 6.429883649724435e-06,
      "loss": 1.4884,
      "step": 210
    },
    {
      "epoch": 0.13465952563121653,
      "grad_norm": 7.419331073760986,
      "learning_rate": 6.736068585425598e-06,
      "loss": 1.4816,
      "step": 220
    },
    {
      "epoch": 0.1407804131599082,
      "grad_norm": 6.6504411697387695,
      "learning_rate": 7.042253521126762e-06,
      "loss": 1.4639,
      "step": 230
    },
    {
      "epoch": 0.14690130068859986,
      "grad_norm": 19.93020248413086,
      "learning_rate": 7.348438456827924e-06,
      "loss": 1.4748,
      "step": 240
    },
    {
      "epoch": 0.1530221882172915,
      "grad_norm": 212.81414794921875,
      "learning_rate": 7.654623392529089e-06,
      "loss": 1.4663,
      "step": 250
    },
    {
      "epoch": 0.15914307574598316,
      "grad_norm": 6.855514049530029,
      "learning_rate": 7.960808328230251e-06,
      "loss": 1.4681,
      "step": 260
    },
    {
      "epoch": 0.16526396327467482,
      "grad_norm": 6.292046546936035,
      "learning_rate": 8.266993263931416e-06,
      "loss": 1.4505,
      "step": 270
    },
    {
      "epoch": 0.1713848508033665,
      "grad_norm": 95.416015625,
      "learning_rate": 8.573178199632578e-06,
      "loss": 1.4735,
      "step": 280
    },
    {
      "epoch": 0.17750573833205815,
      "grad_norm": 9.463255882263184,
      "learning_rate": 8.879363135333741e-06,
      "loss": 1.4572,
      "step": 290
    },
    {
      "epoch": 0.18362662586074982,
      "grad_norm": 5.731226921081543,
      "learning_rate": 9.185548071034905e-06,
      "loss": 1.4478,
      "step": 300
    },
    {
      "epoch": 0.18974751338944146,
      "grad_norm": 7.399662494659424,
      "learning_rate": 9.491733006736068e-06,
      "loss": 1.4655,
      "step": 310
    },
    {
      "epoch": 0.19586840091813312,
      "grad_norm": 10.838027954101562,
      "learning_rate": 9.797917942437232e-06,
      "loss": 1.4375,
      "step": 320
    },
    {
      "epoch": 0.2019892884468248,
      "grad_norm": 7.435635566711426,
      "learning_rate": 1.0104102878138397e-05,
      "loss": 1.431,
      "step": 330
    },
    {
      "epoch": 0.20811017597551645,
      "grad_norm": 12.052409172058105,
      "learning_rate": 1.041028781383956e-05,
      "loss": 1.4273,
      "step": 340
    },
    {
      "epoch": 0.21423106350420812,
      "grad_norm": 10.6692476272583,
      "learning_rate": 1.0716472749540724e-05,
      "loss": 1.4286,
      "step": 350
    },
    {
      "epoch": 0.22035195103289976,
      "grad_norm": 6.245419025421143,
      "learning_rate": 1.1022657685241886e-05,
      "loss": 1.4376,
      "step": 360
    },
    {
      "epoch": 0.22647283856159142,
      "grad_norm": 6.799665451049805,
      "learning_rate": 1.132884262094305e-05,
      "loss": 1.4288,
      "step": 370
    },
    {
      "epoch": 0.2325937260902831,
      "grad_norm": 10.520297050476074,
      "learning_rate": 1.1635027556644213e-05,
      "loss": 1.4242,
      "step": 380
    },
    {
      "epoch": 0.23871461361897475,
      "grad_norm": 6.719942092895508,
      "learning_rate": 1.1941212492345378e-05,
      "loss": 1.4336,
      "step": 390
    },
    {
      "epoch": 0.24483550114766642,
      "grad_norm": 7.194707870483398,
      "learning_rate": 1.224739742804654e-05,
      "loss": 1.4121,
      "step": 400
    },
    {
      "epoch": 0.25095638867635806,
      "grad_norm": 18.913164138793945,
      "learning_rate": 1.2553582363747705e-05,
      "loss": 1.4399,
      "step": 410
    },
    {
      "epoch": 0.2570772762050497,
      "grad_norm": 3.8806185722351074,
      "learning_rate": 1.285976729944887e-05,
      "loss": 1.428,
      "step": 420
    },
    {
      "epoch": 0.2631981637337414,
      "grad_norm": 9.427650451660156,
      "learning_rate": 1.316595223515003e-05,
      "loss": 1.4204,
      "step": 430
    },
    {
      "epoch": 0.26931905126243305,
      "grad_norm": 4.577441692352295,
      "learning_rate": 1.3472137170851196e-05,
      "loss": 1.4168,
      "step": 440
    },
    {
      "epoch": 0.2754399387911247,
      "grad_norm": 6.408951759338379,
      "learning_rate": 1.3778322106552357e-05,
      "loss": 1.4402,
      "step": 450
    },
    {
      "epoch": 0.2815608263198164,
      "grad_norm": 7.625643253326416,
      "learning_rate": 1.4084507042253523e-05,
      "loss": 1.389,
      "step": 460
    },
    {
      "epoch": 0.28768171384850805,
      "grad_norm": 11.841383934020996,
      "learning_rate": 1.4390691977954684e-05,
      "loss": 1.4048,
      "step": 470
    },
    {
      "epoch": 0.2938026013771997,
      "grad_norm": 18.034799575805664,
      "learning_rate": 1.4696876913655849e-05,
      "loss": 1.4152,
      "step": 480
    },
    {
      "epoch": 0.2999234889058914,
      "grad_norm": 8.04079532623291,
      "learning_rate": 1.5003061849357011e-05,
      "loss": 1.413,
      "step": 490
    },
    {
      "epoch": 0.306044376434583,
      "grad_norm": 10.082817077636719,
      "learning_rate": 1.5309246785058177e-05,
      "loss": 1.3984,
      "step": 500
    },
    {
      "epoch": 0.31216526396327465,
      "grad_norm": 3.9042136669158936,
      "learning_rate": 1.5615431720759337e-05,
      "loss": 1.4192,
      "step": 510
    },
    {
      "epoch": 0.3182861514919663,
      "grad_norm": 5.863616466522217,
      "learning_rate": 1.5921616656460503e-05,
      "loss": 1.4082,
      "step": 520
    },
    {
      "epoch": 0.324407039020658,
      "grad_norm": 4.442514419555664,
      "learning_rate": 1.622780159216167e-05,
      "loss": 1.4038,
      "step": 530
    },
    {
      "epoch": 0.33052792654934965,
      "grad_norm": 3.5633575916290283,
      "learning_rate": 1.653398652786283e-05,
      "loss": 1.4019,
      "step": 540
    },
    {
      "epoch": 0.3366488140780413,
      "grad_norm": 6.105249404907227,
      "learning_rate": 1.6840171463563994e-05,
      "loss": 1.408,
      "step": 550
    },
    {
      "epoch": 0.342769701606733,
      "grad_norm": 30.103477478027344,
      "learning_rate": 1.7146356399265157e-05,
      "loss": 1.3889,
      "step": 560
    },
    {
      "epoch": 0.34889058913542464,
      "grad_norm": 6.706213474273682,
      "learning_rate": 1.7452541334966323e-05,
      "loss": 1.3868,
      "step": 570
    },
    {
      "epoch": 0.3550114766641163,
      "grad_norm": 6.28405237197876,
      "learning_rate": 1.7758726270667482e-05,
      "loss": 1.4019,
      "step": 580
    },
    {
      "epoch": 0.361132364192808,
      "grad_norm": 8.72684097290039,
      "learning_rate": 1.8064911206368648e-05,
      "loss": 1.3877,
      "step": 590
    },
    {
      "epoch": 0.36725325172149964,
      "grad_norm": 4.952666759490967,
      "learning_rate": 1.837109614206981e-05,
      "loss": 1.3992,
      "step": 600
    },
    {
      "epoch": 0.37337413925019125,
      "grad_norm": 31.551925659179688,
      "learning_rate": 1.8677281077770973e-05,
      "loss": 1.3858,
      "step": 610
    },
    {
      "epoch": 0.3794950267788829,
      "grad_norm": 4.013786792755127,
      "learning_rate": 1.8983466013472136e-05,
      "loss": 1.3955,
      "step": 620
    },
    {
      "epoch": 0.3856159143075746,
      "grad_norm": 6.236915111541748,
      "learning_rate": 1.9289650949173302e-05,
      "loss": 1.4099,
      "step": 630
    },
    {
      "epoch": 0.39173680183626625,
      "grad_norm": 18.471105575561523,
      "learning_rate": 1.9595835884874465e-05,
      "loss": 1.3936,
      "step": 640
    },
    {
      "epoch": 0.3978576893649579,
      "grad_norm": 8.124337196350098,
      "learning_rate": 1.9902020820575627e-05,
      "loss": 1.412,
      "step": 650
    },
    {
      "epoch": 0.4039785768936496,
      "grad_norm": 10.235346794128418,
      "learning_rate": 2.0208205756276793e-05,
      "loss": 1.3985,
      "step": 660
    },
    {
      "epoch": 0.41009946442234124,
      "grad_norm": 15.285164833068848,
      "learning_rate": 2.0514390691977956e-05,
      "loss": 1.3585,
      "step": 670
    },
    {
      "epoch": 0.4162203519510329,
      "grad_norm": 7.845749855041504,
      "learning_rate": 2.082057562767912e-05,
      "loss": 1.4062,
      "step": 680
    },
    {
      "epoch": 0.4223412394797246,
      "grad_norm": 9.471783638000488,
      "learning_rate": 2.112676056338028e-05,
      "loss": 1.3831,
      "step": 690
    },
    {
      "epoch": 0.42846212700841624,
      "grad_norm": 3.549377679824829,
      "learning_rate": 2.1432945499081448e-05,
      "loss": 1.3984,
      "step": 700
    },
    {
      "epoch": 0.4345830145371079,
      "grad_norm": 4.166463375091553,
      "learning_rate": 2.173913043478261e-05,
      "loss": 1.3931,
      "step": 710
    },
    {
      "epoch": 0.4407039020657995,
      "grad_norm": 21.529024124145508,
      "learning_rate": 2.2045315370483773e-05,
      "loss": 1.389,
      "step": 720
    },
    {
      "epoch": 0.4468247895944912,
      "grad_norm": 3.551419258117676,
      "learning_rate": 2.2351500306184936e-05,
      "loss": 1.3993,
      "step": 730
    },
    {
      "epoch": 0.45294567712318284,
      "grad_norm": 39.104515075683594,
      "learning_rate": 2.26576852418861e-05,
      "loss": 1.373,
      "step": 740
    },
    {
      "epoch": 0.4590665646518745,
      "grad_norm": 6.086831092834473,
      "learning_rate": 2.2963870177587264e-05,
      "loss": 1.4084,
      "step": 750
    },
    {
      "epoch": 0.4651874521805662,
      "grad_norm": 16.3785457611084,
      "learning_rate": 2.3270055113288427e-05,
      "loss": 1.387,
      "step": 760
    },
    {
      "epoch": 0.47130833970925784,
      "grad_norm": 10.629354476928711,
      "learning_rate": 2.357624004898959e-05,
      "loss": 1.3784,
      "step": 770
    },
    {
      "epoch": 0.4774292272379495,
      "grad_norm": 21.637479782104492,
      "learning_rate": 2.3882424984690756e-05,
      "loss": 1.3841,
      "step": 780
    },
    {
      "epoch": 0.48355011476664117,
      "grad_norm": 9.542509078979492,
      "learning_rate": 2.4188609920391918e-05,
      "loss": 1.3896,
      "step": 790
    },
    {
      "epoch": 0.48967100229533284,
      "grad_norm": 4.976247787475586,
      "learning_rate": 2.449479485609308e-05,
      "loss": 1.3781,
      "step": 800
    },
    {
      "epoch": 0.4957918898240245,
      "grad_norm": 4.468277454376221,
      "learning_rate": 2.4800979791794247e-05,
      "loss": 1.3793,
      "step": 810
    },
    {
      "epoch": 0.5019127773527161,
      "grad_norm": 4.785976409912109,
      "learning_rate": 2.510716472749541e-05,
      "loss": 1.3981,
      "step": 820
    },
    {
      "epoch": 0.5080336648814078,
      "grad_norm": 30.538387298583984,
      "learning_rate": 2.5413349663196572e-05,
      "loss": 1.384,
      "step": 830
    },
    {
      "epoch": 0.5141545524100994,
      "grad_norm": 3.1451311111450195,
      "learning_rate": 2.571953459889774e-05,
      "loss": 1.3803,
      "step": 840
    },
    {
      "epoch": 0.5202754399387911,
      "grad_norm": 3.084407091140747,
      "learning_rate": 2.6025719534598898e-05,
      "loss": 1.3821,
      "step": 850
    },
    {
      "epoch": 0.5263963274674828,
      "grad_norm": 2.694706439971924,
      "learning_rate": 2.633190447030006e-05,
      "loss": 1.3875,
      "step": 860
    },
    {
      "epoch": 0.5325172149961744,
      "grad_norm": 9.250505447387695,
      "learning_rate": 2.6638089406001226e-05,
      "loss": 1.3763,
      "step": 870
    },
    {
      "epoch": 0.5386381025248661,
      "grad_norm": 2.440683364868164,
      "learning_rate": 2.6944274341702392e-05,
      "loss": 1.3596,
      "step": 880
    },
    {
      "epoch": 0.5447589900535578,
      "grad_norm": 4.1361165046691895,
      "learning_rate": 2.7250459277403552e-05,
      "loss": 1.3616,
      "step": 890
    },
    {
      "epoch": 0.5508798775822494,
      "grad_norm": 3.9648489952087402,
      "learning_rate": 2.7556644213104714e-05,
      "loss": 1.361,
      "step": 900
    },
    {
      "epoch": 0.5570007651109411,
      "grad_norm": 3.9357821941375732,
      "learning_rate": 2.786282914880588e-05,
      "loss": 1.3763,
      "step": 910
    },
    {
      "epoch": 0.5631216526396328,
      "grad_norm": 4.0848493576049805,
      "learning_rate": 2.8169014084507046e-05,
      "loss": 1.3744,
      "step": 920
    },
    {
      "epoch": 0.5692425401683244,
      "grad_norm": 172.5287628173828,
      "learning_rate": 2.847519902020821e-05,
      "loss": 1.3632,
      "step": 930
    },
    {
      "epoch": 0.5753634276970161,
      "grad_norm": 3.403939723968506,
      "learning_rate": 2.878138395590937e-05,
      "loss": 1.3731,
      "step": 940
    },
    {
      "epoch": 0.5814843152257078,
      "grad_norm": 3.526984930038452,
      "learning_rate": 2.9087568891610534e-05,
      "loss": 1.3827,
      "step": 950
    },
    {
      "epoch": 0.5876052027543994,
      "grad_norm": 23.79243278503418,
      "learning_rate": 2.9393753827311697e-05,
      "loss": 1.3808,
      "step": 960
    },
    {
      "epoch": 0.5937260902830911,
      "grad_norm": 4.624916076660156,
      "learning_rate": 2.9699938763012863e-05,
      "loss": 1.3669,
      "step": 970
    },
    {
      "epoch": 0.5998469778117828,
      "grad_norm": 3.8356945514678955,
      "learning_rate": 3.0006123698714022e-05,
      "loss": 1.3378,
      "step": 980
    },
    {
      "epoch": 0.6059678653404744,
      "grad_norm": 4.029366493225098,
      "learning_rate": 3.031230863441519e-05,
      "loss": 1.3691,
      "step": 990
    },
    {
      "epoch": 0.612088752869166,
      "grad_norm": 4.599860191345215,
      "learning_rate": 3.0618493570116355e-05,
      "loss": 1.3704,
      "step": 1000
    },
    {
      "epoch": 0.6182096403978576,
      "grad_norm": 9.668410301208496,
      "learning_rate": 3.0924678505817514e-05,
      "loss": 1.3767,
      "step": 1010
    },
    {
      "epoch": 0.6243305279265493,
      "grad_norm": 7.55600643157959,
      "learning_rate": 3.123086344151867e-05,
      "loss": 1.3661,
      "step": 1020
    },
    {
      "epoch": 0.630451415455241,
      "grad_norm": 4.51859712600708,
      "learning_rate": 3.153704837721984e-05,
      "loss": 1.3559,
      "step": 1030
    },
    {
      "epoch": 0.6365723029839326,
      "grad_norm": 3.436450242996216,
      "learning_rate": 3.1843233312921005e-05,
      "loss": 1.345,
      "step": 1040
    },
    {
      "epoch": 0.6426931905126243,
      "grad_norm": 3.319326877593994,
      "learning_rate": 3.214941824862217e-05,
      "loss": 1.3485,
      "step": 1050
    },
    {
      "epoch": 0.648814078041316,
      "grad_norm": 5.369412422180176,
      "learning_rate": 3.245560318432334e-05,
      "loss": 1.3516,
      "step": 1060
    },
    {
      "epoch": 0.6549349655700076,
      "grad_norm": 2.8182034492492676,
      "learning_rate": 3.2761788120024497e-05,
      "loss": 1.3703,
      "step": 1070
    },
    {
      "epoch": 0.6610558530986993,
      "grad_norm": 4.989190101623535,
      "learning_rate": 3.306797305572566e-05,
      "loss": 1.3815,
      "step": 1080
    },
    {
      "epoch": 0.667176740627391,
      "grad_norm": 6.330949783325195,
      "learning_rate": 3.337415799142682e-05,
      "loss": 1.3733,
      "step": 1090
    },
    {
      "epoch": 0.6732976281560826,
      "grad_norm": 2.702294111251831,
      "learning_rate": 3.368034292712799e-05,
      "loss": 1.3526,
      "step": 1100
    },
    {
      "epoch": 0.6794185156847743,
      "grad_norm": 4.195952892303467,
      "learning_rate": 3.398652786282915e-05,
      "loss": 1.339,
      "step": 1110
    },
    {
      "epoch": 0.685539403213466,
      "grad_norm": 6.287045478820801,
      "learning_rate": 3.429271279853031e-05,
      "loss": 1.3479,
      "step": 1120
    },
    {
      "epoch": 0.6916602907421576,
      "grad_norm": 2.7063398361206055,
      "learning_rate": 3.459889773423148e-05,
      "loss": 1.3689,
      "step": 1130
    },
    {
      "epoch": 0.6977811782708493,
      "grad_norm": 2.84989595413208,
      "learning_rate": 3.4905082669932645e-05,
      "loss": 1.3589,
      "step": 1140
    },
    {
      "epoch": 0.703902065799541,
      "grad_norm": 4.503838062286377,
      "learning_rate": 3.5211267605633805e-05,
      "loss": 1.3399,
      "step": 1150
    },
    {
      "epoch": 0.7100229533282326,
      "grad_norm": 2.6337504386901855,
      "learning_rate": 3.5517452541334964e-05,
      "loss": 1.3604,
      "step": 1160
    },
    {
      "epoch": 0.7161438408569243,
      "grad_norm": 3.980837821960449,
      "learning_rate": 3.582363747703613e-05,
      "loss": 1.3529,
      "step": 1170
    },
    {
      "epoch": 0.722264728385616,
      "grad_norm": 3.845048189163208,
      "learning_rate": 3.6129822412737296e-05,
      "loss": 1.3456,
      "step": 1180
    },
    {
      "epoch": 0.7283856159143076,
      "grad_norm": 2.598834753036499,
      "learning_rate": 3.643600734843846e-05,
      "loss": 1.3482,
      "step": 1190
    },
    {
      "epoch": 0.7345065034429993,
      "grad_norm": 2.160224676132202,
      "learning_rate": 3.674219228413962e-05,
      "loss": 1.3303,
      "step": 1200
    },
    {
      "epoch": 0.740627390971691,
      "grad_norm": 5.0525360107421875,
      "learning_rate": 3.704837721984079e-05,
      "loss": 1.3836,
      "step": 1210
    },
    {
      "epoch": 0.7467482785003825,
      "grad_norm": 2.637212038040161,
      "learning_rate": 3.735456215554195e-05,
      "loss": 1.3535,
      "step": 1220
    },
    {
      "epoch": 0.7528691660290742,
      "grad_norm": 2.752631425857544,
      "learning_rate": 3.766074709124311e-05,
      "loss": 1.351,
      "step": 1230
    },
    {
      "epoch": 0.7589900535577658,
      "grad_norm": 3.4024102687835693,
      "learning_rate": 3.796693202694427e-05,
      "loss": 1.3518,
      "step": 1240
    },
    {
      "epoch": 0.7651109410864575,
      "grad_norm": 3.2056336402893066,
      "learning_rate": 3.827311696264544e-05,
      "loss": 1.3415,
      "step": 1250
    },
    {
      "epoch": 0.7712318286151492,
      "grad_norm": 3.5055158138275146,
      "learning_rate": 3.8579301898346604e-05,
      "loss": 1.3502,
      "step": 1260
    },
    {
      "epoch": 0.7773527161438408,
      "grad_norm": 2.2886886596679688,
      "learning_rate": 3.888548683404777e-05,
      "loss": 1.3664,
      "step": 1270
    },
    {
      "epoch": 0.7834736036725325,
      "grad_norm": 3.2076241970062256,
      "learning_rate": 3.919167176974893e-05,
      "loss": 1.3293,
      "step": 1280
    },
    {
      "epoch": 0.7895944912012242,
      "grad_norm": 3.1704225540161133,
      "learning_rate": 3.9497856705450096e-05,
      "loss": 1.351,
      "step": 1290
    },
    {
      "epoch": 0.7957153787299158,
      "grad_norm": 2.2499842643737793,
      "learning_rate": 3.9804041641151255e-05,
      "loss": 1.3509,
      "step": 1300
    },
    {
      "epoch": 0.8018362662586075,
      "grad_norm": 2.698070526123047,
      "learning_rate": 4.011022657685242e-05,
      "loss": 1.3463,
      "step": 1310
    },
    {
      "epoch": 0.8079571537872992,
      "grad_norm": 1.8415290117263794,
      "learning_rate": 4.041641151255359e-05,
      "loss": 1.3214,
      "step": 1320
    },
    {
      "epoch": 0.8140780413159908,
      "grad_norm": 2.9317309856414795,
      "learning_rate": 4.0722596448254746e-05,
      "loss": 1.3581,
      "step": 1330
    },
    {
      "epoch": 0.8201989288446825,
      "grad_norm": 4.546383857727051,
      "learning_rate": 4.102878138395591e-05,
      "loss": 1.3413,
      "step": 1340
    },
    {
      "epoch": 0.8263198163733741,
      "grad_norm": 2.5486834049224854,
      "learning_rate": 4.133496631965708e-05,
      "loss": 1.3686,
      "step": 1350
    },
    {
      "epoch": 0.8324407039020658,
      "grad_norm": 2.265129804611206,
      "learning_rate": 4.164115125535824e-05,
      "loss": 1.3125,
      "step": 1360
    },
    {
      "epoch": 0.8385615914307575,
      "grad_norm": 2.1609714031219482,
      "learning_rate": 4.19473361910594e-05,
      "loss": 1.351,
      "step": 1370
    },
    {
      "epoch": 0.8446824789594491,
      "grad_norm": 3.51118540763855,
      "learning_rate": 4.225352112676056e-05,
      "loss": 1.3366,
      "step": 1380
    },
    {
      "epoch": 0.8508033664881408,
      "grad_norm": 3.0679023265838623,
      "learning_rate": 4.255970606246173e-05,
      "loss": 1.3472,
      "step": 1390
    },
    {
      "epoch": 0.8569242540168325,
      "grad_norm": 2.597992181777954,
      "learning_rate": 4.2865890998162895e-05,
      "loss": 1.3632,
      "step": 1400
    },
    {
      "epoch": 0.8630451415455241,
      "grad_norm": 2.0379843711853027,
      "learning_rate": 4.3172075933864054e-05,
      "loss": 1.3615,
      "step": 1410
    },
    {
      "epoch": 0.8691660290742158,
      "grad_norm": 2.303506374359131,
      "learning_rate": 4.347826086956522e-05,
      "loss": 1.3273,
      "step": 1420
    },
    {
      "epoch": 0.8752869166029075,
      "grad_norm": 2.795226812362671,
      "learning_rate": 4.3784445805266386e-05,
      "loss": 1.3475,
      "step": 1430
    },
    {
      "epoch": 0.881407804131599,
      "grad_norm": 2.218827486038208,
      "learning_rate": 4.4090630740967546e-05,
      "loss": 1.3346,
      "step": 1440
    },
    {
      "epoch": 0.8875286916602907,
      "grad_norm": 1.8111498355865479,
      "learning_rate": 4.439681567666871e-05,
      "loss": 1.3135,
      "step": 1450
    },
    {
      "epoch": 0.8936495791889824,
      "grad_norm": 2.2712912559509277,
      "learning_rate": 4.470300061236987e-05,
      "loss": 1.342,
      "step": 1460
    },
    {
      "epoch": 0.899770466717674,
      "grad_norm": 2.1051836013793945,
      "learning_rate": 4.500918554807104e-05,
      "loss": 1.3249,
      "step": 1470
    },
    {
      "epoch": 0.9058913542463657,
      "grad_norm": 3.4877002239227295,
      "learning_rate": 4.53153704837722e-05,
      "loss": 1.3476,
      "step": 1480
    },
    {
      "epoch": 0.9120122417750574,
      "grad_norm": 1.686214566230774,
      "learning_rate": 4.562155541947337e-05,
      "loss": 1.3266,
      "step": 1490
    },
    {
      "epoch": 0.918133129303749,
      "grad_norm": 2.8292617797851562,
      "learning_rate": 4.592774035517453e-05,
      "loss": 1.3308,
      "step": 1500
    },
    {
      "epoch": 0.9242540168324407,
      "grad_norm": 1.6535699367523193,
      "learning_rate": 4.623392529087569e-05,
      "loss": 1.3368,
      "step": 1510
    },
    {
      "epoch": 0.9303749043611323,
      "grad_norm": 2.361920118331909,
      "learning_rate": 4.6540110226576854e-05,
      "loss": 1.3164,
      "step": 1520
    },
    {
      "epoch": 0.936495791889824,
      "grad_norm": 1.7258111238479614,
      "learning_rate": 4.684629516227802e-05,
      "loss": 1.3396,
      "step": 1530
    },
    {
      "epoch": 0.9426166794185157,
      "grad_norm": 1.9547419548034668,
      "learning_rate": 4.715248009797918e-05,
      "loss": 1.3432,
      "step": 1540
    },
    {
      "epoch": 0.9487375669472073,
      "grad_norm": 3.920701742172241,
      "learning_rate": 4.7458665033680345e-05,
      "loss": 1.3296,
      "step": 1550
    },
    {
      "epoch": 0.954858454475899,
      "grad_norm": 2.1728620529174805,
      "learning_rate": 4.776484996938151e-05,
      "loss": 1.3278,
      "step": 1560
    },
    {
      "epoch": 0.9609793420045907,
      "grad_norm": 4.015989780426025,
      "learning_rate": 4.807103490508267e-05,
      "loss": 1.3238,
      "step": 1570
    },
    {
      "epoch": 0.9671002295332823,
      "grad_norm": 3.3949761390686035,
      "learning_rate": 4.8377219840783837e-05,
      "loss": 1.3326,
      "step": 1580
    },
    {
      "epoch": 0.973221117061974,
      "grad_norm": 2.3282153606414795,
      "learning_rate": 4.8683404776484996e-05,
      "loss": 1.3495,
      "step": 1590
    },
    {
      "epoch": 0.9793420045906657,
      "grad_norm": 2.321676731109619,
      "learning_rate": 4.898958971218616e-05,
      "loss": 1.3381,
      "step": 1600
    },
    {
      "epoch": 0.9854628921193573,
      "grad_norm": 2.4791972637176514,
      "learning_rate": 4.929577464788733e-05,
      "loss": 1.3193,
      "step": 1610
    },
    {
      "epoch": 0.991583779648049,
      "grad_norm": 3.545558214187622,
      "learning_rate": 4.9601959583588494e-05,
      "loss": 1.3263,
      "step": 1620
    },
    {
      "epoch": 0.9977046671767407,
      "grad_norm": 2.3265316486358643,
      "learning_rate": 4.990814451928965e-05,
      "loss": 1.3156,
      "step": 1630
    },
    {
      "epoch": 0.9995409334353481,
      "eval_accuracy": 0.44772257718942493,
      "eval_loss": 1.2975871562957764,
      "eval_runtime": 2616.4379,
      "eval_samples_per_second": 79.916,
      "eval_steps_per_second": 0.625,
      "step": 1633
    },
    {
      "epoch": 1.0038255547054322,
      "grad_norm": 2.3401131629943848,
      "learning_rate": 4.997618561611214e-05,
      "loss": 1.3378,
      "step": 1640
    },
    {
      "epoch": 1.009946442234124,
      "grad_norm": 2.454505443572998,
      "learning_rate": 4.994216506770089e-05,
      "loss": 1.3076,
      "step": 1650
    },
    {
      "epoch": 1.0160673297628156,
      "grad_norm": 2.544475793838501,
      "learning_rate": 4.990814451928965e-05,
      "loss": 1.3284,
      "step": 1660
    },
    {
      "epoch": 1.0221882172915073,
      "grad_norm": 3.767465114593506,
      "learning_rate": 4.9874123970878414e-05,
      "loss": 1.3108,
      "step": 1670
    },
    {
      "epoch": 1.0283091048201989,
      "grad_norm": 3.772216796875,
      "learning_rate": 4.984010342246717e-05,
      "loss": 1.309,
      "step": 1680
    },
    {
      "epoch": 1.0344299923488907,
      "grad_norm": 1.745509147644043,
      "learning_rate": 4.980608287405593e-05,
      "loss": 1.337,
      "step": 1690
    },
    {
      "epoch": 1.0405508798775822,
      "grad_norm": 2.45389723777771,
      "learning_rate": 4.977206232564469e-05,
      "loss": 1.3282,
      "step": 1700
    },
    {
      "epoch": 1.046671767406274,
      "grad_norm": 2.846754789352417,
      "learning_rate": 4.973804177723345e-05,
      "loss": 1.3403,
      "step": 1710
    },
    {
      "epoch": 1.0527926549349655,
      "grad_norm": 2.496070146560669,
      "learning_rate": 4.970402122882221e-05,
      "loss": 1.3135,
      "step": 1720
    },
    {
      "epoch": 1.0589135424636573,
      "grad_norm": 3.4331891536712646,
      "learning_rate": 4.967000068041097e-05,
      "loss": 1.3359,
      "step": 1730
    },
    {
      "epoch": 1.0650344299923489,
      "grad_norm": 2.686443328857422,
      "learning_rate": 4.963598013199973e-05,
      "loss": 1.3241,
      "step": 1740
    },
    {
      "epoch": 1.0711553175210407,
      "grad_norm": 2.017503261566162,
      "learning_rate": 4.9601959583588494e-05,
      "loss": 1.3314,
      "step": 1750
    },
    {
      "epoch": 1.0772762050497322,
      "grad_norm": 3.1370229721069336,
      "learning_rate": 4.9567939035177255e-05,
      "loss": 1.3204,
      "step": 1760
    },
    {
      "epoch": 1.0833970925784238,
      "grad_norm": 2.5512239933013916,
      "learning_rate": 4.953391848676601e-05,
      "loss": 1.3321,
      "step": 1770
    },
    {
      "epoch": 1.0895179801071155,
      "grad_norm": 2.360032558441162,
      "learning_rate": 4.949989793835477e-05,
      "loss": 1.3228,
      "step": 1780
    },
    {
      "epoch": 1.095638867635807,
      "grad_norm": 2.947443962097168,
      "learning_rate": 4.9465877389943524e-05,
      "loss": 1.3247,
      "step": 1790
    },
    {
      "epoch": 1.1017597551644989,
      "grad_norm": 2.6693966388702393,
      "learning_rate": 4.9431856841532285e-05,
      "loss": 1.3259,
      "step": 1800
    },
    {
      "epoch": 1.1078806426931904,
      "grad_norm": 3.112257957458496,
      "learning_rate": 4.9397836293121045e-05,
      "loss": 1.3237,
      "step": 1810
    },
    {
      "epoch": 1.1140015302218822,
      "grad_norm": 2.609961986541748,
      "learning_rate": 4.9363815744709806e-05,
      "loss": 1.3246,
      "step": 1820
    },
    {
      "epoch": 1.1201224177505738,
      "grad_norm": 3.4597411155700684,
      "learning_rate": 4.932979519629857e-05,
      "loss": 1.299,
      "step": 1830
    },
    {
      "epoch": 1.1262433052792655,
      "grad_norm": 2.242790937423706,
      "learning_rate": 4.929577464788733e-05,
      "loss": 1.3098,
      "step": 1840
    },
    {
      "epoch": 1.132364192807957,
      "grad_norm": 2.5282511711120605,
      "learning_rate": 4.926175409947609e-05,
      "loss": 1.3345,
      "step": 1850
    },
    {
      "epoch": 1.1384850803366489,
      "grad_norm": 2.2986130714416504,
      "learning_rate": 4.922773355106485e-05,
      "loss": 1.3339,
      "step": 1860
    },
    {
      "epoch": 1.1446059678653404,
      "grad_norm": 5.559305191040039,
      "learning_rate": 4.919371300265361e-05,
      "loss": 1.3338,
      "step": 1870
    },
    {
      "epoch": 1.1507268553940322,
      "grad_norm": 2.9165046215057373,
      "learning_rate": 4.9159692454242364e-05,
      "loss": 1.322,
      "step": 1880
    },
    {
      "epoch": 1.1568477429227237,
      "grad_norm": 2.179335594177246,
      "learning_rate": 4.9125671905831125e-05,
      "loss": 1.3289,
      "step": 1890
    },
    {
      "epoch": 1.1629686304514155,
      "grad_norm": 3.7216742038726807,
      "learning_rate": 4.909165135741988e-05,
      "loss": 1.3033,
      "step": 1900
    },
    {
      "epoch": 1.169089517980107,
      "grad_norm": 2.7910213470458984,
      "learning_rate": 4.905763080900864e-05,
      "loss": 1.3086,
      "step": 1910
    },
    {
      "epoch": 1.1752104055087988,
      "grad_norm": 3.217575788497925,
      "learning_rate": 4.90236102605974e-05,
      "loss": 1.3194,
      "step": 1920
    },
    {
      "epoch": 1.1813312930374904,
      "grad_norm": 6.850343704223633,
      "learning_rate": 4.898958971218616e-05,
      "loss": 1.3253,
      "step": 1930
    },
    {
      "epoch": 1.1874521805661822,
      "grad_norm": 2.523886203765869,
      "learning_rate": 4.895556916377492e-05,
      "loss": 1.3291,
      "step": 1940
    },
    {
      "epoch": 1.1935730680948737,
      "grad_norm": 2.4011008739471436,
      "learning_rate": 4.8921548615363684e-05,
      "loss": 1.3201,
      "step": 1950
    },
    {
      "epoch": 1.1996939556235655,
      "grad_norm": 2.7910492420196533,
      "learning_rate": 4.8887528066952444e-05,
      "loss": 1.3376,
      "step": 1960
    },
    {
      "epoch": 1.205814843152257,
      "grad_norm": 2.3480148315429688,
      "learning_rate": 4.8853507518541205e-05,
      "loss": 1.315,
      "step": 1970
    },
    {
      "epoch": 1.2119357306809486,
      "grad_norm": 3.9389758110046387,
      "learning_rate": 4.8819486970129966e-05,
      "loss": 1.3054,
      "step": 1980
    },
    {
      "epoch": 1.2180566182096404,
      "grad_norm": 2.182918071746826,
      "learning_rate": 4.878546642171872e-05,
      "loss": 1.3312,
      "step": 1990
    },
    {
      "epoch": 1.2241775057383322,
      "grad_norm": 5.687019348144531,
      "learning_rate": 4.8751445873307474e-05,
      "loss": 1.3041,
      "step": 2000
    },
    {
      "epoch": 1.2302983932670237,
      "grad_norm": 1.9981379508972168,
      "learning_rate": 4.8717425324896235e-05,
      "loss": 1.3046,
      "step": 2010
    },
    {
      "epoch": 1.2364192807957153,
      "grad_norm": 2.221470832824707,
      "learning_rate": 4.8683404776484996e-05,
      "loss": 1.2983,
      "step": 2020
    },
    {
      "epoch": 1.242540168324407,
      "grad_norm": 3.1240041255950928,
      "learning_rate": 4.864938422807376e-05,
      "loss": 1.2943,
      "step": 2030
    },
    {
      "epoch": 1.2486610558530986,
      "grad_norm": 2.941997528076172,
      "learning_rate": 4.861536367966252e-05,
      "loss": 1.301,
      "step": 2040
    },
    {
      "epoch": 1.2547819433817904,
      "grad_norm": 2.6538472175598145,
      "learning_rate": 4.858134313125128e-05,
      "loss": 1.3127,
      "step": 2050
    },
    {
      "epoch": 1.260902830910482,
      "grad_norm": 4.265965938568115,
      "learning_rate": 4.854732258284004e-05,
      "loss": 1.3172,
      "step": 2060
    },
    {
      "epoch": 1.2670237184391737,
      "grad_norm": 3.689883232116699,
      "learning_rate": 4.85133020344288e-05,
      "loss": 1.3123,
      "step": 2070
    },
    {
      "epoch": 1.2731446059678653,
      "grad_norm": 2.039571523666382,
      "learning_rate": 4.847928148601756e-05,
      "loss": 1.3197,
      "step": 2080
    },
    {
      "epoch": 1.279265493496557,
      "grad_norm": 3.8511271476745605,
      "learning_rate": 4.8445260937606315e-05,
      "loss": 1.3123,
      "step": 2090
    },
    {
      "epoch": 1.2853863810252486,
      "grad_norm": 1.8921153545379639,
      "learning_rate": 4.8411240389195076e-05,
      "loss": 1.3335,
      "step": 2100
    },
    {
      "epoch": 1.2915072685539404,
      "grad_norm": 2.2640275955200195,
      "learning_rate": 4.8377219840783837e-05,
      "loss": 1.3187,
      "step": 2110
    },
    {
      "epoch": 1.297628156082632,
      "grad_norm": 1.9218261241912842,
      "learning_rate": 4.834319929237259e-05,
      "loss": 1.3124,
      "step": 2120
    },
    {
      "epoch": 1.3037490436113237,
      "grad_norm": 8.252876281738281,
      "learning_rate": 4.830917874396135e-05,
      "loss": 1.316,
      "step": 2130
    },
    {
      "epoch": 1.3098699311400153,
      "grad_norm": 4.976735591888428,
      "learning_rate": 4.827515819555011e-05,
      "loss": 1.3164,
      "step": 2140
    },
    {
      "epoch": 1.315990818668707,
      "grad_norm": 4.07997989654541,
      "learning_rate": 4.824113764713887e-05,
      "loss": 1.3076,
      "step": 2150
    },
    {
      "epoch": 1.3221117061973986,
      "grad_norm": 4.033842086791992,
      "learning_rate": 4.8207117098727634e-05,
      "loss": 1.3206,
      "step": 2160
    },
    {
      "epoch": 1.3282325937260904,
      "grad_norm": 2.3117096424102783,
      "learning_rate": 4.8173096550316395e-05,
      "loss": 1.3358,
      "step": 2170
    },
    {
      "epoch": 1.334353481254782,
      "grad_norm": 2.9266395568847656,
      "learning_rate": 4.8139076001905156e-05,
      "loss": 1.3044,
      "step": 2180
    },
    {
      "epoch": 1.3404743687834735,
      "grad_norm": 2.5126564502716064,
      "learning_rate": 4.8105055453493916e-05,
      "loss": 1.3028,
      "step": 2190
    },
    {
      "epoch": 1.3465952563121653,
      "grad_norm": 3.317023277282715,
      "learning_rate": 4.807103490508267e-05,
      "loss": 1.3223,
      "step": 2200
    },
    {
      "epoch": 1.352716143840857,
      "grad_norm": 2.9920036792755127,
      "learning_rate": 4.803701435667143e-05,
      "loss": 1.3312,
      "step": 2210
    },
    {
      "epoch": 1.3588370313695486,
      "grad_norm": 2.6706953048706055,
      "learning_rate": 4.800299380826019e-05,
      "loss": 1.2947,
      "step": 2220
    },
    {
      "epoch": 1.3649579188982401,
      "grad_norm": 1.970794439315796,
      "learning_rate": 4.7968973259848946e-05,
      "loss": 1.3094,
      "step": 2230
    },
    {
      "epoch": 1.371078806426932,
      "grad_norm": 3.3308029174804688,
      "learning_rate": 4.793495271143771e-05,
      "loss": 1.3207,
      "step": 2240
    },
    {
      "epoch": 1.3771996939556237,
      "grad_norm": 2.6538379192352295,
      "learning_rate": 4.790093216302647e-05,
      "loss": 1.2917,
      "step": 2250
    },
    {
      "epoch": 1.3833205814843152,
      "grad_norm": 1.9290739297866821,
      "learning_rate": 4.786691161461523e-05,
      "loss": 1.3194,
      "step": 2260
    },
    {
      "epoch": 1.3894414690130068,
      "grad_norm": 2.5794217586517334,
      "learning_rate": 4.783289106620399e-05,
      "loss": 1.3156,
      "step": 2270
    },
    {
      "epoch": 1.3955623565416986,
      "grad_norm": 2.9146995544433594,
      "learning_rate": 4.779887051779275e-05,
      "loss": 1.3009,
      "step": 2280
    },
    {
      "epoch": 1.4016832440703901,
      "grad_norm": 3.2218544483184814,
      "learning_rate": 4.776484996938151e-05,
      "loss": 1.3091,
      "step": 2290
    },
    {
      "epoch": 1.407804131599082,
      "grad_norm": 3.2916765213012695,
      "learning_rate": 4.773082942097027e-05,
      "loss": 1.2991,
      "step": 2300
    },
    {
      "epoch": 1.4139250191277735,
      "grad_norm": 2.5758864879608154,
      "learning_rate": 4.7696808872559026e-05,
      "loss": 1.3166,
      "step": 2310
    },
    {
      "epoch": 1.4200459066564652,
      "grad_norm": 2.873842239379883,
      "learning_rate": 4.766278832414779e-05,
      "loss": 1.3322,
      "step": 2320
    },
    {
      "epoch": 1.4261667941851568,
      "grad_norm": 2.3912835121154785,
      "learning_rate": 4.762876777573655e-05,
      "loss": 1.3195,
      "step": 2330
    },
    {
      "epoch": 1.4322876817138486,
      "grad_norm": 1.9144972562789917,
      "learning_rate": 4.759474722732531e-05,
      "loss": 1.3046,
      "step": 2340
    },
    {
      "epoch": 1.4384085692425401,
      "grad_norm": 3.2760283946990967,
      "learning_rate": 4.756072667891406e-05,
      "loss": 1.2808,
      "step": 2350
    },
    {
      "epoch": 1.444529456771232,
      "grad_norm": 2.616299629211426,
      "learning_rate": 4.7526706130502823e-05,
      "loss": 1.2997,
      "step": 2360
    },
    {
      "epoch": 1.4506503442999235,
      "grad_norm": 2.84455943107605,
      "learning_rate": 4.7492685582091584e-05,
      "loss": 1.3005,
      "step": 2370
    },
    {
      "epoch": 1.4567712318286152,
      "grad_norm": 3.6488397121429443,
      "learning_rate": 4.7458665033680345e-05,
      "loss": 1.284,
      "step": 2380
    },
    {
      "epoch": 1.4628921193573068,
      "grad_norm": 3.8589773178100586,
      "learning_rate": 4.7424644485269106e-05,
      "loss": 1.329,
      "step": 2390
    },
    {
      "epoch": 1.4690130068859983,
      "grad_norm": 2.23037052154541,
      "learning_rate": 4.739062393685787e-05,
      "loss": 1.3123,
      "step": 2400
    },
    {
      "epoch": 1.4751338944146901,
      "grad_norm": 1.8849884271621704,
      "learning_rate": 4.735660338844662e-05,
      "loss": 1.3062,
      "step": 2410
    },
    {
      "epoch": 1.481254781943382,
      "grad_norm": 1.9689738750457764,
      "learning_rate": 4.732258284003538e-05,
      "loss": 1.3102,
      "step": 2420
    },
    {
      "epoch": 1.4873756694720734,
      "grad_norm": 2.845853328704834,
      "learning_rate": 4.728856229162414e-05,
      "loss": 1.3123,
      "step": 2430
    },
    {
      "epoch": 1.493496557000765,
      "grad_norm": 2.8484694957733154,
      "learning_rate": 4.72545417432129e-05,
      "loss": 1.3163,
      "step": 2440
    },
    {
      "epoch": 1.4996174445294568,
      "grad_norm": 1.8601232767105103,
      "learning_rate": 4.7220521194801664e-05,
      "loss": 1.3182,
      "step": 2450
    },
    {
      "epoch": 1.5057383320581486,
      "grad_norm": 2.9465630054473877,
      "learning_rate": 4.7186500646390425e-05,
      "loss": 1.2982,
      "step": 2460
    },
    {
      "epoch": 1.51185921958684,
      "grad_norm": 2.2005507946014404,
      "learning_rate": 4.715248009797918e-05,
      "loss": 1.3187,
      "step": 2470
    },
    {
      "epoch": 1.5179801071155317,
      "grad_norm": 2.624022960662842,
      "learning_rate": 4.711845954956794e-05,
      "loss": 1.3099,
      "step": 2480
    },
    {
      "epoch": 1.5241009946442234,
      "grad_norm": 2.030693531036377,
      "learning_rate": 4.70844390011567e-05,
      "loss": 1.3149,
      "step": 2490
    },
    {
      "epoch": 1.5302218821729152,
      "grad_norm": 2.416512966156006,
      "learning_rate": 4.705041845274546e-05,
      "loss": 1.2993,
      "step": 2500
    },
    {
      "epoch": 1.5363427697016068,
      "grad_norm": 3.0876128673553467,
      "learning_rate": 4.701639790433422e-05,
      "loss": 1.3011,
      "step": 2510
    },
    {
      "epoch": 1.5424636572302983,
      "grad_norm": 2.3045902252197266,
      "learning_rate": 4.6982377355922977e-05,
      "loss": 1.3061,
      "step": 2520
    },
    {
      "epoch": 1.54858454475899,
      "grad_norm": 2.438272476196289,
      "learning_rate": 4.694835680751174e-05,
      "loss": 1.3118,
      "step": 2530
    },
    {
      "epoch": 1.5547054322876819,
      "grad_norm": 2.2852370738983154,
      "learning_rate": 4.69143362591005e-05,
      "loss": 1.3074,
      "step": 2540
    },
    {
      "epoch": 1.5608263198163734,
      "grad_norm": 3.027848958969116,
      "learning_rate": 4.688031571068926e-05,
      "loss": 1.2948,
      "step": 2550
    },
    {
      "epoch": 1.566947207345065,
      "grad_norm": 2.1234147548675537,
      "learning_rate": 4.684629516227802e-05,
      "loss": 1.299,
      "step": 2560
    },
    {
      "epoch": 1.5730680948737565,
      "grad_norm": 2.619126558303833,
      "learning_rate": 4.681227461386678e-05,
      "loss": 1.298,
      "step": 2570
    },
    {
      "epoch": 1.5791889824024483,
      "grad_norm": 2.670396327972412,
      "learning_rate": 4.677825406545554e-05,
      "loss": 1.2962,
      "step": 2580
    },
    {
      "epoch": 1.58530986993114,
      "grad_norm": 2.637221097946167,
      "learning_rate": 4.6744233517044296e-05,
      "loss": 1.2873,
      "step": 2590
    },
    {
      "epoch": 1.5914307574598316,
      "grad_norm": 6.313690185546875,
      "learning_rate": 4.6710212968633056e-05,
      "loss": 1.3156,
      "step": 2600
    },
    {
      "epoch": 1.5975516449885232,
      "grad_norm": 2.477229356765747,
      "learning_rate": 4.667619242022182e-05,
      "loss": 1.3134,
      "step": 2610
    },
    {
      "epoch": 1.603672532517215,
      "grad_norm": 2.9845306873321533,
      "learning_rate": 4.664217187181058e-05,
      "loss": 1.333,
      "step": 2620
    },
    {
      "epoch": 1.6097934200459068,
      "grad_norm": 2.5738213062286377,
      "learning_rate": 4.660815132339933e-05,
      "loss": 1.3083,
      "step": 2630
    },
    {
      "epoch": 1.6159143075745983,
      "grad_norm": 2.4613943099975586,
      "learning_rate": 4.657413077498809e-05,
      "loss": 1.3116,
      "step": 2640
    },
    {
      "epoch": 1.6220351951032899,
      "grad_norm": 2.152653932571411,
      "learning_rate": 4.6540110226576854e-05,
      "loss": 1.3051,
      "step": 2650
    },
    {
      "epoch": 1.6281560826319816,
      "grad_norm": 2.0958948135375977,
      "learning_rate": 4.6506089678165615e-05,
      "loss": 1.3,
      "step": 2660
    },
    {
      "epoch": 1.6342769701606734,
      "grad_norm": 2.582873582839966,
      "learning_rate": 4.6472069129754375e-05,
      "loss": 1.3144,
      "step": 2670
    },
    {
      "epoch": 1.640397857689365,
      "grad_norm": 2.145533561706543,
      "learning_rate": 4.6438048581343136e-05,
      "loss": 1.3013,
      "step": 2680
    },
    {
      "epoch": 1.6465187452180565,
      "grad_norm": 1.912956714630127,
      "learning_rate": 4.64040280329319e-05,
      "loss": 1.3054,
      "step": 2690
    },
    {
      "epoch": 1.6526396327467483,
      "grad_norm": 1.9682807922363281,
      "learning_rate": 4.637000748452065e-05,
      "loss": 1.3341,
      "step": 2700
    },
    {
      "epoch": 1.65876052027544,
      "grad_norm": 1.8502757549285889,
      "learning_rate": 4.633598693610941e-05,
      "loss": 1.3353,
      "step": 2710
    },
    {
      "epoch": 1.6648814078041316,
      "grad_norm": 2.60172176361084,
      "learning_rate": 4.630196638769817e-05,
      "loss": 1.2979,
      "step": 2720
    },
    {
      "epoch": 1.6710022953328232,
      "grad_norm": 1.7690106630325317,
      "learning_rate": 4.626794583928693e-05,
      "loss": 1.2969,
      "step": 2730
    },
    {
      "epoch": 1.677123182861515,
      "grad_norm": 1.817347526550293,
      "learning_rate": 4.623392529087569e-05,
      "loss": 1.306,
      "step": 2740
    },
    {
      "epoch": 1.6832440703902067,
      "grad_norm": 1.666766881942749,
      "learning_rate": 4.619990474246445e-05,
      "loss": 1.3031,
      "step": 2750
    },
    {
      "epoch": 1.6893649579188983,
      "grad_norm": 2.351947784423828,
      "learning_rate": 4.616588419405321e-05,
      "loss": 1.2978,
      "step": 2760
    },
    {
      "epoch": 1.6954858454475898,
      "grad_norm": 2.820448875427246,
      "learning_rate": 4.613186364564197e-05,
      "loss": 1.3231,
      "step": 2770
    },
    {
      "epoch": 1.7016067329762814,
      "grad_norm": 1.8438854217529297,
      "learning_rate": 4.609784309723073e-05,
      "loss": 1.3081,
      "step": 2780
    },
    {
      "epoch": 1.7077276205049732,
      "grad_norm": 4.818421840667725,
      "learning_rate": 4.606382254881949e-05,
      "loss": 1.318,
      "step": 2790
    },
    {
      "epoch": 1.713848508033665,
      "grad_norm": 1.912014365196228,
      "learning_rate": 4.602980200040825e-05,
      "loss": 1.2886,
      "step": 2800
    },
    {
      "epoch": 1.7199693955623565,
      "grad_norm": 3.0061912536621094,
      "learning_rate": 4.5995781451997014e-05,
      "loss": 1.2966,
      "step": 2810
    },
    {
      "epoch": 1.726090283091048,
      "grad_norm": 1.7573401927947998,
      "learning_rate": 4.596176090358577e-05,
      "loss": 1.3127,
      "step": 2820
    },
    {
      "epoch": 1.7322111706197398,
      "grad_norm": 1.9873417615890503,
      "learning_rate": 4.592774035517453e-05,
      "loss": 1.3052,
      "step": 2830
    },
    {
      "epoch": 1.7383320581484316,
      "grad_norm": 2.0930285453796387,
      "learning_rate": 4.589371980676328e-05,
      "loss": 1.2923,
      "step": 2840
    },
    {
      "epoch": 1.7444529456771232,
      "grad_norm": 1.8706746101379395,
      "learning_rate": 4.585969925835204e-05,
      "loss": 1.2831,
      "step": 2850
    },
    {
      "epoch": 1.7505738332058147,
      "grad_norm": 1.7447609901428223,
      "learning_rate": 4.5825678709940804e-05,
      "loss": 1.3013,
      "step": 2860
    },
    {
      "epoch": 1.7566947207345065,
      "grad_norm": 1.7126678228378296,
      "learning_rate": 4.5791658161529565e-05,
      "loss": 1.2795,
      "step": 2870
    },
    {
      "epoch": 1.7628156082631983,
      "grad_norm": 3.166419267654419,
      "learning_rate": 4.5757637613118326e-05,
      "loss": 1.298,
      "step": 2880
    },
    {
      "epoch": 1.7689364957918898,
      "grad_norm": 1.741654396057129,
      "learning_rate": 4.572361706470709e-05,
      "loss": 1.3289,
      "step": 2890
    },
    {
      "epoch": 1.7750573833205814,
      "grad_norm": 3.99870228767395,
      "learning_rate": 4.568959651629585e-05,
      "loss": 1.3163,
      "step": 2900
    },
    {
      "epoch": 1.7811782708492732,
      "grad_norm": 2.293135404586792,
      "learning_rate": 4.565557596788461e-05,
      "loss": 1.2951,
      "step": 2910
    },
    {
      "epoch": 1.787299158377965,
      "grad_norm": 4.952125549316406,
      "learning_rate": 4.562155541947337e-05,
      "loss": 1.2998,
      "step": 2920
    },
    {
      "epoch": 1.7934200459066565,
      "grad_norm": 1.812955379486084,
      "learning_rate": 4.558753487106212e-05,
      "loss": 1.2733,
      "step": 2930
    },
    {
      "epoch": 1.799540933435348,
      "grad_norm": 1.9405971765518188,
      "learning_rate": 4.5553514322650884e-05,
      "loss": 1.3216,
      "step": 2940
    },
    {
      "epoch": 1.8056618209640398,
      "grad_norm": 2.513551712036133,
      "learning_rate": 4.551949377423964e-05,
      "loss": 1.2977,
      "step": 2950
    },
    {
      "epoch": 1.8117827084927316,
      "grad_norm": 2.0316007137298584,
      "learning_rate": 4.54854732258284e-05,
      "loss": 1.2958,
      "step": 2960
    },
    {
      "epoch": 1.8179035960214232,
      "grad_norm": 2.4681851863861084,
      "learning_rate": 4.545145267741716e-05,
      "loss": 1.2789,
      "step": 2970
    },
    {
      "epoch": 1.8240244835501147,
      "grad_norm": 2.1180663108825684,
      "learning_rate": 4.541743212900592e-05,
      "loss": 1.2687,
      "step": 2980
    },
    {
      "epoch": 1.8301453710788065,
      "grad_norm": 1.6326196193695068,
      "learning_rate": 4.538341158059468e-05,
      "loss": 1.314,
      "step": 2990
    },
    {
      "epoch": 1.836266258607498,
      "grad_norm": 1.7477805614471436,
      "learning_rate": 4.534939103218344e-05,
      "loss": 1.3038,
      "step": 3000
    },
    {
      "epoch": 1.8423871461361898,
      "grad_norm": 1.446861743927002,
      "learning_rate": 4.53153704837722e-05,
      "loss": 1.3192,
      "step": 3010
    },
    {
      "epoch": 1.8485080336648814,
      "grad_norm": 2.2364675998687744,
      "learning_rate": 4.5281349935360964e-05,
      "loss": 1.2723,
      "step": 3020
    },
    {
      "epoch": 1.854628921193573,
      "grad_norm": 1.59337317943573,
      "learning_rate": 4.5247329386949725e-05,
      "loss": 1.2792,
      "step": 3030
    },
    {
      "epoch": 1.8607498087222647,
      "grad_norm": 2.401291847229004,
      "learning_rate": 4.521330883853848e-05,
      "loss": 1.2842,
      "step": 3040
    },
    {
      "epoch": 1.8668706962509565,
      "grad_norm": 2.394993782043457,
      "learning_rate": 4.517928829012724e-05,
      "loss": 1.3205,
      "step": 3050
    },
    {
      "epoch": 1.872991583779648,
      "grad_norm": 2.0789761543273926,
      "learning_rate": 4.5145267741715994e-05,
      "loss": 1.2946,
      "step": 3060
    },
    {
      "epoch": 1.8791124713083396,
      "grad_norm": 4.516623020172119,
      "learning_rate": 4.5111247193304755e-05,
      "loss": 1.2772,
      "step": 3070
    },
    {
      "epoch": 1.8852333588370314,
      "grad_norm": 2.66314959526062,
      "learning_rate": 4.5077226644893515e-05,
      "loss": 1.2999,
      "step": 3080
    },
    {
      "epoch": 1.8913542463657231,
      "grad_norm": 2.047537088394165,
      "learning_rate": 4.5043206096482276e-05,
      "loss": 1.2849,
      "step": 3090
    },
    {
      "epoch": 1.8974751338944147,
      "grad_norm": 2.709646701812744,
      "learning_rate": 4.500918554807104e-05,
      "loss": 1.2807,
      "step": 3100
    },
    {
      "epoch": 1.9035960214231062,
      "grad_norm": 3.5094571113586426,
      "learning_rate": 4.49751649996598e-05,
      "loss": 1.2962,
      "step": 3110
    },
    {
      "epoch": 1.909716908951798,
      "grad_norm": 2.9054393768310547,
      "learning_rate": 4.494114445124856e-05,
      "loss": 1.2791,
      "step": 3120
    },
    {
      "epoch": 1.9158377964804898,
      "grad_norm": 2.601428747177124,
      "learning_rate": 4.490712390283732e-05,
      "loss": 1.3076,
      "step": 3130
    },
    {
      "epoch": 1.9219586840091814,
      "grad_norm": 2.667839527130127,
      "learning_rate": 4.487310335442608e-05,
      "loss": 1.2874,
      "step": 3140
    },
    {
      "epoch": 1.928079571537873,
      "grad_norm": 1.9442615509033203,
      "learning_rate": 4.4839082806014834e-05,
      "loss": 1.3045,
      "step": 3150
    },
    {
      "epoch": 1.9342004590665647,
      "grad_norm": 1.9131063222885132,
      "learning_rate": 4.4805062257603595e-05,
      "loss": 1.2915,
      "step": 3160
    },
    {
      "epoch": 1.9403213465952565,
      "grad_norm": 2.0807864665985107,
      "learning_rate": 4.477104170919235e-05,
      "loss": 1.2856,
      "step": 3170
    },
    {
      "epoch": 1.946442234123948,
      "grad_norm": 2.696155071258545,
      "learning_rate": 4.473702116078111e-05,
      "loss": 1.2945,
      "step": 3180
    },
    {
      "epoch": 1.9525631216526396,
      "grad_norm": 2.16047739982605,
      "learning_rate": 4.470300061236987e-05,
      "loss": 1.2889,
      "step": 3190
    },
    {
      "epoch": 1.9586840091813313,
      "grad_norm": 1.872473120689392,
      "learning_rate": 4.466898006395863e-05,
      "loss": 1.2869,
      "step": 3200
    },
    {
      "epoch": 1.9648048967100231,
      "grad_norm": 1.9989575147628784,
      "learning_rate": 4.463495951554739e-05,
      "loss": 1.2999,
      "step": 3210
    },
    {
      "epoch": 1.9709257842387147,
      "grad_norm": 1.451534390449524,
      "learning_rate": 4.4600938967136154e-05,
      "loss": 1.2875,
      "step": 3220
    },
    {
      "epoch": 1.9770466717674062,
      "grad_norm": 1.6723949909210205,
      "learning_rate": 4.4566918418724914e-05,
      "loss": 1.2951,
      "step": 3230
    },
    {
      "epoch": 1.9831675592960978,
      "grad_norm": 1.913266658782959,
      "learning_rate": 4.4532897870313675e-05,
      "loss": 1.2909,
      "step": 3240
    },
    {
      "epoch": 1.9892884468247896,
      "grad_norm": 2.985137462615967,
      "learning_rate": 4.449887732190243e-05,
      "loss": 1.2973,
      "step": 3250
    },
    {
      "epoch": 1.9954093343534813,
      "grad_norm": 2.2191321849823,
      "learning_rate": 4.446485677349119e-05,
      "loss": 1.2943,
      "step": 3260
    },
    {
      "epoch": 1.9996939556235653,
      "eval_accuracy": 0.4667951562918468,
      "eval_loss": 1.244300365447998,
      "eval_runtime": 2619.0656,
      "eval_samples_per_second": 79.836,
      "eval_steps_per_second": 0.624,
      "step": 3267
    },
    {
      "epoch": 2.001530221882173,
      "grad_norm": 1.887938380241394,
      "learning_rate": 4.443083622507995e-05,
      "loss": 1.2935,
      "step": 3270
    },
    {
      "epoch": 2.0076511094108644,
      "grad_norm": 2.1216981410980225,
      "learning_rate": 4.439681567666871e-05,
      "loss": 1.2836,
      "step": 3280
    },
    {
      "epoch": 2.0137719969395564,
      "grad_norm": 2.093498706817627,
      "learning_rate": 4.4362795128257466e-05,
      "loss": 1.2785,
      "step": 3290
    },
    {
      "epoch": 2.019892884468248,
      "grad_norm": 1.6856573820114136,
      "learning_rate": 4.432877457984623e-05,
      "loss": 1.3046,
      "step": 3300
    },
    {
      "epoch": 2.0260137719969395,
      "grad_norm": 2.1252543926239014,
      "learning_rate": 4.429475403143499e-05,
      "loss": 1.2838,
      "step": 3310
    },
    {
      "epoch": 2.032134659525631,
      "grad_norm": 2.125027894973755,
      "learning_rate": 4.426073348302375e-05,
      "loss": 1.2772,
      "step": 3320
    },
    {
      "epoch": 2.0382555470543227,
      "grad_norm": 2.1360585689544678,
      "learning_rate": 4.422671293461251e-05,
      "loss": 1.2974,
      "step": 3330
    },
    {
      "epoch": 2.0443764345830147,
      "grad_norm": 1.6038954257965088,
      "learning_rate": 4.419269238620127e-05,
      "loss": 1.2707,
      "step": 3340
    },
    {
      "epoch": 2.050497322111706,
      "grad_norm": 2.2439558506011963,
      "learning_rate": 4.415867183779003e-05,
      "loss": 1.2755,
      "step": 3350
    },
    {
      "epoch": 2.0566182096403978,
      "grad_norm": 1.75458824634552,
      "learning_rate": 4.4124651289378785e-05,
      "loss": 1.2673,
      "step": 3360
    },
    {
      "epoch": 2.0627390971690893,
      "grad_norm": 2.192709445953369,
      "learning_rate": 4.4090630740967546e-05,
      "loss": 1.2809,
      "step": 3370
    },
    {
      "epoch": 2.0688599846977813,
      "grad_norm": 2.699859142303467,
      "learning_rate": 4.4056610192556307e-05,
      "loss": 1.3,
      "step": 3380
    },
    {
      "epoch": 2.074980872226473,
      "grad_norm": 2.0490965843200684,
      "learning_rate": 4.402258964414507e-05,
      "loss": 1.2842,
      "step": 3390
    },
    {
      "epoch": 2.0811017597551644,
      "grad_norm": 2.2830417156219482,
      "learning_rate": 4.398856909573382e-05,
      "loss": 1.2604,
      "step": 3400
    },
    {
      "epoch": 2.087222647283856,
      "grad_norm": 1.8715858459472656,
      "learning_rate": 4.395454854732258e-05,
      "loss": 1.2829,
      "step": 3410
    },
    {
      "epoch": 2.093343534812548,
      "grad_norm": 3.053288221359253,
      "learning_rate": 4.392052799891134e-05,
      "loss": 1.2969,
      "step": 3420
    },
    {
      "epoch": 2.0994644223412395,
      "grad_norm": 2.098418712615967,
      "learning_rate": 4.3886507450500104e-05,
      "loss": 1.2879,
      "step": 3430
    },
    {
      "epoch": 2.105585309869931,
      "grad_norm": 2.1844096183776855,
      "learning_rate": 4.3852486902088865e-05,
      "loss": 1.2907,
      "step": 3440
    },
    {
      "epoch": 2.1117061973986226,
      "grad_norm": 2.036256790161133,
      "learning_rate": 4.3818466353677626e-05,
      "loss": 1.2921,
      "step": 3450
    },
    {
      "epoch": 2.1178270849273146,
      "grad_norm": 2.0535950660705566,
      "learning_rate": 4.3784445805266386e-05,
      "loss": 1.2739,
      "step": 3460
    },
    {
      "epoch": 2.123947972456006,
      "grad_norm": 2.0060815811157227,
      "learning_rate": 4.375042525685514e-05,
      "loss": 1.2891,
      "step": 3470
    },
    {
      "epoch": 2.1300688599846977,
      "grad_norm": 2.4420292377471924,
      "learning_rate": 4.37164047084439e-05,
      "loss": 1.2854,
      "step": 3480
    },
    {
      "epoch": 2.1361897475133893,
      "grad_norm": 1.8335957527160645,
      "learning_rate": 4.368238416003266e-05,
      "loss": 1.3028,
      "step": 3490
    },
    {
      "epoch": 2.1423106350420813,
      "grad_norm": 1.709681749343872,
      "learning_rate": 4.364836361162142e-05,
      "loss": 1.288,
      "step": 3500
    },
    {
      "epoch": 2.148431522570773,
      "grad_norm": 1.943090558052063,
      "learning_rate": 4.3614343063210184e-05,
      "loss": 1.2677,
      "step": 3510
    },
    {
      "epoch": 2.1545524100994644,
      "grad_norm": 2.027233600616455,
      "learning_rate": 4.358032251479894e-05,
      "loss": 1.283,
      "step": 3520
    },
    {
      "epoch": 2.160673297628156,
      "grad_norm": 1.6251754760742188,
      "learning_rate": 4.35463019663877e-05,
      "loss": 1.2895,
      "step": 3530
    },
    {
      "epoch": 2.1667941851568475,
      "grad_norm": 1.8065128326416016,
      "learning_rate": 4.351228141797646e-05,
      "loss": 1.2942,
      "step": 3540
    },
    {
      "epoch": 2.1729150726855395,
      "grad_norm": 2.0807602405548096,
      "learning_rate": 4.347826086956522e-05,
      "loss": 1.274,
      "step": 3550
    },
    {
      "epoch": 2.179035960214231,
      "grad_norm": 1.7792989015579224,
      "learning_rate": 4.344424032115398e-05,
      "loss": 1.2992,
      "step": 3560
    },
    {
      "epoch": 2.1851568477429226,
      "grad_norm": 2.005281448364258,
      "learning_rate": 4.3410219772742735e-05,
      "loss": 1.2703,
      "step": 3570
    },
    {
      "epoch": 2.191277735271614,
      "grad_norm": 2.485933780670166,
      "learning_rate": 4.3376199224331496e-05,
      "loss": 1.2949,
      "step": 3580
    },
    {
      "epoch": 2.197398622800306,
      "grad_norm": 1.4569703340530396,
      "learning_rate": 4.334217867592026e-05,
      "loss": 1.2788,
      "step": 3590
    },
    {
      "epoch": 2.2035195103289977,
      "grad_norm": 1.8452506065368652,
      "learning_rate": 4.330815812750902e-05,
      "loss": 1.2861,
      "step": 3600
    },
    {
      "epoch": 2.2096403978576893,
      "grad_norm": 1.921706199645996,
      "learning_rate": 4.327413757909778e-05,
      "loss": 1.2833,
      "step": 3610
    },
    {
      "epoch": 2.215761285386381,
      "grad_norm": 1.9338184595108032,
      "learning_rate": 4.324011703068654e-05,
      "loss": 1.2811,
      "step": 3620
    },
    {
      "epoch": 2.221882172915073,
      "grad_norm": 1.666922926902771,
      "learning_rate": 4.32060964822753e-05,
      "loss": 1.2631,
      "step": 3630
    },
    {
      "epoch": 2.2280030604437644,
      "grad_norm": 2.1670820713043213,
      "learning_rate": 4.3172075933864054e-05,
      "loss": 1.2949,
      "step": 3640
    },
    {
      "epoch": 2.234123947972456,
      "grad_norm": 1.980981707572937,
      "learning_rate": 4.3138055385452815e-05,
      "loss": 1.2569,
      "step": 3650
    },
    {
      "epoch": 2.2402448355011475,
      "grad_norm": 1.6385728120803833,
      "learning_rate": 4.3104034837041576e-05,
      "loss": 1.2807,
      "step": 3660
    },
    {
      "epoch": 2.2463657230298395,
      "grad_norm": 3.8424932956695557,
      "learning_rate": 4.307001428863034e-05,
      "loss": 1.2776,
      "step": 3670
    },
    {
      "epoch": 2.252486610558531,
      "grad_norm": 1.75303316116333,
      "learning_rate": 4.303599374021909e-05,
      "loss": 1.26,
      "step": 3680
    },
    {
      "epoch": 2.2586074980872226,
      "grad_norm": 2.4019761085510254,
      "learning_rate": 4.300197319180785e-05,
      "loss": 1.2846,
      "step": 3690
    },
    {
      "epoch": 2.264728385615914,
      "grad_norm": 1.8176641464233398,
      "learning_rate": 4.296795264339661e-05,
      "loss": 1.2877,
      "step": 3700
    },
    {
      "epoch": 2.270849273144606,
      "grad_norm": 3.1902916431427,
      "learning_rate": 4.293393209498537e-05,
      "loss": 1.2552,
      "step": 3710
    },
    {
      "epoch": 2.2769701606732977,
      "grad_norm": 1.9043002128601074,
      "learning_rate": 4.2899911546574134e-05,
      "loss": 1.2898,
      "step": 3720
    },
    {
      "epoch": 2.2830910482019893,
      "grad_norm": 1.6254369020462036,
      "learning_rate": 4.2865890998162895e-05,
      "loss": 1.2753,
      "step": 3730
    },
    {
      "epoch": 2.289211935730681,
      "grad_norm": 1.8241372108459473,
      "learning_rate": 4.2831870449751656e-05,
      "loss": 1.2722,
      "step": 3740
    },
    {
      "epoch": 2.295332823259373,
      "grad_norm": 1.5528113842010498,
      "learning_rate": 4.279784990134042e-05,
      "loss": 1.2576,
      "step": 3750
    },
    {
      "epoch": 2.3014537107880644,
      "grad_norm": 2.23053240776062,
      "learning_rate": 4.276382935292917e-05,
      "loss": 1.2991,
      "step": 3760
    },
    {
      "epoch": 2.307574598316756,
      "grad_norm": 2.2314915657043457,
      "learning_rate": 4.272980880451793e-05,
      "loss": 1.2812,
      "step": 3770
    },
    {
      "epoch": 2.3136954858454475,
      "grad_norm": 1.9923975467681885,
      "learning_rate": 4.269578825610669e-05,
      "loss": 1.3072,
      "step": 3780
    },
    {
      "epoch": 2.319816373374139,
      "grad_norm": 2.52502179145813,
      "learning_rate": 4.2661767707695446e-05,
      "loss": 1.2686,
      "step": 3790
    },
    {
      "epoch": 2.325937260902831,
      "grad_norm": 1.7113980054855347,
      "learning_rate": 4.262774715928421e-05,
      "loss": 1.2644,
      "step": 3800
    },
    {
      "epoch": 2.3320581484315226,
      "grad_norm": 1.640830636024475,
      "learning_rate": 4.259372661087297e-05,
      "loss": 1.2841,
      "step": 3810
    },
    {
      "epoch": 2.338179035960214,
      "grad_norm": 2.1533923149108887,
      "learning_rate": 4.255970606246173e-05,
      "loss": 1.2569,
      "step": 3820
    },
    {
      "epoch": 2.3442999234889057,
      "grad_norm": 1.8957128524780273,
      "learning_rate": 4.252568551405049e-05,
      "loss": 1.2557,
      "step": 3830
    },
    {
      "epoch": 2.3504208110175977,
      "grad_norm": 2.399122476577759,
      "learning_rate": 4.249166496563925e-05,
      "loss": 1.2755,
      "step": 3840
    },
    {
      "epoch": 2.3565416985462893,
      "grad_norm": 1.7818849086761475,
      "learning_rate": 4.245764441722801e-05,
      "loss": 1.2749,
      "step": 3850
    },
    {
      "epoch": 2.362662586074981,
      "grad_norm": 1.9884240627288818,
      "learning_rate": 4.242362386881677e-05,
      "loss": 1.2851,
      "step": 3860
    },
    {
      "epoch": 2.3687834736036724,
      "grad_norm": 1.8245644569396973,
      "learning_rate": 4.2389603320405526e-05,
      "loss": 1.2736,
      "step": 3870
    },
    {
      "epoch": 2.3749043611323644,
      "grad_norm": 2.0622308254241943,
      "learning_rate": 4.235558277199429e-05,
      "loss": 1.2661,
      "step": 3880
    },
    {
      "epoch": 2.381025248661056,
      "grad_norm": 1.7899513244628906,
      "learning_rate": 4.232156222358304e-05,
      "loss": 1.2735,
      "step": 3890
    },
    {
      "epoch": 2.3871461361897475,
      "grad_norm": 2.266698122024536,
      "learning_rate": 4.22875416751718e-05,
      "loss": 1.2875,
      "step": 3900
    },
    {
      "epoch": 2.393267023718439,
      "grad_norm": 1.6049247980117798,
      "learning_rate": 4.225352112676056e-05,
      "loss": 1.2709,
      "step": 3910
    },
    {
      "epoch": 2.399387911247131,
      "grad_norm": 2.069531202316284,
      "learning_rate": 4.2219500578349324e-05,
      "loss": 1.2725,
      "step": 3920
    },
    {
      "epoch": 2.4055087987758226,
      "grad_norm": 2.038214683532715,
      "learning_rate": 4.2185480029938085e-05,
      "loss": 1.284,
      "step": 3930
    },
    {
      "epoch": 2.411629686304514,
      "grad_norm": 1.7478989362716675,
      "learning_rate": 4.2151459481526845e-05,
      "loss": 1.2845,
      "step": 3940
    },
    {
      "epoch": 2.4177505738332057,
      "grad_norm": 1.536227822303772,
      "learning_rate": 4.2117438933115606e-05,
      "loss": 1.2724,
      "step": 3950
    },
    {
      "epoch": 2.4238714613618972,
      "grad_norm": 2.473824977874756,
      "learning_rate": 4.208341838470437e-05,
      "loss": 1.2793,
      "step": 3960
    },
    {
      "epoch": 2.4299923488905892,
      "grad_norm": 2.2828357219696045,
      "learning_rate": 4.204939783629313e-05,
      "loss": 1.2595,
      "step": 3970
    },
    {
      "epoch": 2.436113236419281,
      "grad_norm": 2.004650831222534,
      "learning_rate": 4.201537728788188e-05,
      "loss": 1.2778,
      "step": 3980
    },
    {
      "epoch": 2.4422341239479723,
      "grad_norm": 2.862926959991455,
      "learning_rate": 4.198135673947064e-05,
      "loss": 1.2625,
      "step": 3990
    },
    {
      "epoch": 2.4483550114766643,
      "grad_norm": 3.5888588428497314,
      "learning_rate": 4.19473361910594e-05,
      "loss": 1.2608,
      "step": 4000
    },
    {
      "epoch": 2.454475899005356,
      "grad_norm": 1.9902139902114868,
      "learning_rate": 4.191331564264816e-05,
      "loss": 1.2607,
      "step": 4010
    },
    {
      "epoch": 2.4605967865340475,
      "grad_norm": 2.3070342540740967,
      "learning_rate": 4.187929509423692e-05,
      "loss": 1.2555,
      "step": 4020
    },
    {
      "epoch": 2.466717674062739,
      "grad_norm": 1.9157483577728271,
      "learning_rate": 4.184527454582568e-05,
      "loss": 1.2881,
      "step": 4030
    },
    {
      "epoch": 2.4728385615914306,
      "grad_norm": 2.6677842140197754,
      "learning_rate": 4.181125399741444e-05,
      "loss": 1.2861,
      "step": 4040
    },
    {
      "epoch": 2.4789594491201226,
      "grad_norm": 1.9452474117279053,
      "learning_rate": 4.17772334490032e-05,
      "loss": 1.2648,
      "step": 4050
    },
    {
      "epoch": 2.485080336648814,
      "grad_norm": 2.034928798675537,
      "learning_rate": 4.174321290059196e-05,
      "loss": 1.2693,
      "step": 4060
    },
    {
      "epoch": 2.4912012241775057,
      "grad_norm": 2.1892433166503906,
      "learning_rate": 4.170919235218072e-05,
      "loss": 1.2807,
      "step": 4070
    },
    {
      "epoch": 2.4973221117061972,
      "grad_norm": 2.4424924850463867,
      "learning_rate": 4.1675171803769484e-05,
      "loss": 1.2849,
      "step": 4080
    },
    {
      "epoch": 2.503442999234889,
      "grad_norm": 2.0010828971862793,
      "learning_rate": 4.164115125535824e-05,
      "loss": 1.2695,
      "step": 4090
    },
    {
      "epoch": 2.5095638867635808,
      "grad_norm": 2.1089091300964355,
      "learning_rate": 4.1607130706947e-05,
      "loss": 1.2908,
      "step": 4100
    },
    {
      "epoch": 2.5156847742922723,
      "grad_norm": 2.363715887069702,
      "learning_rate": 4.157311015853575e-05,
      "loss": 1.2743,
      "step": 4110
    },
    {
      "epoch": 2.521805661820964,
      "grad_norm": 1.6981701850891113,
      "learning_rate": 4.153908961012451e-05,
      "loss": 1.2672,
      "step": 4120
    },
    {
      "epoch": 2.5279265493496554,
      "grad_norm": 1.7963359355926514,
      "learning_rate": 4.1505069061713274e-05,
      "loss": 1.2625,
      "step": 4130
    },
    {
      "epoch": 2.5340474368783474,
      "grad_norm": 2.267949104309082,
      "learning_rate": 4.1471048513302035e-05,
      "loss": 1.2701,
      "step": 4140
    },
    {
      "epoch": 2.540168324407039,
      "grad_norm": 1.7686315774917603,
      "learning_rate": 4.1437027964890796e-05,
      "loss": 1.2587,
      "step": 4150
    },
    {
      "epoch": 2.5462892119357305,
      "grad_norm": 1.8550716638565063,
      "learning_rate": 4.140300741647956e-05,
      "loss": 1.2511,
      "step": 4160
    },
    {
      "epoch": 2.5524100994644225,
      "grad_norm": 3.2846624851226807,
      "learning_rate": 4.136898686806832e-05,
      "loss": 1.2627,
      "step": 4170
    },
    {
      "epoch": 2.558530986993114,
      "grad_norm": 2.268218994140625,
      "learning_rate": 4.133496631965708e-05,
      "loss": 1.2658,
      "step": 4180
    },
    {
      "epoch": 2.5646518745218057,
      "grad_norm": 1.6908515691757202,
      "learning_rate": 4.130094577124584e-05,
      "loss": 1.2674,
      "step": 4190
    },
    {
      "epoch": 2.570772762050497,
      "grad_norm": 1.8642390966415405,
      "learning_rate": 4.126692522283459e-05,
      "loss": 1.2874,
      "step": 4200
    },
    {
      "epoch": 2.5768936495791888,
      "grad_norm": 2.945605516433716,
      "learning_rate": 4.1232904674423354e-05,
      "loss": 1.2729,
      "step": 4210
    },
    {
      "epoch": 2.5830145371078808,
      "grad_norm": 2.022144079208374,
      "learning_rate": 4.119888412601211e-05,
      "loss": 1.2687,
      "step": 4220
    },
    {
      "epoch": 2.5891354246365723,
      "grad_norm": 1.9110801219940186,
      "learning_rate": 4.116486357760087e-05,
      "loss": 1.2739,
      "step": 4230
    },
    {
      "epoch": 2.595256312165264,
      "grad_norm": 2.433802366256714,
      "learning_rate": 4.113084302918963e-05,
      "loss": 1.2572,
      "step": 4240
    },
    {
      "epoch": 2.601377199693956,
      "grad_norm": 1.9494119882583618,
      "learning_rate": 4.109682248077839e-05,
      "loss": 1.2653,
      "step": 4250
    },
    {
      "epoch": 2.6074980872226474,
      "grad_norm": 1.64384126663208,
      "learning_rate": 4.106280193236715e-05,
      "loss": 1.2791,
      "step": 4260
    },
    {
      "epoch": 2.613618974751339,
      "grad_norm": 2.202897787094116,
      "learning_rate": 4.102878138395591e-05,
      "loss": 1.2564,
      "step": 4270
    },
    {
      "epoch": 2.6197398622800305,
      "grad_norm": 2.0833334922790527,
      "learning_rate": 4.099476083554467e-05,
      "loss": 1.2915,
      "step": 4280
    },
    {
      "epoch": 2.625860749808722,
      "grad_norm": 2.360783815383911,
      "learning_rate": 4.0960740287133434e-05,
      "loss": 1.2879,
      "step": 4290
    },
    {
      "epoch": 2.631981637337414,
      "grad_norm": 1.80355966091156,
      "learning_rate": 4.092671973872219e-05,
      "loss": 1.2798,
      "step": 4300
    },
    {
      "epoch": 2.6381025248661056,
      "grad_norm": 2.43011212348938,
      "learning_rate": 4.089269919031095e-05,
      "loss": 1.2694,
      "step": 4310
    },
    {
      "epoch": 2.644223412394797,
      "grad_norm": 1.8871304988861084,
      "learning_rate": 4.085867864189971e-05,
      "loss": 1.2641,
      "step": 4320
    },
    {
      "epoch": 2.6503442999234887,
      "grad_norm": 2.0215072631835938,
      "learning_rate": 4.082465809348847e-05,
      "loss": 1.2608,
      "step": 4330
    },
    {
      "epoch": 2.6564651874521807,
      "grad_norm": 1.9137364625930786,
      "learning_rate": 4.0790637545077225e-05,
      "loss": 1.27,
      "step": 4340
    },
    {
      "epoch": 2.6625860749808723,
      "grad_norm": 1.7641111612319946,
      "learning_rate": 4.0756616996665985e-05,
      "loss": 1.3004,
      "step": 4350
    },
    {
      "epoch": 2.668706962509564,
      "grad_norm": 1.8382567167282104,
      "learning_rate": 4.0722596448254746e-05,
      "loss": 1.2596,
      "step": 4360
    },
    {
      "epoch": 2.6748278500382554,
      "grad_norm": 2.095576047897339,
      "learning_rate": 4.068857589984351e-05,
      "loss": 1.2685,
      "step": 4370
    },
    {
      "epoch": 2.680948737566947,
      "grad_norm": 1.7252328395843506,
      "learning_rate": 4.065455535143227e-05,
      "loss": 1.3021,
      "step": 4380
    },
    {
      "epoch": 2.687069625095639,
      "grad_norm": 3.130185604095459,
      "learning_rate": 4.062053480302103e-05,
      "loss": 1.2793,
      "step": 4390
    },
    {
      "epoch": 2.6931905126243305,
      "grad_norm": 1.8197978734970093,
      "learning_rate": 4.058651425460979e-05,
      "loss": 1.2679,
      "step": 4400
    },
    {
      "epoch": 2.699311400153022,
      "grad_norm": 2.7329905033111572,
      "learning_rate": 4.0552493706198544e-05,
      "loss": 1.2627,
      "step": 4410
    },
    {
      "epoch": 2.705432287681714,
      "grad_norm": 2.3365578651428223,
      "learning_rate": 4.0518473157787304e-05,
      "loss": 1.2552,
      "step": 4420
    },
    {
      "epoch": 2.7115531752104056,
      "grad_norm": 2.006134510040283,
      "learning_rate": 4.0484452609376065e-05,
      "loss": 1.2712,
      "step": 4430
    },
    {
      "epoch": 2.717674062739097,
      "grad_norm": 1.6508187055587769,
      "learning_rate": 4.0450432060964826e-05,
      "loss": 1.2739,
      "step": 4440
    },
    {
      "epoch": 2.7237949502677887,
      "grad_norm": 2.231863021850586,
      "learning_rate": 4.041641151255359e-05,
      "loss": 1.2694,
      "step": 4450
    },
    {
      "epoch": 2.7299158377964803,
      "grad_norm": 1.7678385972976685,
      "learning_rate": 4.038239096414234e-05,
      "loss": 1.2696,
      "step": 4460
    },
    {
      "epoch": 2.7360367253251723,
      "grad_norm": 1.7382951974868774,
      "learning_rate": 4.03483704157311e-05,
      "loss": 1.2863,
      "step": 4470
    },
    {
      "epoch": 2.742157612853864,
      "grad_norm": 1.6372942924499512,
      "learning_rate": 4.031434986731986e-05,
      "loss": 1.2688,
      "step": 4480
    },
    {
      "epoch": 2.7482785003825554,
      "grad_norm": 1.9079464673995972,
      "learning_rate": 4.0280329318908623e-05,
      "loss": 1.2702,
      "step": 4490
    },
    {
      "epoch": 2.7543993879112474,
      "grad_norm": 1.8411625623703003,
      "learning_rate": 4.0246308770497384e-05,
      "loss": 1.2515,
      "step": 4500
    },
    {
      "epoch": 2.760520275439939,
      "grad_norm": 1.8902119398117065,
      "learning_rate": 4.0212288222086145e-05,
      "loss": 1.2837,
      "step": 4510
    },
    {
      "epoch": 2.7666411629686305,
      "grad_norm": 1.4213835000991821,
      "learning_rate": 4.01782676736749e-05,
      "loss": 1.261,
      "step": 4520
    },
    {
      "epoch": 2.772762050497322,
      "grad_norm": 2.356781244277954,
      "learning_rate": 4.014424712526366e-05,
      "loss": 1.2697,
      "step": 4530
    },
    {
      "epoch": 2.7788829380260136,
      "grad_norm": 2.5330777168273926,
      "learning_rate": 4.011022657685242e-05,
      "loss": 1.2639,
      "step": 4540
    },
    {
      "epoch": 2.785003825554705,
      "grad_norm": 2.177523374557495,
      "learning_rate": 4.007620602844118e-05,
      "loss": 1.2547,
      "step": 4550
    },
    {
      "epoch": 2.791124713083397,
      "grad_norm": 2.4313042163848877,
      "learning_rate": 4.004218548002994e-05,
      "loss": 1.2772,
      "step": 4560
    },
    {
      "epoch": 2.7972456006120887,
      "grad_norm": 2.4263336658477783,
      "learning_rate": 4.00081649316187e-05,
      "loss": 1.2484,
      "step": 4570
    },
    {
      "epoch": 2.8033664881407803,
      "grad_norm": 1.936612606048584,
      "learning_rate": 3.997414438320746e-05,
      "loss": 1.267,
      "step": 4580
    },
    {
      "epoch": 2.8094873756694723,
      "grad_norm": 2.0780820846557617,
      "learning_rate": 3.994012383479622e-05,
      "loss": 1.2625,
      "step": 4590
    },
    {
      "epoch": 2.815608263198164,
      "grad_norm": 1.8207616806030273,
      "learning_rate": 3.990610328638498e-05,
      "loss": 1.2544,
      "step": 4600
    },
    {
      "epoch": 2.8217291507268554,
      "grad_norm": 2.180629014968872,
      "learning_rate": 3.987208273797374e-05,
      "loss": 1.2701,
      "step": 4610
    },
    {
      "epoch": 2.827850038255547,
      "grad_norm": 2.3692867755889893,
      "learning_rate": 3.9838062189562494e-05,
      "loss": 1.2784,
      "step": 4620
    },
    {
      "epoch": 2.8339709257842385,
      "grad_norm": 1.676451563835144,
      "learning_rate": 3.9804041641151255e-05,
      "loss": 1.28,
      "step": 4630
    },
    {
      "epoch": 2.8400918133129305,
      "grad_norm": 2.219745397567749,
      "learning_rate": 3.9770021092740016e-05,
      "loss": 1.256,
      "step": 4640
    },
    {
      "epoch": 2.846212700841622,
      "grad_norm": 2.0797510147094727,
      "learning_rate": 3.9736000544328776e-05,
      "loss": 1.2526,
      "step": 4650
    },
    {
      "epoch": 2.8523335883703136,
      "grad_norm": 2.238093614578247,
      "learning_rate": 3.970197999591754e-05,
      "loss": 1.2936,
      "step": 4660
    },
    {
      "epoch": 2.8584544758990056,
      "grad_norm": 2.133490562438965,
      "learning_rate": 3.96679594475063e-05,
      "loss": 1.2578,
      "step": 4670
    },
    {
      "epoch": 2.864575363427697,
      "grad_norm": 2.0491223335266113,
      "learning_rate": 3.963393889909506e-05,
      "loss": 1.2564,
      "step": 4680
    },
    {
      "epoch": 2.8706962509563887,
      "grad_norm": 1.5369951725006104,
      "learning_rate": 3.959991835068381e-05,
      "loss": 1.2806,
      "step": 4690
    },
    {
      "epoch": 2.8768171384850802,
      "grad_norm": 1.8396649360656738,
      "learning_rate": 3.9565897802272574e-05,
      "loss": 1.2763,
      "step": 4700
    },
    {
      "epoch": 2.882938026013772,
      "grad_norm": 1.60837721824646,
      "learning_rate": 3.9531877253861335e-05,
      "loss": 1.2606,
      "step": 4710
    },
    {
      "epoch": 2.889058913542464,
      "grad_norm": 2.0090882778167725,
      "learning_rate": 3.9497856705450096e-05,
      "loss": 1.2633,
      "step": 4720
    },
    {
      "epoch": 2.8951798010711554,
      "grad_norm": 3.1234018802642822,
      "learning_rate": 3.946383615703885e-05,
      "loss": 1.2743,
      "step": 4730
    },
    {
      "epoch": 2.901300688599847,
      "grad_norm": 1.919801115989685,
      "learning_rate": 3.942981560862761e-05,
      "loss": 1.2815,
      "step": 4740
    },
    {
      "epoch": 2.907421576128539,
      "grad_norm": 1.6320799589157104,
      "learning_rate": 3.939579506021637e-05,
      "loss": 1.2651,
      "step": 4750
    },
    {
      "epoch": 2.9135424636572305,
      "grad_norm": 1.6664077043533325,
      "learning_rate": 3.936177451180513e-05,
      "loss": 1.267,
      "step": 4760
    },
    {
      "epoch": 2.919663351185922,
      "grad_norm": 1.7831507921218872,
      "learning_rate": 3.932775396339389e-05,
      "loss": 1.2613,
      "step": 4770
    },
    {
      "epoch": 2.9257842387146136,
      "grad_norm": 1.8862683773040771,
      "learning_rate": 3.9293733414982654e-05,
      "loss": 1.2512,
      "step": 4780
    },
    {
      "epoch": 2.931905126243305,
      "grad_norm": 1.8569462299346924,
      "learning_rate": 3.9259712866571415e-05,
      "loss": 1.2508,
      "step": 4790
    },
    {
      "epoch": 2.9380260137719967,
      "grad_norm": 2.7612645626068115,
      "learning_rate": 3.9225692318160175e-05,
      "loss": 1.2685,
      "step": 4800
    },
    {
      "epoch": 2.9441469013006887,
      "grad_norm": 3.243058204650879,
      "learning_rate": 3.919167176974893e-05,
      "loss": 1.2725,
      "step": 4810
    },
    {
      "epoch": 2.9502677888293802,
      "grad_norm": 1.8038846254348755,
      "learning_rate": 3.915765122133769e-05,
      "loss": 1.2791,
      "step": 4820
    },
    {
      "epoch": 2.956388676358072,
      "grad_norm": 2.171898603439331,
      "learning_rate": 3.912363067292645e-05,
      "loss": 1.2533,
      "step": 4830
    },
    {
      "epoch": 2.962509563886764,
      "grad_norm": 2.8486342430114746,
      "learning_rate": 3.9089610124515205e-05,
      "loss": 1.2483,
      "step": 4840
    },
    {
      "epoch": 2.9686304514154553,
      "grad_norm": 1.8373701572418213,
      "learning_rate": 3.9055589576103966e-05,
      "loss": 1.2397,
      "step": 4850
    },
    {
      "epoch": 2.974751338944147,
      "grad_norm": 1.858534336090088,
      "learning_rate": 3.902156902769273e-05,
      "loss": 1.2703,
      "step": 4860
    },
    {
      "epoch": 2.9808722264728384,
      "grad_norm": 2.107698678970337,
      "learning_rate": 3.898754847928149e-05,
      "loss": 1.266,
      "step": 4870
    },
    {
      "epoch": 2.98699311400153,
      "grad_norm": 1.9265609979629517,
      "learning_rate": 3.895352793087025e-05,
      "loss": 1.2691,
      "step": 4880
    },
    {
      "epoch": 2.993114001530222,
      "grad_norm": 2.060009241104126,
      "learning_rate": 3.891950738245901e-05,
      "loss": 1.2656,
      "step": 4890
    },
    {
      "epoch": 2.9992348890589136,
      "grad_norm": 1.5703758001327515,
      "learning_rate": 3.888548683404777e-05,
      "loss": 1.2411,
      "step": 4900
    },
    {
      "epoch": 2.9998469778117824,
      "eval_accuracy": 0.47869399701572485,
      "eval_loss": 1.2228519916534424,
      "eval_runtime": 2617.8718,
      "eval_samples_per_second": 79.873,
      "eval_steps_per_second": 0.624,
      "step": 4901
    },
    {
      "epoch": 3.005355776587605,
      "grad_norm": 1.9858216047286987,
      "learning_rate": 3.885146628563653e-05,
      "loss": 1.273,
      "step": 4910
    },
    {
      "epoch": 3.0114766641162967,
      "grad_norm": 2.257009744644165,
      "learning_rate": 3.881744573722529e-05,
      "loss": 1.2604,
      "step": 4920
    },
    {
      "epoch": 3.0175975516449887,
      "grad_norm": 2.2428131103515625,
      "learning_rate": 3.8783425188814046e-05,
      "loss": 1.2407,
      "step": 4930
    },
    {
      "epoch": 3.02371843917368,
      "grad_norm": 1.81501042842865,
      "learning_rate": 3.87494046404028e-05,
      "loss": 1.249,
      "step": 4940
    },
    {
      "epoch": 3.0298393267023718,
      "grad_norm": 2.2534372806549072,
      "learning_rate": 3.871538409199156e-05,
      "loss": 1.2608,
      "step": 4950
    },
    {
      "epoch": 3.0359602142310633,
      "grad_norm": 1.6115994453430176,
      "learning_rate": 3.868136354358032e-05,
      "loss": 1.2425,
      "step": 4960
    },
    {
      "epoch": 3.0420811017597553,
      "grad_norm": 2.0571775436401367,
      "learning_rate": 3.864734299516908e-05,
      "loss": 1.2551,
      "step": 4970
    },
    {
      "epoch": 3.048201989288447,
      "grad_norm": 2.081515073776245,
      "learning_rate": 3.861332244675784e-05,
      "loss": 1.2524,
      "step": 4980
    },
    {
      "epoch": 3.0543228768171384,
      "grad_norm": 2.2645437717437744,
      "learning_rate": 3.8579301898346604e-05,
      "loss": 1.2413,
      "step": 4990
    },
    {
      "epoch": 3.06044376434583,
      "grad_norm": 2.3221487998962402,
      "learning_rate": 3.8545281349935365e-05,
      "loss": 1.2519,
      "step": 5000
    },
    {
      "epoch": 3.066564651874522,
      "grad_norm": 3.2614245414733887,
      "learning_rate": 3.8511260801524126e-05,
      "loss": 1.2543,
      "step": 5010
    },
    {
      "epoch": 3.0726855394032135,
      "grad_norm": 1.9711055755615234,
      "learning_rate": 3.847724025311289e-05,
      "loss": 1.2583,
      "step": 5020
    },
    {
      "epoch": 3.078806426931905,
      "grad_norm": 2.410093069076538,
      "learning_rate": 3.844321970470164e-05,
      "loss": 1.2661,
      "step": 5030
    },
    {
      "epoch": 3.0849273144605966,
      "grad_norm": 2.4068052768707275,
      "learning_rate": 3.84091991562904e-05,
      "loss": 1.2298,
      "step": 5040
    },
    {
      "epoch": 3.0910482019892886,
      "grad_norm": 1.658258080482483,
      "learning_rate": 3.8375178607879156e-05,
      "loss": 1.2351,
      "step": 5050
    },
    {
      "epoch": 3.09716908951798,
      "grad_norm": 1.8933178186416626,
      "learning_rate": 3.8341158059467916e-05,
      "loss": 1.2665,
      "step": 5060
    },
    {
      "epoch": 3.1032899770466718,
      "grad_norm": 2.4667043685913086,
      "learning_rate": 3.830713751105668e-05,
      "loss": 1.2583,
      "step": 5070
    },
    {
      "epoch": 3.1094108645753633,
      "grad_norm": 2.2437615394592285,
      "learning_rate": 3.827311696264544e-05,
      "loss": 1.2442,
      "step": 5080
    },
    {
      "epoch": 3.1155317521040553,
      "grad_norm": 2.0303754806518555,
      "learning_rate": 3.82390964142342e-05,
      "loss": 1.2395,
      "step": 5090
    },
    {
      "epoch": 3.121652639632747,
      "grad_norm": 2.256101608276367,
      "learning_rate": 3.820507586582296e-05,
      "loss": 1.2621,
      "step": 5100
    },
    {
      "epoch": 3.1277735271614384,
      "grad_norm": 4.909579753875732,
      "learning_rate": 3.817105531741172e-05,
      "loss": 1.2733,
      "step": 5110
    },
    {
      "epoch": 3.13389441469013,
      "grad_norm": 1.6804922819137573,
      "learning_rate": 3.813703476900048e-05,
      "loss": 1.2408,
      "step": 5120
    },
    {
      "epoch": 3.1400153022188215,
      "grad_norm": 1.8485102653503418,
      "learning_rate": 3.810301422058924e-05,
      "loss": 1.2511,
      "step": 5130
    },
    {
      "epoch": 3.1461361897475135,
      "grad_norm": 2.0070157051086426,
      "learning_rate": 3.8068993672177996e-05,
      "loss": 1.2407,
      "step": 5140
    },
    {
      "epoch": 3.152257077276205,
      "grad_norm": 2.0214972496032715,
      "learning_rate": 3.803497312376676e-05,
      "loss": 1.2527,
      "step": 5150
    },
    {
      "epoch": 3.1583779648048966,
      "grad_norm": 1.8533238172531128,
      "learning_rate": 3.800095257535551e-05,
      "loss": 1.2293,
      "step": 5160
    },
    {
      "epoch": 3.164498852333588,
      "grad_norm": 1.751216173171997,
      "learning_rate": 3.796693202694427e-05,
      "loss": 1.2299,
      "step": 5170
    },
    {
      "epoch": 3.17061973986228,
      "grad_norm": 2.6898555755615234,
      "learning_rate": 3.793291147853303e-05,
      "loss": 1.2641,
      "step": 5180
    },
    {
      "epoch": 3.1767406273909717,
      "grad_norm": 1.6866509914398193,
      "learning_rate": 3.7898890930121794e-05,
      "loss": 1.2722,
      "step": 5190
    },
    {
      "epoch": 3.1828615149196633,
      "grad_norm": 3.089430093765259,
      "learning_rate": 3.7864870381710555e-05,
      "loss": 1.2548,
      "step": 5200
    },
    {
      "epoch": 3.188982402448355,
      "grad_norm": 2.683788776397705,
      "learning_rate": 3.7830849833299315e-05,
      "loss": 1.2477,
      "step": 5210
    },
    {
      "epoch": 3.195103289977047,
      "grad_norm": 2.5055129528045654,
      "learning_rate": 3.7796829284888076e-05,
      "loss": 1.2566,
      "step": 5220
    },
    {
      "epoch": 3.2012241775057384,
      "grad_norm": 2.4237446784973145,
      "learning_rate": 3.776280873647684e-05,
      "loss": 1.2662,
      "step": 5230
    },
    {
      "epoch": 3.20734506503443,
      "grad_norm": 1.907581090927124,
      "learning_rate": 3.77287881880656e-05,
      "loss": 1.2348,
      "step": 5240
    },
    {
      "epoch": 3.2134659525631215,
      "grad_norm": 2.126953363418579,
      "learning_rate": 3.769476763965435e-05,
      "loss": 1.2582,
      "step": 5250
    },
    {
      "epoch": 3.2195868400918135,
      "grad_norm": 2.169069766998291,
      "learning_rate": 3.766074709124311e-05,
      "loss": 1.2571,
      "step": 5260
    },
    {
      "epoch": 3.225707727620505,
      "grad_norm": 2.7466557025909424,
      "learning_rate": 3.7626726542831874e-05,
      "loss": 1.2542,
      "step": 5270
    },
    {
      "epoch": 3.2318286151491966,
      "grad_norm": 2.4945733547210693,
      "learning_rate": 3.759270599442063e-05,
      "loss": 1.2634,
      "step": 5280
    },
    {
      "epoch": 3.237949502677888,
      "grad_norm": 2.4939398765563965,
      "learning_rate": 3.755868544600939e-05,
      "loss": 1.2249,
      "step": 5290
    },
    {
      "epoch": 3.2440703902065797,
      "grad_norm": 2.0732944011688232,
      "learning_rate": 3.752466489759815e-05,
      "loss": 1.2717,
      "step": 5300
    },
    {
      "epoch": 3.2501912777352717,
      "grad_norm": 2.3293912410736084,
      "learning_rate": 3.749064434918691e-05,
      "loss": 1.2467,
      "step": 5310
    },
    {
      "epoch": 3.2563121652639633,
      "grad_norm": 2.104963779449463,
      "learning_rate": 3.745662380077567e-05,
      "loss": 1.2682,
      "step": 5320
    },
    {
      "epoch": 3.262433052792655,
      "grad_norm": 2.453064203262329,
      "learning_rate": 3.742260325236443e-05,
      "loss": 1.2593,
      "step": 5330
    },
    {
      "epoch": 3.268553940321347,
      "grad_norm": 2.3136184215545654,
      "learning_rate": 3.738858270395319e-05,
      "loss": 1.2457,
      "step": 5340
    },
    {
      "epoch": 3.2746748278500384,
      "grad_norm": 2.3127260208129883,
      "learning_rate": 3.735456215554195e-05,
      "loss": 1.2452,
      "step": 5350
    },
    {
      "epoch": 3.28079571537873,
      "grad_norm": 2.367079496383667,
      "learning_rate": 3.732054160713071e-05,
      "loss": 1.2439,
      "step": 5360
    },
    {
      "epoch": 3.2869166029074215,
      "grad_norm": 2.3883674144744873,
      "learning_rate": 3.728652105871947e-05,
      "loss": 1.2428,
      "step": 5370
    },
    {
      "epoch": 3.293037490436113,
      "grad_norm": 2.274970531463623,
      "learning_rate": 3.725250051030823e-05,
      "loss": 1.2567,
      "step": 5380
    },
    {
      "epoch": 3.299158377964805,
      "grad_norm": 2.298823356628418,
      "learning_rate": 3.721847996189698e-05,
      "loss": 1.2633,
      "step": 5390
    },
    {
      "epoch": 3.3052792654934966,
      "grad_norm": 2.3997528553009033,
      "learning_rate": 3.7184459413485744e-05,
      "loss": 1.2277,
      "step": 5400
    },
    {
      "epoch": 3.311400153022188,
      "grad_norm": 2.273841619491577,
      "learning_rate": 3.7150438865074505e-05,
      "loss": 1.2629,
      "step": 5410
    },
    {
      "epoch": 3.3175210405508797,
      "grad_norm": 1.885188102722168,
      "learning_rate": 3.7116418316663266e-05,
      "loss": 1.2464,
      "step": 5420
    },
    {
      "epoch": 3.3236419280795717,
      "grad_norm": 2.231029987335205,
      "learning_rate": 3.708239776825203e-05,
      "loss": 1.2562,
      "step": 5430
    },
    {
      "epoch": 3.3297628156082633,
      "grad_norm": 2.6225924491882324,
      "learning_rate": 3.704837721984079e-05,
      "loss": 1.2352,
      "step": 5440
    },
    {
      "epoch": 3.335883703136955,
      "grad_norm": 3.8559517860412598,
      "learning_rate": 3.701435667142955e-05,
      "loss": 1.2598,
      "step": 5450
    },
    {
      "epoch": 3.3420045906656464,
      "grad_norm": 1.7077224254608154,
      "learning_rate": 3.69803361230183e-05,
      "loss": 1.2587,
      "step": 5460
    },
    {
      "epoch": 3.3481254781943384,
      "grad_norm": 1.9027427434921265,
      "learning_rate": 3.694631557460706e-05,
      "loss": 1.2417,
      "step": 5470
    },
    {
      "epoch": 3.35424636572303,
      "grad_norm": 2.166884183883667,
      "learning_rate": 3.6912295026195824e-05,
      "loss": 1.2559,
      "step": 5480
    },
    {
      "epoch": 3.3603672532517215,
      "grad_norm": 2.6949479579925537,
      "learning_rate": 3.6878274477784585e-05,
      "loss": 1.2529,
      "step": 5490
    },
    {
      "epoch": 3.366488140780413,
      "grad_norm": 2.176804304122925,
      "learning_rate": 3.6844253929373346e-05,
      "loss": 1.2583,
      "step": 5500
    },
    {
      "epoch": 3.372609028309105,
      "grad_norm": 2.2006351947784424,
      "learning_rate": 3.68102333809621e-05,
      "loss": 1.2481,
      "step": 5510
    },
    {
      "epoch": 3.3787299158377966,
      "grad_norm": 1.8229804039001465,
      "learning_rate": 3.677621283255086e-05,
      "loss": 1.2492,
      "step": 5520
    },
    {
      "epoch": 3.384850803366488,
      "grad_norm": 2.4698712825775146,
      "learning_rate": 3.674219228413962e-05,
      "loss": 1.246,
      "step": 5530
    },
    {
      "epoch": 3.3909716908951797,
      "grad_norm": 2.333577871322632,
      "learning_rate": 3.670817173572838e-05,
      "loss": 1.2416,
      "step": 5540
    },
    {
      "epoch": 3.3970925784238712,
      "grad_norm": 1.945576548576355,
      "learning_rate": 3.667415118731714e-05,
      "loss": 1.2501,
      "step": 5550
    },
    {
      "epoch": 3.4032134659525632,
      "grad_norm": 2.6930272579193115,
      "learning_rate": 3.6640130638905904e-05,
      "loss": 1.237,
      "step": 5560
    },
    {
      "epoch": 3.409334353481255,
      "grad_norm": 1.8951822519302368,
      "learning_rate": 3.660611009049466e-05,
      "loss": 1.247,
      "step": 5570
    },
    {
      "epoch": 3.4154552410099464,
      "grad_norm": 2.351973533630371,
      "learning_rate": 3.657208954208342e-05,
      "loss": 1.2406,
      "step": 5580
    },
    {
      "epoch": 3.4215761285386384,
      "grad_norm": 2.6603872776031494,
      "learning_rate": 3.653806899367218e-05,
      "loss": 1.2575,
      "step": 5590
    },
    {
      "epoch": 3.42769701606733,
      "grad_norm": 2.3119587898254395,
      "learning_rate": 3.650404844526094e-05,
      "loss": 1.2376,
      "step": 5600
    },
    {
      "epoch": 3.4338179035960215,
      "grad_norm": 2.6269590854644775,
      "learning_rate": 3.64700278968497e-05,
      "loss": 1.2413,
      "step": 5610
    },
    {
      "epoch": 3.439938791124713,
      "grad_norm": 2.115171194076538,
      "learning_rate": 3.643600734843846e-05,
      "loss": 1.2224,
      "step": 5620
    },
    {
      "epoch": 3.4460596786534046,
      "grad_norm": 1.861360788345337,
      "learning_rate": 3.6401986800027216e-05,
      "loss": 1.2664,
      "step": 5630
    },
    {
      "epoch": 3.4521805661820966,
      "grad_norm": 2.1996967792510986,
      "learning_rate": 3.636796625161598e-05,
      "loss": 1.2546,
      "step": 5640
    },
    {
      "epoch": 3.458301453710788,
      "grad_norm": 2.7611207962036133,
      "learning_rate": 3.633394570320474e-05,
      "loss": 1.2629,
      "step": 5650
    },
    {
      "epoch": 3.4644223412394797,
      "grad_norm": 2.022286891937256,
      "learning_rate": 3.62999251547935e-05,
      "loss": 1.2454,
      "step": 5660
    },
    {
      "epoch": 3.4705432287681712,
      "grad_norm": 1.778696894645691,
      "learning_rate": 3.626590460638225e-05,
      "loss": 1.2577,
      "step": 5670
    },
    {
      "epoch": 3.4766641162968632,
      "grad_norm": 2.090994358062744,
      "learning_rate": 3.6231884057971014e-05,
      "loss": 1.2511,
      "step": 5680
    },
    {
      "epoch": 3.482785003825555,
      "grad_norm": 2.744858741760254,
      "learning_rate": 3.6197863509559774e-05,
      "loss": 1.2579,
      "step": 5690
    },
    {
      "epoch": 3.4889058913542463,
      "grad_norm": 2.450564384460449,
      "learning_rate": 3.6163842961148535e-05,
      "loss": 1.2393,
      "step": 5700
    },
    {
      "epoch": 3.495026778882938,
      "grad_norm": 1.8227015733718872,
      "learning_rate": 3.6129822412737296e-05,
      "loss": 1.2361,
      "step": 5710
    },
    {
      "epoch": 3.5011476664116294,
      "grad_norm": 1.8969275951385498,
      "learning_rate": 3.609580186432606e-05,
      "loss": 1.2275,
      "step": 5720
    },
    {
      "epoch": 3.5072685539403214,
      "grad_norm": 1.7384281158447266,
      "learning_rate": 3.606178131591482e-05,
      "loss": 1.2522,
      "step": 5730
    },
    {
      "epoch": 3.513389441469013,
      "grad_norm": 2.120198965072632,
      "learning_rate": 3.602776076750358e-05,
      "loss": 1.2322,
      "step": 5740
    },
    {
      "epoch": 3.5195103289977046,
      "grad_norm": 1.9926379919052124,
      "learning_rate": 3.599374021909233e-05,
      "loss": 1.2302,
      "step": 5750
    },
    {
      "epoch": 3.5256312165263965,
      "grad_norm": 2.081388235092163,
      "learning_rate": 3.5959719670681093e-05,
      "loss": 1.2464,
      "step": 5760
    },
    {
      "epoch": 3.531752104055088,
      "grad_norm": 2.0220718383789062,
      "learning_rate": 3.5925699122269854e-05,
      "loss": 1.2502,
      "step": 5770
    },
    {
      "epoch": 3.5378729915837797,
      "grad_norm": 1.9582314491271973,
      "learning_rate": 3.589167857385861e-05,
      "loss": 1.2281,
      "step": 5780
    },
    {
      "epoch": 3.543993879112471,
      "grad_norm": 1.8635259866714478,
      "learning_rate": 3.585765802544737e-05,
      "loss": 1.2109,
      "step": 5790
    },
    {
      "epoch": 3.5501147666411628,
      "grad_norm": 1.8320813179016113,
      "learning_rate": 3.582363747703613e-05,
      "loss": 1.2441,
      "step": 5800
    },
    {
      "epoch": 3.5562356541698548,
      "grad_norm": 2.1509382724761963,
      "learning_rate": 3.578961692862489e-05,
      "loss": 1.2561,
      "step": 5810
    },
    {
      "epoch": 3.5623565416985463,
      "grad_norm": 1.8632539510726929,
      "learning_rate": 3.575559638021365e-05,
      "loss": 1.2481,
      "step": 5820
    },
    {
      "epoch": 3.568477429227238,
      "grad_norm": 1.9509693384170532,
      "learning_rate": 3.572157583180241e-05,
      "loss": 1.2618,
      "step": 5830
    },
    {
      "epoch": 3.57459831675593,
      "grad_norm": 2.189829111099243,
      "learning_rate": 3.568755528339117e-05,
      "loss": 1.2455,
      "step": 5840
    },
    {
      "epoch": 3.5807192042846214,
      "grad_norm": 2.2501230239868164,
      "learning_rate": 3.5653534734979934e-05,
      "loss": 1.2442,
      "step": 5850
    },
    {
      "epoch": 3.586840091813313,
      "grad_norm": 2.8913943767547607,
      "learning_rate": 3.561951418656869e-05,
      "loss": 1.2615,
      "step": 5860
    },
    {
      "epoch": 3.5929609793420045,
      "grad_norm": 1.5915802717208862,
      "learning_rate": 3.558549363815745e-05,
      "loss": 1.2502,
      "step": 5870
    },
    {
      "epoch": 3.599081866870696,
      "grad_norm": 2.0232715606689453,
      "learning_rate": 3.555147308974621e-05,
      "loss": 1.2592,
      "step": 5880
    },
    {
      "epoch": 3.6052027543993876,
      "grad_norm": 2.3988993167877197,
      "learning_rate": 3.5517452541334964e-05,
      "loss": 1.2568,
      "step": 5890
    },
    {
      "epoch": 3.6113236419280796,
      "grad_norm": 1.941627025604248,
      "learning_rate": 3.5483431992923725e-05,
      "loss": 1.2435,
      "step": 5900
    },
    {
      "epoch": 3.617444529456771,
      "grad_norm": 1.9982330799102783,
      "learning_rate": 3.5449411444512486e-05,
      "loss": 1.2441,
      "step": 5910
    },
    {
      "epoch": 3.6235654169854628,
      "grad_norm": 1.9015318155288696,
      "learning_rate": 3.5415390896101246e-05,
      "loss": 1.247,
      "step": 5920
    },
    {
      "epoch": 3.6296863045141547,
      "grad_norm": 1.683561086654663,
      "learning_rate": 3.538137034769001e-05,
      "loss": 1.2401,
      "step": 5930
    },
    {
      "epoch": 3.6358071920428463,
      "grad_norm": 2.5195016860961914,
      "learning_rate": 3.534734979927877e-05,
      "loss": 1.2342,
      "step": 5940
    },
    {
      "epoch": 3.641928079571538,
      "grad_norm": 2.2849786281585693,
      "learning_rate": 3.531332925086753e-05,
      "loss": 1.232,
      "step": 5950
    },
    {
      "epoch": 3.6480489671002294,
      "grad_norm": 2.70056414604187,
      "learning_rate": 3.527930870245629e-05,
      "loss": 1.2573,
      "step": 5960
    },
    {
      "epoch": 3.654169854628921,
      "grad_norm": 5.275028228759766,
      "learning_rate": 3.524528815404505e-05,
      "loss": 1.2237,
      "step": 5970
    },
    {
      "epoch": 3.660290742157613,
      "grad_norm": 1.7736239433288574,
      "learning_rate": 3.5211267605633805e-05,
      "loss": 1.2314,
      "step": 5980
    },
    {
      "epoch": 3.6664116296863045,
      "grad_norm": 1.8925633430480957,
      "learning_rate": 3.517724705722256e-05,
      "loss": 1.2476,
      "step": 5990
    },
    {
      "epoch": 3.672532517214996,
      "grad_norm": 1.8658711910247803,
      "learning_rate": 3.514322650881132e-05,
      "loss": 1.2448,
      "step": 6000
    },
    {
      "epoch": 3.678653404743688,
      "grad_norm": 2.2303245067596436,
      "learning_rate": 3.510920596040008e-05,
      "loss": 1.2566,
      "step": 6010
    },
    {
      "epoch": 3.6847742922723796,
      "grad_norm": 1.967257022857666,
      "learning_rate": 3.507518541198884e-05,
      "loss": 1.2791,
      "step": 6020
    },
    {
      "epoch": 3.690895179801071,
      "grad_norm": 2.001021146774292,
      "learning_rate": 3.50411648635776e-05,
      "loss": 1.2275,
      "step": 6030
    },
    {
      "epoch": 3.6970160673297627,
      "grad_norm": 2.0317656993865967,
      "learning_rate": 3.500714431516636e-05,
      "loss": 1.2223,
      "step": 6040
    },
    {
      "epoch": 3.7031369548584543,
      "grad_norm": 2.7534890174865723,
      "learning_rate": 3.4973123766755124e-05,
      "loss": 1.2521,
      "step": 6050
    },
    {
      "epoch": 3.7092578423871463,
      "grad_norm": 1.9250917434692383,
      "learning_rate": 3.4939103218343885e-05,
      "loss": 1.2426,
      "step": 6060
    },
    {
      "epoch": 3.715378729915838,
      "grad_norm": 2.119258165359497,
      "learning_rate": 3.4905082669932645e-05,
      "loss": 1.2613,
      "step": 6070
    },
    {
      "epoch": 3.7214996174445294,
      "grad_norm": 2.6487741470336914,
      "learning_rate": 3.48710621215214e-05,
      "loss": 1.2531,
      "step": 6080
    },
    {
      "epoch": 3.7276205049732214,
      "grad_norm": 2.616588830947876,
      "learning_rate": 3.483704157311016e-05,
      "loss": 1.2603,
      "step": 6090
    },
    {
      "epoch": 3.733741392501913,
      "grad_norm": 1.95454740524292,
      "learning_rate": 3.4803021024698914e-05,
      "loss": 1.2393,
      "step": 6100
    },
    {
      "epoch": 3.7398622800306045,
      "grad_norm": 2.1213455200195312,
      "learning_rate": 3.4769000476287675e-05,
      "loss": 1.2269,
      "step": 6110
    },
    {
      "epoch": 3.745983167559296,
      "grad_norm": 2.3341705799102783,
      "learning_rate": 3.4734979927876436e-05,
      "loss": 1.236,
      "step": 6120
    },
    {
      "epoch": 3.7521040550879876,
      "grad_norm": 2.3794469833374023,
      "learning_rate": 3.47009593794652e-05,
      "loss": 1.2483,
      "step": 6130
    },
    {
      "epoch": 3.758224942616679,
      "grad_norm": 1.9347407817840576,
      "learning_rate": 3.466693883105396e-05,
      "loss": 1.255,
      "step": 6140
    },
    {
      "epoch": 3.764345830145371,
      "grad_norm": 3.7383499145507812,
      "learning_rate": 3.463291828264272e-05,
      "loss": 1.2169,
      "step": 6150
    },
    {
      "epoch": 3.7704667176740627,
      "grad_norm": 1.8134688138961792,
      "learning_rate": 3.459889773423148e-05,
      "loss": 1.2588,
      "step": 6160
    },
    {
      "epoch": 3.7765876052027543,
      "grad_norm": 1.9380251169204712,
      "learning_rate": 3.456487718582024e-05,
      "loss": 1.248,
      "step": 6170
    },
    {
      "epoch": 3.7827084927314463,
      "grad_norm": 2.238302230834961,
      "learning_rate": 3.4530856637409e-05,
      "loss": 1.2323,
      "step": 6180
    },
    {
      "epoch": 3.788829380260138,
      "grad_norm": 1.9604918956756592,
      "learning_rate": 3.4496836088997755e-05,
      "loss": 1.2261,
      "step": 6190
    },
    {
      "epoch": 3.7949502677888294,
      "grad_norm": 1.7650002241134644,
      "learning_rate": 3.4462815540586516e-05,
      "loss": 1.2346,
      "step": 6200
    },
    {
      "epoch": 3.801071155317521,
      "grad_norm": 2.498751401901245,
      "learning_rate": 3.442879499217527e-05,
      "loss": 1.2297,
      "step": 6210
    },
    {
      "epoch": 3.8071920428462125,
      "grad_norm": 3.40084171295166,
      "learning_rate": 3.439477444376403e-05,
      "loss": 1.2549,
      "step": 6220
    },
    {
      "epoch": 3.8133129303749045,
      "grad_norm": 1.700204610824585,
      "learning_rate": 3.436075389535279e-05,
      "loss": 1.2558,
      "step": 6230
    },
    {
      "epoch": 3.819433817903596,
      "grad_norm": 2.0135040283203125,
      "learning_rate": 3.432673334694155e-05,
      "loss": 1.2576,
      "step": 6240
    },
    {
      "epoch": 3.8255547054322876,
      "grad_norm": 2.295555353164673,
      "learning_rate": 3.429271279853031e-05,
      "loss": 1.2233,
      "step": 6250
    },
    {
      "epoch": 3.8316755929609796,
      "grad_norm": 1.7916216850280762,
      "learning_rate": 3.4258692250119074e-05,
      "loss": 1.2267,
      "step": 6260
    },
    {
      "epoch": 3.837796480489671,
      "grad_norm": 1.6943750381469727,
      "learning_rate": 3.4224671701707835e-05,
      "loss": 1.2482,
      "step": 6270
    },
    {
      "epoch": 3.8439173680183627,
      "grad_norm": 1.9428541660308838,
      "learning_rate": 3.4190651153296596e-05,
      "loss": 1.2324,
      "step": 6280
    },
    {
      "epoch": 3.8500382555470543,
      "grad_norm": 2.07246994972229,
      "learning_rate": 3.415663060488536e-05,
      "loss": 1.2335,
      "step": 6290
    },
    {
      "epoch": 3.856159143075746,
      "grad_norm": 1.881109356880188,
      "learning_rate": 3.412261005647411e-05,
      "loss": 1.2594,
      "step": 6300
    },
    {
      "epoch": 3.862280030604438,
      "grad_norm": 2.3518142700195312,
      "learning_rate": 3.408858950806287e-05,
      "loss": 1.2397,
      "step": 6310
    },
    {
      "epoch": 3.8684009181331294,
      "grad_norm": 2.0112175941467285,
      "learning_rate": 3.405456895965163e-05,
      "loss": 1.2563,
      "step": 6320
    },
    {
      "epoch": 3.874521805661821,
      "grad_norm": 1.5001269578933716,
      "learning_rate": 3.4020548411240386e-05,
      "loss": 1.2371,
      "step": 6330
    },
    {
      "epoch": 3.8806426931905125,
      "grad_norm": 1.9393521547317505,
      "learning_rate": 3.398652786282915e-05,
      "loss": 1.2375,
      "step": 6340
    },
    {
      "epoch": 3.8867635807192045,
      "grad_norm": 2.159621238708496,
      "learning_rate": 3.395250731441791e-05,
      "loss": 1.2535,
      "step": 6350
    },
    {
      "epoch": 3.892884468247896,
      "grad_norm": 2.0031697750091553,
      "learning_rate": 3.391848676600667e-05,
      "loss": 1.2502,
      "step": 6360
    },
    {
      "epoch": 3.8990053557765876,
      "grad_norm": 1.7993634939193726,
      "learning_rate": 3.388446621759543e-05,
      "loss": 1.2323,
      "step": 6370
    },
    {
      "epoch": 3.905126243305279,
      "grad_norm": 1.918763279914856,
      "learning_rate": 3.385044566918419e-05,
      "loss": 1.2363,
      "step": 6380
    },
    {
      "epoch": 3.9112471308339707,
      "grad_norm": 1.641964316368103,
      "learning_rate": 3.381642512077295e-05,
      "loss": 1.2507,
      "step": 6390
    },
    {
      "epoch": 3.9173680183626627,
      "grad_norm": 2.025327444076538,
      "learning_rate": 3.3782404572361705e-05,
      "loss": 1.2392,
      "step": 6400
    },
    {
      "epoch": 3.9234889058913542,
      "grad_norm": 2.53024959564209,
      "learning_rate": 3.3748384023950466e-05,
      "loss": 1.2323,
      "step": 6410
    },
    {
      "epoch": 3.929609793420046,
      "grad_norm": 2.083070755004883,
      "learning_rate": 3.371436347553923e-05,
      "loss": 1.2503,
      "step": 6420
    },
    {
      "epoch": 3.935730680948738,
      "grad_norm": 2.3890929222106934,
      "learning_rate": 3.368034292712799e-05,
      "loss": 1.2482,
      "step": 6430
    },
    {
      "epoch": 3.9418515684774293,
      "grad_norm": 2.910482168197632,
      "learning_rate": 3.364632237871675e-05,
      "loss": 1.2236,
      "step": 6440
    },
    {
      "epoch": 3.947972456006121,
      "grad_norm": 1.9107235670089722,
      "learning_rate": 3.36123018303055e-05,
      "loss": 1.2466,
      "step": 6450
    },
    {
      "epoch": 3.9540933435348125,
      "grad_norm": 1.7786065340042114,
      "learning_rate": 3.3578281281894264e-05,
      "loss": 1.247,
      "step": 6460
    },
    {
      "epoch": 3.960214231063504,
      "grad_norm": 1.7764980792999268,
      "learning_rate": 3.3544260733483025e-05,
      "loss": 1.2401,
      "step": 6470
    },
    {
      "epoch": 3.966335118592196,
      "grad_norm": 2.0788002014160156,
      "learning_rate": 3.3510240185071785e-05,
      "loss": 1.2405,
      "step": 6480
    },
    {
      "epoch": 3.9724560061208876,
      "grad_norm": 2.2174031734466553,
      "learning_rate": 3.3476219636660546e-05,
      "loss": 1.2499,
      "step": 6490
    },
    {
      "epoch": 3.978576893649579,
      "grad_norm": 1.71248459815979,
      "learning_rate": 3.344219908824931e-05,
      "loss": 1.2697,
      "step": 6500
    },
    {
      "epoch": 3.984697781178271,
      "grad_norm": 2.434459686279297,
      "learning_rate": 3.340817853983806e-05,
      "loss": 1.2351,
      "step": 6510
    },
    {
      "epoch": 3.9908186687069627,
      "grad_norm": 3.131270408630371,
      "learning_rate": 3.337415799142682e-05,
      "loss": 1.2279,
      "step": 6520
    },
    {
      "epoch": 3.9969395562356542,
      "grad_norm": 1.993688702583313,
      "learning_rate": 3.334013744301558e-05,
      "loss": 1.2368,
      "step": 6530
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.49007154608409537,
      "eval_loss": 1.1966572999954224,
      "eval_runtime": 2703.3706,
      "eval_samples_per_second": 77.346,
      "eval_steps_per_second": 0.604,
      "step": 6535
    },
    {
      "epoch": 4.003060443764346,
      "grad_norm": 3.0211377143859863,
      "learning_rate": 3.3306116894604344e-05,
      "loss": 1.2443,
      "step": 6540
    },
    {
      "epoch": 4.009181331293037,
      "grad_norm": 2.0056445598602295,
      "learning_rate": 3.3272096346193104e-05,
      "loss": 1.2186,
      "step": 6550
    },
    {
      "epoch": 4.015302218821729,
      "grad_norm": 2.565847396850586,
      "learning_rate": 3.323807579778186e-05,
      "loss": 1.2187,
      "step": 6560
    },
    {
      "epoch": 4.02142310635042,
      "grad_norm": 2.171513319015503,
      "learning_rate": 3.320405524937062e-05,
      "loss": 1.2271,
      "step": 6570
    },
    {
      "epoch": 4.027543993879113,
      "grad_norm": 2.016981840133667,
      "learning_rate": 3.317003470095938e-05,
      "loss": 1.2409,
      "step": 6580
    },
    {
      "epoch": 4.033664881407804,
      "grad_norm": 2.211911916732788,
      "learning_rate": 3.313601415254814e-05,
      "loss": 1.2206,
      "step": 6590
    },
    {
      "epoch": 4.039785768936496,
      "grad_norm": 2.5300450325012207,
      "learning_rate": 3.31019936041369e-05,
      "loss": 1.2272,
      "step": 6600
    },
    {
      "epoch": 4.0459066564651875,
      "grad_norm": 1.8287904262542725,
      "learning_rate": 3.306797305572566e-05,
      "loss": 1.2423,
      "step": 6610
    },
    {
      "epoch": 4.052027543993879,
      "grad_norm": 2.4838716983795166,
      "learning_rate": 3.303395250731442e-05,
      "loss": 1.2215,
      "step": 6620
    },
    {
      "epoch": 4.058148431522571,
      "grad_norm": 2.263672113418579,
      "learning_rate": 3.299993195890318e-05,
      "loss": 1.2403,
      "step": 6630
    },
    {
      "epoch": 4.064269319051262,
      "grad_norm": 1.7068171501159668,
      "learning_rate": 3.296591141049194e-05,
      "loss": 1.2132,
      "step": 6640
    },
    {
      "epoch": 4.070390206579954,
      "grad_norm": 2.154141664505005,
      "learning_rate": 3.29318908620807e-05,
      "loss": 1.209,
      "step": 6650
    },
    {
      "epoch": 4.076511094108645,
      "grad_norm": 2.417459011077881,
      "learning_rate": 3.289787031366946e-05,
      "loss": 1.2118,
      "step": 6660
    },
    {
      "epoch": 4.082631981637338,
      "grad_norm": 3.0920281410217285,
      "learning_rate": 3.286384976525822e-05,
      "loss": 1.2296,
      "step": 6670
    },
    {
      "epoch": 4.088752869166029,
      "grad_norm": 2.09197998046875,
      "learning_rate": 3.2829829216846975e-05,
      "loss": 1.2097,
      "step": 6680
    },
    {
      "epoch": 4.094873756694721,
      "grad_norm": 2.23848819732666,
      "learning_rate": 3.2795808668435736e-05,
      "loss": 1.2354,
      "step": 6690
    },
    {
      "epoch": 4.100994644223412,
      "grad_norm": 2.4771714210510254,
      "learning_rate": 3.2761788120024497e-05,
      "loss": 1.2477,
      "step": 6700
    },
    {
      "epoch": 4.107115531752104,
      "grad_norm": 2.599956750869751,
      "learning_rate": 3.272776757161326e-05,
      "loss": 1.242,
      "step": 6710
    },
    {
      "epoch": 4.1132364192807955,
      "grad_norm": 2.1151208877563477,
      "learning_rate": 3.269374702320202e-05,
      "loss": 1.2182,
      "step": 6720
    },
    {
      "epoch": 4.119357306809487,
      "grad_norm": 2.415768623352051,
      "learning_rate": 3.265972647479077e-05,
      "loss": 1.214,
      "step": 6730
    },
    {
      "epoch": 4.125478194338179,
      "grad_norm": 2.391017436981201,
      "learning_rate": 3.262570592637953e-05,
      "loss": 1.2449,
      "step": 6740
    },
    {
      "epoch": 4.131599081866871,
      "grad_norm": 2.170161724090576,
      "learning_rate": 3.2591685377968294e-05,
      "loss": 1.2286,
      "step": 6750
    },
    {
      "epoch": 4.137719969395563,
      "grad_norm": 2.0515055656433105,
      "learning_rate": 3.2557664829557055e-05,
      "loss": 1.2426,
      "step": 6760
    },
    {
      "epoch": 4.143840856924254,
      "grad_norm": 2.469785213470459,
      "learning_rate": 3.2523644281145816e-05,
      "loss": 1.2284,
      "step": 6770
    },
    {
      "epoch": 4.149961744452946,
      "grad_norm": 2.2527389526367188,
      "learning_rate": 3.2489623732734576e-05,
      "loss": 1.2186,
      "step": 6780
    },
    {
      "epoch": 4.156082631981637,
      "grad_norm": 1.9241255521774292,
      "learning_rate": 3.245560318432334e-05,
      "loss": 1.2133,
      "step": 6790
    },
    {
      "epoch": 4.162203519510329,
      "grad_norm": 2.233799457550049,
      "learning_rate": 3.242158263591209e-05,
      "loss": 1.2279,
      "step": 6800
    },
    {
      "epoch": 4.16832440703902,
      "grad_norm": 2.2666685581207275,
      "learning_rate": 3.238756208750085e-05,
      "loss": 1.2359,
      "step": 6810
    },
    {
      "epoch": 4.174445294567712,
      "grad_norm": 2.9180006980895996,
      "learning_rate": 3.235354153908961e-05,
      "loss": 1.2105,
      "step": 6820
    },
    {
      "epoch": 4.180566182096404,
      "grad_norm": 3.030034303665161,
      "learning_rate": 3.231952099067837e-05,
      "loss": 1.2184,
      "step": 6830
    },
    {
      "epoch": 4.186687069625096,
      "grad_norm": 1.8429760932922363,
      "learning_rate": 3.228550044226713e-05,
      "loss": 1.2197,
      "step": 6840
    },
    {
      "epoch": 4.1928079571537875,
      "grad_norm": 2.205509662628174,
      "learning_rate": 3.225147989385589e-05,
      "loss": 1.2274,
      "step": 6850
    },
    {
      "epoch": 4.198928844682479,
      "grad_norm": 2.7778589725494385,
      "learning_rate": 3.221745934544465e-05,
      "loss": 1.2541,
      "step": 6860
    },
    {
      "epoch": 4.205049732211171,
      "grad_norm": 2.7423245906829834,
      "learning_rate": 3.218343879703341e-05,
      "loss": 1.2135,
      "step": 6870
    },
    {
      "epoch": 4.211170619739862,
      "grad_norm": 2.1378300189971924,
      "learning_rate": 3.214941824862217e-05,
      "loss": 1.2166,
      "step": 6880
    },
    {
      "epoch": 4.217291507268554,
      "grad_norm": 2.2552590370178223,
      "learning_rate": 3.211539770021093e-05,
      "loss": 1.2267,
      "step": 6890
    },
    {
      "epoch": 4.223412394797245,
      "grad_norm": 2.3378567695617676,
      "learning_rate": 3.208137715179969e-05,
      "loss": 1.2274,
      "step": 6900
    },
    {
      "epoch": 4.229533282325937,
      "grad_norm": 2.442488193511963,
      "learning_rate": 3.2047356603388454e-05,
      "loss": 1.2178,
      "step": 6910
    },
    {
      "epoch": 4.235654169854629,
      "grad_norm": 2.7005996704101562,
      "learning_rate": 3.201333605497721e-05,
      "loss": 1.2279,
      "step": 6920
    },
    {
      "epoch": 4.241775057383321,
      "grad_norm": 1.9639949798583984,
      "learning_rate": 3.197931550656597e-05,
      "loss": 1.2372,
      "step": 6930
    },
    {
      "epoch": 4.247895944912012,
      "grad_norm": 2.424386739730835,
      "learning_rate": 3.194529495815472e-05,
      "loss": 1.2217,
      "step": 6940
    },
    {
      "epoch": 4.254016832440704,
      "grad_norm": 2.461184501647949,
      "learning_rate": 3.1911274409743484e-05,
      "loss": 1.2157,
      "step": 6950
    },
    {
      "epoch": 4.2601377199693955,
      "grad_norm": 2.4525201320648193,
      "learning_rate": 3.1877253861332244e-05,
      "loss": 1.2272,
      "step": 6960
    },
    {
      "epoch": 4.266258607498087,
      "grad_norm": 2.1750197410583496,
      "learning_rate": 3.1843233312921005e-05,
      "loss": 1.2195,
      "step": 6970
    },
    {
      "epoch": 4.272379495026779,
      "grad_norm": 2.52929949760437,
      "learning_rate": 3.1809212764509766e-05,
      "loss": 1.2455,
      "step": 6980
    },
    {
      "epoch": 4.27850038255547,
      "grad_norm": 1.8435324430465698,
      "learning_rate": 3.177519221609853e-05,
      "loss": 1.1996,
      "step": 6990
    },
    {
      "epoch": 4.284621270084163,
      "grad_norm": 2.5383806228637695,
      "learning_rate": 3.174117166768729e-05,
      "loss": 1.2393,
      "step": 7000
    },
    {
      "epoch": 4.290742157612854,
      "grad_norm": 2.3538296222686768,
      "learning_rate": 3.170715111927605e-05,
      "loss": 1.2237,
      "step": 7010
    },
    {
      "epoch": 4.296863045141546,
      "grad_norm": 1.8341562747955322,
      "learning_rate": 3.167313057086481e-05,
      "loss": 1.2217,
      "step": 7020
    },
    {
      "epoch": 4.302983932670237,
      "grad_norm": 1.9721349477767944,
      "learning_rate": 3.1639110022453563e-05,
      "loss": 1.2232,
      "step": 7030
    },
    {
      "epoch": 4.309104820198929,
      "grad_norm": 2.715494155883789,
      "learning_rate": 3.1605089474042324e-05,
      "loss": 1.2342,
      "step": 7040
    },
    {
      "epoch": 4.31522570772762,
      "grad_norm": 2.1433260440826416,
      "learning_rate": 3.157106892563108e-05,
      "loss": 1.2411,
      "step": 7050
    },
    {
      "epoch": 4.321346595256312,
      "grad_norm": 2.1917879581451416,
      "learning_rate": 3.153704837721984e-05,
      "loss": 1.2175,
      "step": 7060
    },
    {
      "epoch": 4.3274674827850035,
      "grad_norm": 1.9488756656646729,
      "learning_rate": 3.15030278288086e-05,
      "loss": 1.2296,
      "step": 7070
    },
    {
      "epoch": 4.333588370313695,
      "grad_norm": 2.587290048599243,
      "learning_rate": 3.146900728039736e-05,
      "loss": 1.2451,
      "step": 7080
    },
    {
      "epoch": 4.3397092578423875,
      "grad_norm": 2.3056092262268066,
      "learning_rate": 3.143498673198612e-05,
      "loss": 1.2075,
      "step": 7090
    },
    {
      "epoch": 4.345830145371079,
      "grad_norm": 2.641049385070801,
      "learning_rate": 3.140096618357488e-05,
      "loss": 1.2404,
      "step": 7100
    },
    {
      "epoch": 4.351951032899771,
      "grad_norm": 2.5124692916870117,
      "learning_rate": 3.136694563516364e-05,
      "loss": 1.2183,
      "step": 7110
    },
    {
      "epoch": 4.358071920428462,
      "grad_norm": 2.4015398025512695,
      "learning_rate": 3.1332925086752404e-05,
      "loss": 1.2212,
      "step": 7120
    },
    {
      "epoch": 4.364192807957154,
      "grad_norm": 2.2219901084899902,
      "learning_rate": 3.1298904538341165e-05,
      "loss": 1.2227,
      "step": 7130
    },
    {
      "epoch": 4.370313695485845,
      "grad_norm": 1.9921913146972656,
      "learning_rate": 3.126488398992992e-05,
      "loss": 1.2294,
      "step": 7140
    },
    {
      "epoch": 4.376434583014537,
      "grad_norm": 2.351640224456787,
      "learning_rate": 3.123086344151867e-05,
      "loss": 1.2379,
      "step": 7150
    },
    {
      "epoch": 4.382555470543228,
      "grad_norm": 2.6132214069366455,
      "learning_rate": 3.1196842893107434e-05,
      "loss": 1.2235,
      "step": 7160
    },
    {
      "epoch": 4.388676358071921,
      "grad_norm": 2.056541681289673,
      "learning_rate": 3.1162822344696195e-05,
      "loss": 1.2133,
      "step": 7170
    },
    {
      "epoch": 4.394797245600612,
      "grad_norm": 2.450434684753418,
      "learning_rate": 3.1128801796284956e-05,
      "loss": 1.2085,
      "step": 7180
    },
    {
      "epoch": 4.400918133129304,
      "grad_norm": 1.9336761236190796,
      "learning_rate": 3.1094781247873716e-05,
      "loss": 1.2293,
      "step": 7190
    },
    {
      "epoch": 4.4070390206579955,
      "grad_norm": 2.7180116176605225,
      "learning_rate": 3.106076069946248e-05,
      "loss": 1.2234,
      "step": 7200
    },
    {
      "epoch": 4.413159908186687,
      "grad_norm": 2.412187337875366,
      "learning_rate": 3.102674015105124e-05,
      "loss": 1.2163,
      "step": 7210
    },
    {
      "epoch": 4.419280795715379,
      "grad_norm": 2.206754684448242,
      "learning_rate": 3.099271960264e-05,
      "loss": 1.2197,
      "step": 7220
    },
    {
      "epoch": 4.42540168324407,
      "grad_norm": 1.9868717193603516,
      "learning_rate": 3.095869905422876e-05,
      "loss": 1.2371,
      "step": 7230
    },
    {
      "epoch": 4.431522570772762,
      "grad_norm": 2.1179041862487793,
      "learning_rate": 3.0924678505817514e-05,
      "loss": 1.207,
      "step": 7240
    },
    {
      "epoch": 4.437643458301454,
      "grad_norm": 2.341188907623291,
      "learning_rate": 3.0890657957406275e-05,
      "loss": 1.2212,
      "step": 7250
    },
    {
      "epoch": 4.443764345830146,
      "grad_norm": 2.979339361190796,
      "learning_rate": 3.0856637408995036e-05,
      "loss": 1.2373,
      "step": 7260
    },
    {
      "epoch": 4.449885233358837,
      "grad_norm": 2.224384307861328,
      "learning_rate": 3.082261686058379e-05,
      "loss": 1.2274,
      "step": 7270
    },
    {
      "epoch": 4.456006120887529,
      "grad_norm": 1.9843226671218872,
      "learning_rate": 3.078859631217255e-05,
      "loss": 1.2236,
      "step": 7280
    },
    {
      "epoch": 4.46212700841622,
      "grad_norm": 2.021461248397827,
      "learning_rate": 3.075457576376131e-05,
      "loss": 1.2296,
      "step": 7290
    },
    {
      "epoch": 4.468247895944912,
      "grad_norm": 2.517347812652588,
      "learning_rate": 3.072055521535007e-05,
      "loss": 1.1913,
      "step": 7300
    },
    {
      "epoch": 4.4743687834736035,
      "grad_norm": 2.3766560554504395,
      "learning_rate": 3.068653466693883e-05,
      "loss": 1.2233,
      "step": 7310
    },
    {
      "epoch": 4.480489671002295,
      "grad_norm": 2.126357316970825,
      "learning_rate": 3.0652514118527594e-05,
      "loss": 1.2336,
      "step": 7320
    },
    {
      "epoch": 4.4866105585309874,
      "grad_norm": 2.1979329586029053,
      "learning_rate": 3.0618493570116355e-05,
      "loss": 1.2091,
      "step": 7330
    },
    {
      "epoch": 4.492731446059679,
      "grad_norm": 2.0365238189697266,
      "learning_rate": 3.0584473021705115e-05,
      "loss": 1.2131,
      "step": 7340
    },
    {
      "epoch": 4.4988523335883706,
      "grad_norm": 2.6603610515594482,
      "learning_rate": 3.055045247329387e-05,
      "loss": 1.1893,
      "step": 7350
    },
    {
      "epoch": 4.504973221117062,
      "grad_norm": 2.6453335285186768,
      "learning_rate": 3.0516431924882634e-05,
      "loss": 1.2219,
      "step": 7360
    },
    {
      "epoch": 4.511094108645754,
      "grad_norm": 2.1196227073669434,
      "learning_rate": 3.048241137647139e-05,
      "loss": 1.2368,
      "step": 7370
    },
    {
      "epoch": 4.517214996174445,
      "grad_norm": 1.869175672531128,
      "learning_rate": 3.044839082806015e-05,
      "loss": 1.2139,
      "step": 7380
    },
    {
      "epoch": 4.523335883703137,
      "grad_norm": 2.3623552322387695,
      "learning_rate": 3.0414370279648906e-05,
      "loss": 1.2032,
      "step": 7390
    },
    {
      "epoch": 4.529456771231828,
      "grad_norm": 2.8302743434906006,
      "learning_rate": 3.0380349731237667e-05,
      "loss": 1.2228,
      "step": 7400
    },
    {
      "epoch": 4.53557765876052,
      "grad_norm": 2.22745418548584,
      "learning_rate": 3.0346329182826428e-05,
      "loss": 1.21,
      "step": 7410
    },
    {
      "epoch": 4.541698546289212,
      "grad_norm": 2.6589863300323486,
      "learning_rate": 3.031230863441519e-05,
      "loss": 1.2401,
      "step": 7420
    },
    {
      "epoch": 4.547819433817904,
      "grad_norm": 2.0868048667907715,
      "learning_rate": 3.0278288086003946e-05,
      "loss": 1.2353,
      "step": 7430
    },
    {
      "epoch": 4.553940321346595,
      "grad_norm": 2.1262969970703125,
      "learning_rate": 3.0244267537592707e-05,
      "loss": 1.2056,
      "step": 7440
    },
    {
      "epoch": 4.560061208875287,
      "grad_norm": 2.061774730682373,
      "learning_rate": 3.0210246989181468e-05,
      "loss": 1.2139,
      "step": 7450
    },
    {
      "epoch": 4.5661820964039785,
      "grad_norm": 2.2398433685302734,
      "learning_rate": 3.017622644077023e-05,
      "loss": 1.2223,
      "step": 7460
    },
    {
      "epoch": 4.57230298393267,
      "grad_norm": 2.264333963394165,
      "learning_rate": 3.014220589235899e-05,
      "loss": 1.2289,
      "step": 7470
    },
    {
      "epoch": 4.578423871461362,
      "grad_norm": 2.1285994052886963,
      "learning_rate": 3.0108185343947747e-05,
      "loss": 1.22,
      "step": 7480
    },
    {
      "epoch": 4.584544758990053,
      "grad_norm": 2.1647374629974365,
      "learning_rate": 3.0074164795536508e-05,
      "loss": 1.2366,
      "step": 7490
    },
    {
      "epoch": 4.590665646518746,
      "grad_norm": 2.071086883544922,
      "learning_rate": 3.004014424712526e-05,
      "loss": 1.1946,
      "step": 7500
    },
    {
      "epoch": 4.596786534047437,
      "grad_norm": 2.2825191020965576,
      "learning_rate": 3.0006123698714022e-05,
      "loss": 1.2258,
      "step": 7510
    },
    {
      "epoch": 4.602907421576129,
      "grad_norm": 2.1768798828125,
      "learning_rate": 2.9972103150302783e-05,
      "loss": 1.2357,
      "step": 7520
    },
    {
      "epoch": 4.60902830910482,
      "grad_norm": 1.8341929912567139,
      "learning_rate": 2.9938082601891544e-05,
      "loss": 1.2337,
      "step": 7530
    },
    {
      "epoch": 4.615149196633512,
      "grad_norm": 2.1512627601623535,
      "learning_rate": 2.99040620534803e-05,
      "loss": 1.2289,
      "step": 7540
    },
    {
      "epoch": 4.621270084162203,
      "grad_norm": 3.288439989089966,
      "learning_rate": 2.9870041505069062e-05,
      "loss": 1.2379,
      "step": 7550
    },
    {
      "epoch": 4.627390971690895,
      "grad_norm": 2.537337064743042,
      "learning_rate": 2.9836020956657823e-05,
      "loss": 1.2334,
      "step": 7560
    },
    {
      "epoch": 4.6335118592195865,
      "grad_norm": 2.0168588161468506,
      "learning_rate": 2.9802000408246584e-05,
      "loss": 1.2124,
      "step": 7570
    },
    {
      "epoch": 4.639632746748278,
      "grad_norm": 2.4690663814544678,
      "learning_rate": 2.9767979859835345e-05,
      "loss": 1.2069,
      "step": 7580
    },
    {
      "epoch": 4.6457536342769705,
      "grad_norm": 2.4409384727478027,
      "learning_rate": 2.9733959311424102e-05,
      "loss": 1.2304,
      "step": 7590
    },
    {
      "epoch": 4.651874521805662,
      "grad_norm": 1.9961847066879272,
      "learning_rate": 2.9699938763012863e-05,
      "loss": 1.2159,
      "step": 7600
    },
    {
      "epoch": 4.657995409334354,
      "grad_norm": 2.5575664043426514,
      "learning_rate": 2.9665918214601624e-05,
      "loss": 1.2252,
      "step": 7610
    },
    {
      "epoch": 4.664116296863045,
      "grad_norm": 2.107255220413208,
      "learning_rate": 2.9631897666190378e-05,
      "loss": 1.1979,
      "step": 7620
    },
    {
      "epoch": 4.670237184391737,
      "grad_norm": 2.145972490310669,
      "learning_rate": 2.959787711777914e-05,
      "loss": 1.2467,
      "step": 7630
    },
    {
      "epoch": 4.676358071920428,
      "grad_norm": 1.9515827894210815,
      "learning_rate": 2.95638565693679e-05,
      "loss": 1.2142,
      "step": 7640
    },
    {
      "epoch": 4.68247895944912,
      "grad_norm": 2.8625612258911133,
      "learning_rate": 2.9529836020956657e-05,
      "loss": 1.238,
      "step": 7650
    },
    {
      "epoch": 4.688599846977811,
      "grad_norm": 2.255072832107544,
      "learning_rate": 2.9495815472545418e-05,
      "loss": 1.232,
      "step": 7660
    },
    {
      "epoch": 4.694720734506504,
      "grad_norm": 2.5473906993865967,
      "learning_rate": 2.946179492413418e-05,
      "loss": 1.2325,
      "step": 7670
    },
    {
      "epoch": 4.700841622035195,
      "grad_norm": 2.642747640609741,
      "learning_rate": 2.942777437572294e-05,
      "loss": 1.2431,
      "step": 7680
    },
    {
      "epoch": 4.706962509563887,
      "grad_norm": 2.264507532119751,
      "learning_rate": 2.9393753827311697e-05,
      "loss": 1.221,
      "step": 7690
    },
    {
      "epoch": 4.7130833970925785,
      "grad_norm": 1.8713449239730835,
      "learning_rate": 2.9359733278900458e-05,
      "loss": 1.217,
      "step": 7700
    },
    {
      "epoch": 4.71920428462127,
      "grad_norm": 3.092440128326416,
      "learning_rate": 2.932571273048922e-05,
      "loss": 1.2544,
      "step": 7710
    },
    {
      "epoch": 4.725325172149962,
      "grad_norm": 2.9618237018585205,
      "learning_rate": 2.929169218207798e-05,
      "loss": 1.2355,
      "step": 7720
    },
    {
      "epoch": 4.731446059678653,
      "grad_norm": 1.8672477006912231,
      "learning_rate": 2.9257671633666734e-05,
      "loss": 1.1962,
      "step": 7730
    },
    {
      "epoch": 4.737566947207345,
      "grad_norm": 2.159982442855835,
      "learning_rate": 2.9223651085255495e-05,
      "loss": 1.2319,
      "step": 7740
    },
    {
      "epoch": 4.743687834736036,
      "grad_norm": 2.338762044906616,
      "learning_rate": 2.9189630536844252e-05,
      "loss": 1.2143,
      "step": 7750
    },
    {
      "epoch": 4.749808722264729,
      "grad_norm": 2.3189380168914795,
      "learning_rate": 2.9155609988433013e-05,
      "loss": 1.239,
      "step": 7760
    },
    {
      "epoch": 4.75592960979342,
      "grad_norm": 1.9343451261520386,
      "learning_rate": 2.9121589440021774e-05,
      "loss": 1.2121,
      "step": 7770
    },
    {
      "epoch": 4.762050497322112,
      "grad_norm": 1.9574135541915894,
      "learning_rate": 2.9087568891610534e-05,
      "loss": 1.228,
      "step": 7780
    },
    {
      "epoch": 4.768171384850803,
      "grad_norm": 2.6530020236968994,
      "learning_rate": 2.9053548343199295e-05,
      "loss": 1.2365,
      "step": 7790
    },
    {
      "epoch": 4.774292272379495,
      "grad_norm": 1.8905853033065796,
      "learning_rate": 2.9019527794788053e-05,
      "loss": 1.2071,
      "step": 7800
    },
    {
      "epoch": 4.7804131599081865,
      "grad_norm": 2.568509101867676,
      "learning_rate": 2.8985507246376814e-05,
      "loss": 1.2149,
      "step": 7810
    },
    {
      "epoch": 4.786534047436878,
      "grad_norm": 2.2760283946990967,
      "learning_rate": 2.8951486697965574e-05,
      "loss": 1.2287,
      "step": 7820
    },
    {
      "epoch": 4.7926549349655705,
      "grad_norm": 2.0909883975982666,
      "learning_rate": 2.8917466149554335e-05,
      "loss": 1.2221,
      "step": 7830
    },
    {
      "epoch": 4.798775822494262,
      "grad_norm": 2.4575860500335693,
      "learning_rate": 2.8883445601143093e-05,
      "loss": 1.2271,
      "step": 7840
    },
    {
      "epoch": 4.804896710022954,
      "grad_norm": 1.9734315872192383,
      "learning_rate": 2.884942505273185e-05,
      "loss": 1.2501,
      "step": 7850
    },
    {
      "epoch": 4.811017597551645,
      "grad_norm": 3.550708532333374,
      "learning_rate": 2.8815404504320608e-05,
      "loss": 1.2128,
      "step": 7860
    },
    {
      "epoch": 4.817138485080337,
      "grad_norm": 2.462048053741455,
      "learning_rate": 2.878138395590937e-05,
      "loss": 1.2109,
      "step": 7870
    },
    {
      "epoch": 4.823259372609028,
      "grad_norm": 1.9959014654159546,
      "learning_rate": 2.874736340749813e-05,
      "loss": 1.2379,
      "step": 7880
    },
    {
      "epoch": 4.82938026013772,
      "grad_norm": 2.04964542388916,
      "learning_rate": 2.871334285908689e-05,
      "loss": 1.2224,
      "step": 7890
    },
    {
      "epoch": 4.835501147666411,
      "grad_norm": 2.9247825145721436,
      "learning_rate": 2.867932231067565e-05,
      "loss": 1.2303,
      "step": 7900
    },
    {
      "epoch": 4.841622035195103,
      "grad_norm": 2.2298436164855957,
      "learning_rate": 2.864530176226441e-05,
      "loss": 1.243,
      "step": 7910
    },
    {
      "epoch": 4.8477429227237945,
      "grad_norm": 1.946879267692566,
      "learning_rate": 2.861128121385317e-05,
      "loss": 1.2206,
      "step": 7920
    },
    {
      "epoch": 4.853863810252487,
      "grad_norm": 2.5111823081970215,
      "learning_rate": 2.857726066544193e-05,
      "loss": 1.2002,
      "step": 7930
    },
    {
      "epoch": 4.8599846977811785,
      "grad_norm": 2.873385190963745,
      "learning_rate": 2.854324011703069e-05,
      "loss": 1.2167,
      "step": 7940
    },
    {
      "epoch": 4.86610558530987,
      "grad_norm": 2.1272952556610107,
      "learning_rate": 2.8509219568619448e-05,
      "loss": 1.2131,
      "step": 7950
    },
    {
      "epoch": 4.872226472838562,
      "grad_norm": 2.3787174224853516,
      "learning_rate": 2.847519902020821e-05,
      "loss": 1.2121,
      "step": 7960
    },
    {
      "epoch": 4.878347360367253,
      "grad_norm": 2.747269868850708,
      "learning_rate": 2.8441178471796963e-05,
      "loss": 1.227,
      "step": 7970
    },
    {
      "epoch": 4.884468247895945,
      "grad_norm": 3.0929527282714844,
      "learning_rate": 2.8407157923385724e-05,
      "loss": 1.2276,
      "step": 7980
    },
    {
      "epoch": 4.890589135424636,
      "grad_norm": 2.1043317317962646,
      "learning_rate": 2.8373137374974485e-05,
      "loss": 1.2188,
      "step": 7990
    },
    {
      "epoch": 4.896710022953329,
      "grad_norm": 2.1406781673431396,
      "learning_rate": 2.8339116826563246e-05,
      "loss": 1.2234,
      "step": 8000
    },
    {
      "epoch": 4.90283091048202,
      "grad_norm": 2.324378728866577,
      "learning_rate": 2.8305096278152003e-05,
      "loss": 1.2438,
      "step": 8010
    },
    {
      "epoch": 4.908951798010712,
      "grad_norm": 1.8091750144958496,
      "learning_rate": 2.8271075729740764e-05,
      "loss": 1.2426,
      "step": 8020
    },
    {
      "epoch": 4.915072685539403,
      "grad_norm": 2.3250181674957275,
      "learning_rate": 2.8237055181329525e-05,
      "loss": 1.2115,
      "step": 8030
    },
    {
      "epoch": 4.921193573068095,
      "grad_norm": 2.8805532455444336,
      "learning_rate": 2.8203034632918286e-05,
      "loss": 1.2214,
      "step": 8040
    },
    {
      "epoch": 4.9273144605967865,
      "grad_norm": 1.797640323638916,
      "learning_rate": 2.8169014084507046e-05,
      "loss": 1.2072,
      "step": 8050
    },
    {
      "epoch": 4.933435348125478,
      "grad_norm": 2.2686753273010254,
      "learning_rate": 2.8134993536095804e-05,
      "loss": 1.1998,
      "step": 8060
    },
    {
      "epoch": 4.93955623565417,
      "grad_norm": 2.1427371501922607,
      "learning_rate": 2.8100972987684565e-05,
      "loss": 1.2291,
      "step": 8070
    },
    {
      "epoch": 4.945677123182861,
      "grad_norm": 1.856122374534607,
      "learning_rate": 2.8066952439273326e-05,
      "loss": 1.2189,
      "step": 8080
    },
    {
      "epoch": 4.951798010711553,
      "grad_norm": 1.8572533130645752,
      "learning_rate": 2.803293189086208e-05,
      "loss": 1.2252,
      "step": 8090
    },
    {
      "epoch": 4.957918898240245,
      "grad_norm": 2.145905017852783,
      "learning_rate": 2.799891134245084e-05,
      "loss": 1.2065,
      "step": 8100
    },
    {
      "epoch": 4.964039785768937,
      "grad_norm": 2.219866991043091,
      "learning_rate": 2.79648907940396e-05,
      "loss": 1.2106,
      "step": 8110
    },
    {
      "epoch": 4.970160673297628,
      "grad_norm": 1.9409259557724,
      "learning_rate": 2.793087024562836e-05,
      "loss": 1.2306,
      "step": 8120
    },
    {
      "epoch": 4.97628156082632,
      "grad_norm": 1.977242350578308,
      "learning_rate": 2.789684969721712e-05,
      "loss": 1.216,
      "step": 8130
    },
    {
      "epoch": 4.982402448355011,
      "grad_norm": 2.118283271789551,
      "learning_rate": 2.786282914880588e-05,
      "loss": 1.2048,
      "step": 8140
    },
    {
      "epoch": 4.988523335883703,
      "grad_norm": 2.632073402404785,
      "learning_rate": 2.782880860039464e-05,
      "loss": 1.2132,
      "step": 8150
    },
    {
      "epoch": 4.9946442234123944,
      "grad_norm": 2.513634204864502,
      "learning_rate": 2.77947880519834e-05,
      "loss": 1.1973,
      "step": 8160
    },
    {
      "epoch": 4.999540933435348,
      "eval_accuracy": 0.49269235183839,
      "eval_loss": 1.191017508506775,
      "eval_runtime": 2626.2477,
      "eval_samples_per_second": 79.618,
      "eval_steps_per_second": 0.622,
      "step": 8168
    },
    {
      "epoch": 5.000765110941087,
      "grad_norm": 2.578747034072876,
      "learning_rate": 2.776076750357216e-05,
      "loss": 1.2102,
      "step": 8170
    },
    {
      "epoch": 5.006885998469778,
      "grad_norm": 2.2281763553619385,
      "learning_rate": 2.772674695516092e-05,
      "loss": 1.2144,
      "step": 8180
    },
    {
      "epoch": 5.01300688599847,
      "grad_norm": 2.208728551864624,
      "learning_rate": 2.769272640674968e-05,
      "loss": 1.1976,
      "step": 8190
    },
    {
      "epoch": 5.0191277735271616,
      "grad_norm": 2.4716689586639404,
      "learning_rate": 2.7658705858338435e-05,
      "loss": 1.1719,
      "step": 8200
    },
    {
      "epoch": 5.025248661055853,
      "grad_norm": 2.048522472381592,
      "learning_rate": 2.7624685309927196e-05,
      "loss": 1.2016,
      "step": 8210
    },
    {
      "epoch": 5.031369548584545,
      "grad_norm": 2.6114094257354736,
      "learning_rate": 2.7590664761515957e-05,
      "loss": 1.1989,
      "step": 8220
    },
    {
      "epoch": 5.037490436113236,
      "grad_norm": 2.179579734802246,
      "learning_rate": 2.7556644213104714e-05,
      "loss": 1.2051,
      "step": 8230
    },
    {
      "epoch": 5.043611323641928,
      "grad_norm": 2.6294515132904053,
      "learning_rate": 2.7522623664693475e-05,
      "loss": 1.2084,
      "step": 8240
    },
    {
      "epoch": 5.049732211170619,
      "grad_norm": 2.3146214485168457,
      "learning_rate": 2.7488603116282236e-05,
      "loss": 1.1932,
      "step": 8250
    },
    {
      "epoch": 5.055853098699312,
      "grad_norm": 3.6110377311706543,
      "learning_rate": 2.7454582567870997e-05,
      "loss": 1.2052,
      "step": 8260
    },
    {
      "epoch": 5.061973986228003,
      "grad_norm": 3.290837049484253,
      "learning_rate": 2.7420562019459754e-05,
      "loss": 1.2179,
      "step": 8270
    },
    {
      "epoch": 5.068094873756695,
      "grad_norm": 1.990761637687683,
      "learning_rate": 2.7386541471048515e-05,
      "loss": 1.2253,
      "step": 8280
    },
    {
      "epoch": 5.074215761285386,
      "grad_norm": 2.85910964012146,
      "learning_rate": 2.7352520922637276e-05,
      "loss": 1.1992,
      "step": 8290
    },
    {
      "epoch": 5.080336648814078,
      "grad_norm": 6.976127624511719,
      "learning_rate": 2.7318500374226037e-05,
      "loss": 1.2065,
      "step": 8300
    },
    {
      "epoch": 5.0864575363427695,
      "grad_norm": 2.3096303939819336,
      "learning_rate": 2.7284479825814798e-05,
      "loss": 1.1973,
      "step": 8310
    },
    {
      "epoch": 5.092578423871461,
      "grad_norm": 6.251516342163086,
      "learning_rate": 2.7250459277403552e-05,
      "loss": 1.187,
      "step": 8320
    },
    {
      "epoch": 5.098699311400153,
      "grad_norm": 2.466515064239502,
      "learning_rate": 2.721643872899231e-05,
      "loss": 1.1905,
      "step": 8330
    },
    {
      "epoch": 5.104820198928845,
      "grad_norm": 2.5167441368103027,
      "learning_rate": 2.718241818058107e-05,
      "loss": 1.2128,
      "step": 8340
    },
    {
      "epoch": 5.110941086457537,
      "grad_norm": 2.935549736022949,
      "learning_rate": 2.714839763216983e-05,
      "loss": 1.2005,
      "step": 8350
    },
    {
      "epoch": 5.117061973986228,
      "grad_norm": 3.1587438583374023,
      "learning_rate": 2.711437708375859e-05,
      "loss": 1.194,
      "step": 8360
    },
    {
      "epoch": 5.12318286151492,
      "grad_norm": 2.3066651821136475,
      "learning_rate": 2.7080356535347352e-05,
      "loss": 1.1774,
      "step": 8370
    },
    {
      "epoch": 5.129303749043611,
      "grad_norm": 2.729510545730591,
      "learning_rate": 2.704633598693611e-05,
      "loss": 1.2187,
      "step": 8380
    },
    {
      "epoch": 5.135424636572303,
      "grad_norm": 2.3249425888061523,
      "learning_rate": 2.701231543852487e-05,
      "loss": 1.1844,
      "step": 8390
    },
    {
      "epoch": 5.141545524100994,
      "grad_norm": 2.5726587772369385,
      "learning_rate": 2.697829489011363e-05,
      "loss": 1.1862,
      "step": 8400
    },
    {
      "epoch": 5.147666411629686,
      "grad_norm": 2.2208099365234375,
      "learning_rate": 2.6944274341702392e-05,
      "loss": 1.2141,
      "step": 8410
    },
    {
      "epoch": 5.1537872991583775,
      "grad_norm": 2.545058012008667,
      "learning_rate": 2.691025379329115e-05,
      "loss": 1.2017,
      "step": 8420
    },
    {
      "epoch": 5.15990818668707,
      "grad_norm": 2.34517502784729,
      "learning_rate": 2.687623324487991e-05,
      "loss": 1.1668,
      "step": 8430
    },
    {
      "epoch": 5.1660290742157615,
      "grad_norm": 3.2366106510162354,
      "learning_rate": 2.6842212696468665e-05,
      "loss": 1.2014,
      "step": 8440
    },
    {
      "epoch": 5.172149961744453,
      "grad_norm": 2.4992125034332275,
      "learning_rate": 2.6808192148057426e-05,
      "loss": 1.2127,
      "step": 8450
    },
    {
      "epoch": 5.178270849273145,
      "grad_norm": 2.8871748447418213,
      "learning_rate": 2.6774171599646186e-05,
      "loss": 1.202,
      "step": 8460
    },
    {
      "epoch": 5.184391736801836,
      "grad_norm": 2.683912515640259,
      "learning_rate": 2.6740151051234947e-05,
      "loss": 1.2157,
      "step": 8470
    },
    {
      "epoch": 5.190512624330528,
      "grad_norm": 2.367983102798462,
      "learning_rate": 2.6706130502823705e-05,
      "loss": 1.2233,
      "step": 8480
    },
    {
      "epoch": 5.196633511859219,
      "grad_norm": 2.0448033809661865,
      "learning_rate": 2.6672109954412466e-05,
      "loss": 1.2066,
      "step": 8490
    },
    {
      "epoch": 5.202754399387911,
      "grad_norm": 2.684910774230957,
      "learning_rate": 2.6638089406001226e-05,
      "loss": 1.1947,
      "step": 8500
    },
    {
      "epoch": 5.208875286916603,
      "grad_norm": 1.977565884590149,
      "learning_rate": 2.6604068857589987e-05,
      "loss": 1.1967,
      "step": 8510
    },
    {
      "epoch": 5.214996174445295,
      "grad_norm": 2.5993175506591797,
      "learning_rate": 2.6570048309178748e-05,
      "loss": 1.1995,
      "step": 8520
    },
    {
      "epoch": 5.221117061973986,
      "grad_norm": 2.7591464519500732,
      "learning_rate": 2.6536027760767505e-05,
      "loss": 1.1821,
      "step": 8530
    },
    {
      "epoch": 5.227237949502678,
      "grad_norm": 3.1257448196411133,
      "learning_rate": 2.6502007212356266e-05,
      "loss": 1.2133,
      "step": 8540
    },
    {
      "epoch": 5.2333588370313695,
      "grad_norm": 2.6627302169799805,
      "learning_rate": 2.646798666394502e-05,
      "loss": 1.2123,
      "step": 8550
    },
    {
      "epoch": 5.239479724560061,
      "grad_norm": 2.7436108589172363,
      "learning_rate": 2.643396611553378e-05,
      "loss": 1.1839,
      "step": 8560
    },
    {
      "epoch": 5.245600612088753,
      "grad_norm": 2.5480778217315674,
      "learning_rate": 2.6399945567122542e-05,
      "loss": 1.2051,
      "step": 8570
    },
    {
      "epoch": 5.251721499617444,
      "grad_norm": 2.8406577110290527,
      "learning_rate": 2.6365925018711303e-05,
      "loss": 1.2,
      "step": 8580
    },
    {
      "epoch": 5.257842387146136,
      "grad_norm": 2.1096248626708984,
      "learning_rate": 2.633190447030006e-05,
      "loss": 1.2257,
      "step": 8590
    },
    {
      "epoch": 5.263963274674828,
      "grad_norm": 2.118366003036499,
      "learning_rate": 2.629788392188882e-05,
      "loss": 1.1917,
      "step": 8600
    },
    {
      "epoch": 5.27008416220352,
      "grad_norm": 2.6410601139068604,
      "learning_rate": 2.6263863373477582e-05,
      "loss": 1.199,
      "step": 8610
    },
    {
      "epoch": 5.276205049732211,
      "grad_norm": 2.6622838973999023,
      "learning_rate": 2.6229842825066343e-05,
      "loss": 1.2393,
      "step": 8620
    },
    {
      "epoch": 5.282325937260903,
      "grad_norm": 2.324550151824951,
      "learning_rate": 2.6195822276655104e-05,
      "loss": 1.1996,
      "step": 8630
    },
    {
      "epoch": 5.288446824789594,
      "grad_norm": 2.170179843902588,
      "learning_rate": 2.616180172824386e-05,
      "loss": 1.2034,
      "step": 8640
    },
    {
      "epoch": 5.294567712318286,
      "grad_norm": 2.926877737045288,
      "learning_rate": 2.6127781179832622e-05,
      "loss": 1.2017,
      "step": 8650
    },
    {
      "epoch": 5.3006885998469775,
      "grad_norm": 2.6772279739379883,
      "learning_rate": 2.6093760631421383e-05,
      "loss": 1.2162,
      "step": 8660
    },
    {
      "epoch": 5.30680948737567,
      "grad_norm": 2.1664867401123047,
      "learning_rate": 2.6059740083010137e-05,
      "loss": 1.2046,
      "step": 8670
    },
    {
      "epoch": 5.3129303749043615,
      "grad_norm": 3.135590076446533,
      "learning_rate": 2.6025719534598898e-05,
      "loss": 1.2098,
      "step": 8680
    },
    {
      "epoch": 5.319051262433053,
      "grad_norm": 2.273746967315674,
      "learning_rate": 2.599169898618766e-05,
      "loss": 1.1882,
      "step": 8690
    },
    {
      "epoch": 5.325172149961745,
      "grad_norm": 2.1665115356445312,
      "learning_rate": 2.5957678437776416e-05,
      "loss": 1.2069,
      "step": 8700
    },
    {
      "epoch": 5.331293037490436,
      "grad_norm": 2.255600929260254,
      "learning_rate": 2.5923657889365177e-05,
      "loss": 1.1852,
      "step": 8710
    },
    {
      "epoch": 5.337413925019128,
      "grad_norm": 2.3505523204803467,
      "learning_rate": 2.5889637340953938e-05,
      "loss": 1.198,
      "step": 8720
    },
    {
      "epoch": 5.343534812547819,
      "grad_norm": 2.702526330947876,
      "learning_rate": 2.58556167925427e-05,
      "loss": 1.1917,
      "step": 8730
    },
    {
      "epoch": 5.349655700076511,
      "grad_norm": 2.1754770278930664,
      "learning_rate": 2.5821596244131456e-05,
      "loss": 1.2049,
      "step": 8740
    },
    {
      "epoch": 5.355776587605202,
      "grad_norm": 2.5231616497039795,
      "learning_rate": 2.5787575695720217e-05,
      "loss": 1.1991,
      "step": 8750
    },
    {
      "epoch": 5.361897475133895,
      "grad_norm": 2.6360552310943604,
      "learning_rate": 2.5753555147308978e-05,
      "loss": 1.2092,
      "step": 8760
    },
    {
      "epoch": 5.368018362662586,
      "grad_norm": 2.3326311111450195,
      "learning_rate": 2.571953459889774e-05,
      "loss": 1.198,
      "step": 8770
    },
    {
      "epoch": 5.374139250191278,
      "grad_norm": 2.2699687480926514,
      "learning_rate": 2.56855140504865e-05,
      "loss": 1.2005,
      "step": 8780
    },
    {
      "epoch": 5.3802601377199695,
      "grad_norm": 2.333057165145874,
      "learning_rate": 2.5651493502075253e-05,
      "loss": 1.2159,
      "step": 8790
    },
    {
      "epoch": 5.386381025248661,
      "grad_norm": 2.36719012260437,
      "learning_rate": 2.561747295366401e-05,
      "loss": 1.184,
      "step": 8800
    },
    {
      "epoch": 5.392501912777353,
      "grad_norm": 2.606476306915283,
      "learning_rate": 2.558345240525277e-05,
      "loss": 1.2024,
      "step": 8810
    },
    {
      "epoch": 5.398622800306044,
      "grad_norm": 2.2720136642456055,
      "learning_rate": 2.5549431856841532e-05,
      "loss": 1.2019,
      "step": 8820
    },
    {
      "epoch": 5.404743687834736,
      "grad_norm": 2.720341205596924,
      "learning_rate": 2.5515411308430293e-05,
      "loss": 1.1857,
      "step": 8830
    },
    {
      "epoch": 5.410864575363428,
      "grad_norm": 2.564955234527588,
      "learning_rate": 2.5481390760019054e-05,
      "loss": 1.1948,
      "step": 8840
    },
    {
      "epoch": 5.41698546289212,
      "grad_norm": 2.1959404945373535,
      "learning_rate": 2.544737021160781e-05,
      "loss": 1.209,
      "step": 8850
    },
    {
      "epoch": 5.423106350420811,
      "grad_norm": 2.346018075942993,
      "learning_rate": 2.5413349663196572e-05,
      "loss": 1.2147,
      "step": 8860
    },
    {
      "epoch": 5.429227237949503,
      "grad_norm": 3.420417547225952,
      "learning_rate": 2.5379329114785333e-05,
      "loss": 1.2052,
      "step": 8870
    },
    {
      "epoch": 5.435348125478194,
      "grad_norm": 2.6672658920288086,
      "learning_rate": 2.5345308566374094e-05,
      "loss": 1.1976,
      "step": 8880
    },
    {
      "epoch": 5.441469013006886,
      "grad_norm": 2.4706966876983643,
      "learning_rate": 2.531128801796285e-05,
      "loss": 1.1956,
      "step": 8890
    },
    {
      "epoch": 5.4475899005355775,
      "grad_norm": 2.0787277221679688,
      "learning_rate": 2.527726746955161e-05,
      "loss": 1.2117,
      "step": 8900
    },
    {
      "epoch": 5.453710788064269,
      "grad_norm": 2.4532337188720703,
      "learning_rate": 2.5243246921140366e-05,
      "loss": 1.2004,
      "step": 8910
    },
    {
      "epoch": 5.459831675592961,
      "grad_norm": 2.5366806983947754,
      "learning_rate": 2.5209226372729127e-05,
      "loss": 1.2241,
      "step": 8920
    },
    {
      "epoch": 5.465952563121653,
      "grad_norm": 2.249305486679077,
      "learning_rate": 2.5175205824317888e-05,
      "loss": 1.1777,
      "step": 8930
    },
    {
      "epoch": 5.472073450650345,
      "grad_norm": 2.310941457748413,
      "learning_rate": 2.514118527590665e-05,
      "loss": 1.2017,
      "step": 8940
    },
    {
      "epoch": 5.478194338179036,
      "grad_norm": 2.3156533241271973,
      "learning_rate": 2.510716472749541e-05,
      "loss": 1.2053,
      "step": 8950
    },
    {
      "epoch": 5.484315225707728,
      "grad_norm": 2.2076308727264404,
      "learning_rate": 2.5073144179084167e-05,
      "loss": 1.197,
      "step": 8960
    },
    {
      "epoch": 5.490436113236419,
      "grad_norm": 2.4675791263580322,
      "learning_rate": 2.5039123630672928e-05,
      "loss": 1.2096,
      "step": 8970
    },
    {
      "epoch": 5.496557000765111,
      "grad_norm": 2.278066396713257,
      "learning_rate": 2.500510308226169e-05,
      "loss": 1.2003,
      "step": 8980
    },
    {
      "epoch": 5.502677888293802,
      "grad_norm": 2.8391642570495605,
      "learning_rate": 2.4971082533850446e-05,
      "loss": 1.1821,
      "step": 8990
    },
    {
      "epoch": 5.508798775822494,
      "grad_norm": 2.1658828258514404,
      "learning_rate": 2.4937061985439207e-05,
      "loss": 1.1976,
      "step": 9000
    },
    {
      "epoch": 5.514919663351186,
      "grad_norm": 2.391244888305664,
      "learning_rate": 2.4903041437027964e-05,
      "loss": 1.1906,
      "step": 9010
    },
    {
      "epoch": 5.521040550879878,
      "grad_norm": 3.277883529663086,
      "learning_rate": 2.4869020888616725e-05,
      "loss": 1.2039,
      "step": 9020
    },
    {
      "epoch": 5.527161438408569,
      "grad_norm": 2.536834716796875,
      "learning_rate": 2.4835000340205486e-05,
      "loss": 1.1884,
      "step": 9030
    },
    {
      "epoch": 5.533282325937261,
      "grad_norm": 2.8371753692626953,
      "learning_rate": 2.4800979791794247e-05,
      "loss": 1.1964,
      "step": 9040
    },
    {
      "epoch": 5.5394032134659525,
      "grad_norm": 2.2327940464019775,
      "learning_rate": 2.4766959243383004e-05,
      "loss": 1.2042,
      "step": 9050
    },
    {
      "epoch": 5.545524100994644,
      "grad_norm": 2.28041934967041,
      "learning_rate": 2.4732938694971762e-05,
      "loss": 1.2076,
      "step": 9060
    },
    {
      "epoch": 5.551644988523336,
      "grad_norm": 2.6812901496887207,
      "learning_rate": 2.4698918146560523e-05,
      "loss": 1.197,
      "step": 9070
    },
    {
      "epoch": 5.557765876052027,
      "grad_norm": 2.6180360317230225,
      "learning_rate": 2.4664897598149284e-05,
      "loss": 1.2023,
      "step": 9080
    },
    {
      "epoch": 5.563886763580719,
      "grad_norm": 2.2223103046417236,
      "learning_rate": 2.4630877049738044e-05,
      "loss": 1.2079,
      "step": 9090
    },
    {
      "epoch": 5.570007651109411,
      "grad_norm": 2.5299456119537354,
      "learning_rate": 2.4596856501326805e-05,
      "loss": 1.1983,
      "step": 9100
    },
    {
      "epoch": 5.576128538638103,
      "grad_norm": 2.541381359100342,
      "learning_rate": 2.4562835952915563e-05,
      "loss": 1.211,
      "step": 9110
    },
    {
      "epoch": 5.582249426166794,
      "grad_norm": 2.687126636505127,
      "learning_rate": 2.452881540450432e-05,
      "loss": 1.2138,
      "step": 9120
    },
    {
      "epoch": 5.588370313695486,
      "grad_norm": 2.71907377243042,
      "learning_rate": 2.449479485609308e-05,
      "loss": 1.213,
      "step": 9130
    },
    {
      "epoch": 5.594491201224177,
      "grad_norm": 2.3322901725769043,
      "learning_rate": 2.4460774307681842e-05,
      "loss": 1.1948,
      "step": 9140
    },
    {
      "epoch": 5.600612088752869,
      "grad_norm": 2.353484630584717,
      "learning_rate": 2.4426753759270603e-05,
      "loss": 1.1968,
      "step": 9150
    },
    {
      "epoch": 5.6067329762815605,
      "grad_norm": 2.285670042037964,
      "learning_rate": 2.439273321085936e-05,
      "loss": 1.216,
      "step": 9160
    },
    {
      "epoch": 5.612853863810253,
      "grad_norm": 2.336120128631592,
      "learning_rate": 2.4358712662448117e-05,
      "loss": 1.1931,
      "step": 9170
    },
    {
      "epoch": 5.6189747513389445,
      "grad_norm": 2.471015453338623,
      "learning_rate": 2.432469211403688e-05,
      "loss": 1.1919,
      "step": 9180
    },
    {
      "epoch": 5.625095638867636,
      "grad_norm": 2.1988890171051025,
      "learning_rate": 2.429067156562564e-05,
      "loss": 1.2223,
      "step": 9190
    },
    {
      "epoch": 5.631216526396328,
      "grad_norm": 1.9588499069213867,
      "learning_rate": 2.42566510172144e-05,
      "loss": 1.2036,
      "step": 9200
    },
    {
      "epoch": 5.637337413925019,
      "grad_norm": 2.280674934387207,
      "learning_rate": 2.4222630468803157e-05,
      "loss": 1.1847,
      "step": 9210
    },
    {
      "epoch": 5.643458301453711,
      "grad_norm": 2.6746459007263184,
      "learning_rate": 2.4188609920391918e-05,
      "loss": 1.2244,
      "step": 9220
    },
    {
      "epoch": 5.649579188982402,
      "grad_norm": 2.370379686355591,
      "learning_rate": 2.4154589371980676e-05,
      "loss": 1.204,
      "step": 9230
    },
    {
      "epoch": 5.655700076511094,
      "grad_norm": 2.2550511360168457,
      "learning_rate": 2.4120568823569437e-05,
      "loss": 1.1869,
      "step": 9240
    },
    {
      "epoch": 5.661820964039785,
      "grad_norm": 2.5164883136749268,
      "learning_rate": 2.4086548275158197e-05,
      "loss": 1.2165,
      "step": 9250
    },
    {
      "epoch": 5.667941851568477,
      "grad_norm": 2.711061954498291,
      "learning_rate": 2.4052527726746958e-05,
      "loss": 1.2242,
      "step": 9260
    },
    {
      "epoch": 5.674062739097169,
      "grad_norm": 2.4196157455444336,
      "learning_rate": 2.4018507178335716e-05,
      "loss": 1.2015,
      "step": 9270
    },
    {
      "epoch": 5.680183626625861,
      "grad_norm": 3.0094361305236816,
      "learning_rate": 2.3984486629924473e-05,
      "loss": 1.2095,
      "step": 9280
    },
    {
      "epoch": 5.6863045141545525,
      "grad_norm": 2.66207218170166,
      "learning_rate": 2.3950466081513234e-05,
      "loss": 1.1981,
      "step": 9290
    },
    {
      "epoch": 5.692425401683244,
      "grad_norm": 2.4135472774505615,
      "learning_rate": 2.3916445533101995e-05,
      "loss": 1.2121,
      "step": 9300
    },
    {
      "epoch": 5.698546289211936,
      "grad_norm": 2.1922266483306885,
      "learning_rate": 2.3882424984690756e-05,
      "loss": 1.2119,
      "step": 9310
    },
    {
      "epoch": 5.704667176740627,
      "grad_norm": 2.5734925270080566,
      "learning_rate": 2.3848404436279513e-05,
      "loss": 1.1831,
      "step": 9320
    },
    {
      "epoch": 5.710788064269319,
      "grad_norm": 2.7392802238464355,
      "learning_rate": 2.3814383887868274e-05,
      "loss": 1.2075,
      "step": 9330
    },
    {
      "epoch": 5.716908951798011,
      "grad_norm": 2.732255458831787,
      "learning_rate": 2.378036333945703e-05,
      "loss": 1.2072,
      "step": 9340
    },
    {
      "epoch": 5.723029839326703,
      "grad_norm": 2.7654104232788086,
      "learning_rate": 2.3746342791045792e-05,
      "loss": 1.2062,
      "step": 9350
    },
    {
      "epoch": 5.729150726855394,
      "grad_norm": 2.4640185832977295,
      "learning_rate": 2.3712322242634553e-05,
      "loss": 1.2067,
      "step": 9360
    },
    {
      "epoch": 5.735271614384086,
      "grad_norm": 2.2052786350250244,
      "learning_rate": 2.367830169422331e-05,
      "loss": 1.1818,
      "step": 9370
    },
    {
      "epoch": 5.741392501912777,
      "grad_norm": 2.3217477798461914,
      "learning_rate": 2.364428114581207e-05,
      "loss": 1.2122,
      "step": 9380
    },
    {
      "epoch": 5.747513389441469,
      "grad_norm": 2.3705313205718994,
      "learning_rate": 2.3610260597400832e-05,
      "loss": 1.1926,
      "step": 9390
    },
    {
      "epoch": 5.7536342769701605,
      "grad_norm": 2.948225259780884,
      "learning_rate": 2.357624004898959e-05,
      "loss": 1.2,
      "step": 9400
    },
    {
      "epoch": 5.759755164498852,
      "grad_norm": 2.4769287109375,
      "learning_rate": 2.354221950057835e-05,
      "loss": 1.1832,
      "step": 9410
    },
    {
      "epoch": 5.765876052027544,
      "grad_norm": 2.542545795440674,
      "learning_rate": 2.350819895216711e-05,
      "loss": 1.2188,
      "step": 9420
    },
    {
      "epoch": 5.771996939556235,
      "grad_norm": 2.491767406463623,
      "learning_rate": 2.347417840375587e-05,
      "loss": 1.1808,
      "step": 9430
    },
    {
      "epoch": 5.778117827084928,
      "grad_norm": 2.5319976806640625,
      "learning_rate": 2.344015785534463e-05,
      "loss": 1.2085,
      "step": 9440
    },
    {
      "epoch": 5.784238714613619,
      "grad_norm": 2.477506399154663,
      "learning_rate": 2.340613730693339e-05,
      "loss": 1.1946,
      "step": 9450
    },
    {
      "epoch": 5.790359602142311,
      "grad_norm": 2.290252685546875,
      "learning_rate": 2.3372116758522148e-05,
      "loss": 1.1919,
      "step": 9460
    },
    {
      "epoch": 5.796480489671002,
      "grad_norm": 2.647599697113037,
      "learning_rate": 2.333809621011091e-05,
      "loss": 1.2116,
      "step": 9470
    },
    {
      "epoch": 5.802601377199694,
      "grad_norm": 2.6891698837280273,
      "learning_rate": 2.3304075661699666e-05,
      "loss": 1.1808,
      "step": 9480
    },
    {
      "epoch": 5.808722264728385,
      "grad_norm": 2.669074773788452,
      "learning_rate": 2.3270055113288427e-05,
      "loss": 1.1937,
      "step": 9490
    },
    {
      "epoch": 5.814843152257077,
      "grad_norm": 2.0883407592773438,
      "learning_rate": 2.3236034564877188e-05,
      "loss": 1.1748,
      "step": 9500
    },
    {
      "epoch": 5.820964039785769,
      "grad_norm": 2.735302448272705,
      "learning_rate": 2.320201401646595e-05,
      "loss": 1.2035,
      "step": 9510
    },
    {
      "epoch": 5.827084927314461,
      "grad_norm": 2.263305187225342,
      "learning_rate": 2.3167993468054706e-05,
      "loss": 1.204,
      "step": 9520
    },
    {
      "epoch": 5.8332058148431525,
      "grad_norm": 2.3452515602111816,
      "learning_rate": 2.3133972919643463e-05,
      "loss": 1.2145,
      "step": 9530
    },
    {
      "epoch": 5.839326702371844,
      "grad_norm": 2.134207010269165,
      "learning_rate": 2.3099952371232224e-05,
      "loss": 1.1987,
      "step": 9540
    },
    {
      "epoch": 5.845447589900536,
      "grad_norm": 2.2933249473571777,
      "learning_rate": 2.3065931822820985e-05,
      "loss": 1.2018,
      "step": 9550
    },
    {
      "epoch": 5.851568477429227,
      "grad_norm": 3.017359972000122,
      "learning_rate": 2.3031911274409746e-05,
      "loss": 1.1982,
      "step": 9560
    },
    {
      "epoch": 5.857689364957919,
      "grad_norm": 2.179676055908203,
      "learning_rate": 2.2997890725998507e-05,
      "loss": 1.1911,
      "step": 9570
    },
    {
      "epoch": 5.86381025248661,
      "grad_norm": 3.0648555755615234,
      "learning_rate": 2.2963870177587264e-05,
      "loss": 1.1962,
      "step": 9580
    },
    {
      "epoch": 5.869931140015302,
      "grad_norm": 2.395503044128418,
      "learning_rate": 2.292984962917602e-05,
      "loss": 1.2117,
      "step": 9590
    },
    {
      "epoch": 5.876052027543994,
      "grad_norm": 2.281404972076416,
      "learning_rate": 2.2895829080764783e-05,
      "loss": 1.2103,
      "step": 9600
    },
    {
      "epoch": 5.882172915072686,
      "grad_norm": 2.796386957168579,
      "learning_rate": 2.2861808532353543e-05,
      "loss": 1.2088,
      "step": 9610
    },
    {
      "epoch": 5.888293802601377,
      "grad_norm": 3.063288927078247,
      "learning_rate": 2.2827787983942304e-05,
      "loss": 1.2001,
      "step": 9620
    },
    {
      "epoch": 5.894414690130069,
      "grad_norm": 2.6174395084381104,
      "learning_rate": 2.279376743553106e-05,
      "loss": 1.2125,
      "step": 9630
    },
    {
      "epoch": 5.9005355776587605,
      "grad_norm": 2.5155296325683594,
      "learning_rate": 2.275974688711982e-05,
      "loss": 1.1971,
      "step": 9640
    },
    {
      "epoch": 5.906656465187452,
      "grad_norm": 2.342050790786743,
      "learning_rate": 2.272572633870858e-05,
      "loss": 1.2212,
      "step": 9650
    },
    {
      "epoch": 5.912777352716144,
      "grad_norm": 2.788961172103882,
      "learning_rate": 2.269170579029734e-05,
      "loss": 1.1955,
      "step": 9660
    },
    {
      "epoch": 5.918898240244835,
      "grad_norm": 2.815035581588745,
      "learning_rate": 2.26576852418861e-05,
      "loss": 1.2054,
      "step": 9670
    },
    {
      "epoch": 5.925019127773528,
      "grad_norm": 2.6851422786712646,
      "learning_rate": 2.2623664693474862e-05,
      "loss": 1.2034,
      "step": 9680
    },
    {
      "epoch": 5.931140015302219,
      "grad_norm": 2.8981006145477295,
      "learning_rate": 2.258964414506362e-05,
      "loss": 1.2022,
      "step": 9690
    },
    {
      "epoch": 5.937260902830911,
      "grad_norm": 2.1799545288085938,
      "learning_rate": 2.2555623596652377e-05,
      "loss": 1.1863,
      "step": 9700
    },
    {
      "epoch": 5.943381790359602,
      "grad_norm": 2.2509403228759766,
      "learning_rate": 2.2521603048241138e-05,
      "loss": 1.1952,
      "step": 9710
    },
    {
      "epoch": 5.949502677888294,
      "grad_norm": 2.3222877979278564,
      "learning_rate": 2.24875824998299e-05,
      "loss": 1.2116,
      "step": 9720
    },
    {
      "epoch": 5.955623565416985,
      "grad_norm": 2.506871223449707,
      "learning_rate": 2.245356195141866e-05,
      "loss": 1.1901,
      "step": 9730
    },
    {
      "epoch": 5.961744452945677,
      "grad_norm": 2.736382484436035,
      "learning_rate": 2.2419541403007417e-05,
      "loss": 1.2051,
      "step": 9740
    },
    {
      "epoch": 5.9678653404743685,
      "grad_norm": 2.47483491897583,
      "learning_rate": 2.2385520854596175e-05,
      "loss": 1.2044,
      "step": 9750
    },
    {
      "epoch": 5.97398622800306,
      "grad_norm": 2.4884045124053955,
      "learning_rate": 2.2351500306184936e-05,
      "loss": 1.1871,
      "step": 9760
    },
    {
      "epoch": 5.9801071155317524,
      "grad_norm": 3.008117437362671,
      "learning_rate": 2.2317479757773696e-05,
      "loss": 1.2276,
      "step": 9770
    },
    {
      "epoch": 5.986228003060444,
      "grad_norm": 2.610323429107666,
      "learning_rate": 2.2283459209362457e-05,
      "loss": 1.1854,
      "step": 9780
    },
    {
      "epoch": 5.992348890589136,
      "grad_norm": 2.8609611988067627,
      "learning_rate": 2.2249438660951215e-05,
      "loss": 1.1797,
      "step": 9790
    },
    {
      "epoch": 5.998469778117827,
      "grad_norm": 2.5393433570861816,
      "learning_rate": 2.2215418112539975e-05,
      "loss": 1.2124,
      "step": 9800
    },
    {
      "epoch": 5.999693955623566,
      "eval_accuracy": 0.49892393924321843,
      "eval_loss": 1.181107997894287,
      "eval_runtime": 2638.7084,
      "eval_samples_per_second": 79.242,
      "eval_steps_per_second": 0.619,
      "step": 9802
    },
    {
      "epoch": 6.004590665646519,
      "grad_norm": 2.1775569915771484,
      "learning_rate": 2.2181397564128733e-05,
      "loss": 1.1978,
      "step": 9810
    },
    {
      "epoch": 6.01071155317521,
      "grad_norm": 2.870755434036255,
      "learning_rate": 2.2147377015717494e-05,
      "loss": 1.1981,
      "step": 9820
    },
    {
      "epoch": 6.016832440703902,
      "grad_norm": 2.644536018371582,
      "learning_rate": 2.2113356467306255e-05,
      "loss": 1.1809,
      "step": 9830
    },
    {
      "epoch": 6.022953328232593,
      "grad_norm": 4.089147090911865,
      "learning_rate": 2.2079335918895015e-05,
      "loss": 1.1818,
      "step": 9840
    },
    {
      "epoch": 6.029074215761286,
      "grad_norm": 3.8916430473327637,
      "learning_rate": 2.2045315370483773e-05,
      "loss": 1.1734,
      "step": 9850
    },
    {
      "epoch": 6.035195103289977,
      "grad_norm": 2.7396483421325684,
      "learning_rate": 2.2011294822072534e-05,
      "loss": 1.1856,
      "step": 9860
    },
    {
      "epoch": 6.041315990818669,
      "grad_norm": 3.0049068927764893,
      "learning_rate": 2.197727427366129e-05,
      "loss": 1.1725,
      "step": 9870
    },
    {
      "epoch": 6.04743687834736,
      "grad_norm": 5.130383014678955,
      "learning_rate": 2.1943253725250052e-05,
      "loss": 1.1805,
      "step": 9880
    },
    {
      "epoch": 6.053557765876052,
      "grad_norm": 2.8813374042510986,
      "learning_rate": 2.1909233176838813e-05,
      "loss": 1.1973,
      "step": 9890
    },
    {
      "epoch": 6.0596786534047435,
      "grad_norm": 2.5018324851989746,
      "learning_rate": 2.187521262842757e-05,
      "loss": 1.1796,
      "step": 9900
    },
    {
      "epoch": 6.065799540933435,
      "grad_norm": 3.9136996269226074,
      "learning_rate": 2.184119208001633e-05,
      "loss": 1.1864,
      "step": 9910
    },
    {
      "epoch": 6.071920428462127,
      "grad_norm": 2.804731845855713,
      "learning_rate": 2.1807171531605092e-05,
      "loss": 1.1714,
      "step": 9920
    },
    {
      "epoch": 6.078041315990819,
      "grad_norm": 2.595283269882202,
      "learning_rate": 2.177315098319385e-05,
      "loss": 1.2029,
      "step": 9930
    },
    {
      "epoch": 6.084162203519511,
      "grad_norm": 3.3707568645477295,
      "learning_rate": 2.173913043478261e-05,
      "loss": 1.1851,
      "step": 9940
    },
    {
      "epoch": 6.090283091048202,
      "grad_norm": 2.980250597000122,
      "learning_rate": 2.1705109886371368e-05,
      "loss": 1.1693,
      "step": 9950
    },
    {
      "epoch": 6.096403978576894,
      "grad_norm": 2.76073956489563,
      "learning_rate": 2.167108933796013e-05,
      "loss": 1.1947,
      "step": 9960
    },
    {
      "epoch": 6.102524866105585,
      "grad_norm": 2.2360520362854004,
      "learning_rate": 2.163706878954889e-05,
      "loss": 1.185,
      "step": 9970
    },
    {
      "epoch": 6.108645753634277,
      "grad_norm": 11.473621368408203,
      "learning_rate": 2.160304824113765e-05,
      "loss": 1.1858,
      "step": 9980
    },
    {
      "epoch": 6.114766641162968,
      "grad_norm": 2.612234115600586,
      "learning_rate": 2.1569027692726408e-05,
      "loss": 1.1713,
      "step": 9990
    },
    {
      "epoch": 6.12088752869166,
      "grad_norm": 2.6785049438476562,
      "learning_rate": 2.153500714431517e-05,
      "loss": 1.1796,
      "step": 10000
    },
    {
      "epoch": 6.1270084162203515,
      "grad_norm": 2.6284472942352295,
      "learning_rate": 2.1500986595903926e-05,
      "loss": 1.1747,
      "step": 10010
    },
    {
      "epoch": 6.133129303749044,
      "grad_norm": 3.1232402324676514,
      "learning_rate": 2.1466966047492687e-05,
      "loss": 1.1757,
      "step": 10020
    },
    {
      "epoch": 6.1392501912777355,
      "grad_norm": 2.471266508102417,
      "learning_rate": 2.1432945499081448e-05,
      "loss": 1.1757,
      "step": 10030
    },
    {
      "epoch": 6.145371078806427,
      "grad_norm": 3.1765995025634766,
      "learning_rate": 2.139892495067021e-05,
      "loss": 1.1865,
      "step": 10040
    },
    {
      "epoch": 6.151491966335119,
      "grad_norm": 2.3057191371917725,
      "learning_rate": 2.1364904402258966e-05,
      "loss": 1.1802,
      "step": 10050
    },
    {
      "epoch": 6.15761285386381,
      "grad_norm": 3.5264956951141357,
      "learning_rate": 2.1330883853847723e-05,
      "loss": 1.1786,
      "step": 10060
    },
    {
      "epoch": 6.163733741392502,
      "grad_norm": 2.909853219985962,
      "learning_rate": 2.1296863305436484e-05,
      "loss": 1.1902,
      "step": 10070
    },
    {
      "epoch": 6.169854628921193,
      "grad_norm": 2.270792007446289,
      "learning_rate": 2.1262842757025245e-05,
      "loss": 1.1738,
      "step": 10080
    },
    {
      "epoch": 6.175975516449885,
      "grad_norm": 2.9542202949523926,
      "learning_rate": 2.1228822208614006e-05,
      "loss": 1.1752,
      "step": 10090
    },
    {
      "epoch": 6.182096403978577,
      "grad_norm": 3.13840913772583,
      "learning_rate": 2.1194801660202763e-05,
      "loss": 1.1809,
      "step": 10100
    },
    {
      "epoch": 6.188217291507269,
      "grad_norm": 2.623544216156006,
      "learning_rate": 2.116078111179152e-05,
      "loss": 1.1945,
      "step": 10110
    },
    {
      "epoch": 6.19433817903596,
      "grad_norm": 2.469536304473877,
      "learning_rate": 2.112676056338028e-05,
      "loss": 1.1781,
      "step": 10120
    },
    {
      "epoch": 6.200459066564652,
      "grad_norm": 3.681007146835327,
      "learning_rate": 2.1092740014969042e-05,
      "loss": 1.1644,
      "step": 10130
    },
    {
      "epoch": 6.2065799540933435,
      "grad_norm": 2.528697967529297,
      "learning_rate": 2.1058719466557803e-05,
      "loss": 1.1842,
      "step": 10140
    },
    {
      "epoch": 6.212700841622035,
      "grad_norm": 2.5676088333129883,
      "learning_rate": 2.1024698918146564e-05,
      "loss": 1.1688,
      "step": 10150
    },
    {
      "epoch": 6.218821729150727,
      "grad_norm": 2.5315933227539062,
      "learning_rate": 2.099067836973532e-05,
      "loss": 1.1532,
      "step": 10160
    },
    {
      "epoch": 6.224942616679418,
      "grad_norm": 2.9567384719848633,
      "learning_rate": 2.095665782132408e-05,
      "loss": 1.1692,
      "step": 10170
    },
    {
      "epoch": 6.231063504208111,
      "grad_norm": 2.8193509578704834,
      "learning_rate": 2.092263727291284e-05,
      "loss": 1.1883,
      "step": 10180
    },
    {
      "epoch": 6.237184391736802,
      "grad_norm": 2.6887598037719727,
      "learning_rate": 2.08886167245016e-05,
      "loss": 1.1845,
      "step": 10190
    },
    {
      "epoch": 6.243305279265494,
      "grad_norm": 2.5700228214263916,
      "learning_rate": 2.085459617609036e-05,
      "loss": 1.1845,
      "step": 10200
    },
    {
      "epoch": 6.249426166794185,
      "grad_norm": 2.4592013359069824,
      "learning_rate": 2.082057562767912e-05,
      "loss": 1.1918,
      "step": 10210
    },
    {
      "epoch": 6.255547054322877,
      "grad_norm": 2.9663164615631104,
      "learning_rate": 2.0786555079267876e-05,
      "loss": 1.1804,
      "step": 10220
    },
    {
      "epoch": 6.261667941851568,
      "grad_norm": 3.259889841079712,
      "learning_rate": 2.0752534530856637e-05,
      "loss": 1.2066,
      "step": 10230
    },
    {
      "epoch": 6.26778882938026,
      "grad_norm": 2.5240590572357178,
      "learning_rate": 2.0718513982445398e-05,
      "loss": 1.1703,
      "step": 10240
    },
    {
      "epoch": 6.2739097169089515,
      "grad_norm": 2.6424803733825684,
      "learning_rate": 2.068449343403416e-05,
      "loss": 1.1812,
      "step": 10250
    },
    {
      "epoch": 6.280030604437643,
      "grad_norm": 2.5683681964874268,
      "learning_rate": 2.065047288562292e-05,
      "loss": 1.207,
      "step": 10260
    },
    {
      "epoch": 6.2861514919663355,
      "grad_norm": 3.324641227722168,
      "learning_rate": 2.0616452337211677e-05,
      "loss": 1.1763,
      "step": 10270
    },
    {
      "epoch": 6.292272379495027,
      "grad_norm": 2.700704574584961,
      "learning_rate": 2.0582431788800434e-05,
      "loss": 1.1748,
      "step": 10280
    },
    {
      "epoch": 6.298393267023719,
      "grad_norm": 2.4409584999084473,
      "learning_rate": 2.0548411240389195e-05,
      "loss": 1.1982,
      "step": 10290
    },
    {
      "epoch": 6.30451415455241,
      "grad_norm": 2.4599087238311768,
      "learning_rate": 2.0514390691977956e-05,
      "loss": 1.1868,
      "step": 10300
    },
    {
      "epoch": 6.310635042081102,
      "grad_norm": 2.6646196842193604,
      "learning_rate": 2.0480370143566717e-05,
      "loss": 1.1888,
      "step": 10310
    },
    {
      "epoch": 6.316755929609793,
      "grad_norm": 2.526289463043213,
      "learning_rate": 2.0446349595155474e-05,
      "loss": 1.1866,
      "step": 10320
    },
    {
      "epoch": 6.322876817138485,
      "grad_norm": 2.692807912826538,
      "learning_rate": 2.0412329046744235e-05,
      "loss": 1.1929,
      "step": 10330
    },
    {
      "epoch": 6.328997704667176,
      "grad_norm": 4.2312846183776855,
      "learning_rate": 2.0378308498332993e-05,
      "loss": 1.1613,
      "step": 10340
    },
    {
      "epoch": 6.335118592195869,
      "grad_norm": 2.276723861694336,
      "learning_rate": 2.0344287949921754e-05,
      "loss": 1.1885,
      "step": 10350
    },
    {
      "epoch": 6.34123947972456,
      "grad_norm": 3.342312812805176,
      "learning_rate": 2.0310267401510514e-05,
      "loss": 1.1896,
      "step": 10360
    },
    {
      "epoch": 6.347360367253252,
      "grad_norm": 2.6276004314422607,
      "learning_rate": 2.0276246853099272e-05,
      "loss": 1.1788,
      "step": 10370
    },
    {
      "epoch": 6.3534812547819435,
      "grad_norm": 2.7103192806243896,
      "learning_rate": 2.0242226304688033e-05,
      "loss": 1.1777,
      "step": 10380
    },
    {
      "epoch": 6.359602142310635,
      "grad_norm": 3.6030147075653076,
      "learning_rate": 2.0208205756276793e-05,
      "loss": 1.1764,
      "step": 10390
    },
    {
      "epoch": 6.365723029839327,
      "grad_norm": 2.7777633666992188,
      "learning_rate": 2.017418520786555e-05,
      "loss": 1.1853,
      "step": 10400
    },
    {
      "epoch": 6.371843917368018,
      "grad_norm": 3.4900357723236084,
      "learning_rate": 2.0140164659454312e-05,
      "loss": 1.1888,
      "step": 10410
    },
    {
      "epoch": 6.37796480489671,
      "grad_norm": 3.1087987422943115,
      "learning_rate": 2.0106144111043073e-05,
      "loss": 1.1711,
      "step": 10420
    },
    {
      "epoch": 6.384085692425401,
      "grad_norm": 3.3824617862701416,
      "learning_rate": 2.007212356263183e-05,
      "loss": 1.1865,
      "step": 10430
    },
    {
      "epoch": 6.390206579954094,
      "grad_norm": 4.719346046447754,
      "learning_rate": 2.003810301422059e-05,
      "loss": 1.195,
      "step": 10440
    },
    {
      "epoch": 6.396327467482785,
      "grad_norm": 2.5060925483703613,
      "learning_rate": 2.000408246580935e-05,
      "loss": 1.171,
      "step": 10450
    },
    {
      "epoch": 6.402448355011477,
      "grad_norm": 2.3634145259857178,
      "learning_rate": 1.997006191739811e-05,
      "loss": 1.177,
      "step": 10460
    },
    {
      "epoch": 6.408569242540168,
      "grad_norm": 2.408721685409546,
      "learning_rate": 1.993604136898687e-05,
      "loss": 1.1695,
      "step": 10470
    },
    {
      "epoch": 6.41469013006886,
      "grad_norm": 2.7625339031219482,
      "learning_rate": 1.9902020820575627e-05,
      "loss": 1.1819,
      "step": 10480
    },
    {
      "epoch": 6.4208110175975515,
      "grad_norm": 2.217163562774658,
      "learning_rate": 1.9868000272164388e-05,
      "loss": 1.1802,
      "step": 10490
    },
    {
      "epoch": 6.426931905126243,
      "grad_norm": 2.648359775543213,
      "learning_rate": 1.983397972375315e-05,
      "loss": 1.1587,
      "step": 10500
    },
    {
      "epoch": 6.433052792654935,
      "grad_norm": 2.5863988399505615,
      "learning_rate": 1.9799959175341907e-05,
      "loss": 1.1944,
      "step": 10510
    },
    {
      "epoch": 6.439173680183627,
      "grad_norm": 2.959826707839966,
      "learning_rate": 1.9765938626930667e-05,
      "loss": 1.1812,
      "step": 10520
    },
    {
      "epoch": 6.445294567712319,
      "grad_norm": 2.727541208267212,
      "learning_rate": 1.9731918078519425e-05,
      "loss": 1.1546,
      "step": 10530
    },
    {
      "epoch": 6.45141545524101,
      "grad_norm": 2.1876068115234375,
      "learning_rate": 1.9697897530108186e-05,
      "loss": 1.1914,
      "step": 10540
    },
    {
      "epoch": 6.457536342769702,
      "grad_norm": 3.7542362213134766,
      "learning_rate": 1.9663876981696946e-05,
      "loss": 1.1692,
      "step": 10550
    },
    {
      "epoch": 6.463657230298393,
      "grad_norm": 2.905738353729248,
      "learning_rate": 1.9629856433285707e-05,
      "loss": 1.1724,
      "step": 10560
    },
    {
      "epoch": 6.469778117827085,
      "grad_norm": 3.0283126831054688,
      "learning_rate": 1.9595835884874465e-05,
      "loss": 1.1866,
      "step": 10570
    },
    {
      "epoch": 6.475899005355776,
      "grad_norm": 2.3837485313415527,
      "learning_rate": 1.9561815336463226e-05,
      "loss": 1.2031,
      "step": 10580
    },
    {
      "epoch": 6.482019892884468,
      "grad_norm": 2.557549476623535,
      "learning_rate": 1.9527794788051983e-05,
      "loss": 1.184,
      "step": 10590
    },
    {
      "epoch": 6.4881407804131594,
      "grad_norm": 2.638843059539795,
      "learning_rate": 1.9493774239640744e-05,
      "loss": 1.165,
      "step": 10600
    },
    {
      "epoch": 6.494261667941852,
      "grad_norm": 2.693385601043701,
      "learning_rate": 1.9459753691229505e-05,
      "loss": 1.1862,
      "step": 10610
    },
    {
      "epoch": 6.5003825554705434,
      "grad_norm": 3.040398597717285,
      "learning_rate": 1.9425733142818266e-05,
      "loss": 1.1936,
      "step": 10620
    },
    {
      "epoch": 6.506503442999235,
      "grad_norm": 2.483701229095459,
      "learning_rate": 1.9391712594407023e-05,
      "loss": 1.1766,
      "step": 10630
    },
    {
      "epoch": 6.5126243305279266,
      "grad_norm": 3.3055386543273926,
      "learning_rate": 1.935769204599578e-05,
      "loss": 1.1818,
      "step": 10640
    },
    {
      "epoch": 6.518745218056618,
      "grad_norm": 2.60319447517395,
      "learning_rate": 1.932367149758454e-05,
      "loss": 1.1672,
      "step": 10650
    },
    {
      "epoch": 6.52486610558531,
      "grad_norm": 2.968649387359619,
      "learning_rate": 1.9289650949173302e-05,
      "loss": 1.1689,
      "step": 10660
    },
    {
      "epoch": 6.530986993114001,
      "grad_norm": 2.3525383472442627,
      "learning_rate": 1.9255630400762063e-05,
      "loss": 1.174,
      "step": 10670
    },
    {
      "epoch": 6.537107880642694,
      "grad_norm": 2.5817339420318604,
      "learning_rate": 1.922160985235082e-05,
      "loss": 1.1747,
      "step": 10680
    },
    {
      "epoch": 6.543228768171385,
      "grad_norm": 2.791980743408203,
      "learning_rate": 1.9187589303939578e-05,
      "loss": 1.168,
      "step": 10690
    },
    {
      "epoch": 6.549349655700077,
      "grad_norm": 2.451387405395508,
      "learning_rate": 1.915356875552834e-05,
      "loss": 1.1625,
      "step": 10700
    },
    {
      "epoch": 6.555470543228768,
      "grad_norm": 3.5441620349884033,
      "learning_rate": 1.91195482071171e-05,
      "loss": 1.1739,
      "step": 10710
    },
    {
      "epoch": 6.56159143075746,
      "grad_norm": 2.8768322467803955,
      "learning_rate": 1.908552765870586e-05,
      "loss": 1.1802,
      "step": 10720
    },
    {
      "epoch": 6.567712318286151,
      "grad_norm": 3.1151416301727295,
      "learning_rate": 1.905150711029462e-05,
      "loss": 1.2072,
      "step": 10730
    },
    {
      "epoch": 6.573833205814843,
      "grad_norm": 2.7969307899475098,
      "learning_rate": 1.901748656188338e-05,
      "loss": 1.1929,
      "step": 10740
    },
    {
      "epoch": 6.5799540933435345,
      "grad_norm": 2.526097059249878,
      "learning_rate": 1.8983466013472136e-05,
      "loss": 1.1804,
      "step": 10750
    },
    {
      "epoch": 6.586074980872226,
      "grad_norm": 3.3882269859313965,
      "learning_rate": 1.8949445465060897e-05,
      "loss": 1.1857,
      "step": 10760
    },
    {
      "epoch": 6.592195868400918,
      "grad_norm": 2.8102879524230957,
      "learning_rate": 1.8915424916649658e-05,
      "loss": 1.1749,
      "step": 10770
    },
    {
      "epoch": 6.59831675592961,
      "grad_norm": 2.504945755004883,
      "learning_rate": 1.888140436823842e-05,
      "loss": 1.1949,
      "step": 10780
    },
    {
      "epoch": 6.604437643458302,
      "grad_norm": 2.4723777770996094,
      "learning_rate": 1.8847383819827176e-05,
      "loss": 1.1984,
      "step": 10790
    },
    {
      "epoch": 6.610558530986993,
      "grad_norm": 2.8720338344573975,
      "learning_rate": 1.8813363271415937e-05,
      "loss": 1.1748,
      "step": 10800
    },
    {
      "epoch": 6.616679418515685,
      "grad_norm": 2.2344985008239746,
      "learning_rate": 1.8779342723004694e-05,
      "loss": 1.1727,
      "step": 10810
    },
    {
      "epoch": 6.622800306044376,
      "grad_norm": 2.382798433303833,
      "learning_rate": 1.8745322174593455e-05,
      "loss": 1.1895,
      "step": 10820
    },
    {
      "epoch": 6.628921193573068,
      "grad_norm": 3.054856777191162,
      "learning_rate": 1.8711301626182216e-05,
      "loss": 1.1806,
      "step": 10830
    },
    {
      "epoch": 6.635042081101759,
      "grad_norm": 2.8418335914611816,
      "learning_rate": 1.8677281077770973e-05,
      "loss": 1.1927,
      "step": 10840
    },
    {
      "epoch": 6.641162968630452,
      "grad_norm": 3.4169845581054688,
      "learning_rate": 1.8643260529359734e-05,
      "loss": 1.1948,
      "step": 10850
    },
    {
      "epoch": 6.647283856159143,
      "grad_norm": 2.521684169769287,
      "learning_rate": 1.860923998094849e-05,
      "loss": 1.1669,
      "step": 10860
    },
    {
      "epoch": 6.653404743687835,
      "grad_norm": 2.434762954711914,
      "learning_rate": 1.8575219432537252e-05,
      "loss": 1.1578,
      "step": 10870
    },
    {
      "epoch": 6.6595256312165265,
      "grad_norm": 2.8318793773651123,
      "learning_rate": 1.8541198884126013e-05,
      "loss": 1.1674,
      "step": 10880
    },
    {
      "epoch": 6.665646518745218,
      "grad_norm": 2.788933038711548,
      "learning_rate": 1.8507178335714774e-05,
      "loss": 1.169,
      "step": 10890
    },
    {
      "epoch": 6.67176740627391,
      "grad_norm": 2.764143943786621,
      "learning_rate": 1.847315778730353e-05,
      "loss": 1.1838,
      "step": 10900
    },
    {
      "epoch": 6.677888293802601,
      "grad_norm": 2.8891654014587402,
      "learning_rate": 1.8439137238892292e-05,
      "loss": 1.175,
      "step": 10910
    },
    {
      "epoch": 6.684009181331293,
      "grad_norm": 2.8268089294433594,
      "learning_rate": 1.840511669048105e-05,
      "loss": 1.1808,
      "step": 10920
    },
    {
      "epoch": 6.690130068859984,
      "grad_norm": 2.4628899097442627,
      "learning_rate": 1.837109614206981e-05,
      "loss": 1.187,
      "step": 10930
    },
    {
      "epoch": 6.696250956388677,
      "grad_norm": 2.6242809295654297,
      "learning_rate": 1.833707559365857e-05,
      "loss": 1.1686,
      "step": 10940
    },
    {
      "epoch": 6.702371843917368,
      "grad_norm": 2.548687696456909,
      "learning_rate": 1.830305504524733e-05,
      "loss": 1.1736,
      "step": 10950
    },
    {
      "epoch": 6.70849273144606,
      "grad_norm": 3.2288942337036133,
      "learning_rate": 1.826903449683609e-05,
      "loss": 1.174,
      "step": 10960
    },
    {
      "epoch": 6.714613618974751,
      "grad_norm": 3.222327947616577,
      "learning_rate": 1.823501394842485e-05,
      "loss": 1.1773,
      "step": 10970
    },
    {
      "epoch": 6.720734506503443,
      "grad_norm": 3.697409152984619,
      "learning_rate": 1.8200993400013608e-05,
      "loss": 1.1763,
      "step": 10980
    },
    {
      "epoch": 6.7268553940321345,
      "grad_norm": 2.843804359436035,
      "learning_rate": 1.816697285160237e-05,
      "loss": 1.1649,
      "step": 10990
    },
    {
      "epoch": 6.732976281560826,
      "grad_norm": 2.6715619564056396,
      "learning_rate": 1.8132952303191126e-05,
      "loss": 1.1723,
      "step": 11000
    },
    {
      "epoch": 6.739097169089518,
      "grad_norm": 6.662703514099121,
      "learning_rate": 1.8098931754779887e-05,
      "loss": 1.1733,
      "step": 11010
    },
    {
      "epoch": 6.74521805661821,
      "grad_norm": 2.99680757522583,
      "learning_rate": 1.8064911206368648e-05,
      "loss": 1.1768,
      "step": 11020
    },
    {
      "epoch": 6.751338944146902,
      "grad_norm": 2.3245575428009033,
      "learning_rate": 1.803089065795741e-05,
      "loss": 1.1777,
      "step": 11030
    },
    {
      "epoch": 6.757459831675593,
      "grad_norm": 2.462326765060425,
      "learning_rate": 1.7996870109546166e-05,
      "loss": 1.1786,
      "step": 11040
    },
    {
      "epoch": 6.763580719204285,
      "grad_norm": 2.6256535053253174,
      "learning_rate": 1.7962849561134927e-05,
      "loss": 1.1995,
      "step": 11050
    },
    {
      "epoch": 6.769701606732976,
      "grad_norm": 3.1199820041656494,
      "learning_rate": 1.7928829012723685e-05,
      "loss": 1.1864,
      "step": 11060
    },
    {
      "epoch": 6.775822494261668,
      "grad_norm": 2.832915782928467,
      "learning_rate": 1.7894808464312445e-05,
      "loss": 1.1835,
      "step": 11070
    },
    {
      "epoch": 6.781943381790359,
      "grad_norm": 3.149301052093506,
      "learning_rate": 1.7860787915901206e-05,
      "loss": 1.1758,
      "step": 11080
    },
    {
      "epoch": 6.788064269319051,
      "grad_norm": 2.3810904026031494,
      "learning_rate": 1.7826767367489967e-05,
      "loss": 1.1814,
      "step": 11090
    },
    {
      "epoch": 6.7941851568477425,
      "grad_norm": 3.023635149002075,
      "learning_rate": 1.7792746819078725e-05,
      "loss": 1.2116,
      "step": 11100
    },
    {
      "epoch": 6.800306044376435,
      "grad_norm": 2.3528966903686523,
      "learning_rate": 1.7758726270667482e-05,
      "loss": 1.1843,
      "step": 11110
    },
    {
      "epoch": 6.8064269319051265,
      "grad_norm": 2.8839402198791504,
      "learning_rate": 1.7724705722256243e-05,
      "loss": 1.1801,
      "step": 11120
    },
    {
      "epoch": 6.812547819433818,
      "grad_norm": 2.401827573776245,
      "learning_rate": 1.7690685173845004e-05,
      "loss": 1.1914,
      "step": 11130
    },
    {
      "epoch": 6.81866870696251,
      "grad_norm": 2.917004346847534,
      "learning_rate": 1.7656664625433764e-05,
      "loss": 1.1822,
      "step": 11140
    },
    {
      "epoch": 6.824789594491201,
      "grad_norm": 2.9332504272460938,
      "learning_rate": 1.7622644077022525e-05,
      "loss": 1.1879,
      "step": 11150
    },
    {
      "epoch": 6.830910482019893,
      "grad_norm": 2.532586097717285,
      "learning_rate": 1.758862352861128e-05,
      "loss": 1.1712,
      "step": 11160
    },
    {
      "epoch": 6.837031369548584,
      "grad_norm": 2.5997772216796875,
      "learning_rate": 1.755460298020004e-05,
      "loss": 1.1806,
      "step": 11170
    },
    {
      "epoch": 6.843152257077277,
      "grad_norm": 2.8061439990997314,
      "learning_rate": 1.75205824317888e-05,
      "loss": 1.1824,
      "step": 11180
    },
    {
      "epoch": 6.849273144605968,
      "grad_norm": 2.2987356185913086,
      "learning_rate": 1.7486561883377562e-05,
      "loss": 1.1849,
      "step": 11190
    },
    {
      "epoch": 6.85539403213466,
      "grad_norm": 2.5757126808166504,
      "learning_rate": 1.7452541334966323e-05,
      "loss": 1.1794,
      "step": 11200
    },
    {
      "epoch": 6.861514919663351,
      "grad_norm": 3.3306543827056885,
      "learning_rate": 1.741852078655508e-05,
      "loss": 1.1942,
      "step": 11210
    },
    {
      "epoch": 6.867635807192043,
      "grad_norm": 2.8044543266296387,
      "learning_rate": 1.7384500238143838e-05,
      "loss": 1.1695,
      "step": 11220
    },
    {
      "epoch": 6.8737566947207345,
      "grad_norm": 2.9289653301239014,
      "learning_rate": 1.73504796897326e-05,
      "loss": 1.165,
      "step": 11230
    },
    {
      "epoch": 6.879877582249426,
      "grad_norm": 2.7337334156036377,
      "learning_rate": 1.731645914132136e-05,
      "loss": 1.1732,
      "step": 11240
    },
    {
      "epoch": 6.885998469778118,
      "grad_norm": 2.9211790561676025,
      "learning_rate": 1.728243859291012e-05,
      "loss": 1.1905,
      "step": 11250
    },
    {
      "epoch": 6.892119357306809,
      "grad_norm": 2.608776092529297,
      "learning_rate": 1.7248418044498878e-05,
      "loss": 1.1794,
      "step": 11260
    },
    {
      "epoch": 6.898240244835501,
      "grad_norm": 2.365649938583374,
      "learning_rate": 1.7214397496087635e-05,
      "loss": 1.1882,
      "step": 11270
    },
    {
      "epoch": 6.904361132364193,
      "grad_norm": 2.889016628265381,
      "learning_rate": 1.7180376947676396e-05,
      "loss": 1.1644,
      "step": 11280
    },
    {
      "epoch": 6.910482019892885,
      "grad_norm": 2.8026652336120605,
      "learning_rate": 1.7146356399265157e-05,
      "loss": 1.1785,
      "step": 11290
    },
    {
      "epoch": 6.916602907421576,
      "grad_norm": 2.359157085418701,
      "learning_rate": 1.7112335850853917e-05,
      "loss": 1.1581,
      "step": 11300
    },
    {
      "epoch": 6.922723794950268,
      "grad_norm": 2.2410833835601807,
      "learning_rate": 1.707831530244268e-05,
      "loss": 1.1835,
      "step": 11310
    },
    {
      "epoch": 6.928844682478959,
      "grad_norm": 3.33797287940979,
      "learning_rate": 1.7044294754031436e-05,
      "loss": 1.2033,
      "step": 11320
    },
    {
      "epoch": 6.934965570007651,
      "grad_norm": 2.6676926612854004,
      "learning_rate": 1.7010274205620193e-05,
      "loss": 1.1759,
      "step": 11330
    },
    {
      "epoch": 6.9410864575363425,
      "grad_norm": 2.5464632511138916,
      "learning_rate": 1.6976253657208954e-05,
      "loss": 1.178,
      "step": 11340
    },
    {
      "epoch": 6.947207345065035,
      "grad_norm": 2.017395496368408,
      "learning_rate": 1.6942233108797715e-05,
      "loss": 1.1843,
      "step": 11350
    },
    {
      "epoch": 6.9533282325937265,
      "grad_norm": 3.047525405883789,
      "learning_rate": 1.6908212560386476e-05,
      "loss": 1.1589,
      "step": 11360
    },
    {
      "epoch": 6.959449120122418,
      "grad_norm": 3.0717241764068604,
      "learning_rate": 1.6874192011975233e-05,
      "loss": 1.1871,
      "step": 11370
    },
    {
      "epoch": 6.96557000765111,
      "grad_norm": 3.1424989700317383,
      "learning_rate": 1.6840171463563994e-05,
      "loss": 1.179,
      "step": 11380
    },
    {
      "epoch": 6.971690895179801,
      "grad_norm": 2.526454210281372,
      "learning_rate": 1.680615091515275e-05,
      "loss": 1.1832,
      "step": 11390
    },
    {
      "epoch": 6.977811782708493,
      "grad_norm": 2.4816434383392334,
      "learning_rate": 1.6772130366741512e-05,
      "loss": 1.1773,
      "step": 11400
    },
    {
      "epoch": 6.983932670237184,
      "grad_norm": 2.4473142623901367,
      "learning_rate": 1.6738109818330273e-05,
      "loss": 1.1609,
      "step": 11410
    },
    {
      "epoch": 6.990053557765876,
      "grad_norm": 4.933506488800049,
      "learning_rate": 1.670408926991903e-05,
      "loss": 1.1895,
      "step": 11420
    },
    {
      "epoch": 6.996174445294567,
      "grad_norm": 2.4855034351348877,
      "learning_rate": 1.667006872150779e-05,
      "loss": 1.1753,
      "step": 11430
    },
    {
      "epoch": 6.9998469778117824,
      "eval_accuracy": 0.506212457435819,
      "eval_loss": 1.1684563159942627,
      "eval_runtime": 2638.0369,
      "eval_samples_per_second": 79.262,
      "eval_steps_per_second": 0.619,
      "step": 11436
    },
    {
      "epoch": 7.00229533282326,
      "grad_norm": 2.4456984996795654,
      "learning_rate": 1.6636048173096552e-05,
      "loss": 1.1607,
      "step": 11440
    },
    {
      "epoch": 7.008416220351951,
      "grad_norm": 2.768742322921753,
      "learning_rate": 1.660202762468531e-05,
      "loss": 1.1586,
      "step": 11450
    },
    {
      "epoch": 7.014537107880643,
      "grad_norm": 6.783010482788086,
      "learning_rate": 1.656800707627407e-05,
      "loss": 1.1458,
      "step": 11460
    },
    {
      "epoch": 7.020657995409334,
      "grad_norm": 3.2379367351531982,
      "learning_rate": 1.653398652786283e-05,
      "loss": 1.156,
      "step": 11470
    },
    {
      "epoch": 7.026778882938026,
      "grad_norm": 2.605741262435913,
      "learning_rate": 1.649996597945159e-05,
      "loss": 1.1538,
      "step": 11480
    },
    {
      "epoch": 7.0328997704667175,
      "grad_norm": 2.8403258323669434,
      "learning_rate": 1.646594543104035e-05,
      "loss": 1.1369,
      "step": 11490
    },
    {
      "epoch": 7.039020657995409,
      "grad_norm": 2.7052204608917236,
      "learning_rate": 1.643192488262911e-05,
      "loss": 1.1392,
      "step": 11500
    },
    {
      "epoch": 7.045141545524101,
      "grad_norm": 2.6772308349609375,
      "learning_rate": 1.6397904334217868e-05,
      "loss": 1.159,
      "step": 11510
    },
    {
      "epoch": 7.051262433052793,
      "grad_norm": 2.957548141479492,
      "learning_rate": 1.636388378580663e-05,
      "loss": 1.147,
      "step": 11520
    },
    {
      "epoch": 7.057383320581485,
      "grad_norm": 3.451127767562866,
      "learning_rate": 1.6329863237395386e-05,
      "loss": 1.1674,
      "step": 11530
    },
    {
      "epoch": 7.063504208110176,
      "grad_norm": 2.7239480018615723,
      "learning_rate": 1.6295842688984147e-05,
      "loss": 1.1764,
      "step": 11540
    },
    {
      "epoch": 7.069625095638868,
      "grad_norm": 2.3374781608581543,
      "learning_rate": 1.6261822140572908e-05,
      "loss": 1.1757,
      "step": 11550
    },
    {
      "epoch": 7.075745983167559,
      "grad_norm": 3.032133102416992,
      "learning_rate": 1.622780159216167e-05,
      "loss": 1.1412,
      "step": 11560
    },
    {
      "epoch": 7.081866870696251,
      "grad_norm": 2.7997188568115234,
      "learning_rate": 1.6193781043750426e-05,
      "loss": 1.1475,
      "step": 11570
    },
    {
      "epoch": 7.087987758224942,
      "grad_norm": 2.812303304672241,
      "learning_rate": 1.6159760495339184e-05,
      "loss": 1.1512,
      "step": 11580
    },
    {
      "epoch": 7.094108645753634,
      "grad_norm": 3.3279213905334473,
      "learning_rate": 1.6125739946927944e-05,
      "loss": 1.1652,
      "step": 11590
    },
    {
      "epoch": 7.1002295332823255,
      "grad_norm": 3.0579566955566406,
      "learning_rate": 1.6091719398516705e-05,
      "loss": 1.1648,
      "step": 11600
    },
    {
      "epoch": 7.106350420811018,
      "grad_norm": 3.063218355178833,
      "learning_rate": 1.6057698850105466e-05,
      "loss": 1.1553,
      "step": 11610
    },
    {
      "epoch": 7.1124713083397095,
      "grad_norm": 4.288387775421143,
      "learning_rate": 1.6023678301694227e-05,
      "loss": 1.1675,
      "step": 11620
    },
    {
      "epoch": 7.118592195868401,
      "grad_norm": 2.884408473968506,
      "learning_rate": 1.5989657753282984e-05,
      "loss": 1.1605,
      "step": 11630
    },
    {
      "epoch": 7.124713083397093,
      "grad_norm": 3.101125478744507,
      "learning_rate": 1.5955637204871742e-05,
      "loss": 1.1643,
      "step": 11640
    },
    {
      "epoch": 7.130833970925784,
      "grad_norm": 2.6853866577148438,
      "learning_rate": 1.5921616656460503e-05,
      "loss": 1.1542,
      "step": 11650
    },
    {
      "epoch": 7.136954858454476,
      "grad_norm": 3.852677822113037,
      "learning_rate": 1.5887596108049263e-05,
      "loss": 1.1571,
      "step": 11660
    },
    {
      "epoch": 7.143075745983167,
      "grad_norm": 3.694631814956665,
      "learning_rate": 1.5853575559638024e-05,
      "loss": 1.1841,
      "step": 11670
    },
    {
      "epoch": 7.149196633511859,
      "grad_norm": 3.0455989837646484,
      "learning_rate": 1.5819555011226782e-05,
      "loss": 1.1337,
      "step": 11680
    },
    {
      "epoch": 7.155317521040551,
      "grad_norm": 2.740424394607544,
      "learning_rate": 1.578553446281554e-05,
      "loss": 1.1515,
      "step": 11690
    },
    {
      "epoch": 7.161438408569243,
      "grad_norm": 2.8961853981018066,
      "learning_rate": 1.57515139144043e-05,
      "loss": 1.1851,
      "step": 11700
    },
    {
      "epoch": 7.167559296097934,
      "grad_norm": 2.4369847774505615,
      "learning_rate": 1.571749336599306e-05,
      "loss": 1.1595,
      "step": 11710
    },
    {
      "epoch": 7.173680183626626,
      "grad_norm": 2.7445836067199707,
      "learning_rate": 1.568347281758182e-05,
      "loss": 1.15,
      "step": 11720
    },
    {
      "epoch": 7.1798010711553175,
      "grad_norm": 2.5091135501861572,
      "learning_rate": 1.5649452269170582e-05,
      "loss": 1.173,
      "step": 11730
    },
    {
      "epoch": 7.185921958684009,
      "grad_norm": 2.7890584468841553,
      "learning_rate": 1.5615431720759337e-05,
      "loss": 1.1538,
      "step": 11740
    },
    {
      "epoch": 7.192042846212701,
      "grad_norm": 2.6841843128204346,
      "learning_rate": 1.5581411172348097e-05,
      "loss": 1.1627,
      "step": 11750
    },
    {
      "epoch": 7.198163733741392,
      "grad_norm": 3.1628477573394775,
      "learning_rate": 1.5547390623936858e-05,
      "loss": 1.1598,
      "step": 11760
    },
    {
      "epoch": 7.204284621270084,
      "grad_norm": 3.3740761280059814,
      "learning_rate": 1.551337007552562e-05,
      "loss": 1.1385,
      "step": 11770
    },
    {
      "epoch": 7.210405508798776,
      "grad_norm": 3.2839245796203613,
      "learning_rate": 1.547934952711438e-05,
      "loss": 1.1818,
      "step": 11780
    },
    {
      "epoch": 7.216526396327468,
      "grad_norm": 2.728667736053467,
      "learning_rate": 1.5445328978703137e-05,
      "loss": 1.1555,
      "step": 11790
    },
    {
      "epoch": 7.222647283856159,
      "grad_norm": 2.8254268169403076,
      "learning_rate": 1.5411308430291895e-05,
      "loss": 1.1651,
      "step": 11800
    },
    {
      "epoch": 7.228768171384851,
      "grad_norm": 2.361497640609741,
      "learning_rate": 1.5377287881880656e-05,
      "loss": 1.1478,
      "step": 11810
    },
    {
      "epoch": 7.234889058913542,
      "grad_norm": 2.7829697132110596,
      "learning_rate": 1.5343267333469416e-05,
      "loss": 1.1575,
      "step": 11820
    },
    {
      "epoch": 7.241009946442234,
      "grad_norm": 3.025606870651245,
      "learning_rate": 1.5309246785058177e-05,
      "loss": 1.1693,
      "step": 11830
    },
    {
      "epoch": 7.2471308339709255,
      "grad_norm": 2.929812431335449,
      "learning_rate": 1.5275226236646935e-05,
      "loss": 1.1697,
      "step": 11840
    },
    {
      "epoch": 7.253251721499617,
      "grad_norm": 2.4072935581207275,
      "learning_rate": 1.5241205688235696e-05,
      "loss": 1.1605,
      "step": 11850
    },
    {
      "epoch": 7.2593726090283095,
      "grad_norm": 3.0431549549102783,
      "learning_rate": 1.5207185139824453e-05,
      "loss": 1.1653,
      "step": 11860
    },
    {
      "epoch": 7.265493496557001,
      "grad_norm": 2.6219193935394287,
      "learning_rate": 1.5173164591413214e-05,
      "loss": 1.1629,
      "step": 11870
    },
    {
      "epoch": 7.271614384085693,
      "grad_norm": 4.558614253997803,
      "learning_rate": 1.5139144043001973e-05,
      "loss": 1.1462,
      "step": 11880
    },
    {
      "epoch": 7.277735271614384,
      "grad_norm": 3.1963789463043213,
      "learning_rate": 1.5105123494590734e-05,
      "loss": 1.1594,
      "step": 11890
    },
    {
      "epoch": 7.283856159143076,
      "grad_norm": 2.740114212036133,
      "learning_rate": 1.5071102946179495e-05,
      "loss": 1.1599,
      "step": 11900
    },
    {
      "epoch": 7.289977046671767,
      "grad_norm": 2.688758134841919,
      "learning_rate": 1.5037082397768254e-05,
      "loss": 1.1705,
      "step": 11910
    },
    {
      "epoch": 7.296097934200459,
      "grad_norm": 3.089334011077881,
      "learning_rate": 1.5003061849357011e-05,
      "loss": 1.1837,
      "step": 11920
    },
    {
      "epoch": 7.30221882172915,
      "grad_norm": 2.151132822036743,
      "learning_rate": 1.4969041300945772e-05,
      "loss": 1.1793,
      "step": 11930
    },
    {
      "epoch": 7.308339709257842,
      "grad_norm": 2.6404690742492676,
      "learning_rate": 1.4935020752534531e-05,
      "loss": 1.1483,
      "step": 11940
    },
    {
      "epoch": 7.314460596786534,
      "grad_norm": 3.6337945461273193,
      "learning_rate": 1.4901000204123292e-05,
      "loss": 1.178,
      "step": 11950
    },
    {
      "epoch": 7.320581484315226,
      "grad_norm": 3.0208799839019775,
      "learning_rate": 1.4866979655712051e-05,
      "loss": 1.1774,
      "step": 11960
    },
    {
      "epoch": 7.3267023718439175,
      "grad_norm": 2.9319422245025635,
      "learning_rate": 1.4832959107300812e-05,
      "loss": 1.1627,
      "step": 11970
    },
    {
      "epoch": 7.332823259372609,
      "grad_norm": 2.71775484085083,
      "learning_rate": 1.479893855888957e-05,
      "loss": 1.1509,
      "step": 11980
    },
    {
      "epoch": 7.338944146901301,
      "grad_norm": 2.7247722148895264,
      "learning_rate": 1.4764918010478329e-05,
      "loss": 1.151,
      "step": 11990
    },
    {
      "epoch": 7.345065034429992,
      "grad_norm": 2.7740538120269775,
      "learning_rate": 1.473089746206709e-05,
      "loss": 1.1558,
      "step": 12000
    },
    {
      "epoch": 7.351185921958684,
      "grad_norm": 2.2844157218933105,
      "learning_rate": 1.4696876913655849e-05,
      "loss": 1.1709,
      "step": 12010
    },
    {
      "epoch": 7.357306809487375,
      "grad_norm": 2.582582950592041,
      "learning_rate": 1.466285636524461e-05,
      "loss": 1.1666,
      "step": 12020
    },
    {
      "epoch": 7.363427697016068,
      "grad_norm": 2.9742329120635986,
      "learning_rate": 1.4628835816833367e-05,
      "loss": 1.1523,
      "step": 12030
    },
    {
      "epoch": 7.369548584544759,
      "grad_norm": 3.2235240936279297,
      "learning_rate": 1.4594815268422126e-05,
      "loss": 1.1655,
      "step": 12040
    },
    {
      "epoch": 7.375669472073451,
      "grad_norm": 2.9418814182281494,
      "learning_rate": 1.4560794720010887e-05,
      "loss": 1.1478,
      "step": 12050
    },
    {
      "epoch": 7.381790359602142,
      "grad_norm": 2.813707113265991,
      "learning_rate": 1.4526774171599648e-05,
      "loss": 1.1548,
      "step": 12060
    },
    {
      "epoch": 7.387911247130834,
      "grad_norm": 3.019864082336426,
      "learning_rate": 1.4492753623188407e-05,
      "loss": 1.1583,
      "step": 12070
    },
    {
      "epoch": 7.3940321346595255,
      "grad_norm": 2.753734588623047,
      "learning_rate": 1.4458733074777168e-05,
      "loss": 1.1695,
      "step": 12080
    },
    {
      "epoch": 7.400153022188217,
      "grad_norm": 2.878682851791382,
      "learning_rate": 1.4424712526365925e-05,
      "loss": 1.1662,
      "step": 12090
    },
    {
      "epoch": 7.406273909716909,
      "grad_norm": 2.875643730163574,
      "learning_rate": 1.4390691977954684e-05,
      "loss": 1.1695,
      "step": 12100
    },
    {
      "epoch": 7.412394797245601,
      "grad_norm": 2.6274614334106445,
      "learning_rate": 1.4356671429543445e-05,
      "loss": 1.1682,
      "step": 12110
    },
    {
      "epoch": 7.418515684774293,
      "grad_norm": 3.541102647781372,
      "learning_rate": 1.4322650881132204e-05,
      "loss": 1.1555,
      "step": 12120
    },
    {
      "epoch": 7.424636572302984,
      "grad_norm": 2.542127847671509,
      "learning_rate": 1.4288630332720965e-05,
      "loss": 1.1637,
      "step": 12130
    },
    {
      "epoch": 7.430757459831676,
      "grad_norm": 3.612274169921875,
      "learning_rate": 1.4254609784309724e-05,
      "loss": 1.1533,
      "step": 12140
    },
    {
      "epoch": 7.436878347360367,
      "grad_norm": 2.997279405593872,
      "learning_rate": 1.4220589235898482e-05,
      "loss": 1.1724,
      "step": 12150
    },
    {
      "epoch": 7.442999234889059,
      "grad_norm": 2.5861494541168213,
      "learning_rate": 1.4186568687487242e-05,
      "loss": 1.1835,
      "step": 12160
    },
    {
      "epoch": 7.44912012241775,
      "grad_norm": 2.768721342086792,
      "learning_rate": 1.4152548139076002e-05,
      "loss": 1.1371,
      "step": 12170
    },
    {
      "epoch": 7.455241009946442,
      "grad_norm": 3.0678112506866455,
      "learning_rate": 1.4118527590664762e-05,
      "loss": 1.1829,
      "step": 12180
    },
    {
      "epoch": 7.461361897475134,
      "grad_norm": 2.529768228530884,
      "learning_rate": 1.4084507042253523e-05,
      "loss": 1.1469,
      "step": 12190
    },
    {
      "epoch": 7.467482785003826,
      "grad_norm": 2.7529709339141846,
      "learning_rate": 1.4050486493842282e-05,
      "loss": 1.1494,
      "step": 12200
    },
    {
      "epoch": 7.4736036725325175,
      "grad_norm": 3.012122631072998,
      "learning_rate": 1.401646594543104e-05,
      "loss": 1.1716,
      "step": 12210
    },
    {
      "epoch": 7.479724560061209,
      "grad_norm": 2.9628913402557373,
      "learning_rate": 1.39824453970198e-05,
      "loss": 1.1623,
      "step": 12220
    },
    {
      "epoch": 7.485845447589901,
      "grad_norm": 2.6262001991271973,
      "learning_rate": 1.394842484860856e-05,
      "loss": 1.1684,
      "step": 12230
    },
    {
      "epoch": 7.491966335118592,
      "grad_norm": 3.49383544921875,
      "learning_rate": 1.391440430019732e-05,
      "loss": 1.1686,
      "step": 12240
    },
    {
      "epoch": 7.498087222647284,
      "grad_norm": 2.6639795303344727,
      "learning_rate": 1.388038375178608e-05,
      "loss": 1.1402,
      "step": 12250
    },
    {
      "epoch": 7.504208110175975,
      "grad_norm": 3.079808235168457,
      "learning_rate": 1.384636320337484e-05,
      "loss": 1.168,
      "step": 12260
    },
    {
      "epoch": 7.510328997704667,
      "grad_norm": 2.770954132080078,
      "learning_rate": 1.3812342654963598e-05,
      "loss": 1.1693,
      "step": 12270
    },
    {
      "epoch": 7.516449885233359,
      "grad_norm": 3.519986629486084,
      "learning_rate": 1.3778322106552357e-05,
      "loss": 1.147,
      "step": 12280
    },
    {
      "epoch": 7.522570772762051,
      "grad_norm": 2.8663604259490967,
      "learning_rate": 1.3744301558141118e-05,
      "loss": 1.1609,
      "step": 12290
    },
    {
      "epoch": 7.528691660290742,
      "grad_norm": 3.6388792991638184,
      "learning_rate": 1.3710281009729877e-05,
      "loss": 1.1953,
      "step": 12300
    },
    {
      "epoch": 7.534812547819434,
      "grad_norm": 2.5936012268066406,
      "learning_rate": 1.3676260461318638e-05,
      "loss": 1.1634,
      "step": 12310
    },
    {
      "epoch": 7.540933435348125,
      "grad_norm": 2.83449387550354,
      "learning_rate": 1.3642239912907399e-05,
      "loss": 1.1716,
      "step": 12320
    },
    {
      "epoch": 7.547054322876817,
      "grad_norm": 3.5120432376861572,
      "learning_rate": 1.3608219364496155e-05,
      "loss": 1.179,
      "step": 12330
    },
    {
      "epoch": 7.5531752104055085,
      "grad_norm": 2.879821300506592,
      "learning_rate": 1.3574198816084915e-05,
      "loss": 1.1607,
      "step": 12340
    },
    {
      "epoch": 7.5592960979342,
      "grad_norm": 3.0477402210235596,
      "learning_rate": 1.3540178267673676e-05,
      "loss": 1.1605,
      "step": 12350
    },
    {
      "epoch": 7.5654169854628925,
      "grad_norm": 2.9386744499206543,
      "learning_rate": 1.3506157719262435e-05,
      "loss": 1.1582,
      "step": 12360
    },
    {
      "epoch": 7.571537872991584,
      "grad_norm": 2.8157691955566406,
      "learning_rate": 1.3472137170851196e-05,
      "loss": 1.1514,
      "step": 12370
    },
    {
      "epoch": 7.577658760520276,
      "grad_norm": 3.2182819843292236,
      "learning_rate": 1.3438116622439955e-05,
      "loss": 1.1571,
      "step": 12380
    },
    {
      "epoch": 7.583779648048967,
      "grad_norm": 3.252556562423706,
      "learning_rate": 1.3404096074028713e-05,
      "loss": 1.1616,
      "step": 12390
    },
    {
      "epoch": 7.589900535577659,
      "grad_norm": 2.7979633808135986,
      "learning_rate": 1.3370075525617474e-05,
      "loss": 1.1643,
      "step": 12400
    },
    {
      "epoch": 7.59602142310635,
      "grad_norm": 2.8204915523529053,
      "learning_rate": 1.3336054977206233e-05,
      "loss": 1.1598,
      "step": 12410
    },
    {
      "epoch": 7.602142310635042,
      "grad_norm": 2.7400918006896973,
      "learning_rate": 1.3302034428794994e-05,
      "loss": 1.1792,
      "step": 12420
    },
    {
      "epoch": 7.608263198163733,
      "grad_norm": 2.467127561569214,
      "learning_rate": 1.3268013880383753e-05,
      "loss": 1.1724,
      "step": 12430
    },
    {
      "epoch": 7.614384085692425,
      "grad_norm": 2.994223117828369,
      "learning_rate": 1.323399333197251e-05,
      "loss": 1.1749,
      "step": 12440
    },
    {
      "epoch": 7.620504973221117,
      "grad_norm": 2.5678718090057373,
      "learning_rate": 1.3199972783561271e-05,
      "loss": 1.1516,
      "step": 12450
    },
    {
      "epoch": 7.626625860749809,
      "grad_norm": 2.963792085647583,
      "learning_rate": 1.316595223515003e-05,
      "loss": 1.1915,
      "step": 12460
    },
    {
      "epoch": 7.6327467482785005,
      "grad_norm": 3.1967992782592773,
      "learning_rate": 1.3131931686738791e-05,
      "loss": 1.1713,
      "step": 12470
    },
    {
      "epoch": 7.638867635807192,
      "grad_norm": 2.8124046325683594,
      "learning_rate": 1.3097911138327552e-05,
      "loss": 1.1524,
      "step": 12480
    },
    {
      "epoch": 7.644988523335884,
      "grad_norm": 2.9094791412353516,
      "learning_rate": 1.3063890589916311e-05,
      "loss": 1.1482,
      "step": 12490
    },
    {
      "epoch": 7.651109410864575,
      "grad_norm": 3.05389142036438,
      "learning_rate": 1.3029870041505068e-05,
      "loss": 1.1529,
      "step": 12500
    },
    {
      "epoch": 7.657230298393267,
      "grad_norm": 2.79539155960083,
      "learning_rate": 1.299584949309383e-05,
      "loss": 1.1594,
      "step": 12510
    },
    {
      "epoch": 7.663351185921958,
      "grad_norm": 2.77883243560791,
      "learning_rate": 1.2961828944682588e-05,
      "loss": 1.1402,
      "step": 12520
    },
    {
      "epoch": 7.669472073450651,
      "grad_norm": 3.6547787189483643,
      "learning_rate": 1.292780839627135e-05,
      "loss": 1.1764,
      "step": 12530
    },
    {
      "epoch": 7.675592960979342,
      "grad_norm": 3.068427801132202,
      "learning_rate": 1.2893787847860108e-05,
      "loss": 1.1619,
      "step": 12540
    },
    {
      "epoch": 7.681713848508034,
      "grad_norm": 3.1389434337615967,
      "learning_rate": 1.285976729944887e-05,
      "loss": 1.1603,
      "step": 12550
    },
    {
      "epoch": 7.687834736036725,
      "grad_norm": 2.6703555583953857,
      "learning_rate": 1.2825746751037627e-05,
      "loss": 1.1638,
      "step": 12560
    },
    {
      "epoch": 7.693955623565417,
      "grad_norm": 3.204761028289795,
      "learning_rate": 1.2791726202626386e-05,
      "loss": 1.1702,
      "step": 12570
    },
    {
      "epoch": 7.7000765110941085,
      "grad_norm": 2.758955717086792,
      "learning_rate": 1.2757705654215147e-05,
      "loss": 1.1335,
      "step": 12580
    },
    {
      "epoch": 7.7061973986228,
      "grad_norm": 2.848846197128296,
      "learning_rate": 1.2723685105803906e-05,
      "loss": 1.1589,
      "step": 12590
    },
    {
      "epoch": 7.712318286151492,
      "grad_norm": 4.4077630043029785,
      "learning_rate": 1.2689664557392667e-05,
      "loss": 1.128,
      "step": 12600
    },
    {
      "epoch": 7.718439173680183,
      "grad_norm": 2.898097038269043,
      "learning_rate": 1.2655644008981426e-05,
      "loss": 1.1513,
      "step": 12610
    },
    {
      "epoch": 7.724560061208876,
      "grad_norm": 3.664088249206543,
      "learning_rate": 1.2621623460570183e-05,
      "loss": 1.1615,
      "step": 12620
    },
    {
      "epoch": 7.730680948737567,
      "grad_norm": 2.5493152141571045,
      "learning_rate": 1.2587602912158944e-05,
      "loss": 1.1498,
      "step": 12630
    },
    {
      "epoch": 7.736801836266259,
      "grad_norm": 3.5458126068115234,
      "learning_rate": 1.2553582363747705e-05,
      "loss": 1.1375,
      "step": 12640
    },
    {
      "epoch": 7.74292272379495,
      "grad_norm": 2.8673930168151855,
      "learning_rate": 1.2519561815336464e-05,
      "loss": 1.1696,
      "step": 12650
    },
    {
      "epoch": 7.749043611323642,
      "grad_norm": 2.6049365997314453,
      "learning_rate": 1.2485541266925223e-05,
      "loss": 1.1786,
      "step": 12660
    },
    {
      "epoch": 7.755164498852333,
      "grad_norm": 3.1647322177886963,
      "learning_rate": 1.2451520718513982e-05,
      "loss": 1.1593,
      "step": 12670
    },
    {
      "epoch": 7.761285386381025,
      "grad_norm": 4.2167649269104,
      "learning_rate": 1.2417500170102743e-05,
      "loss": 1.1604,
      "step": 12680
    },
    {
      "epoch": 7.767406273909717,
      "grad_norm": 3.039738655090332,
      "learning_rate": 1.2383479621691502e-05,
      "loss": 1.162,
      "step": 12690
    },
    {
      "epoch": 7.773527161438409,
      "grad_norm": 4.248793125152588,
      "learning_rate": 1.2349459073280261e-05,
      "loss": 1.1537,
      "step": 12700
    },
    {
      "epoch": 7.7796480489671005,
      "grad_norm": 2.4704842567443848,
      "learning_rate": 1.2315438524869022e-05,
      "loss": 1.1563,
      "step": 12710
    },
    {
      "epoch": 7.785768936495792,
      "grad_norm": 3.5454394817352295,
      "learning_rate": 1.2281417976457781e-05,
      "loss": 1.1546,
      "step": 12720
    },
    {
      "epoch": 7.791889824024484,
      "grad_norm": 2.821631669998169,
      "learning_rate": 1.224739742804654e-05,
      "loss": 1.1563,
      "step": 12730
    },
    {
      "epoch": 7.798010711553175,
      "grad_norm": 2.692189931869507,
      "learning_rate": 1.2213376879635301e-05,
      "loss": 1.1658,
      "step": 12740
    },
    {
      "epoch": 7.804131599081867,
      "grad_norm": 2.9162681102752686,
      "learning_rate": 1.2179356331224059e-05,
      "loss": 1.176,
      "step": 12750
    },
    {
      "epoch": 7.810252486610558,
      "grad_norm": 3.638213872909546,
      "learning_rate": 1.214533578281282e-05,
      "loss": 1.165,
      "step": 12760
    },
    {
      "epoch": 7.81637337413925,
      "grad_norm": 3.419288158416748,
      "learning_rate": 1.2111315234401579e-05,
      "loss": 1.1561,
      "step": 12770
    },
    {
      "epoch": 7.822494261667941,
      "grad_norm": 2.933861255645752,
      "learning_rate": 1.2077294685990338e-05,
      "loss": 1.1663,
      "step": 12780
    },
    {
      "epoch": 7.828615149196634,
      "grad_norm": 2.9365577697753906,
      "learning_rate": 1.2043274137579099e-05,
      "loss": 1.1514,
      "step": 12790
    },
    {
      "epoch": 7.834736036725325,
      "grad_norm": 2.736703634262085,
      "learning_rate": 1.2009253589167858e-05,
      "loss": 1.1579,
      "step": 12800
    },
    {
      "epoch": 7.840856924254017,
      "grad_norm": 2.8209228515625,
      "learning_rate": 1.1975233040756617e-05,
      "loss": 1.164,
      "step": 12810
    },
    {
      "epoch": 7.8469778117827085,
      "grad_norm": 4.310190200805664,
      "learning_rate": 1.1941212492345378e-05,
      "loss": 1.1767,
      "step": 12820
    },
    {
      "epoch": 7.8530986993114,
      "grad_norm": 2.7291927337646484,
      "learning_rate": 1.1907191943934137e-05,
      "loss": 1.1503,
      "step": 12830
    },
    {
      "epoch": 7.859219586840092,
      "grad_norm": 2.8844997882843018,
      "learning_rate": 1.1873171395522896e-05,
      "loss": 1.1516,
      "step": 12840
    },
    {
      "epoch": 7.865340474368783,
      "grad_norm": 3.1168551445007324,
      "learning_rate": 1.1839150847111655e-05,
      "loss": 1.1593,
      "step": 12850
    },
    {
      "epoch": 7.871461361897476,
      "grad_norm": 2.552144765853882,
      "learning_rate": 1.1805130298700416e-05,
      "loss": 1.133,
      "step": 12860
    },
    {
      "epoch": 7.877582249426167,
      "grad_norm": 3.00880765914917,
      "learning_rate": 1.1771109750289175e-05,
      "loss": 1.1585,
      "step": 12870
    },
    {
      "epoch": 7.883703136954859,
      "grad_norm": 3.129321813583374,
      "learning_rate": 1.1737089201877934e-05,
      "loss": 1.1643,
      "step": 12880
    },
    {
      "epoch": 7.88982402448355,
      "grad_norm": 2.6133196353912354,
      "learning_rate": 1.1703068653466695e-05,
      "loss": 1.1933,
      "step": 12890
    },
    {
      "epoch": 7.895944912012242,
      "grad_norm": 3.4052469730377197,
      "learning_rate": 1.1669048105055454e-05,
      "loss": 1.1478,
      "step": 12900
    },
    {
      "epoch": 7.902065799540933,
      "grad_norm": 3.1864941120147705,
      "learning_rate": 1.1635027556644213e-05,
      "loss": 1.1558,
      "step": 12910
    },
    {
      "epoch": 7.908186687069625,
      "grad_norm": 2.8696846961975098,
      "learning_rate": 1.1601007008232974e-05,
      "loss": 1.1498,
      "step": 12920
    },
    {
      "epoch": 7.9143075745983165,
      "grad_norm": 3.1243202686309814,
      "learning_rate": 1.1566986459821732e-05,
      "loss": 1.1532,
      "step": 12930
    },
    {
      "epoch": 7.920428462127008,
      "grad_norm": 3.030133008956909,
      "learning_rate": 1.1532965911410493e-05,
      "loss": 1.168,
      "step": 12940
    },
    {
      "epoch": 7.9265493496557005,
      "grad_norm": 5.5496602058410645,
      "learning_rate": 1.1498945362999253e-05,
      "loss": 1.1728,
      "step": 12950
    },
    {
      "epoch": 7.932670237184392,
      "grad_norm": 2.774750232696533,
      "learning_rate": 1.146492481458801e-05,
      "loss": 1.1698,
      "step": 12960
    },
    {
      "epoch": 7.938791124713084,
      "grad_norm": 2.4071033000946045,
      "learning_rate": 1.1430904266176772e-05,
      "loss": 1.1621,
      "step": 12970
    },
    {
      "epoch": 7.944912012241775,
      "grad_norm": 3.9501190185546875,
      "learning_rate": 1.139688371776553e-05,
      "loss": 1.1595,
      "step": 12980
    },
    {
      "epoch": 7.951032899770467,
      "grad_norm": 2.919703483581543,
      "learning_rate": 1.136286316935429e-05,
      "loss": 1.1575,
      "step": 12990
    },
    {
      "epoch": 7.957153787299158,
      "grad_norm": 2.7440402507781982,
      "learning_rate": 1.132884262094305e-05,
      "loss": 1.1659,
      "step": 13000
    },
    {
      "epoch": 7.96327467482785,
      "grad_norm": 2.6349709033966064,
      "learning_rate": 1.129482207253181e-05,
      "loss": 1.148,
      "step": 13010
    },
    {
      "epoch": 7.969395562356541,
      "grad_norm": 3.7944247722625732,
      "learning_rate": 1.1260801524120569e-05,
      "loss": 1.1524,
      "step": 13020
    },
    {
      "epoch": 7.975516449885234,
      "grad_norm": 3.455249071121216,
      "learning_rate": 1.122678097570933e-05,
      "loss": 1.1796,
      "step": 13030
    },
    {
      "epoch": 7.981637337413925,
      "grad_norm": 3.802947759628296,
      "learning_rate": 1.1192760427298087e-05,
      "loss": 1.163,
      "step": 13040
    },
    {
      "epoch": 7.987758224942617,
      "grad_norm": 3.643726348876953,
      "learning_rate": 1.1158739878886848e-05,
      "loss": 1.1361,
      "step": 13050
    },
    {
      "epoch": 7.9938791124713084,
      "grad_norm": 3.0998475551605225,
      "learning_rate": 1.1124719330475607e-05,
      "loss": 1.1557,
      "step": 13060
    },
    {
      "epoch": 8.0,
      "grad_norm": 4.292564392089844,
      "learning_rate": 1.1090698782064366e-05,
      "loss": 1.1554,
      "step": 13070
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.5080250219994643,
      "eval_loss": 1.168102741241455,
      "eval_runtime": 2685.348,
      "eval_samples_per_second": 77.866,
      "eval_steps_per_second": 0.608,
      "step": 13070
    },
    {
      "epoch": 8.006120887528692,
      "grad_norm": 2.589630603790283,
      "learning_rate": 1.1056678233653127e-05,
      "loss": 1.124,
      "step": 13080
    },
    {
      "epoch": 8.012241775057383,
      "grad_norm": 2.709681749343872,
      "learning_rate": 1.1022657685241886e-05,
      "loss": 1.1433,
      "step": 13090
    },
    {
      "epoch": 8.018362662586075,
      "grad_norm": 3.8132050037384033,
      "learning_rate": 1.0988637136830646e-05,
      "loss": 1.1654,
      "step": 13100
    },
    {
      "epoch": 8.024483550114766,
      "grad_norm": 3.2627689838409424,
      "learning_rate": 1.0954616588419406e-05,
      "loss": 1.1345,
      "step": 13110
    },
    {
      "epoch": 8.030604437643458,
      "grad_norm": 3.628002405166626,
      "learning_rate": 1.0920596040008166e-05,
      "loss": 1.1702,
      "step": 13120
    },
    {
      "epoch": 8.03672532517215,
      "grad_norm": 2.8849120140075684,
      "learning_rate": 1.0886575491596925e-05,
      "loss": 1.1406,
      "step": 13130
    },
    {
      "epoch": 8.04284621270084,
      "grad_norm": 3.8340485095977783,
      "learning_rate": 1.0852554943185684e-05,
      "loss": 1.1339,
      "step": 13140
    },
    {
      "epoch": 8.048967100229532,
      "grad_norm": 3.1332828998565674,
      "learning_rate": 1.0818534394774445e-05,
      "loss": 1.1345,
      "step": 13150
    },
    {
      "epoch": 8.055087987758226,
      "grad_norm": 3.195749044418335,
      "learning_rate": 1.0784513846363204e-05,
      "loss": 1.1504,
      "step": 13160
    },
    {
      "epoch": 8.061208875286917,
      "grad_norm": 3.9875683784484863,
      "learning_rate": 1.0750493297951963e-05,
      "loss": 1.1356,
      "step": 13170
    },
    {
      "epoch": 8.067329762815609,
      "grad_norm": 4.1312384605407715,
      "learning_rate": 1.0716472749540724e-05,
      "loss": 1.1529,
      "step": 13180
    },
    {
      "epoch": 8.0734506503443,
      "grad_norm": 4.20302152633667,
      "learning_rate": 1.0682452201129483e-05,
      "loss": 1.1317,
      "step": 13190
    },
    {
      "epoch": 8.079571537872992,
      "grad_norm": 3.356748580932617,
      "learning_rate": 1.0648431652718242e-05,
      "loss": 1.1307,
      "step": 13200
    },
    {
      "epoch": 8.085692425401684,
      "grad_norm": 2.7438583374023438,
      "learning_rate": 1.0614411104307003e-05,
      "loss": 1.1565,
      "step": 13210
    },
    {
      "epoch": 8.091813312930375,
      "grad_norm": 2.8637239933013916,
      "learning_rate": 1.058039055589576e-05,
      "loss": 1.1327,
      "step": 13220
    },
    {
      "epoch": 8.097934200459067,
      "grad_norm": 2.9302523136138916,
      "learning_rate": 1.0546370007484521e-05,
      "loss": 1.1527,
      "step": 13230
    },
    {
      "epoch": 8.104055087987758,
      "grad_norm": 3.599940538406372,
      "learning_rate": 1.0512349459073282e-05,
      "loss": 1.1311,
      "step": 13240
    },
    {
      "epoch": 8.11017597551645,
      "grad_norm": 3.4082233905792236,
      "learning_rate": 1.047832891066204e-05,
      "loss": 1.1282,
      "step": 13250
    },
    {
      "epoch": 8.116296863045141,
      "grad_norm": 3.8687469959259033,
      "learning_rate": 1.04443083622508e-05,
      "loss": 1.141,
      "step": 13260
    },
    {
      "epoch": 8.122417750573833,
      "grad_norm": 3.1596381664276123,
      "learning_rate": 1.041028781383956e-05,
      "loss": 1.1433,
      "step": 13270
    },
    {
      "epoch": 8.128538638102524,
      "grad_norm": 3.489830255508423,
      "learning_rate": 1.0376267265428319e-05,
      "loss": 1.1456,
      "step": 13280
    },
    {
      "epoch": 8.134659525631216,
      "grad_norm": 3.366809844970703,
      "learning_rate": 1.034224671701708e-05,
      "loss": 1.1499,
      "step": 13290
    },
    {
      "epoch": 8.140780413159908,
      "grad_norm": 3.154359817504883,
      "learning_rate": 1.0308226168605839e-05,
      "loss": 1.1721,
      "step": 13300
    },
    {
      "epoch": 8.146901300688599,
      "grad_norm": 3.2491090297698975,
      "learning_rate": 1.0274205620194598e-05,
      "loss": 1.1306,
      "step": 13310
    },
    {
      "epoch": 8.15302218821729,
      "grad_norm": 3.3683319091796875,
      "learning_rate": 1.0240185071783358e-05,
      "loss": 1.1425,
      "step": 13320
    },
    {
      "epoch": 8.159143075745984,
      "grad_norm": 3.144820213317871,
      "learning_rate": 1.0206164523372118e-05,
      "loss": 1.1222,
      "step": 13330
    },
    {
      "epoch": 8.165263963274676,
      "grad_norm": 3.5433125495910645,
      "learning_rate": 1.0172143974960877e-05,
      "loss": 1.1324,
      "step": 13340
    },
    {
      "epoch": 8.171384850803367,
      "grad_norm": 3.311105728149414,
      "learning_rate": 1.0138123426549636e-05,
      "loss": 1.1394,
      "step": 13350
    },
    {
      "epoch": 8.177505738332059,
      "grad_norm": 2.810060739517212,
      "learning_rate": 1.0104102878138397e-05,
      "loss": 1.1454,
      "step": 13360
    },
    {
      "epoch": 8.18362662586075,
      "grad_norm": 3.0078938007354736,
      "learning_rate": 1.0070082329727156e-05,
      "loss": 1.1376,
      "step": 13370
    },
    {
      "epoch": 8.189747513389442,
      "grad_norm": 2.672898769378662,
      "learning_rate": 1.0036061781315915e-05,
      "loss": 1.1611,
      "step": 13380
    },
    {
      "epoch": 8.195868400918133,
      "grad_norm": 2.5500643253326416,
      "learning_rate": 1.0002041232904676e-05,
      "loss": 1.1525,
      "step": 13390
    },
    {
      "epoch": 8.201989288446825,
      "grad_norm": 3.041093587875366,
      "learning_rate": 9.968020684493435e-06,
      "loss": 1.1366,
      "step": 13400
    },
    {
      "epoch": 8.208110175975516,
      "grad_norm": 3.1694231033325195,
      "learning_rate": 9.934000136082194e-06,
      "loss": 1.1638,
      "step": 13410
    },
    {
      "epoch": 8.214231063504208,
      "grad_norm": 3.110459089279175,
      "learning_rate": 9.899979587670953e-06,
      "loss": 1.1397,
      "step": 13420
    },
    {
      "epoch": 8.2203519510329,
      "grad_norm": 3.1803088188171387,
      "learning_rate": 9.865959039259712e-06,
      "loss": 1.1641,
      "step": 13430
    },
    {
      "epoch": 8.226472838561591,
      "grad_norm": 4.122441291809082,
      "learning_rate": 9.831938490848473e-06,
      "loss": 1.1562,
      "step": 13440
    },
    {
      "epoch": 8.232593726090283,
      "grad_norm": 2.7772445678710938,
      "learning_rate": 9.797917942437232e-06,
      "loss": 1.154,
      "step": 13450
    },
    {
      "epoch": 8.238714613618974,
      "grad_norm": 2.789196014404297,
      "learning_rate": 9.763897394025992e-06,
      "loss": 1.1483,
      "step": 13460
    },
    {
      "epoch": 8.244835501147666,
      "grad_norm": 3.0457327365875244,
      "learning_rate": 9.729876845614752e-06,
      "loss": 1.1212,
      "step": 13470
    },
    {
      "epoch": 8.250956388676357,
      "grad_norm": 3.0723581314086914,
      "learning_rate": 9.695856297203511e-06,
      "loss": 1.1467,
      "step": 13480
    },
    {
      "epoch": 8.257077276205049,
      "grad_norm": 3.5150368213653564,
      "learning_rate": 9.66183574879227e-06,
      "loss": 1.1353,
      "step": 13490
    },
    {
      "epoch": 8.263198163733742,
      "grad_norm": 3.7156941890716553,
      "learning_rate": 9.627815200381031e-06,
      "loss": 1.1536,
      "step": 13500
    },
    {
      "epoch": 8.269319051262434,
      "grad_norm": 3.0745482444763184,
      "learning_rate": 9.593794651969789e-06,
      "loss": 1.1182,
      "step": 13510
    },
    {
      "epoch": 8.275439938791125,
      "grad_norm": 3.3888468742370605,
      "learning_rate": 9.55977410355855e-06,
      "loss": 1.1684,
      "step": 13520
    },
    {
      "epoch": 8.281560826319817,
      "grad_norm": 3.1190714836120605,
      "learning_rate": 9.52575355514731e-06,
      "loss": 1.1388,
      "step": 13530
    },
    {
      "epoch": 8.287681713848508,
      "grad_norm": 3.361152172088623,
      "learning_rate": 9.491733006736068e-06,
      "loss": 1.166,
      "step": 13540
    },
    {
      "epoch": 8.2938026013772,
      "grad_norm": 3.003202199935913,
      "learning_rate": 9.457712458324829e-06,
      "loss": 1.1659,
      "step": 13550
    },
    {
      "epoch": 8.299923488905891,
      "grad_norm": 3.121140718460083,
      "learning_rate": 9.423691909913588e-06,
      "loss": 1.131,
      "step": 13560
    },
    {
      "epoch": 8.306044376434583,
      "grad_norm": 4.083986282348633,
      "learning_rate": 9.389671361502347e-06,
      "loss": 1.1433,
      "step": 13570
    },
    {
      "epoch": 8.312165263963275,
      "grad_norm": 2.6656112670898438,
      "learning_rate": 9.355650813091108e-06,
      "loss": 1.1422,
      "step": 13580
    },
    {
      "epoch": 8.318286151491966,
      "grad_norm": 3.0381946563720703,
      "learning_rate": 9.321630264679867e-06,
      "loss": 1.1543,
      "step": 13590
    },
    {
      "epoch": 8.324407039020658,
      "grad_norm": 2.631458282470703,
      "learning_rate": 9.287609716268626e-06,
      "loss": 1.1442,
      "step": 13600
    },
    {
      "epoch": 8.33052792654935,
      "grad_norm": 3.234130620956421,
      "learning_rate": 9.253589167857387e-06,
      "loss": 1.1463,
      "step": 13610
    },
    {
      "epoch": 8.33664881407804,
      "grad_norm": 2.735459566116333,
      "learning_rate": 9.219568619446146e-06,
      "loss": 1.1337,
      "step": 13620
    },
    {
      "epoch": 8.342769701606732,
      "grad_norm": 3.386232852935791,
      "learning_rate": 9.185548071034905e-06,
      "loss": 1.1434,
      "step": 13630
    },
    {
      "epoch": 8.348890589135424,
      "grad_norm": 2.588688611984253,
      "learning_rate": 9.151527522623664e-06,
      "loss": 1.1449,
      "step": 13640
    },
    {
      "epoch": 8.355011476664115,
      "grad_norm": 2.9801838397979736,
      "learning_rate": 9.117506974212425e-06,
      "loss": 1.1649,
      "step": 13650
    },
    {
      "epoch": 8.361132364192809,
      "grad_norm": 3.2618236541748047,
      "learning_rate": 9.083486425801184e-06,
      "loss": 1.1522,
      "step": 13660
    },
    {
      "epoch": 8.3672532517215,
      "grad_norm": 3.2830750942230225,
      "learning_rate": 9.049465877389944e-06,
      "loss": 1.1472,
      "step": 13670
    },
    {
      "epoch": 8.373374139250192,
      "grad_norm": 3.4445407390594482,
      "learning_rate": 9.015445328978704e-06,
      "loss": 1.1538,
      "step": 13680
    },
    {
      "epoch": 8.379495026778883,
      "grad_norm": 3.6787800788879395,
      "learning_rate": 8.981424780567464e-06,
      "loss": 1.1503,
      "step": 13690
    },
    {
      "epoch": 8.385615914307575,
      "grad_norm": 3.1863129138946533,
      "learning_rate": 8.947404232156223e-06,
      "loss": 1.1275,
      "step": 13700
    },
    {
      "epoch": 8.391736801836267,
      "grad_norm": 2.996192216873169,
      "learning_rate": 8.913383683744984e-06,
      "loss": 1.1438,
      "step": 13710
    },
    {
      "epoch": 8.397857689364958,
      "grad_norm": 2.6561343669891357,
      "learning_rate": 8.879363135333741e-06,
      "loss": 1.1432,
      "step": 13720
    },
    {
      "epoch": 8.40397857689365,
      "grad_norm": 2.806730031967163,
      "learning_rate": 8.845342586922502e-06,
      "loss": 1.1325,
      "step": 13730
    },
    {
      "epoch": 8.410099464422341,
      "grad_norm": 2.868938446044922,
      "learning_rate": 8.811322038511263e-06,
      "loss": 1.1437,
      "step": 13740
    },
    {
      "epoch": 8.416220351951033,
      "grad_norm": 2.915384292602539,
      "learning_rate": 8.77730149010002e-06,
      "loss": 1.1502,
      "step": 13750
    },
    {
      "epoch": 8.422341239479724,
      "grad_norm": 3.5552773475646973,
      "learning_rate": 8.743280941688781e-06,
      "loss": 1.1418,
      "step": 13760
    },
    {
      "epoch": 8.428462127008416,
      "grad_norm": 3.456284999847412,
      "learning_rate": 8.70926039327754e-06,
      "loss": 1.1539,
      "step": 13770
    },
    {
      "epoch": 8.434583014537107,
      "grad_norm": 3.1513900756835938,
      "learning_rate": 8.6752398448663e-06,
      "loss": 1.1522,
      "step": 13780
    },
    {
      "epoch": 8.440703902065799,
      "grad_norm": 3.183990240097046,
      "learning_rate": 8.64121929645506e-06,
      "loss": 1.1419,
      "step": 13790
    },
    {
      "epoch": 8.44682478959449,
      "grad_norm": 3.3508360385894775,
      "learning_rate": 8.607198748043817e-06,
      "loss": 1.151,
      "step": 13800
    },
    {
      "epoch": 8.452945677123182,
      "grad_norm": 3.314988374710083,
      "learning_rate": 8.573178199632578e-06,
      "loss": 1.1358,
      "step": 13810
    },
    {
      "epoch": 8.459066564651874,
      "grad_norm": 3.3967995643615723,
      "learning_rate": 8.53915765122134e-06,
      "loss": 1.147,
      "step": 13820
    },
    {
      "epoch": 8.465187452180567,
      "grad_norm": 3.0596559047698975,
      "learning_rate": 8.505137102810097e-06,
      "loss": 1.1446,
      "step": 13830
    },
    {
      "epoch": 8.471308339709259,
      "grad_norm": 2.67799973487854,
      "learning_rate": 8.471116554398857e-06,
      "loss": 1.1332,
      "step": 13840
    },
    {
      "epoch": 8.47742922723795,
      "grad_norm": 3.3761391639709473,
      "learning_rate": 8.437096005987617e-06,
      "loss": 1.1318,
      "step": 13850
    },
    {
      "epoch": 8.483550114766642,
      "grad_norm": 7.103214263916016,
      "learning_rate": 8.403075457576376e-06,
      "loss": 1.1476,
      "step": 13860
    },
    {
      "epoch": 8.489671002295333,
      "grad_norm": 3.1061320304870605,
      "learning_rate": 8.369054909165137e-06,
      "loss": 1.1586,
      "step": 13870
    },
    {
      "epoch": 8.495791889824025,
      "grad_norm": 3.3592042922973633,
      "learning_rate": 8.335034360753896e-06,
      "loss": 1.1442,
      "step": 13880
    },
    {
      "epoch": 8.501912777352716,
      "grad_norm": 5.015500545501709,
      "learning_rate": 8.301013812342655e-06,
      "loss": 1.1537,
      "step": 13890
    },
    {
      "epoch": 8.508033664881408,
      "grad_norm": 3.5375547409057617,
      "learning_rate": 8.266993263931416e-06,
      "loss": 1.1601,
      "step": 13900
    },
    {
      "epoch": 8.5141545524101,
      "grad_norm": 3.1028242111206055,
      "learning_rate": 8.232972715520175e-06,
      "loss": 1.1347,
      "step": 13910
    },
    {
      "epoch": 8.520275439938791,
      "grad_norm": 2.954200506210327,
      "learning_rate": 8.198952167108934e-06,
      "loss": 1.1233,
      "step": 13920
    },
    {
      "epoch": 8.526396327467483,
      "grad_norm": 4.131378650665283,
      "learning_rate": 8.164931618697693e-06,
      "loss": 1.1252,
      "step": 13930
    },
    {
      "epoch": 8.532517214996174,
      "grad_norm": 3.189302921295166,
      "learning_rate": 8.130911070286454e-06,
      "loss": 1.143,
      "step": 13940
    },
    {
      "epoch": 8.538638102524866,
      "grad_norm": 4.135822772979736,
      "learning_rate": 8.096890521875213e-06,
      "loss": 1.1489,
      "step": 13950
    },
    {
      "epoch": 8.544758990053557,
      "grad_norm": 3.0186097621917725,
      "learning_rate": 8.062869973463972e-06,
      "loss": 1.1458,
      "step": 13960
    },
    {
      "epoch": 8.550879877582249,
      "grad_norm": 3.5816197395324707,
      "learning_rate": 8.028849425052733e-06,
      "loss": 1.1348,
      "step": 13970
    },
    {
      "epoch": 8.55700076511094,
      "grad_norm": 3.028970956802368,
      "learning_rate": 7.994828876641492e-06,
      "loss": 1.1498,
      "step": 13980
    },
    {
      "epoch": 8.563121652639634,
      "grad_norm": 2.8785955905914307,
      "learning_rate": 7.960808328230251e-06,
      "loss": 1.1504,
      "step": 13990
    },
    {
      "epoch": 8.569242540168325,
      "grad_norm": 3.270470380783081,
      "learning_rate": 7.926787779819012e-06,
      "loss": 1.1361,
      "step": 14000
    },
    {
      "epoch": 8.575363427697017,
      "grad_norm": 2.8553550243377686,
      "learning_rate": 7.89276723140777e-06,
      "loss": 1.1572,
      "step": 14010
    },
    {
      "epoch": 8.581484315225708,
      "grad_norm": 3.0945401191711426,
      "learning_rate": 7.85874668299653e-06,
      "loss": 1.1139,
      "step": 14020
    },
    {
      "epoch": 8.5876052027544,
      "grad_norm": 3.0976555347442627,
      "learning_rate": 7.824726134585291e-06,
      "loss": 1.1414,
      "step": 14030
    },
    {
      "epoch": 8.593726090283091,
      "grad_norm": 3.190762758255005,
      "learning_rate": 7.790705586174049e-06,
      "loss": 1.1243,
      "step": 14040
    },
    {
      "epoch": 8.599846977811783,
      "grad_norm": 3.8209619522094727,
      "learning_rate": 7.75668503776281e-06,
      "loss": 1.1383,
      "step": 14050
    },
    {
      "epoch": 8.605967865340475,
      "grad_norm": 3.3825807571411133,
      "learning_rate": 7.722664489351569e-06,
      "loss": 1.1393,
      "step": 14060
    },
    {
      "epoch": 8.612088752869166,
      "grad_norm": 2.7561495304107666,
      "learning_rate": 7.688643940940328e-06,
      "loss": 1.1503,
      "step": 14070
    },
    {
      "epoch": 8.618209640397858,
      "grad_norm": 2.9515817165374756,
      "learning_rate": 7.654623392529089e-06,
      "loss": 1.1396,
      "step": 14080
    },
    {
      "epoch": 8.62433052792655,
      "grad_norm": 2.807973861694336,
      "learning_rate": 7.620602844117848e-06,
      "loss": 1.1299,
      "step": 14090
    },
    {
      "epoch": 8.63045141545524,
      "grad_norm": 2.8332042694091797,
      "learning_rate": 7.586582295706607e-06,
      "loss": 1.1255,
      "step": 14100
    },
    {
      "epoch": 8.636572302983932,
      "grad_norm": 3.248446226119995,
      "learning_rate": 7.552561747295367e-06,
      "loss": 1.1442,
      "step": 14110
    },
    {
      "epoch": 8.642693190512624,
      "grad_norm": 3.1265931129455566,
      "learning_rate": 7.518541198884127e-06,
      "loss": 1.1516,
      "step": 14120
    },
    {
      "epoch": 8.648814078041315,
      "grad_norm": 2.808518886566162,
      "learning_rate": 7.484520650472886e-06,
      "loss": 1.1333,
      "step": 14130
    },
    {
      "epoch": 8.654934965570007,
      "grad_norm": 4.110435962677002,
      "learning_rate": 7.450500102061646e-06,
      "loss": 1.1439,
      "step": 14140
    },
    {
      "epoch": 8.661055853098699,
      "grad_norm": 3.197977304458618,
      "learning_rate": 7.416479553650406e-06,
      "loss": 1.1325,
      "step": 14150
    },
    {
      "epoch": 8.66717674062739,
      "grad_norm": 3.7765300273895264,
      "learning_rate": 7.382459005239164e-06,
      "loss": 1.158,
      "step": 14160
    },
    {
      "epoch": 8.673297628156083,
      "grad_norm": 3.2383854389190674,
      "learning_rate": 7.348438456827924e-06,
      "loss": 1.1646,
      "step": 14170
    },
    {
      "epoch": 8.679418515684775,
      "grad_norm": 2.7088334560394287,
      "learning_rate": 7.314417908416683e-06,
      "loss": 1.1301,
      "step": 14180
    },
    {
      "epoch": 8.685539403213467,
      "grad_norm": 4.367984771728516,
      "learning_rate": 7.280397360005443e-06,
      "loss": 1.1286,
      "step": 14190
    },
    {
      "epoch": 8.691660290742158,
      "grad_norm": 3.3485443592071533,
      "learning_rate": 7.246376811594203e-06,
      "loss": 1.153,
      "step": 14200
    },
    {
      "epoch": 8.69778117827085,
      "grad_norm": 4.202768802642822,
      "learning_rate": 7.2123562631829625e-06,
      "loss": 1.1468,
      "step": 14210
    },
    {
      "epoch": 8.703902065799541,
      "grad_norm": 3.7260403633117676,
      "learning_rate": 7.1783357147717225e-06,
      "loss": 1.1485,
      "step": 14220
    },
    {
      "epoch": 8.710022953328233,
      "grad_norm": 3.604518413543701,
      "learning_rate": 7.1443151663604825e-06,
      "loss": 1.1551,
      "step": 14230
    },
    {
      "epoch": 8.716143840856924,
      "grad_norm": 2.95175838470459,
      "learning_rate": 7.110294617949241e-06,
      "loss": 1.1454,
      "step": 14240
    },
    {
      "epoch": 8.722264728385616,
      "grad_norm": 3.4083735942840576,
      "learning_rate": 7.076274069538001e-06,
      "loss": 1.1589,
      "step": 14250
    },
    {
      "epoch": 8.728385615914307,
      "grad_norm": 3.287588357925415,
      "learning_rate": 7.042253521126762e-06,
      "loss": 1.1326,
      "step": 14260
    },
    {
      "epoch": 8.734506503442999,
      "grad_norm": 2.924208164215088,
      "learning_rate": 7.00823297271552e-06,
      "loss": 1.1171,
      "step": 14270
    },
    {
      "epoch": 8.74062739097169,
      "grad_norm": 2.9708151817321777,
      "learning_rate": 6.97421242430428e-06,
      "loss": 1.1443,
      "step": 14280
    },
    {
      "epoch": 8.746748278500382,
      "grad_norm": 3.3981306552886963,
      "learning_rate": 6.94019187589304e-06,
      "loss": 1.1377,
      "step": 14290
    },
    {
      "epoch": 8.752869166029074,
      "grad_norm": 4.204988956451416,
      "learning_rate": 6.906171327481799e-06,
      "loss": 1.1445,
      "step": 14300
    },
    {
      "epoch": 8.758990053557765,
      "grad_norm": 3.8043899536132812,
      "learning_rate": 6.872150779070559e-06,
      "loss": 1.1468,
      "step": 14310
    },
    {
      "epoch": 8.765110941086457,
      "grad_norm": 3.0697388648986816,
      "learning_rate": 6.838130230659319e-06,
      "loss": 1.1339,
      "step": 14320
    },
    {
      "epoch": 8.77123182861515,
      "grad_norm": 3.4975595474243164,
      "learning_rate": 6.804109682248077e-06,
      "loss": 1.1465,
      "step": 14330
    },
    {
      "epoch": 8.777352716143842,
      "grad_norm": 3.3167073726654053,
      "learning_rate": 6.770089133836838e-06,
      "loss": 1.1393,
      "step": 14340
    },
    {
      "epoch": 8.783473603672533,
      "grad_norm": 3.330259323120117,
      "learning_rate": 6.736068585425598e-06,
      "loss": 1.1289,
      "step": 14350
    },
    {
      "epoch": 8.789594491201225,
      "grad_norm": 3.079725503921509,
      "learning_rate": 6.702048037014356e-06,
      "loss": 1.1521,
      "step": 14360
    },
    {
      "epoch": 8.795715378729916,
      "grad_norm": 3.1370887756347656,
      "learning_rate": 6.668027488603116e-06,
      "loss": 1.1732,
      "step": 14370
    },
    {
      "epoch": 8.801836266258608,
      "grad_norm": 2.693749189376831,
      "learning_rate": 6.634006940191876e-06,
      "loss": 1.147,
      "step": 14380
    },
    {
      "epoch": 8.8079571537873,
      "grad_norm": 2.9202158451080322,
      "learning_rate": 6.5999863917806355e-06,
      "loss": 1.1401,
      "step": 14390
    },
    {
      "epoch": 8.814078041315991,
      "grad_norm": 2.882169246673584,
      "learning_rate": 6.5659658433693955e-06,
      "loss": 1.1602,
      "step": 14400
    },
    {
      "epoch": 8.820198928844682,
      "grad_norm": 3.2943568229675293,
      "learning_rate": 6.5319452949581555e-06,
      "loss": 1.1259,
      "step": 14410
    },
    {
      "epoch": 8.826319816373374,
      "grad_norm": 3.4829976558685303,
      "learning_rate": 6.497924746546915e-06,
      "loss": 1.1503,
      "step": 14420
    },
    {
      "epoch": 8.832440703902066,
      "grad_norm": 3.278031826019287,
      "learning_rate": 6.463904198135675e-06,
      "loss": 1.1534,
      "step": 14430
    },
    {
      "epoch": 8.838561591430757,
      "grad_norm": 3.270439386367798,
      "learning_rate": 6.429883649724435e-06,
      "loss": 1.1414,
      "step": 14440
    },
    {
      "epoch": 8.844682478959449,
      "grad_norm": 3.0096373558044434,
      "learning_rate": 6.395863101313193e-06,
      "loss": 1.1311,
      "step": 14450
    },
    {
      "epoch": 8.85080336648814,
      "grad_norm": 2.464264392852783,
      "learning_rate": 6.361842552901953e-06,
      "loss": 1.1442,
      "step": 14460
    },
    {
      "epoch": 8.856924254016832,
      "grad_norm": 3.1836256980895996,
      "learning_rate": 6.327822004490713e-06,
      "loss": 1.1667,
      "step": 14470
    },
    {
      "epoch": 8.863045141545523,
      "grad_norm": 2.9464852809906006,
      "learning_rate": 6.293801456079472e-06,
      "loss": 1.1368,
      "step": 14480
    },
    {
      "epoch": 8.869166029074215,
      "grad_norm": 3.198124408721924,
      "learning_rate": 6.259780907668232e-06,
      "loss": 1.1569,
      "step": 14490
    },
    {
      "epoch": 8.875286916602908,
      "grad_norm": 3.599224328994751,
      "learning_rate": 6.225760359256991e-06,
      "loss": 1.1241,
      "step": 14500
    },
    {
      "epoch": 8.8814078041316,
      "grad_norm": 2.898756742477417,
      "learning_rate": 6.191739810845751e-06,
      "loss": 1.144,
      "step": 14510
    },
    {
      "epoch": 8.887528691660291,
      "grad_norm": 2.7918193340301514,
      "learning_rate": 6.157719262434511e-06,
      "loss": 1.1261,
      "step": 14520
    },
    {
      "epoch": 8.893649579188983,
      "grad_norm": 3.0213284492492676,
      "learning_rate": 6.12369871402327e-06,
      "loss": 1.1325,
      "step": 14530
    },
    {
      "epoch": 8.899770466717674,
      "grad_norm": 3.097947359085083,
      "learning_rate": 6.089678165612029e-06,
      "loss": 1.1257,
      "step": 14540
    },
    {
      "epoch": 8.905891354246366,
      "grad_norm": 3.194293975830078,
      "learning_rate": 6.055657617200789e-06,
      "loss": 1.1345,
      "step": 14550
    },
    {
      "epoch": 8.912012241775058,
      "grad_norm": 3.5505549907684326,
      "learning_rate": 6.021637068789549e-06,
      "loss": 1.1222,
      "step": 14560
    },
    {
      "epoch": 8.91813312930375,
      "grad_norm": 3.582949638366699,
      "learning_rate": 5.9876165203783085e-06,
      "loss": 1.1516,
      "step": 14570
    },
    {
      "epoch": 8.92425401683244,
      "grad_norm": 3.011648178100586,
      "learning_rate": 5.9535959719670685e-06,
      "loss": 1.1439,
      "step": 14580
    },
    {
      "epoch": 8.930374904361132,
      "grad_norm": 3.2848992347717285,
      "learning_rate": 5.919575423555828e-06,
      "loss": 1.1335,
      "step": 14590
    },
    {
      "epoch": 8.936495791889824,
      "grad_norm": 3.178999423980713,
      "learning_rate": 5.885554875144588e-06,
      "loss": 1.1529,
      "step": 14600
    },
    {
      "epoch": 8.942616679418515,
      "grad_norm": 7.24118709564209,
      "learning_rate": 5.851534326733348e-06,
      "loss": 1.1363,
      "step": 14610
    },
    {
      "epoch": 8.948737566947207,
      "grad_norm": 2.974168062210083,
      "learning_rate": 5.817513778322107e-06,
      "loss": 1.1155,
      "step": 14620
    },
    {
      "epoch": 8.954858454475898,
      "grad_norm": 3.646303176879883,
      "learning_rate": 5.783493229910866e-06,
      "loss": 1.1383,
      "step": 14630
    },
    {
      "epoch": 8.96097934200459,
      "grad_norm": 3.389556407928467,
      "learning_rate": 5.749472681499627e-06,
      "loss": 1.1449,
      "step": 14640
    },
    {
      "epoch": 8.967100229533282,
      "grad_norm": 3.2630605697631836,
      "learning_rate": 5.715452133088386e-06,
      "loss": 1.1312,
      "step": 14650
    },
    {
      "epoch": 8.973221117061975,
      "grad_norm": 3.2353131771087646,
      "learning_rate": 5.681431584677145e-06,
      "loss": 1.1315,
      "step": 14660
    },
    {
      "epoch": 8.979342004590666,
      "grad_norm": 3.8882558345794678,
      "learning_rate": 5.647411036265905e-06,
      "loss": 1.1425,
      "step": 14670
    },
    {
      "epoch": 8.985462892119358,
      "grad_norm": 2.721869945526123,
      "learning_rate": 5.613390487854665e-06,
      "loss": 1.1308,
      "step": 14680
    },
    {
      "epoch": 8.99158377964805,
      "grad_norm": 3.2869043350219727,
      "learning_rate": 5.579369939443424e-06,
      "loss": 1.1277,
      "step": 14690
    },
    {
      "epoch": 8.997704667176741,
      "grad_norm": 4.439366817474365,
      "learning_rate": 5.545349391032183e-06,
      "loss": 1.1279,
      "step": 14700
    },
    {
      "epoch": 8.999540933435348,
      "eval_accuracy": 0.5099954088074378,
      "eval_loss": 1.168525218963623,
      "eval_runtime": 2629.8,
      "eval_samples_per_second": 79.51,
      "eval_steps_per_second": 0.621,
      "step": 14703
    },
    {
      "epoch": 9.003825554705433,
      "grad_norm": 3.387463092803955,
      "learning_rate": 5.511328842620943e-06,
      "loss": 1.1331,
      "step": 14710
    },
    {
      "epoch": 9.009946442234124,
      "grad_norm": 3.4445242881774902,
      "learning_rate": 5.477308294209703e-06,
      "loss": 1.1233,
      "step": 14720
    },
    {
      "epoch": 9.016067329762816,
      "grad_norm": 3.013448476791382,
      "learning_rate": 5.443287745798462e-06,
      "loss": 1.1391,
      "step": 14730
    },
    {
      "epoch": 9.022188217291507,
      "grad_norm": 2.931241512298584,
      "learning_rate": 5.409267197387222e-06,
      "loss": 1.1187,
      "step": 14740
    },
    {
      "epoch": 9.028309104820199,
      "grad_norm": 3.7724857330322266,
      "learning_rate": 5.3752466489759815e-06,
      "loss": 1.1375,
      "step": 14750
    },
    {
      "epoch": 9.03442999234889,
      "grad_norm": 3.2617881298065186,
      "learning_rate": 5.3412261005647414e-06,
      "loss": 1.1239,
      "step": 14760
    },
    {
      "epoch": 9.040550879877582,
      "grad_norm": 3.163550853729248,
      "learning_rate": 5.3072055521535014e-06,
      "loss": 1.1333,
      "step": 14770
    },
    {
      "epoch": 9.046671767406274,
      "grad_norm": 3.271584987640381,
      "learning_rate": 5.2731850037422606e-06,
      "loss": 1.1241,
      "step": 14780
    },
    {
      "epoch": 9.052792654934965,
      "grad_norm": 2.9983322620391846,
      "learning_rate": 5.23916445533102e-06,
      "loss": 1.1245,
      "step": 14790
    },
    {
      "epoch": 9.058913542463657,
      "grad_norm": 3.0628502368927,
      "learning_rate": 5.20514390691978e-06,
      "loss": 1.1411,
      "step": 14800
    },
    {
      "epoch": 9.065034429992348,
      "grad_norm": 2.9316301345825195,
      "learning_rate": 5.17112335850854e-06,
      "loss": 1.1326,
      "step": 14810
    },
    {
      "epoch": 9.07115531752104,
      "grad_norm": 2.7524290084838867,
      "learning_rate": 5.137102810097299e-06,
      "loss": 1.1024,
      "step": 14820
    },
    {
      "epoch": 9.077276205049731,
      "grad_norm": 3.432013511657715,
      "learning_rate": 5.103082261686059e-06,
      "loss": 1.1071,
      "step": 14830
    },
    {
      "epoch": 9.083397092578425,
      "grad_norm": 3.5625956058502197,
      "learning_rate": 5.069061713274818e-06,
      "loss": 1.098,
      "step": 14840
    },
    {
      "epoch": 9.089517980107116,
      "grad_norm": 3.2693939208984375,
      "learning_rate": 5.035041164863578e-06,
      "loss": 1.1313,
      "step": 14850
    },
    {
      "epoch": 9.095638867635808,
      "grad_norm": 4.684074878692627,
      "learning_rate": 5.001020616452338e-06,
      "loss": 1.1344,
      "step": 14860
    },
    {
      "epoch": 9.1017597551645,
      "grad_norm": 3.109464168548584,
      "learning_rate": 4.967000068041097e-06,
      "loss": 1.1211,
      "step": 14870
    },
    {
      "epoch": 9.10788064269319,
      "grad_norm": 3.3348963260650635,
      "learning_rate": 4.932979519629856e-06,
      "loss": 1.1295,
      "step": 14880
    },
    {
      "epoch": 9.114001530221882,
      "grad_norm": 2.6500282287597656,
      "learning_rate": 4.898958971218616e-06,
      "loss": 1.1262,
      "step": 14890
    },
    {
      "epoch": 9.120122417750574,
      "grad_norm": 4.016347408294678,
      "learning_rate": 4.864938422807376e-06,
      "loss": 1.1288,
      "step": 14900
    },
    {
      "epoch": 9.126243305279266,
      "grad_norm": 3.280271291732788,
      "learning_rate": 4.830917874396135e-06,
      "loss": 1.1441,
      "step": 14910
    },
    {
      "epoch": 9.132364192807957,
      "grad_norm": 2.6855580806732178,
      "learning_rate": 4.7968973259848945e-06,
      "loss": 1.1212,
      "step": 14920
    },
    {
      "epoch": 9.138485080336649,
      "grad_norm": 3.044363021850586,
      "learning_rate": 4.762876777573655e-06,
      "loss": 1.1398,
      "step": 14930
    },
    {
      "epoch": 9.14460596786534,
      "grad_norm": 3.8019986152648926,
      "learning_rate": 4.7288562291624144e-06,
      "loss": 1.1222,
      "step": 14940
    },
    {
      "epoch": 9.150726855394032,
      "grad_norm": 3.0457823276519775,
      "learning_rate": 4.6948356807511736e-06,
      "loss": 1.1531,
      "step": 14950
    },
    {
      "epoch": 9.156847742922723,
      "grad_norm": 3.0033369064331055,
      "learning_rate": 4.6608151323399336e-06,
      "loss": 1.1374,
      "step": 14960
    },
    {
      "epoch": 9.162968630451415,
      "grad_norm": 3.2390763759613037,
      "learning_rate": 4.6267945839286935e-06,
      "loss": 1.124,
      "step": 14970
    },
    {
      "epoch": 9.169089517980106,
      "grad_norm": 3.1349003314971924,
      "learning_rate": 4.592774035517453e-06,
      "loss": 1.1241,
      "step": 14980
    },
    {
      "epoch": 9.175210405508798,
      "grad_norm": 3.171571969985962,
      "learning_rate": 4.558753487106213e-06,
      "loss": 1.1218,
      "step": 14990
    },
    {
      "epoch": 9.181331293037491,
      "grad_norm": 3.28657603263855,
      "learning_rate": 4.524732938694972e-06,
      "loss": 1.1209,
      "step": 15000
    },
    {
      "epoch": 9.187452180566183,
      "grad_norm": 3.4533939361572266,
      "learning_rate": 4.490712390283732e-06,
      "loss": 1.129,
      "step": 15010
    },
    {
      "epoch": 9.193573068094874,
      "grad_norm": 3.1935927867889404,
      "learning_rate": 4.456691841872492e-06,
      "loss": 1.1171,
      "step": 15020
    },
    {
      "epoch": 9.199693955623566,
      "grad_norm": 3.5906758308410645,
      "learning_rate": 4.422671293461251e-06,
      "loss": 1.1337,
      "step": 15030
    },
    {
      "epoch": 9.205814843152258,
      "grad_norm": 3.6400201320648193,
      "learning_rate": 4.38865074505001e-06,
      "loss": 1.1338,
      "step": 15040
    },
    {
      "epoch": 9.211935730680949,
      "grad_norm": 3.086146116256714,
      "learning_rate": 4.35463019663877e-06,
      "loss": 1.133,
      "step": 15050
    },
    {
      "epoch": 9.21805661820964,
      "grad_norm": 3.2579269409179688,
      "learning_rate": 4.32060964822753e-06,
      "loss": 1.1388,
      "step": 15060
    },
    {
      "epoch": 9.224177505738332,
      "grad_norm": 2.8636515140533447,
      "learning_rate": 4.286589099816289e-06,
      "loss": 1.1233,
      "step": 15070
    },
    {
      "epoch": 9.230298393267024,
      "grad_norm": 3.022768497467041,
      "learning_rate": 4.252568551405048e-06,
      "loss": 1.1253,
      "step": 15080
    },
    {
      "epoch": 9.236419280795715,
      "grad_norm": 3.074159860610962,
      "learning_rate": 4.218548002993808e-06,
      "loss": 1.1527,
      "step": 15090
    },
    {
      "epoch": 9.242540168324407,
      "grad_norm": 5.568500518798828,
      "learning_rate": 4.184527454582568e-06,
      "loss": 1.1397,
      "step": 15100
    },
    {
      "epoch": 9.248661055853098,
      "grad_norm": 3.478727102279663,
      "learning_rate": 4.150506906171327e-06,
      "loss": 1.1371,
      "step": 15110
    },
    {
      "epoch": 9.25478194338179,
      "grad_norm": 3.0093882083892822,
      "learning_rate": 4.116486357760087e-06,
      "loss": 1.1308,
      "step": 15120
    },
    {
      "epoch": 9.260902830910481,
      "grad_norm": 2.999668836593628,
      "learning_rate": 4.0824658093488465e-06,
      "loss": 1.1466,
      "step": 15130
    },
    {
      "epoch": 9.267023718439173,
      "grad_norm": 3.1364176273345947,
      "learning_rate": 4.0484452609376065e-06,
      "loss": 1.1379,
      "step": 15140
    },
    {
      "epoch": 9.273144605967865,
      "grad_norm": 3.1165690422058105,
      "learning_rate": 4.0144247125263665e-06,
      "loss": 1.118,
      "step": 15150
    },
    {
      "epoch": 9.279265493496556,
      "grad_norm": 3.0751230716705322,
      "learning_rate": 3.980404164115126e-06,
      "loss": 1.1519,
      "step": 15160
    },
    {
      "epoch": 9.28538638102525,
      "grad_norm": 3.8941943645477295,
      "learning_rate": 3.946383615703885e-06,
      "loss": 1.1118,
      "step": 15170
    },
    {
      "epoch": 9.291507268553941,
      "grad_norm": 2.9752681255340576,
      "learning_rate": 3.912363067292646e-06,
      "loss": 1.1358,
      "step": 15180
    },
    {
      "epoch": 9.297628156082633,
      "grad_norm": 3.164053201675415,
      "learning_rate": 3.878342518881405e-06,
      "loss": 1.1298,
      "step": 15190
    },
    {
      "epoch": 9.303749043611324,
      "grad_norm": 3.064384698867798,
      "learning_rate": 3.844321970470164e-06,
      "loss": 1.1337,
      "step": 15200
    },
    {
      "epoch": 9.309869931140016,
      "grad_norm": 3.1741909980773926,
      "learning_rate": 3.810301422058924e-06,
      "loss": 1.1217,
      "step": 15210
    },
    {
      "epoch": 9.315990818668707,
      "grad_norm": 3.1322076320648193,
      "learning_rate": 3.7762808736476835e-06,
      "loss": 1.1268,
      "step": 15220
    },
    {
      "epoch": 9.322111706197399,
      "grad_norm": 3.045135736465454,
      "learning_rate": 3.742260325236443e-06,
      "loss": 1.1352,
      "step": 15230
    },
    {
      "epoch": 9.32823259372609,
      "grad_norm": 2.932527542114258,
      "learning_rate": 3.708239776825203e-06,
      "loss": 1.1294,
      "step": 15240
    },
    {
      "epoch": 9.334353481254782,
      "grad_norm": 3.5815134048461914,
      "learning_rate": 3.674219228413962e-06,
      "loss": 1.1304,
      "step": 15250
    },
    {
      "epoch": 9.340474368783473,
      "grad_norm": 3.414130449295044,
      "learning_rate": 3.6401986800027217e-06,
      "loss": 1.1232,
      "step": 15260
    },
    {
      "epoch": 9.346595256312165,
      "grad_norm": 5.269806861877441,
      "learning_rate": 3.6061781315914813e-06,
      "loss": 1.121,
      "step": 15270
    },
    {
      "epoch": 9.352716143840857,
      "grad_norm": 3.498656749725342,
      "learning_rate": 3.5721575831802413e-06,
      "loss": 1.1319,
      "step": 15280
    },
    {
      "epoch": 9.358837031369548,
      "grad_norm": 3.1993322372436523,
      "learning_rate": 3.5381370347690004e-06,
      "loss": 1.1244,
      "step": 15290
    },
    {
      "epoch": 9.36495791889824,
      "grad_norm": 3.3193633556365967,
      "learning_rate": 3.50411648635776e-06,
      "loss": 1.1454,
      "step": 15300
    },
    {
      "epoch": 9.371078806426931,
      "grad_norm": 3.172529697418213,
      "learning_rate": 3.47009593794652e-06,
      "loss": 1.1373,
      "step": 15310
    },
    {
      "epoch": 9.377199693955623,
      "grad_norm": 3.223278045654297,
      "learning_rate": 3.4360753895352795e-06,
      "loss": 1.1099,
      "step": 15320
    },
    {
      "epoch": 9.383320581484316,
      "grad_norm": 3.193432569503784,
      "learning_rate": 3.4020548411240386e-06,
      "loss": 1.121,
      "step": 15330
    },
    {
      "epoch": 9.389441469013008,
      "grad_norm": 3.438258647918701,
      "learning_rate": 3.368034292712799e-06,
      "loss": 1.1343,
      "step": 15340
    },
    {
      "epoch": 9.3955623565417,
      "grad_norm": 2.990633964538574,
      "learning_rate": 3.334013744301558e-06,
      "loss": 1.1368,
      "step": 15350
    },
    {
      "epoch": 9.40168324407039,
      "grad_norm": 3.5829241275787354,
      "learning_rate": 3.2999931958903178e-06,
      "loss": 1.1209,
      "step": 15360
    },
    {
      "epoch": 9.407804131599082,
      "grad_norm": 3.2910852432250977,
      "learning_rate": 3.2659726474790777e-06,
      "loss": 1.1385,
      "step": 15370
    },
    {
      "epoch": 9.413925019127774,
      "grad_norm": 2.937903642654419,
      "learning_rate": 3.2319520990678373e-06,
      "loss": 1.124,
      "step": 15380
    },
    {
      "epoch": 9.420045906656465,
      "grad_norm": 3.887308359146118,
      "learning_rate": 3.1979315506565964e-06,
      "loss": 1.1046,
      "step": 15390
    },
    {
      "epoch": 9.426166794185157,
      "grad_norm": 3.3708431720733643,
      "learning_rate": 3.1639110022453564e-06,
      "loss": 1.1138,
      "step": 15400
    },
    {
      "epoch": 9.432287681713849,
      "grad_norm": 3.5277280807495117,
      "learning_rate": 3.129890453834116e-06,
      "loss": 1.1523,
      "step": 15410
    },
    {
      "epoch": 9.43840856924254,
      "grad_norm": 2.8049731254577637,
      "learning_rate": 3.0958699054228756e-06,
      "loss": 1.1184,
      "step": 15420
    },
    {
      "epoch": 9.444529456771232,
      "grad_norm": 3.5323147773742676,
      "learning_rate": 3.061849357011635e-06,
      "loss": 1.1315,
      "step": 15430
    },
    {
      "epoch": 9.450650344299923,
      "grad_norm": 3.5388619899749756,
      "learning_rate": 3.0278288086003947e-06,
      "loss": 1.1142,
      "step": 15440
    },
    {
      "epoch": 9.456771231828615,
      "grad_norm": 3.382925033569336,
      "learning_rate": 2.9938082601891542e-06,
      "loss": 1.1183,
      "step": 15450
    },
    {
      "epoch": 9.462892119357306,
      "grad_norm": 3.5912578105926514,
      "learning_rate": 2.959787711777914e-06,
      "loss": 1.1363,
      "step": 15460
    },
    {
      "epoch": 9.469013006885998,
      "grad_norm": 3.481766700744629,
      "learning_rate": 2.925767163366674e-06,
      "loss": 1.1075,
      "step": 15470
    },
    {
      "epoch": 9.47513389441469,
      "grad_norm": 3.3989980220794678,
      "learning_rate": 2.891746614955433e-06,
      "loss": 1.1367,
      "step": 15480
    },
    {
      "epoch": 9.481254781943381,
      "grad_norm": 3.504020929336548,
      "learning_rate": 2.857726066544193e-06,
      "loss": 1.1343,
      "step": 15490
    },
    {
      "epoch": 9.487375669472073,
      "grad_norm": 3.38145112991333,
      "learning_rate": 2.8237055181329525e-06,
      "loss": 1.1064,
      "step": 15500
    },
    {
      "epoch": 9.493496557000766,
      "grad_norm": 3.2473201751708984,
      "learning_rate": 2.789684969721712e-06,
      "loss": 1.135,
      "step": 15510
    },
    {
      "epoch": 9.499617444529457,
      "grad_norm": 3.069551706314087,
      "learning_rate": 2.7556644213104716e-06,
      "loss": 1.1204,
      "step": 15520
    },
    {
      "epoch": 9.505738332058149,
      "grad_norm": 3.7915170192718506,
      "learning_rate": 2.721643872899231e-06,
      "loss": 1.1449,
      "step": 15530
    },
    {
      "epoch": 9.51185921958684,
      "grad_norm": 3.12088680267334,
      "learning_rate": 2.6876233244879907e-06,
      "loss": 1.1241,
      "step": 15540
    },
    {
      "epoch": 9.517980107115532,
      "grad_norm": 5.423216342926025,
      "learning_rate": 2.6536027760767507e-06,
      "loss": 1.1169,
      "step": 15550
    },
    {
      "epoch": 9.524100994644224,
      "grad_norm": 3.776402473449707,
      "learning_rate": 2.61958222766551e-06,
      "loss": 1.1351,
      "step": 15560
    },
    {
      "epoch": 9.530221882172915,
      "grad_norm": 3.071627616882324,
      "learning_rate": 2.58556167925427e-06,
      "loss": 1.1043,
      "step": 15570
    },
    {
      "epoch": 9.536342769701607,
      "grad_norm": 4.0253825187683105,
      "learning_rate": 2.5515411308430294e-06,
      "loss": 1.1118,
      "step": 15580
    },
    {
      "epoch": 9.542463657230298,
      "grad_norm": 3.3617985248565674,
      "learning_rate": 2.517520582431789e-06,
      "loss": 1.1208,
      "step": 15590
    },
    {
      "epoch": 9.54858454475899,
      "grad_norm": 2.9002251625061035,
      "learning_rate": 2.4835000340205485e-06,
      "loss": 1.1143,
      "step": 15600
    },
    {
      "epoch": 9.554705432287681,
      "grad_norm": 3.4758708477020264,
      "learning_rate": 2.449479485609308e-06,
      "loss": 1.1395,
      "step": 15610
    },
    {
      "epoch": 9.560826319816373,
      "grad_norm": 2.933244228363037,
      "learning_rate": 2.4154589371980677e-06,
      "loss": 1.1125,
      "step": 15620
    },
    {
      "epoch": 9.566947207345065,
      "grad_norm": 3.2921762466430664,
      "learning_rate": 2.3814383887868276e-06,
      "loss": 1.1387,
      "step": 15630
    },
    {
      "epoch": 9.573068094873756,
      "grad_norm": 3.7236812114715576,
      "learning_rate": 2.3474178403755868e-06,
      "loss": 1.1284,
      "step": 15640
    },
    {
      "epoch": 9.579188982402448,
      "grad_norm": 3.9367823600769043,
      "learning_rate": 2.3133972919643468e-06,
      "loss": 1.1343,
      "step": 15650
    },
    {
      "epoch": 9.58530986993114,
      "grad_norm": 3.410693645477295,
      "learning_rate": 2.2793767435531063e-06,
      "loss": 1.1268,
      "step": 15660
    },
    {
      "epoch": 9.591430757459833,
      "grad_norm": 3.306009292602539,
      "learning_rate": 2.245356195141866e-06,
      "loss": 1.1068,
      "step": 15670
    },
    {
      "epoch": 9.597551644988524,
      "grad_norm": 3.4138855934143066,
      "learning_rate": 2.2113356467306255e-06,
      "loss": 1.1455,
      "step": 15680
    },
    {
      "epoch": 9.603672532517216,
      "grad_norm": 3.4779481887817383,
      "learning_rate": 2.177315098319385e-06,
      "loss": 1.1274,
      "step": 15690
    },
    {
      "epoch": 9.609793420045907,
      "grad_norm": 3.2200963497161865,
      "learning_rate": 2.1432945499081446e-06,
      "loss": 1.1384,
      "step": 15700
    },
    {
      "epoch": 9.615914307574599,
      "grad_norm": 3.1388471126556396,
      "learning_rate": 2.109274001496904e-06,
      "loss": 1.1296,
      "step": 15710
    },
    {
      "epoch": 9.62203519510329,
      "grad_norm": 3.209235668182373,
      "learning_rate": 2.0752534530856637e-06,
      "loss": 1.1473,
      "step": 15720
    },
    {
      "epoch": 9.628156082631982,
      "grad_norm": 3.1397557258605957,
      "learning_rate": 2.0412329046744233e-06,
      "loss": 1.1258,
      "step": 15730
    },
    {
      "epoch": 9.634276970160673,
      "grad_norm": 3.0715200901031494,
      "learning_rate": 2.0072123562631833e-06,
      "loss": 1.1275,
      "step": 15740
    },
    {
      "epoch": 9.640397857689365,
      "grad_norm": 2.979800224304199,
      "learning_rate": 1.9731918078519424e-06,
      "loss": 1.1087,
      "step": 15750
    },
    {
      "epoch": 9.646518745218057,
      "grad_norm": 3.0174732208251953,
      "learning_rate": 1.9391712594407024e-06,
      "loss": 1.125,
      "step": 15760
    },
    {
      "epoch": 9.652639632746748,
      "grad_norm": 3.9182639122009277,
      "learning_rate": 1.905150711029462e-06,
      "loss": 1.1404,
      "step": 15770
    },
    {
      "epoch": 9.65876052027544,
      "grad_norm": 3.180032730102539,
      "learning_rate": 1.8711301626182215e-06,
      "loss": 1.1258,
      "step": 15780
    },
    {
      "epoch": 9.664881407804131,
      "grad_norm": 3.5461888313293457,
      "learning_rate": 1.837109614206981e-06,
      "loss": 1.1234,
      "step": 15790
    },
    {
      "epoch": 9.671002295332823,
      "grad_norm": 3.3645544052124023,
      "learning_rate": 1.8030890657957406e-06,
      "loss": 1.1248,
      "step": 15800
    },
    {
      "epoch": 9.677123182861514,
      "grad_norm": 3.192190170288086,
      "learning_rate": 1.7690685173845002e-06,
      "loss": 1.1435,
      "step": 15810
    },
    {
      "epoch": 9.683244070390206,
      "grad_norm": 3.084299325942993,
      "learning_rate": 1.73504796897326e-06,
      "loss": 1.113,
      "step": 15820
    },
    {
      "epoch": 9.689364957918897,
      "grad_norm": 3.075233221054077,
      "learning_rate": 1.7010274205620193e-06,
      "loss": 1.1314,
      "step": 15830
    },
    {
      "epoch": 9.695485845447589,
      "grad_norm": 2.9223477840423584,
      "learning_rate": 1.667006872150779e-06,
      "loss": 1.1218,
      "step": 15840
    },
    {
      "epoch": 9.701606732976282,
      "grad_norm": 2.988662004470825,
      "learning_rate": 1.6329863237395389e-06,
      "loss": 1.1205,
      "step": 15850
    },
    {
      "epoch": 9.707727620504974,
      "grad_norm": 3.0018911361694336,
      "learning_rate": 1.5989657753282982e-06,
      "loss": 1.1387,
      "step": 15860
    },
    {
      "epoch": 9.713848508033665,
      "grad_norm": 3.0322518348693848,
      "learning_rate": 1.564945226917058e-06,
      "loss": 1.1287,
      "step": 15870
    },
    {
      "epoch": 9.719969395562357,
      "grad_norm": 3.4595978260040283,
      "learning_rate": 1.5309246785058176e-06,
      "loss": 1.1386,
      "step": 15880
    },
    {
      "epoch": 9.726090283091049,
      "grad_norm": 3.31239914894104,
      "learning_rate": 1.4969041300945771e-06,
      "loss": 1.1349,
      "step": 15890
    },
    {
      "epoch": 9.73221117061974,
      "grad_norm": 2.9286789894104004,
      "learning_rate": 1.462883581683337e-06,
      "loss": 1.122,
      "step": 15900
    },
    {
      "epoch": 9.738332058148432,
      "grad_norm": 3.4690709114074707,
      "learning_rate": 1.4288630332720965e-06,
      "loss": 1.1368,
      "step": 15910
    },
    {
      "epoch": 9.744452945677123,
      "grad_norm": 3.5552656650543213,
      "learning_rate": 1.394842484860856e-06,
      "loss": 1.1394,
      "step": 15920
    },
    {
      "epoch": 9.750573833205815,
      "grad_norm": 3.3070826530456543,
      "learning_rate": 1.3608219364496156e-06,
      "loss": 1.1426,
      "step": 15930
    },
    {
      "epoch": 9.756694720734506,
      "grad_norm": 3.5823755264282227,
      "learning_rate": 1.3268013880383754e-06,
      "loss": 1.1264,
      "step": 15940
    },
    {
      "epoch": 9.762815608263198,
      "grad_norm": 2.63779354095459,
      "learning_rate": 1.292780839627135e-06,
      "loss": 1.1305,
      "step": 15950
    },
    {
      "epoch": 9.76893649579189,
      "grad_norm": 2.931502103805542,
      "learning_rate": 1.2587602912158945e-06,
      "loss": 1.118,
      "step": 15960
    },
    {
      "epoch": 9.775057383320581,
      "grad_norm": 2.708097457885742,
      "learning_rate": 1.224739742804654e-06,
      "loss": 1.1333,
      "step": 15970
    },
    {
      "epoch": 9.781178270849272,
      "grad_norm": 3.4733378887176514,
      "learning_rate": 1.1907191943934138e-06,
      "loss": 1.1304,
      "step": 15980
    },
    {
      "epoch": 9.787299158377964,
      "grad_norm": 3.208436965942383,
      "learning_rate": 1.1566986459821734e-06,
      "loss": 1.116,
      "step": 15990
    },
    {
      "epoch": 9.793420045906657,
      "grad_norm": 3.083686351776123,
      "learning_rate": 1.122678097570933e-06,
      "loss": 1.1345,
      "step": 16000
    },
    {
      "epoch": 9.799540933435349,
      "grad_norm": 5.919510841369629,
      "learning_rate": 1.0886575491596925e-06,
      "loss": 1.1295,
      "step": 16010
    },
    {
      "epoch": 9.80566182096404,
      "grad_norm": 3.0980031490325928,
      "learning_rate": 1.054637000748452e-06,
      "loss": 1.1264,
      "step": 16020
    },
    {
      "epoch": 9.811782708492732,
      "grad_norm": 2.991114854812622,
      "learning_rate": 1.0206164523372116e-06,
      "loss": 1.1158,
      "step": 16030
    },
    {
      "epoch": 9.817903596021424,
      "grad_norm": 2.762467861175537,
      "learning_rate": 9.865959039259712e-07,
      "loss": 1.1482,
      "step": 16040
    },
    {
      "epoch": 9.824024483550115,
      "grad_norm": 2.762880325317383,
      "learning_rate": 9.52575355514731e-07,
      "loss": 1.1252,
      "step": 16050
    },
    {
      "epoch": 9.830145371078807,
      "grad_norm": 3.7553000450134277,
      "learning_rate": 9.185548071034905e-07,
      "loss": 1.1354,
      "step": 16060
    },
    {
      "epoch": 9.836266258607498,
      "grad_norm": 3.1074488162994385,
      "learning_rate": 8.845342586922501e-07,
      "loss": 1.1433,
      "step": 16070
    },
    {
      "epoch": 9.84238714613619,
      "grad_norm": 3.9977195262908936,
      "learning_rate": 8.505137102810097e-07,
      "loss": 1.1506,
      "step": 16080
    },
    {
      "epoch": 9.848508033664881,
      "grad_norm": 2.973998785018921,
      "learning_rate": 8.164931618697694e-07,
      "loss": 1.138,
      "step": 16090
    },
    {
      "epoch": 9.854628921193573,
      "grad_norm": 2.8410356044769287,
      "learning_rate": 7.82472613458529e-07,
      "loss": 1.1325,
      "step": 16100
    },
    {
      "epoch": 9.860749808722264,
      "grad_norm": 3.599996328353882,
      "learning_rate": 7.484520650472886e-07,
      "loss": 1.1124,
      "step": 16110
    },
    {
      "epoch": 9.866870696250956,
      "grad_norm": 3.0998120307922363,
      "learning_rate": 7.144315166360482e-07,
      "loss": 1.1207,
      "step": 16120
    },
    {
      "epoch": 9.872991583779648,
      "grad_norm": 4.06972599029541,
      "learning_rate": 6.804109682248078e-07,
      "loss": 1.1025,
      "step": 16130
    },
    {
      "epoch": 9.87911247130834,
      "grad_norm": 4.1819329261779785,
      "learning_rate": 6.463904198135675e-07,
      "loss": 1.115,
      "step": 16140
    },
    {
      "epoch": 9.88523335883703,
      "grad_norm": 3.015798568725586,
      "learning_rate": 6.12369871402327e-07,
      "loss": 1.1476,
      "step": 16150
    },
    {
      "epoch": 9.891354246365722,
      "grad_norm": 3.4522454738616943,
      "learning_rate": 5.783493229910867e-07,
      "loss": 1.1165,
      "step": 16160
    },
    {
      "epoch": 9.897475133894414,
      "grad_norm": 3.4204092025756836,
      "learning_rate": 5.443287745798463e-07,
      "loss": 1.1144,
      "step": 16170
    },
    {
      "epoch": 9.903596021423107,
      "grad_norm": 3.0444397926330566,
      "learning_rate": 5.103082261686058e-07,
      "loss": 1.1305,
      "step": 16180
    },
    {
      "epoch": 9.909716908951799,
      "grad_norm": 3.3424997329711914,
      "learning_rate": 4.762876777573655e-07,
      "loss": 1.1561,
      "step": 16190
    },
    {
      "epoch": 9.91583779648049,
      "grad_norm": 2.94327449798584,
      "learning_rate": 4.4226712934612505e-07,
      "loss": 1.1472,
      "step": 16200
    },
    {
      "epoch": 9.921958684009182,
      "grad_norm": 3.0581696033477783,
      "learning_rate": 4.082465809348847e-07,
      "loss": 1.1122,
      "step": 16210
    },
    {
      "epoch": 9.928079571537873,
      "grad_norm": 3.0790395736694336,
      "learning_rate": 3.742260325236443e-07,
      "loss": 1.1427,
      "step": 16220
    },
    {
      "epoch": 9.934200459066565,
      "grad_norm": 3.1817245483398438,
      "learning_rate": 3.402054841124039e-07,
      "loss": 1.1301,
      "step": 16230
    },
    {
      "epoch": 9.940321346595256,
      "grad_norm": 2.9370524883270264,
      "learning_rate": 3.061849357011635e-07,
      "loss": 1.1217,
      "step": 16240
    },
    {
      "epoch": 9.946442234123948,
      "grad_norm": 3.528365135192871,
      "learning_rate": 2.7216438728992313e-07,
      "loss": 1.1252,
      "step": 16250
    },
    {
      "epoch": 9.95256312165264,
      "grad_norm": 2.9326071739196777,
      "learning_rate": 2.3814383887868274e-07,
      "loss": 1.109,
      "step": 16260
    },
    {
      "epoch": 9.958684009181331,
      "grad_norm": 3.28116512298584,
      "learning_rate": 2.0412329046744236e-07,
      "loss": 1.1162,
      "step": 16270
    },
    {
      "epoch": 9.964804896710023,
      "grad_norm": 2.973410129547119,
      "learning_rate": 1.7010274205620195e-07,
      "loss": 1.112,
      "step": 16280
    },
    {
      "epoch": 9.970925784238714,
      "grad_norm": 3.1284003257751465,
      "learning_rate": 1.3608219364496156e-07,
      "loss": 1.1297,
      "step": 16290
    },
    {
      "epoch": 9.977046671767406,
      "grad_norm": 3.262115001678467,
      "learning_rate": 1.0206164523372118e-07,
      "loss": 1.1359,
      "step": 16300
    },
    {
      "epoch": 9.983167559296097,
      "grad_norm": 4.041773796081543,
      "learning_rate": 6.804109682248078e-08,
      "loss": 1.1251,
      "step": 16310
    },
    {
      "epoch": 9.989288446824789,
      "grad_norm": 3.1230151653289795,
      "learning_rate": 3.402054841124039e-08,
      "loss": 1.1696,
      "step": 16320
    },
    {
      "epoch": 9.99540933435348,
      "grad_norm": 5.236699104309082,
      "learning_rate": 0.0,
      "loss": 1.1121,
      "step": 16330
    },
    {
      "epoch": 9.99540933435348,
      "eval_accuracy": 0.5118223208478402,
      "eval_loss": 1.1701384782791138,
      "eval_runtime": 2639.5684,
      "eval_samples_per_second": 79.216,
      "eval_steps_per_second": 0.619,
      "step": 16330
    },
    {
      "epoch": 9.99540933435348,
      "step": 16330,
      "total_flos": 8.570305751657333e+20,
      "train_loss": 1.22920058554692,
      "train_runtime": 204809.1788,
      "train_samples_per_second": 40.836,
      "train_steps_per_second": 0.08
    }
  ],
  "logging_steps": 10,
  "max_steps": 16330,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "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": 8.570305751657333e+20,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}