{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.2401104508073714,
  "eval_steps": 375,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016007363387158093,
      "grad_norm": 5.485477959155105e-05,
      "learning_rate": 2e-05,
      "loss": 11.5,
      "step": 1
    },
    {
      "epoch": 0.00016007363387158093,
      "eval_loss": 11.5,
      "eval_runtime": 61.3751,
      "eval_samples_per_second": 171.438,
      "eval_steps_per_second": 85.719,
      "step": 1
    },
    {
      "epoch": 0.00032014726774316185,
      "grad_norm": 8.732182323001325e-05,
      "learning_rate": 4e-05,
      "loss": 11.5,
      "step": 2
    },
    {
      "epoch": 0.0004802209016147428,
      "grad_norm": 7.794566045049578e-05,
      "learning_rate": 6e-05,
      "loss": 11.5,
      "step": 3
    },
    {
      "epoch": 0.0006402945354863237,
      "grad_norm": 0.00010553398169577122,
      "learning_rate": 8e-05,
      "loss": 11.5,
      "step": 4
    },
    {
      "epoch": 0.0008003681693579046,
      "grad_norm": 9.141851478489116e-05,
      "learning_rate": 0.0001,
      "loss": 11.5,
      "step": 5
    },
    {
      "epoch": 0.0009604418032294856,
      "grad_norm": 7.036867464194074e-05,
      "learning_rate": 0.00012,
      "loss": 11.5,
      "step": 6
    },
    {
      "epoch": 0.0011205154371010666,
      "grad_norm": 0.00010718707926571369,
      "learning_rate": 0.00014,
      "loss": 11.5,
      "step": 7
    },
    {
      "epoch": 0.0012805890709726474,
      "grad_norm": 5.46959818166215e-05,
      "learning_rate": 0.00016,
      "loss": 11.5,
      "step": 8
    },
    {
      "epoch": 0.0014406627048442284,
      "grad_norm": 6.93291804054752e-05,
      "learning_rate": 0.00018,
      "loss": 11.5,
      "step": 9
    },
    {
      "epoch": 0.0016007363387158093,
      "grad_norm": 6.610731361433864e-05,
      "learning_rate": 0.0002,
      "loss": 11.5,
      "step": 10
    },
    {
      "epoch": 0.0017608099725873903,
      "grad_norm": 7.073202141327783e-05,
      "learning_rate": 0.00019999977772170748,
      "loss": 11.5,
      "step": 11
    },
    {
      "epoch": 0.0019208836064589711,
      "grad_norm": 5.89275878155604e-05,
      "learning_rate": 0.00019999911088781805,
      "loss": 11.5,
      "step": 12
    },
    {
      "epoch": 0.002080957240330552,
      "grad_norm": 7.123701652744785e-05,
      "learning_rate": 0.0001999979995012962,
      "loss": 11.5,
      "step": 13
    },
    {
      "epoch": 0.002241030874202133,
      "grad_norm": 7.767384522594512e-05,
      "learning_rate": 0.00019999644356708261,
      "loss": 11.5,
      "step": 14
    },
    {
      "epoch": 0.002401104508073714,
      "grad_norm": 0.00010461104830028489,
      "learning_rate": 0.00019999444309209432,
      "loss": 11.5,
      "step": 15
    },
    {
      "epoch": 0.002561178141945295,
      "grad_norm": 0.00013544182002078742,
      "learning_rate": 0.0001999919980852246,
      "loss": 11.5,
      "step": 16
    },
    {
      "epoch": 0.0027212517758168756,
      "grad_norm": 0.00012140026228735223,
      "learning_rate": 0.00019998910855734288,
      "loss": 11.5,
      "step": 17
    },
    {
      "epoch": 0.002881325409688457,
      "grad_norm": 0.00015228302800096571,
      "learning_rate": 0.0001999857745212947,
      "loss": 11.5,
      "step": 18
    },
    {
      "epoch": 0.0030413990435600377,
      "grad_norm": 0.00013085530372336507,
      "learning_rate": 0.00019998199599190178,
      "loss": 11.5,
      "step": 19
    },
    {
      "epoch": 0.0032014726774316185,
      "grad_norm": 0.00015696501941420138,
      "learning_rate": 0.0001999777729859618,
      "loss": 11.5,
      "step": 20
    },
    {
      "epoch": 0.0033615463113031993,
      "grad_norm": 0.00011811301374109462,
      "learning_rate": 0.00019997310552224846,
      "loss": 11.5,
      "step": 21
    },
    {
      "epoch": 0.0035216199451747806,
      "grad_norm": 0.00017093494534492493,
      "learning_rate": 0.00019996799362151122,
      "loss": 11.5,
      "step": 22
    },
    {
      "epoch": 0.0036816935790463614,
      "grad_norm": 0.00016428295930381864,
      "learning_rate": 0.00019996243730647538,
      "loss": 11.5,
      "step": 23
    },
    {
      "epoch": 0.0038417672129179422,
      "grad_norm": 0.00017610278155189008,
      "learning_rate": 0.00019995643660184191,
      "loss": 11.5,
      "step": 24
    },
    {
      "epoch": 0.0040018408467895235,
      "grad_norm": 0.00022790305956732482,
      "learning_rate": 0.00019994999153428737,
      "loss": 11.5,
      "step": 25
    },
    {
      "epoch": 0.004161914480661104,
      "grad_norm": 0.00025935209123417735,
      "learning_rate": 0.00019994310213246368,
      "loss": 11.5,
      "step": 26
    },
    {
      "epoch": 0.004321988114532685,
      "grad_norm": 0.0002765641256701201,
      "learning_rate": 0.00019993576842699816,
      "loss": 11.5,
      "step": 27
    },
    {
      "epoch": 0.004482061748404266,
      "grad_norm": 0.00021294229372870177,
      "learning_rate": 0.0001999279904504933,
      "loss": 11.5,
      "step": 28
    },
    {
      "epoch": 0.004642135382275847,
      "grad_norm": 0.00019341120787430555,
      "learning_rate": 0.00019991976823752653,
      "loss": 11.5,
      "step": 29
    },
    {
      "epoch": 0.004802209016147428,
      "grad_norm": 0.0002532599901314825,
      "learning_rate": 0.00019991110182465032,
      "loss": 11.5,
      "step": 30
    },
    {
      "epoch": 0.004962282650019008,
      "grad_norm": 0.0003308654122520238,
      "learning_rate": 0.00019990199125039174,
      "loss": 11.5,
      "step": 31
    },
    {
      "epoch": 0.00512235628389059,
      "grad_norm": 0.00029195210663601756,
      "learning_rate": 0.00019989243655525247,
      "loss": 11.5,
      "step": 32
    },
    {
      "epoch": 0.005282429917762171,
      "grad_norm": 0.00023563031572848558,
      "learning_rate": 0.00019988243778170853,
      "loss": 11.5,
      "step": 33
    },
    {
      "epoch": 0.005442503551633751,
      "grad_norm": 0.00020833924645558,
      "learning_rate": 0.0001998719949742101,
      "loss": 11.5,
      "step": 34
    },
    {
      "epoch": 0.0056025771855053325,
      "grad_norm": 0.0004335304838605225,
      "learning_rate": 0.0001998611081791814,
      "loss": 11.5,
      "step": 35
    },
    {
      "epoch": 0.005762650819376914,
      "grad_norm": 0.0003254961338825524,
      "learning_rate": 0.00019984977744502038,
      "loss": 11.5,
      "step": 36
    },
    {
      "epoch": 0.005922724453248494,
      "grad_norm": 0.0003397865511942655,
      "learning_rate": 0.00019983800282209857,
      "loss": 11.5,
      "step": 37
    },
    {
      "epoch": 0.006082798087120075,
      "grad_norm": 0.00024661177303642035,
      "learning_rate": 0.00019982578436276082,
      "loss": 11.5,
      "step": 38
    },
    {
      "epoch": 0.006242871720991656,
      "grad_norm": 0.00032913414179347456,
      "learning_rate": 0.00019981312212132512,
      "loss": 11.5,
      "step": 39
    },
    {
      "epoch": 0.006402945354863237,
      "grad_norm": 0.00048708287067711353,
      "learning_rate": 0.00019980001615408228,
      "loss": 11.5,
      "step": 40
    },
    {
      "epoch": 0.006563018988734818,
      "grad_norm": 0.0004900452331639826,
      "learning_rate": 0.00019978646651929572,
      "loss": 11.5,
      "step": 41
    },
    {
      "epoch": 0.006723092622606399,
      "grad_norm": 0.00046467658830806613,
      "learning_rate": 0.00019977247327720128,
      "loss": 11.5,
      "step": 42
    },
    {
      "epoch": 0.00688316625647798,
      "grad_norm": 0.0006093564443290234,
      "learning_rate": 0.0001997580364900068,
      "loss": 11.5,
      "step": 43
    },
    {
      "epoch": 0.007043239890349561,
      "grad_norm": 0.0009993163403123617,
      "learning_rate": 0.000199743156221892,
      "loss": 11.5,
      "step": 44
    },
    {
      "epoch": 0.007203313524221142,
      "grad_norm": 0.001258009229786694,
      "learning_rate": 0.00019972783253900808,
      "loss": 11.5,
      "step": 45
    },
    {
      "epoch": 0.007363387158092723,
      "grad_norm": 0.001445628353394568,
      "learning_rate": 0.00019971206550947748,
      "loss": 11.5,
      "step": 46
    },
    {
      "epoch": 0.007523460791964303,
      "grad_norm": 0.0013488645199686289,
      "learning_rate": 0.00019969585520339354,
      "loss": 11.5,
      "step": 47
    },
    {
      "epoch": 0.0076835344258358845,
      "grad_norm": 0.0006975385476835072,
      "learning_rate": 0.0001996792016928203,
      "loss": 11.5,
      "step": 48
    },
    {
      "epoch": 0.007843608059707465,
      "grad_norm": 0.000834140635561198,
      "learning_rate": 0.00019966210505179197,
      "loss": 11.5,
      "step": 49
    },
    {
      "epoch": 0.008003681693579047,
      "grad_norm": 0.0004872016725130379,
      "learning_rate": 0.00019964456535631286,
      "loss": 11.5,
      "step": 50
    },
    {
      "epoch": 0.008163755327450627,
      "grad_norm": 0.00016838940791785717,
      "learning_rate": 0.0001996265826843568,
      "loss": 11.5,
      "step": 51
    },
    {
      "epoch": 0.008323828961322208,
      "grad_norm": 0.00023144090664573014,
      "learning_rate": 0.00019960815711586696,
      "loss": 11.5,
      "step": 52
    },
    {
      "epoch": 0.00848390259519379,
      "grad_norm": 0.00019563916430342942,
      "learning_rate": 0.00019958928873275539,
      "loss": 11.5,
      "step": 53
    },
    {
      "epoch": 0.00864397622906537,
      "grad_norm": 0.00020072546612937003,
      "learning_rate": 0.00019956997761890277,
      "loss": 11.5,
      "step": 54
    },
    {
      "epoch": 0.00880404986293695,
      "grad_norm": 0.000214951389352791,
      "learning_rate": 0.00019955022386015792,
      "loss": 11.5,
      "step": 55
    },
    {
      "epoch": 0.008964123496808533,
      "grad_norm": 0.0001576873182784766,
      "learning_rate": 0.00019953002754433743,
      "loss": 11.5,
      "step": 56
    },
    {
      "epoch": 0.009124197130680113,
      "grad_norm": 0.00016661152767483145,
      "learning_rate": 0.00019950938876122542,
      "loss": 11.5,
      "step": 57
    },
    {
      "epoch": 0.009284270764551694,
      "grad_norm": 0.00020268915977794677,
      "learning_rate": 0.00019948830760257291,
      "loss": 11.5,
      "step": 58
    },
    {
      "epoch": 0.009444344398423274,
      "grad_norm": 0.00017972583009395748,
      "learning_rate": 0.0001994667841620976,
      "loss": 11.5,
      "step": 59
    },
    {
      "epoch": 0.009604418032294856,
      "grad_norm": 0.00018623605137690902,
      "learning_rate": 0.00019944481853548335,
      "loss": 11.5,
      "step": 60
    },
    {
      "epoch": 0.009764491666166436,
      "grad_norm": 0.00023312588746193796,
      "learning_rate": 0.00019942241082037982,
      "loss": 11.5,
      "step": 61
    },
    {
      "epoch": 0.009924565300038017,
      "grad_norm": 0.0001705118193058297,
      "learning_rate": 0.00019939956111640197,
      "loss": 11.5,
      "step": 62
    },
    {
      "epoch": 0.010084638933909599,
      "grad_norm": 0.0002978912671096623,
      "learning_rate": 0.00019937626952512964,
      "loss": 11.5,
      "step": 63
    },
    {
      "epoch": 0.01024471256778118,
      "grad_norm": 0.00032203004229813814,
      "learning_rate": 0.0001993525361501072,
      "loss": 11.5,
      "step": 64
    },
    {
      "epoch": 0.01040478620165276,
      "grad_norm": 0.0003070418315473944,
      "learning_rate": 0.00019932836109684286,
      "loss": 11.5,
      "step": 65
    },
    {
      "epoch": 0.010564859835524342,
      "grad_norm": 0.0004711727669928223,
      "learning_rate": 0.00019930374447280845,
      "loss": 11.5,
      "step": 66
    },
    {
      "epoch": 0.010724933469395922,
      "grad_norm": 0.0004554349579848349,
      "learning_rate": 0.00019927868638743875,
      "loss": 11.5,
      "step": 67
    },
    {
      "epoch": 0.010885007103267503,
      "grad_norm": 0.0005644407356157899,
      "learning_rate": 0.0001992531869521312,
      "loss": 11.5,
      "step": 68
    },
    {
      "epoch": 0.011045080737139085,
      "grad_norm": 0.0006445281323976815,
      "learning_rate": 0.00019922724628024515,
      "loss": 11.5,
      "step": 69
    },
    {
      "epoch": 0.011205154371010665,
      "grad_norm": 0.00069785414962098,
      "learning_rate": 0.0001992008644871016,
      "loss": 11.5,
      "step": 70
    },
    {
      "epoch": 0.011365228004882245,
      "grad_norm": 0.0009710849262773991,
      "learning_rate": 0.00019917404168998256,
      "loss": 11.5,
      "step": 71
    },
    {
      "epoch": 0.011525301638753828,
      "grad_norm": 0.0006939312443137169,
      "learning_rate": 0.0001991467780081305,
      "loss": 11.5,
      "step": 72
    },
    {
      "epoch": 0.011685375272625408,
      "grad_norm": 0.0008124214364215732,
      "learning_rate": 0.00019911907356274795,
      "loss": 11.5,
      "step": 73
    },
    {
      "epoch": 0.011845448906496988,
      "grad_norm": 0.0010405608918517828,
      "learning_rate": 0.00019909092847699683,
      "loss": 11.5,
      "step": 74
    },
    {
      "epoch": 0.012005522540368569,
      "grad_norm": 0.0007668936741538346,
      "learning_rate": 0.00019906234287599798,
      "loss": 11.5,
      "step": 75
    },
    {
      "epoch": 0.01216559617424015,
      "grad_norm": 0.0009385758312419057,
      "learning_rate": 0.00019903331688683057,
      "loss": 11.5,
      "step": 76
    },
    {
      "epoch": 0.012325669808111731,
      "grad_norm": 0.0007635114598087966,
      "learning_rate": 0.00019900385063853154,
      "loss": 11.5,
      "step": 77
    },
    {
      "epoch": 0.012485743441983312,
      "grad_norm": 0.0008907768642529845,
      "learning_rate": 0.00019897394426209505,
      "loss": 11.5,
      "step": 78
    },
    {
      "epoch": 0.012645817075854894,
      "grad_norm": 0.0007882064091973007,
      "learning_rate": 0.00019894359789047187,
      "loss": 11.5,
      "step": 79
    },
    {
      "epoch": 0.012805890709726474,
      "grad_norm": 0.0009297032374888659,
      "learning_rate": 0.00019891281165856873,
      "loss": 11.5,
      "step": 80
    },
    {
      "epoch": 0.012965964343598054,
      "grad_norm": 0.0008925965521484613,
      "learning_rate": 0.00019888158570324795,
      "loss": 11.5,
      "step": 81
    },
    {
      "epoch": 0.013126037977469637,
      "grad_norm": 0.0011857181088998914,
      "learning_rate": 0.0001988499201633265,
      "loss": 11.5,
      "step": 82
    },
    {
      "epoch": 0.013286111611341217,
      "grad_norm": 0.0008185437764041126,
      "learning_rate": 0.00019881781517957562,
      "loss": 11.5,
      "step": 83
    },
    {
      "epoch": 0.013446185245212797,
      "grad_norm": 0.0010364435147494078,
      "learning_rate": 0.0001987852708947202,
      "loss": 11.5,
      "step": 84
    },
    {
      "epoch": 0.01360625887908438,
      "grad_norm": 0.0013269423507153988,
      "learning_rate": 0.00019875228745343794,
      "loss": 11.5,
      "step": 85
    },
    {
      "epoch": 0.01376633251295596,
      "grad_norm": 0.0012907687341794372,
      "learning_rate": 0.0001987188650023589,
      "loss": 11.5,
      "step": 86
    },
    {
      "epoch": 0.01392640614682754,
      "grad_norm": 0.0011142597068101168,
      "learning_rate": 0.0001986850036900648,
      "loss": 11.5,
      "step": 87
    },
    {
      "epoch": 0.014086479780699122,
      "grad_norm": 0.0012716882629320025,
      "learning_rate": 0.00019865070366708836,
      "loss": 11.5,
      "step": 88
    },
    {
      "epoch": 0.014246553414570703,
      "grad_norm": 0.0014261571923270822,
      "learning_rate": 0.00019861596508591255,
      "loss": 11.5,
      "step": 89
    },
    {
      "epoch": 0.014406627048442283,
      "grad_norm": 0.0009479870204813778,
      "learning_rate": 0.00019858078810097002,
      "loss": 11.5,
      "step": 90
    },
    {
      "epoch": 0.014566700682313864,
      "grad_norm": 0.0018024398013949394,
      "learning_rate": 0.00019854517286864245,
      "loss": 11.5,
      "step": 91
    },
    {
      "epoch": 0.014726774316185446,
      "grad_norm": 0.0014115448575466871,
      "learning_rate": 0.0001985091195472596,
      "loss": 11.5,
      "step": 92
    },
    {
      "epoch": 0.014886847950057026,
      "grad_norm": 0.0021201353520154953,
      "learning_rate": 0.0001984726282970989,
      "loss": 11.5,
      "step": 93
    },
    {
      "epoch": 0.015046921583928606,
      "grad_norm": 0.001981760375201702,
      "learning_rate": 0.0001984356992803847,
      "loss": 11.5,
      "step": 94
    },
    {
      "epoch": 0.015206995217800189,
      "grad_norm": 0.0031021598260849714,
      "learning_rate": 0.00019839833266128724,
      "loss": 11.5,
      "step": 95
    },
    {
      "epoch": 0.015367068851671769,
      "grad_norm": 0.004780913703143597,
      "learning_rate": 0.00019836052860592237,
      "loss": 11.5,
      "step": 96
    },
    {
      "epoch": 0.01552714248554335,
      "grad_norm": 0.0040374197997152805,
      "learning_rate": 0.0001983222872823505,
      "loss": 11.5,
      "step": 97
    },
    {
      "epoch": 0.01568721611941493,
      "grad_norm": 0.0029658880084753036,
      "learning_rate": 0.00019828360886057594,
      "loss": 11.5,
      "step": 98
    },
    {
      "epoch": 0.01584728975328651,
      "grad_norm": 0.0017225135816261172,
      "learning_rate": 0.00019824449351254616,
      "loss": 11.5,
      "step": 99
    },
    {
      "epoch": 0.016007363387158094,
      "grad_norm": 0.001711854711174965,
      "learning_rate": 0.00019820494141215104,
      "loss": 11.5,
      "step": 100
    },
    {
      "epoch": 0.016167437021029674,
      "grad_norm": 0.0003945448261220008,
      "learning_rate": 0.000198164952735222,
      "loss": 11.5,
      "step": 101
    },
    {
      "epoch": 0.016327510654901255,
      "grad_norm": 0.0003840984427370131,
      "learning_rate": 0.00019812452765953135,
      "loss": 11.5,
      "step": 102
    },
    {
      "epoch": 0.016487584288772835,
      "grad_norm": 0.0004888374824076891,
      "learning_rate": 0.00019808366636479147,
      "loss": 11.5,
      "step": 103
    },
    {
      "epoch": 0.016647657922644415,
      "grad_norm": 0.0005102389259263873,
      "learning_rate": 0.00019804236903265388,
      "loss": 11.5,
      "step": 104
    },
    {
      "epoch": 0.016807731556515996,
      "grad_norm": 0.0005896842922084033,
      "learning_rate": 0.00019800063584670863,
      "loss": 11.5,
      "step": 105
    },
    {
      "epoch": 0.01696780519038758,
      "grad_norm": 0.0005505888257175684,
      "learning_rate": 0.00019795846699248332,
      "loss": 11.5,
      "step": 106
    },
    {
      "epoch": 0.01712787882425916,
      "grad_norm": 0.00038450522697530687,
      "learning_rate": 0.00019791586265744237,
      "loss": 11.5,
      "step": 107
    },
    {
      "epoch": 0.01728795245813074,
      "grad_norm": 0.000713772140443325,
      "learning_rate": 0.00019787282303098617,
      "loss": 11.5,
      "step": 108
    },
    {
      "epoch": 0.01744802609200232,
      "grad_norm": 0.0007358138682320714,
      "learning_rate": 0.0001978293483044502,
      "loss": 11.5,
      "step": 109
    },
    {
      "epoch": 0.0176080997258739,
      "grad_norm": 0.00032157220994122326,
      "learning_rate": 0.00019778543867110426,
      "loss": 11.5,
      "step": 110
    },
    {
      "epoch": 0.01776817335974548,
      "grad_norm": 0.0005707935779355466,
      "learning_rate": 0.00019774109432615147,
      "loss": 11.5,
      "step": 111
    },
    {
      "epoch": 0.017928246993617065,
      "grad_norm": 0.00048133719246834517,
      "learning_rate": 0.00019769631546672756,
      "loss": 11.5,
      "step": 112
    },
    {
      "epoch": 0.018088320627488646,
      "grad_norm": 0.00041901893564499915,
      "learning_rate": 0.00019765110229189988,
      "loss": 11.5,
      "step": 113
    },
    {
      "epoch": 0.018248394261360226,
      "grad_norm": 0.0005528696929104626,
      "learning_rate": 0.00019760545500266657,
      "loss": 11.5,
      "step": 114
    },
    {
      "epoch": 0.018408467895231807,
      "grad_norm": 0.001078005414456129,
      "learning_rate": 0.00019755937380195568,
      "loss": 11.5,
      "step": 115
    },
    {
      "epoch": 0.018568541529103387,
      "grad_norm": 0.001656021922826767,
      "learning_rate": 0.00019751285889462423,
      "loss": 11.5,
      "step": 116
    },
    {
      "epoch": 0.018728615162974967,
      "grad_norm": 0.0010341198649257421,
      "learning_rate": 0.0001974659104874573,
      "loss": 11.5,
      "step": 117
    },
    {
      "epoch": 0.018888688796846548,
      "grad_norm": 0.001869532628916204,
      "learning_rate": 0.0001974185287891671,
      "loss": 11.5,
      "step": 118
    },
    {
      "epoch": 0.01904876243071813,
      "grad_norm": 0.001640613074414432,
      "learning_rate": 0.0001973707140103921,
      "loss": 11.5,
      "step": 119
    },
    {
      "epoch": 0.019208836064589712,
      "grad_norm": 0.0012787277810275555,
      "learning_rate": 0.00019732246636369605,
      "loss": 11.5,
      "step": 120
    },
    {
      "epoch": 0.019368909698461292,
      "grad_norm": 0.001795590273104608,
      "learning_rate": 0.00019727378606356703,
      "loss": 11.5,
      "step": 121
    },
    {
      "epoch": 0.019528983332332873,
      "grad_norm": 0.002011685399338603,
      "learning_rate": 0.00019722467332641656,
      "loss": 11.5,
      "step": 122
    },
    {
      "epoch": 0.019689056966204453,
      "grad_norm": 0.0022899755276739597,
      "learning_rate": 0.00019717512837057855,
      "loss": 11.5,
      "step": 123
    },
    {
      "epoch": 0.019849130600076034,
      "grad_norm": 0.0017873673932626843,
      "learning_rate": 0.0001971251514163083,
      "loss": 11.5,
      "step": 124
    },
    {
      "epoch": 0.020009204233947617,
      "grad_norm": 0.002426765626296401,
      "learning_rate": 0.0001970747426857817,
      "loss": 11.5,
      "step": 125
    },
    {
      "epoch": 0.020169277867819198,
      "grad_norm": 0.0022798413410782814,
      "learning_rate": 0.00019702390240309404,
      "loss": 11.5,
      "step": 126
    },
    {
      "epoch": 0.020329351501690778,
      "grad_norm": 0.0030735242180526257,
      "learning_rate": 0.0001969726307942592,
      "loss": 11.5,
      "step": 127
    },
    {
      "epoch": 0.02048942513556236,
      "grad_norm": 0.001868702471256256,
      "learning_rate": 0.00019692092808720846,
      "loss": 11.5,
      "step": 128
    },
    {
      "epoch": 0.02064949876943394,
      "grad_norm": 0.00234604743309319,
      "learning_rate": 0.0001968687945117896,
      "loss": 11.5,
      "step": 129
    },
    {
      "epoch": 0.02080957240330552,
      "grad_norm": 0.0020907274447381496,
      "learning_rate": 0.00019681623029976588,
      "loss": 11.5,
      "step": 130
    },
    {
      "epoch": 0.020969646037177103,
      "grad_norm": 0.0021262471564114094,
      "learning_rate": 0.00019676323568481498,
      "loss": 11.5,
      "step": 131
    },
    {
      "epoch": 0.021129719671048684,
      "grad_norm": 0.00310887536033988,
      "learning_rate": 0.00019670981090252792,
      "loss": 11.5,
      "step": 132
    },
    {
      "epoch": 0.021289793304920264,
      "grad_norm": 0.002246802905574441,
      "learning_rate": 0.00019665595619040808,
      "loss": 11.5,
      "step": 133
    },
    {
      "epoch": 0.021449866938791844,
      "grad_norm": 0.002843370893970132,
      "learning_rate": 0.0001966016717878702,
      "loss": 11.5,
      "step": 134
    },
    {
      "epoch": 0.021609940572663425,
      "grad_norm": 0.002506015356630087,
      "learning_rate": 0.00019654695793623907,
      "loss": 11.5,
      "step": 135
    },
    {
      "epoch": 0.021770014206535005,
      "grad_norm": 0.0026175500825047493,
      "learning_rate": 0.0001964918148787488,
      "loss": 11.5,
      "step": 136
    },
    {
      "epoch": 0.021930087840406585,
      "grad_norm": 0.002090322319418192,
      "learning_rate": 0.00019643624286054144,
      "loss": 11.5,
      "step": 137
    },
    {
      "epoch": 0.02209016147427817,
      "grad_norm": 0.0033717560581862926,
      "learning_rate": 0.00019638024212866606,
      "loss": 11.5,
      "step": 138
    },
    {
      "epoch": 0.02225023510814975,
      "grad_norm": 0.0035848321858793497,
      "learning_rate": 0.0001963238129320776,
      "loss": 11.5,
      "step": 139
    },
    {
      "epoch": 0.02241030874202133,
      "grad_norm": 0.0018094985280185938,
      "learning_rate": 0.00019626695552163578,
      "loss": 11.5,
      "step": 140
    },
    {
      "epoch": 0.02257038237589291,
      "grad_norm": 0.0034546051174402237,
      "learning_rate": 0.00019620967015010395,
      "loss": 11.5,
      "step": 141
    },
    {
      "epoch": 0.02273045600976449,
      "grad_norm": 0.0036798615474253893,
      "learning_rate": 0.00019615195707214803,
      "loss": 11.5,
      "step": 142
    },
    {
      "epoch": 0.02289052964363607,
      "grad_norm": 0.0025492943823337555,
      "learning_rate": 0.0001960938165443353,
      "loss": 11.5,
      "step": 143
    },
    {
      "epoch": 0.023050603277507655,
      "grad_norm": 0.004303085617721081,
      "learning_rate": 0.00019603524882513327,
      "loss": 11.5,
      "step": 144
    },
    {
      "epoch": 0.023210676911379235,
      "grad_norm": 0.0070602428168058395,
      "learning_rate": 0.0001959762541749086,
      "loss": 11.5,
      "step": 145
    },
    {
      "epoch": 0.023370750545250816,
      "grad_norm": 0.005210757255554199,
      "learning_rate": 0.00019591683285592593,
      "loss": 11.5,
      "step": 146
    },
    {
      "epoch": 0.023530824179122396,
      "grad_norm": 0.007379893679171801,
      "learning_rate": 0.00019585698513234663,
      "loss": 11.5,
      "step": 147
    },
    {
      "epoch": 0.023690897812993977,
      "grad_norm": 0.0037022745236754417,
      "learning_rate": 0.0001957967112702277,
      "loss": 11.5,
      "step": 148
    },
    {
      "epoch": 0.023850971446865557,
      "grad_norm": 0.001964289229363203,
      "learning_rate": 0.00019573601153752052,
      "loss": 11.5,
      "step": 149
    },
    {
      "epoch": 0.024011045080737137,
      "grad_norm": 0.0038277339190244675,
      "learning_rate": 0.00019567488620406983,
      "loss": 11.5,
      "step": 150
    },
    {
      "epoch": 0.02417111871460872,
      "grad_norm": 0.001184475957415998,
      "learning_rate": 0.00019561333554161224,
      "loss": 11.5,
      "step": 151
    },
    {
      "epoch": 0.0243311923484803,
      "grad_norm": 0.0013432531850412488,
      "learning_rate": 0.0001955513598237753,
      "loss": 11.5,
      "step": 152
    },
    {
      "epoch": 0.024491265982351882,
      "grad_norm": 0.0012706320267170668,
      "learning_rate": 0.00019548895932607621,
      "loss": 11.5,
      "step": 153
    },
    {
      "epoch": 0.024651339616223462,
      "grad_norm": 0.0011202878085896373,
      "learning_rate": 0.00019542613432592038,
      "loss": 11.5,
      "step": 154
    },
    {
      "epoch": 0.024811413250095043,
      "grad_norm": 0.0010707036126405,
      "learning_rate": 0.00019536288510260056,
      "loss": 11.5,
      "step": 155
    },
    {
      "epoch": 0.024971486883966623,
      "grad_norm": 0.0010517543414607644,
      "learning_rate": 0.00019529921193729534,
      "loss": 11.5,
      "step": 156
    },
    {
      "epoch": 0.025131560517838207,
      "grad_norm": 0.0010216586524620652,
      "learning_rate": 0.00019523511511306793,
      "loss": 11.5,
      "step": 157
    },
    {
      "epoch": 0.025291634151709787,
      "grad_norm": 0.0010877292370423675,
      "learning_rate": 0.000195170594914865,
      "loss": 11.5,
      "step": 158
    },
    {
      "epoch": 0.025451707785581368,
      "grad_norm": 0.0007502186927013099,
      "learning_rate": 0.00019510565162951537,
      "loss": 11.5,
      "step": 159
    },
    {
      "epoch": 0.025611781419452948,
      "grad_norm": 0.0013300442369654775,
      "learning_rate": 0.00019504028554572864,
      "loss": 11.5,
      "step": 160
    },
    {
      "epoch": 0.02577185505332453,
      "grad_norm": 0.0007744083413854241,
      "learning_rate": 0.00019497449695409408,
      "loss": 11.5,
      "step": 161
    },
    {
      "epoch": 0.02593192868719611,
      "grad_norm": 0.00122166913934052,
      "learning_rate": 0.00019490828614707916,
      "loss": 11.5,
      "step": 162
    },
    {
      "epoch": 0.026092002321067693,
      "grad_norm": 0.0014826179249212146,
      "learning_rate": 0.00019484165341902845,
      "loss": 11.5,
      "step": 163
    },
    {
      "epoch": 0.026252075954939273,
      "grad_norm": 0.001125049777328968,
      "learning_rate": 0.00019477459906616206,
      "loss": 11.5,
      "step": 164
    },
    {
      "epoch": 0.026412149588810854,
      "grad_norm": 0.0015355943469330668,
      "learning_rate": 0.00019470712338657458,
      "loss": 11.5,
      "step": 165
    },
    {
      "epoch": 0.026572223222682434,
      "grad_norm": 0.002622431144118309,
      "learning_rate": 0.0001946392266802336,
      "loss": 11.5,
      "step": 166
    },
    {
      "epoch": 0.026732296856554014,
      "grad_norm": 0.0022032030392438173,
      "learning_rate": 0.0001945709092489783,
      "loss": 11.5,
      "step": 167
    },
    {
      "epoch": 0.026892370490425595,
      "grad_norm": 0.0028216661885380745,
      "learning_rate": 0.00019450217139651844,
      "loss": 11.5,
      "step": 168
    },
    {
      "epoch": 0.027052444124297175,
      "grad_norm": 0.003236501244828105,
      "learning_rate": 0.0001944330134284326,
      "loss": 11.5,
      "step": 169
    },
    {
      "epoch": 0.02721251775816876,
      "grad_norm": 0.004418008495122194,
      "learning_rate": 0.00019436343565216711,
      "loss": 11.5,
      "step": 170
    },
    {
      "epoch": 0.02737259139204034,
      "grad_norm": 0.0035076425410807133,
      "learning_rate": 0.00019429343837703455,
      "loss": 11.5,
      "step": 171
    },
    {
      "epoch": 0.02753266502591192,
      "grad_norm": 0.003910847939550877,
      "learning_rate": 0.0001942230219142124,
      "loss": 11.5,
      "step": 172
    },
    {
      "epoch": 0.0276927386597835,
      "grad_norm": 0.004566975403577089,
      "learning_rate": 0.0001941521865767417,
      "loss": 11.5,
      "step": 173
    },
    {
      "epoch": 0.02785281229365508,
      "grad_norm": 0.0030329632572829723,
      "learning_rate": 0.0001940809326795256,
      "loss": 11.5,
      "step": 174
    },
    {
      "epoch": 0.02801288592752666,
      "grad_norm": 0.0040180860087275505,
      "learning_rate": 0.000194009260539328,
      "loss": 11.5,
      "step": 175
    },
    {
      "epoch": 0.028172959561398245,
      "grad_norm": 0.0033933145459741354,
      "learning_rate": 0.0001939371704747721,
      "loss": 11.5,
      "step": 176
    },
    {
      "epoch": 0.028333033195269825,
      "grad_norm": 0.004221628420054913,
      "learning_rate": 0.00019386466280633906,
      "loss": 11.5,
      "step": 177
    },
    {
      "epoch": 0.028493106829141406,
      "grad_norm": 0.0038972035981714725,
      "learning_rate": 0.00019379173785636646,
      "loss": 11.5,
      "step": 178
    },
    {
      "epoch": 0.028653180463012986,
      "grad_norm": 0.0039198147132992744,
      "learning_rate": 0.000193718395949047,
      "loss": 11.5,
      "step": 179
    },
    {
      "epoch": 0.028813254096884566,
      "grad_norm": 0.005124758463352919,
      "learning_rate": 0.00019364463741042694,
      "loss": 11.5,
      "step": 180
    },
    {
      "epoch": 0.028973327730756147,
      "grad_norm": 0.004067589528858662,
      "learning_rate": 0.00019357046256840473,
      "loss": 11.5,
      "step": 181
    },
    {
      "epoch": 0.029133401364627727,
      "grad_norm": 0.004641098901629448,
      "learning_rate": 0.00019349587175272948,
      "loss": 11.5,
      "step": 182
    },
    {
      "epoch": 0.02929347499849931,
      "grad_norm": 0.004570235963910818,
      "learning_rate": 0.0001934208652949996,
      "loss": 11.5,
      "step": 183
    },
    {
      "epoch": 0.02945354863237089,
      "grad_norm": 0.004834281280636787,
      "learning_rate": 0.00019334544352866127,
      "loss": 11.5,
      "step": 184
    },
    {
      "epoch": 0.02961362226624247,
      "grad_norm": 0.004718173760920763,
      "learning_rate": 0.00019326960678900688,
      "loss": 11.5,
      "step": 185
    },
    {
      "epoch": 0.029773695900114052,
      "grad_norm": 0.0034208674915134907,
      "learning_rate": 0.00019319335541317361,
      "loss": 11.5,
      "step": 186
    },
    {
      "epoch": 0.029933769533985632,
      "grad_norm": 0.004345146007835865,
      "learning_rate": 0.00019311668974014208,
      "loss": 11.5,
      "step": 187
    },
    {
      "epoch": 0.030093843167857213,
      "grad_norm": 0.004935440141707659,
      "learning_rate": 0.00019303961011073447,
      "loss": 11.5,
      "step": 188
    },
    {
      "epoch": 0.030253916801728797,
      "grad_norm": 0.003570390399545431,
      "learning_rate": 0.00019296211686761346,
      "loss": 11.5,
      "step": 189
    },
    {
      "epoch": 0.030413990435600377,
      "grad_norm": 0.004747358616441488,
      "learning_rate": 0.00019288421035528028,
      "loss": 11.5,
      "step": 190
    },
    {
      "epoch": 0.030574064069471957,
      "grad_norm": 0.0070312670432031155,
      "learning_rate": 0.00019280589092007352,
      "loss": 11.5,
      "step": 191
    },
    {
      "epoch": 0.030734137703343538,
      "grad_norm": 0.006528772413730621,
      "learning_rate": 0.00019272715891016735,
      "loss": 11.5,
      "step": 192
    },
    {
      "epoch": 0.030894211337215118,
      "grad_norm": 0.004292914178222418,
      "learning_rate": 0.00019264801467557007,
      "loss": 11.5,
      "step": 193
    },
    {
      "epoch": 0.0310542849710867,
      "grad_norm": 0.003930619452148676,
      "learning_rate": 0.00019256845856812266,
      "loss": 11.5,
      "step": 194
    },
    {
      "epoch": 0.031214358604958282,
      "grad_norm": 0.009428051300346851,
      "learning_rate": 0.000192488490941497,
      "loss": 11.5,
      "step": 195
    },
    {
      "epoch": 0.03137443223882986,
      "grad_norm": 0.011329618282616138,
      "learning_rate": 0.00019240811215119448,
      "loss": 11.5,
      "step": 196
    },
    {
      "epoch": 0.03153450587270144,
      "grad_norm": 0.015292741358280182,
      "learning_rate": 0.00019232732255454422,
      "loss": 11.5,
      "step": 197
    },
    {
      "epoch": 0.03169457950657302,
      "grad_norm": 0.007595112547278404,
      "learning_rate": 0.00019224612251070175,
      "loss": 11.5,
      "step": 198
    },
    {
      "epoch": 0.031854653140444604,
      "grad_norm": 0.00519321346655488,
      "learning_rate": 0.0001921645123806472,
      "loss": 11.5,
      "step": 199
    },
    {
      "epoch": 0.03201472677431619,
      "grad_norm": 0.008084663189947605,
      "learning_rate": 0.0001920824925271838,
      "loss": 11.5,
      "step": 200
    },
    {
      "epoch": 0.032174800408187765,
      "grad_norm": 0.004125955514609814,
      "learning_rate": 0.0001920000633149362,
      "loss": 11.5,
      "step": 201
    },
    {
      "epoch": 0.03233487404205935,
      "grad_norm": 0.004124477505683899,
      "learning_rate": 0.00019191722511034884,
      "loss": 11.5,
      "step": 202
    },
    {
      "epoch": 0.032494947675930926,
      "grad_norm": 0.003988343756645918,
      "learning_rate": 0.00019183397828168448,
      "loss": 11.5,
      "step": 203
    },
    {
      "epoch": 0.03265502130980251,
      "grad_norm": 0.00596611388027668,
      "learning_rate": 0.00019175032319902234,
      "loss": 11.5,
      "step": 204
    },
    {
      "epoch": 0.03281509494367409,
      "grad_norm": 0.004642547108232975,
      "learning_rate": 0.00019166626023425662,
      "loss": 11.5,
      "step": 205
    },
    {
      "epoch": 0.03297516857754567,
      "grad_norm": 0.005087132565677166,
      "learning_rate": 0.00019158178976109476,
      "loss": 11.5,
      "step": 206
    },
    {
      "epoch": 0.033135242211417254,
      "grad_norm": 0.0064563388004899025,
      "learning_rate": 0.0001914969121550558,
      "loss": 11.5,
      "step": 207
    },
    {
      "epoch": 0.03329531584528883,
      "grad_norm": 0.005523744970560074,
      "learning_rate": 0.00019141162779346874,
      "loss": 11.5,
      "step": 208
    },
    {
      "epoch": 0.033455389479160415,
      "grad_norm": 0.005560313351452351,
      "learning_rate": 0.00019132593705547082,
      "loss": 11.5,
      "step": 209
    },
    {
      "epoch": 0.03361546311303199,
      "grad_norm": 0.004362552426755428,
      "learning_rate": 0.00019123984032200586,
      "loss": 11.5,
      "step": 210
    },
    {
      "epoch": 0.033775536746903576,
      "grad_norm": 0.005080255679786205,
      "learning_rate": 0.00019115333797582254,
      "loss": 11.5,
      "step": 211
    },
    {
      "epoch": 0.03393561038077516,
      "grad_norm": 0.0047751725651323795,
      "learning_rate": 0.00019106643040147278,
      "loss": 11.5,
      "step": 212
    },
    {
      "epoch": 0.034095684014646736,
      "grad_norm": 0.0036601570900529623,
      "learning_rate": 0.00019097911798530987,
      "loss": 11.5,
      "step": 213
    },
    {
      "epoch": 0.03425575764851832,
      "grad_norm": 0.0042919619008898735,
      "learning_rate": 0.00019089140111548696,
      "loss": 11.5,
      "step": 214
    },
    {
      "epoch": 0.0344158312823899,
      "grad_norm": 0.004078218713402748,
      "learning_rate": 0.00019080328018195513,
      "loss": 11.5,
      "step": 215
    },
    {
      "epoch": 0.03457590491626148,
      "grad_norm": 0.0035699764266610146,
      "learning_rate": 0.0001907147555764618,
      "loss": 11.5,
      "step": 216
    },
    {
      "epoch": 0.03473597855013306,
      "grad_norm": 0.006953587289899588,
      "learning_rate": 0.00019062582769254895,
      "loss": 11.5,
      "step": 217
    },
    {
      "epoch": 0.03489605218400464,
      "grad_norm": 0.005379633978009224,
      "learning_rate": 0.00019053649692555135,
      "loss": 11.5,
      "step": 218
    },
    {
      "epoch": 0.035056125817876226,
      "grad_norm": 0.005702970549464226,
      "learning_rate": 0.00019044676367259476,
      "loss": 11.5,
      "step": 219
    },
    {
      "epoch": 0.0352161994517478,
      "grad_norm": 0.005873862653970718,
      "learning_rate": 0.00019035662833259432,
      "loss": 11.5,
      "step": 220
    },
    {
      "epoch": 0.035376273085619386,
      "grad_norm": 0.0050443257205188274,
      "learning_rate": 0.00019026609130625257,
      "loss": 11.5,
      "step": 221
    },
    {
      "epoch": 0.03553634671949096,
      "grad_norm": 0.005148936528712511,
      "learning_rate": 0.00019017515299605788,
      "loss": 11.5,
      "step": 222
    },
    {
      "epoch": 0.03569642035336255,
      "grad_norm": 0.004865546710789204,
      "learning_rate": 0.00019008381380628247,
      "loss": 11.5,
      "step": 223
    },
    {
      "epoch": 0.03585649398723413,
      "grad_norm": 0.00710660545155406,
      "learning_rate": 0.00018999207414298067,
      "loss": 11.5,
      "step": 224
    },
    {
      "epoch": 0.03601656762110571,
      "grad_norm": 0.005359895061701536,
      "learning_rate": 0.00018989993441398726,
      "loss": 11.5,
      "step": 225
    },
    {
      "epoch": 0.03617664125497729,
      "grad_norm": 0.006631465628743172,
      "learning_rate": 0.00018980739502891546,
      "loss": 11.5,
      "step": 226
    },
    {
      "epoch": 0.03633671488884887,
      "grad_norm": 0.00704937893897295,
      "learning_rate": 0.0001897144563991552,
      "loss": 11.5,
      "step": 227
    },
    {
      "epoch": 0.03649678852272045,
      "grad_norm": 0.0053662629798054695,
      "learning_rate": 0.00018962111893787128,
      "loss": 11.5,
      "step": 228
    },
    {
      "epoch": 0.03665686215659203,
      "grad_norm": 0.005165701732039452,
      "learning_rate": 0.00018952738306000151,
      "loss": 11.5,
      "step": 229
    },
    {
      "epoch": 0.03681693579046361,
      "grad_norm": 0.005945558659732342,
      "learning_rate": 0.00018943324918225494,
      "loss": 11.5,
      "step": 230
    },
    {
      "epoch": 0.0369770094243352,
      "grad_norm": 0.006662803236395121,
      "learning_rate": 0.0001893387177231099,
      "loss": 11.5,
      "step": 231
    },
    {
      "epoch": 0.037137083058206774,
      "grad_norm": 0.005151469726115465,
      "learning_rate": 0.0001892437891028122,
      "loss": 11.5,
      "step": 232
    },
    {
      "epoch": 0.03729715669207836,
      "grad_norm": 0.007021139841526747,
      "learning_rate": 0.0001891484637433733,
      "loss": 11.5,
      "step": 233
    },
    {
      "epoch": 0.037457230325949935,
      "grad_norm": 0.006236946675926447,
      "learning_rate": 0.00018905274206856837,
      "loss": 11.5,
      "step": 234
    },
    {
      "epoch": 0.03761730395982152,
      "grad_norm": 0.006137041375041008,
      "learning_rate": 0.00018895662450393438,
      "loss": 11.5,
      "step": 235
    },
    {
      "epoch": 0.037777377593693096,
      "grad_norm": 0.005807884503155947,
      "learning_rate": 0.00018886011147676833,
      "loss": 11.5,
      "step": 236
    },
    {
      "epoch": 0.03793745122756468,
      "grad_norm": 0.006715100258588791,
      "learning_rate": 0.00018876320341612522,
      "loss": 11.5,
      "step": 237
    },
    {
      "epoch": 0.03809752486143626,
      "grad_norm": 0.007929547689855099,
      "learning_rate": 0.00018866590075281624,
      "loss": 11.5,
      "step": 238
    },
    {
      "epoch": 0.03825759849530784,
      "grad_norm": 0.007534688338637352,
      "learning_rate": 0.00018856820391940674,
      "loss": 11.5,
      "step": 239
    },
    {
      "epoch": 0.038417672129179424,
      "grad_norm": 0.00604398176074028,
      "learning_rate": 0.00018847011335021449,
      "loss": 11.5,
      "step": 240
    },
    {
      "epoch": 0.038577745763051,
      "grad_norm": 0.008669275790452957,
      "learning_rate": 0.00018837162948130752,
      "loss": 11.5,
      "step": 241
    },
    {
      "epoch": 0.038737819396922585,
      "grad_norm": 0.004927826579660177,
      "learning_rate": 0.00018827275275050233,
      "loss": 11.5,
      "step": 242
    },
    {
      "epoch": 0.03889789303079417,
      "grad_norm": 0.008499566465616226,
      "learning_rate": 0.00018817348359736203,
      "loss": 11.5,
      "step": 243
    },
    {
      "epoch": 0.039057966664665746,
      "grad_norm": 0.014102022163569927,
      "learning_rate": 0.00018807382246319412,
      "loss": 11.5,
      "step": 244
    },
    {
      "epoch": 0.03921804029853733,
      "grad_norm": 0.016383271664381027,
      "learning_rate": 0.00018797376979104872,
      "loss": 11.5,
      "step": 245
    },
    {
      "epoch": 0.039378113932408906,
      "grad_norm": 0.01808208040893078,
      "learning_rate": 0.00018787332602571662,
      "loss": 11.5,
      "step": 246
    },
    {
      "epoch": 0.03953818756628049,
      "grad_norm": 0.01708586886525154,
      "learning_rate": 0.00018777249161372713,
      "loss": 11.5,
      "step": 247
    },
    {
      "epoch": 0.03969826120015207,
      "grad_norm": 0.02052490971982479,
      "learning_rate": 0.00018767126700334634,
      "loss": 11.5,
      "step": 248
    },
    {
      "epoch": 0.03985833483402365,
      "grad_norm": 0.017746634781360626,
      "learning_rate": 0.0001875696526445749,
      "loss": 11.5,
      "step": 249
    },
    {
      "epoch": 0.040018408467895235,
      "grad_norm": 0.019202718511223793,
      "learning_rate": 0.0001874676489891461,
      "loss": 11.5,
      "step": 250
    },
    {
      "epoch": 0.04017848210176681,
      "grad_norm": 0.010096721351146698,
      "learning_rate": 0.00018736525649052394,
      "loss": 11.5,
      "step": 251
    },
    {
      "epoch": 0.040338555735638396,
      "grad_norm": 0.01723094843327999,
      "learning_rate": 0.00018726247560390099,
      "loss": 11.5,
      "step": 252
    },
    {
      "epoch": 0.04049862936950997,
      "grad_norm": 0.011470257304608822,
      "learning_rate": 0.00018715930678619644,
      "loss": 11.5,
      "step": 253
    },
    {
      "epoch": 0.040658703003381556,
      "grad_norm": 0.014084495604038239,
      "learning_rate": 0.00018705575049605413,
      "loss": 11.5,
      "step": 254
    },
    {
      "epoch": 0.04081877663725313,
      "grad_norm": 0.009606949985027313,
      "learning_rate": 0.00018695180719384029,
      "loss": 11.5,
      "step": 255
    },
    {
      "epoch": 0.04097885027112472,
      "grad_norm": 0.013065108098089695,
      "learning_rate": 0.00018684747734164177,
      "loss": 11.5,
      "step": 256
    },
    {
      "epoch": 0.0411389239049963,
      "grad_norm": 0.007634848356246948,
      "learning_rate": 0.00018674276140326376,
      "loss": 11.5,
      "step": 257
    },
    {
      "epoch": 0.04129899753886788,
      "grad_norm": 0.009482149966061115,
      "learning_rate": 0.00018663765984422786,
      "loss": 11.5,
      "step": 258
    },
    {
      "epoch": 0.04145907117273946,
      "grad_norm": 0.007441854570060968,
      "learning_rate": 0.00018653217313177004,
      "loss": 11.5,
      "step": 259
    },
    {
      "epoch": 0.04161914480661104,
      "grad_norm": 0.006992577575147152,
      "learning_rate": 0.00018642630173483832,
      "loss": 11.5,
      "step": 260
    },
    {
      "epoch": 0.04177921844048262,
      "grad_norm": 0.0063720578327775,
      "learning_rate": 0.00018632004612409103,
      "loss": 11.5,
      "step": 261
    },
    {
      "epoch": 0.041939292074354206,
      "grad_norm": 0.004319782834500074,
      "learning_rate": 0.00018621340677189453,
      "loss": 11.5,
      "step": 262
    },
    {
      "epoch": 0.04209936570822578,
      "grad_norm": 0.00529808783903718,
      "learning_rate": 0.00018610638415232097,
      "loss": 11.5,
      "step": 263
    },
    {
      "epoch": 0.04225943934209737,
      "grad_norm": 0.005276544950902462,
      "learning_rate": 0.00018599897874114652,
      "loss": 11.5,
      "step": 264
    },
    {
      "epoch": 0.042419512975968944,
      "grad_norm": 0.004047976806759834,
      "learning_rate": 0.00018589119101584898,
      "loss": 11.5,
      "step": 265
    },
    {
      "epoch": 0.04257958660984053,
      "grad_norm": 0.004963539075106382,
      "learning_rate": 0.00018578302145560584,
      "loss": 11.5,
      "step": 266
    },
    {
      "epoch": 0.042739660243712105,
      "grad_norm": 0.004974601324647665,
      "learning_rate": 0.00018567447054129195,
      "loss": 11.5,
      "step": 267
    },
    {
      "epoch": 0.04289973387758369,
      "grad_norm": 0.005424274131655693,
      "learning_rate": 0.00018556553875547754,
      "loss": 11.5,
      "step": 268
    },
    {
      "epoch": 0.04305980751145527,
      "grad_norm": 0.004362328443676233,
      "learning_rate": 0.00018545622658242607,
      "loss": 11.5,
      "step": 269
    },
    {
      "epoch": 0.04321988114532685,
      "grad_norm": 0.004392134957015514,
      "learning_rate": 0.00018534653450809197,
      "loss": 11.5,
      "step": 270
    },
    {
      "epoch": 0.04337995477919843,
      "grad_norm": 0.004988868720829487,
      "learning_rate": 0.00018523646302011867,
      "loss": 11.5,
      "step": 271
    },
    {
      "epoch": 0.04354002841307001,
      "grad_norm": 0.0036948062479496002,
      "learning_rate": 0.00018512601260783606,
      "loss": 11.5,
      "step": 272
    },
    {
      "epoch": 0.043700102046941594,
      "grad_norm": 0.005431415978819132,
      "learning_rate": 0.00018501518376225887,
      "loss": 11.5,
      "step": 273
    },
    {
      "epoch": 0.04386017568081317,
      "grad_norm": 0.005191940348595381,
      "learning_rate": 0.00018490397697608395,
      "loss": 11.5,
      "step": 274
    },
    {
      "epoch": 0.044020249314684755,
      "grad_norm": 0.005340060219168663,
      "learning_rate": 0.0001847923927436884,
      "loss": 11.5,
      "step": 275
    },
    {
      "epoch": 0.04418032294855634,
      "grad_norm": 0.005495226942002773,
      "learning_rate": 0.00018468043156112728,
      "loss": 11.5,
      "step": 276
    },
    {
      "epoch": 0.044340396582427916,
      "grad_norm": 0.0054380279034376144,
      "learning_rate": 0.0001845680939261314,
      "loss": 11.5,
      "step": 277
    },
    {
      "epoch": 0.0445004702162995,
      "grad_norm": 0.005373974796384573,
      "learning_rate": 0.00018445538033810515,
      "loss": 11.5,
      "step": 278
    },
    {
      "epoch": 0.044660543850171076,
      "grad_norm": 0.006492419634014368,
      "learning_rate": 0.00018434229129812418,
      "loss": 11.5,
      "step": 279
    },
    {
      "epoch": 0.04482061748404266,
      "grad_norm": 0.004978380165994167,
      "learning_rate": 0.0001842288273089332,
      "loss": 11.5,
      "step": 280
    },
    {
      "epoch": 0.04498069111791424,
      "grad_norm": 0.005132167134433985,
      "learning_rate": 0.00018411498887494396,
      "loss": 11.5,
      "step": 281
    },
    {
      "epoch": 0.04514076475178582,
      "grad_norm": 0.004873708356171846,
      "learning_rate": 0.00018400077650223263,
      "loss": 11.5,
      "step": 282
    },
    {
      "epoch": 0.045300838385657405,
      "grad_norm": 0.004690905567258596,
      "learning_rate": 0.0001838861906985379,
      "loss": 11.5,
      "step": 283
    },
    {
      "epoch": 0.04546091201952898,
      "grad_norm": 0.006050188094377518,
      "learning_rate": 0.00018377123197325842,
      "loss": 11.5,
      "step": 284
    },
    {
      "epoch": 0.045620985653400566,
      "grad_norm": 0.008120089769363403,
      "learning_rate": 0.00018365590083745085,
      "loss": 11.5,
      "step": 285
    },
    {
      "epoch": 0.04578105928727214,
      "grad_norm": 0.005599613301455975,
      "learning_rate": 0.00018354019780382735,
      "loss": 11.5,
      "step": 286
    },
    {
      "epoch": 0.045941132921143726,
      "grad_norm": 0.006047028582543135,
      "learning_rate": 0.0001834241233867533,
      "loss": 11.5,
      "step": 287
    },
    {
      "epoch": 0.04610120655501531,
      "grad_norm": 0.005496839992702007,
      "learning_rate": 0.00018330767810224524,
      "loss": 11.5,
      "step": 288
    },
    {
      "epoch": 0.04626128018888689,
      "grad_norm": 0.004045095294713974,
      "learning_rate": 0.0001831908624679683,
      "loss": 11.5,
      "step": 289
    },
    {
      "epoch": 0.04642135382275847,
      "grad_norm": 0.0066123465076088905,
      "learning_rate": 0.0001830736770032341,
      "loss": 11.5,
      "step": 290
    },
    {
      "epoch": 0.04658142745663005,
      "grad_norm": 0.004854083526879549,
      "learning_rate": 0.0001829561222289984,
      "loss": 11.5,
      "step": 291
    },
    {
      "epoch": 0.04674150109050163,
      "grad_norm": 0.004592921119183302,
      "learning_rate": 0.00018283819866785853,
      "loss": 11.5,
      "step": 292
    },
    {
      "epoch": 0.04690157472437321,
      "grad_norm": 0.021113738417625427,
      "learning_rate": 0.0001827199068440516,
      "loss": 11.5,
      "step": 293
    },
    {
      "epoch": 0.04706164835824479,
      "grad_norm": 0.026811130344867706,
      "learning_rate": 0.00018260124728345162,
      "loss": 11.5,
      "step": 294
    },
    {
      "epoch": 0.047221721992116376,
      "grad_norm": 0.011840801686048508,
      "learning_rate": 0.00018248222051356754,
      "loss": 11.5,
      "step": 295
    },
    {
      "epoch": 0.04738179562598795,
      "grad_norm": 0.00733939791098237,
      "learning_rate": 0.00018236282706354063,
      "loss": 11.5,
      "step": 296
    },
    {
      "epoch": 0.04754186925985954,
      "grad_norm": 0.013145462609827518,
      "learning_rate": 0.00018224306746414238,
      "loss": 11.5,
      "step": 297
    },
    {
      "epoch": 0.047701942893731114,
      "grad_norm": 0.009584108367562294,
      "learning_rate": 0.00018212294224777197,
      "loss": 11.5,
      "step": 298
    },
    {
      "epoch": 0.0478620165276027,
      "grad_norm": 0.011472940444946289,
      "learning_rate": 0.00018200245194845399,
      "loss": 11.5,
      "step": 299
    },
    {
      "epoch": 0.048022090161474275,
      "grad_norm": 0.01373722031712532,
      "learning_rate": 0.00018188159710183594,
      "loss": 11.5,
      "step": 300
    },
    {
      "epoch": 0.04818216379534586,
      "grad_norm": 0.007853199727833271,
      "learning_rate": 0.000181760378245186,
      "loss": 11.5,
      "step": 301
    },
    {
      "epoch": 0.04834223742921744,
      "grad_norm": 0.006789433304220438,
      "learning_rate": 0.00018163879591739067,
      "loss": 11.5,
      "step": 302
    },
    {
      "epoch": 0.04850231106308902,
      "grad_norm": 0.00708062294870615,
      "learning_rate": 0.0001815168506589521,
      "loss": 11.5,
      "step": 303
    },
    {
      "epoch": 0.0486623846969606,
      "grad_norm": 0.003183502471074462,
      "learning_rate": 0.000181394543011986,
      "loss": 11.5,
      "step": 304
    },
    {
      "epoch": 0.04882245833083218,
      "grad_norm": 0.005664783995598555,
      "learning_rate": 0.00018127187352021907,
      "loss": 11.5,
      "step": 305
    },
    {
      "epoch": 0.048982531964703764,
      "grad_norm": 0.004790788050740957,
      "learning_rate": 0.0001811488427289866,
      "loss": 11.5,
      "step": 306
    },
    {
      "epoch": 0.04914260559857535,
      "grad_norm": 0.005017767194658518,
      "learning_rate": 0.00018102545118523007,
      "loss": 11.5,
      "step": 307
    },
    {
      "epoch": 0.049302679232446925,
      "grad_norm": 0.004967286717146635,
      "learning_rate": 0.00018090169943749476,
      "loss": 11.5,
      "step": 308
    },
    {
      "epoch": 0.04946275286631851,
      "grad_norm": 0.004688838496804237,
      "learning_rate": 0.00018077758803592718,
      "loss": 11.5,
      "step": 309
    },
    {
      "epoch": 0.049622826500190086,
      "grad_norm": 0.003855897579342127,
      "learning_rate": 0.00018065311753227273,
      "loss": 11.5,
      "step": 310
    },
    {
      "epoch": 0.04978290013406167,
      "grad_norm": 0.003533479291945696,
      "learning_rate": 0.0001805282884798732,
      "loss": 11.5,
      "step": 311
    },
    {
      "epoch": 0.049942973767933246,
      "grad_norm": 0.003911127801984549,
      "learning_rate": 0.00018040310143366446,
      "loss": 11.5,
      "step": 312
    },
    {
      "epoch": 0.05010304740180483,
      "grad_norm": 0.0077298590913414955,
      "learning_rate": 0.00018027755695017368,
      "loss": 11.5,
      "step": 313
    },
    {
      "epoch": 0.050263121035676414,
      "grad_norm": 0.002832760103046894,
      "learning_rate": 0.00018015165558751717,
      "loss": 11.5,
      "step": 314
    },
    {
      "epoch": 0.05042319466954799,
      "grad_norm": 0.004375755321234465,
      "learning_rate": 0.00018002539790539773,
      "loss": 11.5,
      "step": 315
    },
    {
      "epoch": 0.050583268303419575,
      "grad_norm": 0.0037666945718228817,
      "learning_rate": 0.00017989878446510215,
      "loss": 11.5,
      "step": 316
    },
    {
      "epoch": 0.05074334193729115,
      "grad_norm": 0.0037504660431295633,
      "learning_rate": 0.00017977181582949888,
      "loss": 11.5,
      "step": 317
    },
    {
      "epoch": 0.050903415571162736,
      "grad_norm": 0.003293354529887438,
      "learning_rate": 0.0001796444925630353,
      "loss": 11.5,
      "step": 318
    },
    {
      "epoch": 0.05106348920503431,
      "grad_norm": 0.003078571753576398,
      "learning_rate": 0.00017951681523173542,
      "loss": 11.5,
      "step": 319
    },
    {
      "epoch": 0.051223562838905896,
      "grad_norm": 0.004428179934620857,
      "learning_rate": 0.0001793887844031972,
      "loss": 11.5,
      "step": 320
    },
    {
      "epoch": 0.05138363647277748,
      "grad_norm": 0.00518577266484499,
      "learning_rate": 0.00017926040064659014,
      "loss": 11.5,
      "step": 321
    },
    {
      "epoch": 0.05154371010664906,
      "grad_norm": 0.0041250321082770824,
      "learning_rate": 0.0001791316645326526,
      "loss": 11.5,
      "step": 322
    },
    {
      "epoch": 0.05170378374052064,
      "grad_norm": 0.00472582271322608,
      "learning_rate": 0.00017900257663368963,
      "loss": 11.5,
      "step": 323
    },
    {
      "epoch": 0.05186385737439222,
      "grad_norm": 0.004885179456323385,
      "learning_rate": 0.0001788731375235698,
      "loss": 11.5,
      "step": 324
    },
    {
      "epoch": 0.0520239310082638,
      "grad_norm": 0.004233427811414003,
      "learning_rate": 0.00017874334777772327,
      "loss": 11.5,
      "step": 325
    },
    {
      "epoch": 0.052184004642135386,
      "grad_norm": 0.0075990804471075535,
      "learning_rate": 0.00017861320797313892,
      "loss": 11.5,
      "step": 326
    },
    {
      "epoch": 0.05234407827600696,
      "grad_norm": 0.004590803291648626,
      "learning_rate": 0.0001784827186883618,
      "loss": 11.5,
      "step": 327
    },
    {
      "epoch": 0.052504151909878546,
      "grad_norm": 0.007067880593240261,
      "learning_rate": 0.00017835188050349064,
      "loss": 11.5,
      "step": 328
    },
    {
      "epoch": 0.05266422554375012,
      "grad_norm": 0.00449848547577858,
      "learning_rate": 0.00017822069400017516,
      "loss": 11.5,
      "step": 329
    },
    {
      "epoch": 0.05282429917762171,
      "grad_norm": 0.003897727932780981,
      "learning_rate": 0.00017808915976161362,
      "loss": 11.5,
      "step": 330
    },
    {
      "epoch": 0.052984372811493284,
      "grad_norm": 0.003709547221660614,
      "learning_rate": 0.00017795727837255015,
      "loss": 11.5,
      "step": 331
    },
    {
      "epoch": 0.05314444644536487,
      "grad_norm": 0.0030794816557317972,
      "learning_rate": 0.00017782505041927216,
      "loss": 11.5,
      "step": 332
    },
    {
      "epoch": 0.05330452007923645,
      "grad_norm": 0.0032050926238298416,
      "learning_rate": 0.00017769247648960774,
      "loss": 11.5,
      "step": 333
    },
    {
      "epoch": 0.05346459371310803,
      "grad_norm": 0.0033614016138017178,
      "learning_rate": 0.00017755955717292296,
      "loss": 11.5,
      "step": 334
    },
    {
      "epoch": 0.05362466734697961,
      "grad_norm": 0.003943106159567833,
      "learning_rate": 0.00017742629306011944,
      "loss": 11.5,
      "step": 335
    },
    {
      "epoch": 0.05378474098085119,
      "grad_norm": 0.006896649021655321,
      "learning_rate": 0.00017729268474363154,
      "loss": 11.5,
      "step": 336
    },
    {
      "epoch": 0.05394481461472277,
      "grad_norm": 0.004238173831254244,
      "learning_rate": 0.0001771587328174239,
      "loss": 11.5,
      "step": 337
    },
    {
      "epoch": 0.05410488824859435,
      "grad_norm": 0.0035634618252515793,
      "learning_rate": 0.0001770244378769885,
      "loss": 11.5,
      "step": 338
    },
    {
      "epoch": 0.054264961882465934,
      "grad_norm": 0.002922303741797805,
      "learning_rate": 0.0001768898005193425,
      "loss": 11.5,
      "step": 339
    },
    {
      "epoch": 0.05442503551633752,
      "grad_norm": 0.0035376015584915876,
      "learning_rate": 0.000176754821343025,
      "loss": 11.5,
      "step": 340
    },
    {
      "epoch": 0.054585109150209095,
      "grad_norm": 0.0034501487389206886,
      "learning_rate": 0.0001766195009480949,
      "loss": 11.5,
      "step": 341
    },
    {
      "epoch": 0.05474518278408068,
      "grad_norm": 0.012458324432373047,
      "learning_rate": 0.0001764838399361279,
      "loss": 11.5,
      "step": 342
    },
    {
      "epoch": 0.054905256417952256,
      "grad_norm": 0.0368187241256237,
      "learning_rate": 0.00017634783891021393,
      "loss": 11.5,
      "step": 343
    },
    {
      "epoch": 0.05506533005182384,
      "grad_norm": 0.01612844504415989,
      "learning_rate": 0.00017621149847495458,
      "loss": 11.5,
      "step": 344
    },
    {
      "epoch": 0.05522540368569542,
      "grad_norm": 0.006253810133785009,
      "learning_rate": 0.00017607481923646016,
      "loss": 11.5,
      "step": 345
    },
    {
      "epoch": 0.055385477319567,
      "grad_norm": 0.008016599342226982,
      "learning_rate": 0.0001759378018023473,
      "loss": 11.5,
      "step": 346
    },
    {
      "epoch": 0.055545550953438584,
      "grad_norm": 0.008556563407182693,
      "learning_rate": 0.00017580044678173592,
      "loss": 11.5,
      "step": 347
    },
    {
      "epoch": 0.05570562458731016,
      "grad_norm": 0.007470616605132818,
      "learning_rate": 0.00017566275478524693,
      "loss": 11.5,
      "step": 348
    },
    {
      "epoch": 0.055865698221181745,
      "grad_norm": 0.013478902168571949,
      "learning_rate": 0.0001755247264249991,
      "loss": 11.5,
      "step": 349
    },
    {
      "epoch": 0.05602577185505332,
      "grad_norm": 0.011351917870342731,
      "learning_rate": 0.0001753863623146066,
      "loss": 11.5,
      "step": 350
    },
    {
      "epoch": 0.056185845488924906,
      "grad_norm": 0.0046243141405284405,
      "learning_rate": 0.00017524766306917618,
      "loss": 11.5,
      "step": 351
    },
    {
      "epoch": 0.05634591912279649,
      "grad_norm": 0.0038260514847934246,
      "learning_rate": 0.0001751086293053045,
      "loss": 11.5,
      "step": 352
    },
    {
      "epoch": 0.056505992756668066,
      "grad_norm": 0.0032652895897626877,
      "learning_rate": 0.0001749692616410753,
      "loss": 11.5,
      "step": 353
    },
    {
      "epoch": 0.05666606639053965,
      "grad_norm": 0.00571425911039114,
      "learning_rate": 0.00017482956069605668,
      "loss": 11.5,
      "step": 354
    },
    {
      "epoch": 0.05682614002441123,
      "grad_norm": 0.005943847354501486,
      "learning_rate": 0.00017468952709129846,
      "loss": 11.5,
      "step": 355
    },
    {
      "epoch": 0.05698621365828281,
      "grad_norm": 0.00332586164586246,
      "learning_rate": 0.00017454916144932922,
      "loss": 11.5,
      "step": 356
    },
    {
      "epoch": 0.05714628729215439,
      "grad_norm": 0.002929111709818244,
      "learning_rate": 0.0001744084643941536,
      "loss": 11.5,
      "step": 357
    },
    {
      "epoch": 0.05730636092602597,
      "grad_norm": 0.00693755317479372,
      "learning_rate": 0.00017426743655124974,
      "loss": 11.5,
      "step": 358
    },
    {
      "epoch": 0.057466434559897556,
      "grad_norm": 0.004010320641100407,
      "learning_rate": 0.0001741260785475661,
      "loss": 11.5,
      "step": 359
    },
    {
      "epoch": 0.05762650819376913,
      "grad_norm": 0.00563431903719902,
      "learning_rate": 0.00017398439101151905,
      "loss": 11.5,
      "step": 360
    },
    {
      "epoch": 0.057786581827640716,
      "grad_norm": 0.004954980686306953,
      "learning_rate": 0.00017384237457298987,
      "loss": 11.5,
      "step": 361
    },
    {
      "epoch": 0.05794665546151229,
      "grad_norm": 0.0028225225396454334,
      "learning_rate": 0.00017370002986332193,
      "loss": 11.5,
      "step": 362
    },
    {
      "epoch": 0.05810672909538388,
      "grad_norm": 0.0028268226888030767,
      "learning_rate": 0.00017355735751531807,
      "loss": 11.5,
      "step": 363
    },
    {
      "epoch": 0.058266802729255454,
      "grad_norm": 0.0043122912757098675,
      "learning_rate": 0.00017341435816323756,
      "loss": 11.5,
      "step": 364
    },
    {
      "epoch": 0.05842687636312704,
      "grad_norm": 0.004459218587726355,
      "learning_rate": 0.00017327103244279348,
      "loss": 11.5,
      "step": 365
    },
    {
      "epoch": 0.05858694999699862,
      "grad_norm": 0.0035306205973029137,
      "learning_rate": 0.00017312738099114973,
      "loss": 11.5,
      "step": 366
    },
    {
      "epoch": 0.0587470236308702,
      "grad_norm": 0.0034746327437460423,
      "learning_rate": 0.00017298340444691835,
      "loss": 11.5,
      "step": 367
    },
    {
      "epoch": 0.05890709726474178,
      "grad_norm": 0.0036737604532390833,
      "learning_rate": 0.00017283910345015647,
      "loss": 11.5,
      "step": 368
    },
    {
      "epoch": 0.05906717089861336,
      "grad_norm": 0.002505925018340349,
      "learning_rate": 0.0001726944786423637,
      "loss": 11.5,
      "step": 369
    },
    {
      "epoch": 0.05922724453248494,
      "grad_norm": 0.0037393190432339907,
      "learning_rate": 0.00017254953066647913,
      "loss": 11.5,
      "step": 370
    },
    {
      "epoch": 0.05938731816635653,
      "grad_norm": 0.005481667350977659,
      "learning_rate": 0.00017240426016687863,
      "loss": 11.5,
      "step": 371
    },
    {
      "epoch": 0.059547391800228104,
      "grad_norm": 0.0024753627367317677,
      "learning_rate": 0.00017225866778937165,
      "loss": 11.5,
      "step": 372
    },
    {
      "epoch": 0.05970746543409969,
      "grad_norm": 0.0036691934801638126,
      "learning_rate": 0.00017211275418119876,
      "loss": 11.5,
      "step": 373
    },
    {
      "epoch": 0.059867539067971265,
      "grad_norm": 0.006581971887499094,
      "learning_rate": 0.0001719665199910285,
      "loss": 11.5,
      "step": 374
    },
    {
      "epoch": 0.06002761270184285,
      "grad_norm": 0.0032361126504838467,
      "learning_rate": 0.00017181996586895454,
      "loss": 11.5,
      "step": 375
    },
    {
      "epoch": 0.06002761270184285,
      "eval_loss": 11.5,
      "eval_runtime": 60.2134,
      "eval_samples_per_second": 174.745,
      "eval_steps_per_second": 87.373,
      "step": 375
    },
    {
      "epoch": 0.060187686335714426,
      "grad_norm": 0.004640330094844103,
      "learning_rate": 0.00017167309246649297,
      "loss": 11.5,
      "step": 376
    },
    {
      "epoch": 0.06034775996958601,
      "grad_norm": 0.00437448313459754,
      "learning_rate": 0.0001715259004365791,
      "loss": 11.5,
      "step": 377
    },
    {
      "epoch": 0.06050783360345759,
      "grad_norm": 0.003996383864432573,
      "learning_rate": 0.00017137839043356484,
      "loss": 11.5,
      "step": 378
    },
    {
      "epoch": 0.06066790723732917,
      "grad_norm": 0.008043423295021057,
      "learning_rate": 0.00017123056311321562,
      "loss": 11.5,
      "step": 379
    },
    {
      "epoch": 0.060827980871200754,
      "grad_norm": 0.003773126285523176,
      "learning_rate": 0.0001710824191327075,
      "loss": 11.5,
      "step": 380
    },
    {
      "epoch": 0.06098805450507233,
      "grad_norm": 0.0024749895092099905,
      "learning_rate": 0.00017093395915062428,
      "loss": 11.5,
      "step": 381
    },
    {
      "epoch": 0.061148128138943915,
      "grad_norm": 0.0026714850682765245,
      "learning_rate": 0.00017078518382695465,
      "loss": 11.5,
      "step": 382
    },
    {
      "epoch": 0.06130820177281549,
      "grad_norm": 0.00383149110712111,
      "learning_rate": 0.00017063609382308908,
      "loss": 11.5,
      "step": 383
    },
    {
      "epoch": 0.061468275406687076,
      "grad_norm": 0.0031768609769642353,
      "learning_rate": 0.00017048668980181698,
      "loss": 11.5,
      "step": 384
    },
    {
      "epoch": 0.06162834904055866,
      "grad_norm": 0.004482600837945938,
      "learning_rate": 0.00017033697242732377,
      "loss": 11.5,
      "step": 385
    },
    {
      "epoch": 0.061788422674430236,
      "grad_norm": 0.00431401701644063,
      "learning_rate": 0.0001701869423651879,
      "loss": 11.5,
      "step": 386
    },
    {
      "epoch": 0.06194849630830182,
      "grad_norm": 0.004294798243790865,
      "learning_rate": 0.00017003660028237793,
      "loss": 11.5,
      "step": 387
    },
    {
      "epoch": 0.0621085699421734,
      "grad_norm": 0.00420028530061245,
      "learning_rate": 0.00016988594684724947,
      "loss": 11.5,
      "step": 388
    },
    {
      "epoch": 0.06226864357604498,
      "grad_norm": 0.0035517988726496696,
      "learning_rate": 0.00016973498272954222,
      "loss": 11.5,
      "step": 389
    },
    {
      "epoch": 0.062428717209916565,
      "grad_norm": 0.0033941552974283695,
      "learning_rate": 0.00016958370860037717,
      "loss": 11.5,
      "step": 390
    },
    {
      "epoch": 0.06258879084378814,
      "grad_norm": 0.004522206727415323,
      "learning_rate": 0.00016943212513225345,
      "loss": 11.5,
      "step": 391
    },
    {
      "epoch": 0.06274886447765972,
      "grad_norm": 0.0027748660650104284,
      "learning_rate": 0.00016928023299904533,
      "loss": 11.5,
      "step": 392
    },
    {
      "epoch": 0.06290893811153131,
      "grad_norm": 0.015132404863834381,
      "learning_rate": 0.0001691280328759992,
      "loss": 11.5,
      "step": 393
    },
    {
      "epoch": 0.06306901174540289,
      "grad_norm": 0.030538409948349,
      "learning_rate": 0.00016897552543973084,
      "loss": 11.5,
      "step": 394
    },
    {
      "epoch": 0.06322908537927446,
      "grad_norm": 0.009957745671272278,
      "learning_rate": 0.00016882271136822206,
      "loss": 11.5,
      "step": 395
    },
    {
      "epoch": 0.06338915901314604,
      "grad_norm": 0.008293859660625458,
      "learning_rate": 0.0001686695913408179,
      "loss": 11.5,
      "step": 396
    },
    {
      "epoch": 0.06354923264701763,
      "grad_norm": 0.008452611975371838,
      "learning_rate": 0.0001685161660382235,
      "loss": 11.5,
      "step": 397
    },
    {
      "epoch": 0.06370930628088921,
      "grad_norm": 0.012902924790978432,
      "learning_rate": 0.00016836243614250113,
      "loss": 11.5,
      "step": 398
    },
    {
      "epoch": 0.06386937991476078,
      "grad_norm": 0.004804533440619707,
      "learning_rate": 0.00016820840233706719,
      "loss": 11.5,
      "step": 399
    },
    {
      "epoch": 0.06402945354863238,
      "grad_norm": 0.007359676528722048,
      "learning_rate": 0.0001680540653066891,
      "loss": 11.5,
      "step": 400
    },
    {
      "epoch": 0.06418952718250395,
      "grad_norm": 0.004529973026365042,
      "learning_rate": 0.00016789942573748232,
      "loss": 11.5,
      "step": 401
    },
    {
      "epoch": 0.06434960081637553,
      "grad_norm": 0.0040603941306471825,
      "learning_rate": 0.0001677444843169072,
      "loss": 11.5,
      "step": 402
    },
    {
      "epoch": 0.06450967445024712,
      "grad_norm": 0.004139019176363945,
      "learning_rate": 0.00016758924173376603,
      "loss": 11.5,
      "step": 403
    },
    {
      "epoch": 0.0646697480841187,
      "grad_norm": 0.004315665923058987,
      "learning_rate": 0.0001674336986781999,
      "loss": 11.5,
      "step": 404
    },
    {
      "epoch": 0.06482982171799027,
      "grad_norm": 0.0029983599670231342,
      "learning_rate": 0.00016727785584168581,
      "loss": 11.5,
      "step": 405
    },
    {
      "epoch": 0.06498989535186185,
      "grad_norm": 0.003190722782164812,
      "learning_rate": 0.0001671217139170333,
      "loss": 11.5,
      "step": 406
    },
    {
      "epoch": 0.06514996898573344,
      "grad_norm": 0.005884988699108362,
      "learning_rate": 0.00016696527359838154,
      "loss": 11.5,
      "step": 407
    },
    {
      "epoch": 0.06531004261960502,
      "grad_norm": 0.003929830156266689,
      "learning_rate": 0.00016680853558119632,
      "loss": 11.5,
      "step": 408
    },
    {
      "epoch": 0.0654701162534766,
      "grad_norm": 0.003794959979131818,
      "learning_rate": 0.0001666515005622668,
      "loss": 11.5,
      "step": 409
    },
    {
      "epoch": 0.06563018988734819,
      "grad_norm": 0.004015319515019655,
      "learning_rate": 0.0001664941692397025,
      "loss": 11.5,
      "step": 410
    },
    {
      "epoch": 0.06579026352121976,
      "grad_norm": 0.005592698697000742,
      "learning_rate": 0.00016633654231293013,
      "loss": 11.5,
      "step": 411
    },
    {
      "epoch": 0.06595033715509134,
      "grad_norm": 0.006442185491323471,
      "learning_rate": 0.00016617862048269065,
      "loss": 11.5,
      "step": 412
    },
    {
      "epoch": 0.06611041078896292,
      "grad_norm": 0.0037807307671755552,
      "learning_rate": 0.00016602040445103588,
      "loss": 11.5,
      "step": 413
    },
    {
      "epoch": 0.06627048442283451,
      "grad_norm": 0.0030333532486110926,
      "learning_rate": 0.00016586189492132566,
      "loss": 11.5,
      "step": 414
    },
    {
      "epoch": 0.06643055805670608,
      "grad_norm": 0.003252002876251936,
      "learning_rate": 0.00016570309259822453,
      "loss": 11.5,
      "step": 415
    },
    {
      "epoch": 0.06659063169057766,
      "grad_norm": 0.005443210247904062,
      "learning_rate": 0.0001655439981876987,
      "loss": 11.5,
      "step": 416
    },
    {
      "epoch": 0.06675070532444925,
      "grad_norm": 0.004117962904274464,
      "learning_rate": 0.00016538461239701277,
      "loss": 11.5,
      "step": 417
    },
    {
      "epoch": 0.06691077895832083,
      "grad_norm": 0.004072123672813177,
      "learning_rate": 0.00016522493593472683,
      "loss": 11.5,
      "step": 418
    },
    {
      "epoch": 0.0670708525921924,
      "grad_norm": 0.0023909725714474916,
      "learning_rate": 0.0001650649695106931,
      "loss": 11.5,
      "step": 419
    },
    {
      "epoch": 0.06723092622606398,
      "grad_norm": 0.004317248240113258,
      "learning_rate": 0.00016490471383605288,
      "loss": 11.5,
      "step": 420
    },
    {
      "epoch": 0.06739099985993557,
      "grad_norm": 0.003780324012041092,
      "learning_rate": 0.00016474416962323325,
      "loss": 11.5,
      "step": 421
    },
    {
      "epoch": 0.06755107349380715,
      "grad_norm": 0.003763846354559064,
      "learning_rate": 0.00016458333758594414,
      "loss": 11.5,
      "step": 422
    },
    {
      "epoch": 0.06771114712767873,
      "grad_norm": 0.0038902440574020147,
      "learning_rate": 0.00016442221843917496,
      "loss": 11.5,
      "step": 423
    },
    {
      "epoch": 0.06787122076155032,
      "grad_norm": 0.003966157324612141,
      "learning_rate": 0.00016426081289919143,
      "loss": 11.5,
      "step": 424
    },
    {
      "epoch": 0.0680312943954219,
      "grad_norm": 0.004923288710415363,
      "learning_rate": 0.0001640991216835326,
      "loss": 11.5,
      "step": 425
    },
    {
      "epoch": 0.06819136802929347,
      "grad_norm": 0.003366629360243678,
      "learning_rate": 0.00016393714551100734,
      "loss": 11.5,
      "step": 426
    },
    {
      "epoch": 0.06835144166316505,
      "grad_norm": 0.005111074540764093,
      "learning_rate": 0.0001637748851016914,
      "loss": 11.5,
      "step": 427
    },
    {
      "epoch": 0.06851151529703664,
      "grad_norm": 0.0035507299471646547,
      "learning_rate": 0.00016361234117692413,
      "loss": 11.5,
      "step": 428
    },
    {
      "epoch": 0.06867158893090822,
      "grad_norm": 0.005407991353422403,
      "learning_rate": 0.00016344951445930526,
      "loss": 11.5,
      "step": 429
    },
    {
      "epoch": 0.0688316625647798,
      "grad_norm": 0.003740898100659251,
      "learning_rate": 0.0001632864056726917,
      "loss": 11.5,
      "step": 430
    },
    {
      "epoch": 0.06899173619865138,
      "grad_norm": 0.0027963933534920216,
      "learning_rate": 0.00016312301554219426,
      "loss": 11.5,
      "step": 431
    },
    {
      "epoch": 0.06915180983252296,
      "grad_norm": 0.003220595885068178,
      "learning_rate": 0.00016295934479417453,
      "loss": 11.5,
      "step": 432
    },
    {
      "epoch": 0.06931188346639454,
      "grad_norm": 0.002428829437121749,
      "learning_rate": 0.00016279539415624164,
      "loss": 11.5,
      "step": 433
    },
    {
      "epoch": 0.06947195710026612,
      "grad_norm": 0.004904985893517733,
      "learning_rate": 0.0001626311643572489,
      "loss": 11.5,
      "step": 434
    },
    {
      "epoch": 0.0696320307341377,
      "grad_norm": 0.003591889049857855,
      "learning_rate": 0.00016246665612729074,
      "loss": 11.5,
      "step": 435
    },
    {
      "epoch": 0.06979210436800928,
      "grad_norm": 0.0030072187073528767,
      "learning_rate": 0.00016230187019769928,
      "loss": 11.5,
      "step": 436
    },
    {
      "epoch": 0.06995217800188086,
      "grad_norm": 0.0034716983791440725,
      "learning_rate": 0.00016213680730104124,
      "loss": 11.5,
      "step": 437
    },
    {
      "epoch": 0.07011225163575245,
      "grad_norm": 0.004672607406973839,
      "learning_rate": 0.0001619714681711146,
      "loss": 11.5,
      "step": 438
    },
    {
      "epoch": 0.07027232526962403,
      "grad_norm": 0.002551988000050187,
      "learning_rate": 0.00016180585354294536,
      "loss": 11.5,
      "step": 439
    },
    {
      "epoch": 0.0704323989034956,
      "grad_norm": 0.004141319077461958,
      "learning_rate": 0.00016163996415278424,
      "loss": 11.5,
      "step": 440
    },
    {
      "epoch": 0.0705924725373672,
      "grad_norm": 0.0030993500258773565,
      "learning_rate": 0.00016147380073810346,
      "loss": 11.5,
      "step": 441
    },
    {
      "epoch": 0.07075254617123877,
      "grad_norm": 0.0031521443743258715,
      "learning_rate": 0.0001613073640375934,
      "loss": 11.5,
      "step": 442
    },
    {
      "epoch": 0.07091261980511035,
      "grad_norm": 0.024885622784495354,
      "learning_rate": 0.00016114065479115946,
      "loss": 11.5,
      "step": 443
    },
    {
      "epoch": 0.07107269343898193,
      "grad_norm": 0.051207348704338074,
      "learning_rate": 0.00016097367373991842,
      "loss": 11.5,
      "step": 444
    },
    {
      "epoch": 0.07123276707285352,
      "grad_norm": 0.008169927634298801,
      "learning_rate": 0.00016080642162619565,
      "loss": 11.5,
      "step": 445
    },
    {
      "epoch": 0.0713928407067251,
      "grad_norm": 0.007604275830090046,
      "learning_rate": 0.0001606388991935214,
      "loss": 11.5,
      "step": 446
    },
    {
      "epoch": 0.07155291434059667,
      "grad_norm": 0.007420423440635204,
      "learning_rate": 0.0001604711071866277,
      "loss": 11.5,
      "step": 447
    },
    {
      "epoch": 0.07171298797446826,
      "grad_norm": 0.00525730662047863,
      "learning_rate": 0.00016030304635144494,
      "loss": 11.5,
      "step": 448
    },
    {
      "epoch": 0.07187306160833984,
      "grad_norm": 0.005916302558034658,
      "learning_rate": 0.00016013471743509862,
      "loss": 11.5,
      "step": 449
    },
    {
      "epoch": 0.07203313524221142,
      "grad_norm": 0.011136213317513466,
      "learning_rate": 0.00015996612118590603,
      "loss": 11.5,
      "step": 450
    },
    {
      "epoch": 0.07219320887608299,
      "grad_norm": 0.005346770863980055,
      "learning_rate": 0.00015979725835337294,
      "loss": 11.5,
      "step": 451
    },
    {
      "epoch": 0.07235328250995458,
      "grad_norm": 0.0031074946746230125,
      "learning_rate": 0.00015962812968819016,
      "loss": 11.5,
      "step": 452
    },
    {
      "epoch": 0.07251335614382616,
      "grad_norm": 0.0027644415386021137,
      "learning_rate": 0.0001594587359422303,
      "loss": 11.5,
      "step": 453
    },
    {
      "epoch": 0.07267342977769774,
      "grad_norm": 0.006376570090651512,
      "learning_rate": 0.0001592890778685444,
      "loss": 11.5,
      "step": 454
    },
    {
      "epoch": 0.07283350341156933,
      "grad_norm": 0.004349604249000549,
      "learning_rate": 0.00015911915622135862,
      "loss": 11.5,
      "step": 455
    },
    {
      "epoch": 0.0729935770454409,
      "grad_norm": 0.002568909665569663,
      "learning_rate": 0.00015894897175607086,
      "loss": 11.5,
      "step": 456
    },
    {
      "epoch": 0.07315365067931248,
      "grad_norm": 0.002386229345574975,
      "learning_rate": 0.00015877852522924732,
      "loss": 11.5,
      "step": 457
    },
    {
      "epoch": 0.07331372431318406,
      "grad_norm": 0.005009147804230452,
      "learning_rate": 0.00015860781739861928,
      "loss": 11.5,
      "step": 458
    },
    {
      "epoch": 0.07347379794705565,
      "grad_norm": 0.004310858901590109,
      "learning_rate": 0.00015843684902307962,
      "loss": 11.5,
      "step": 459
    },
    {
      "epoch": 0.07363387158092723,
      "grad_norm": 0.002923219930380583,
      "learning_rate": 0.00015826562086267956,
      "loss": 11.5,
      "step": 460
    },
    {
      "epoch": 0.0737939452147988,
      "grad_norm": 0.004609262570738792,
      "learning_rate": 0.00015809413367862512,
      "loss": 11.5,
      "step": 461
    },
    {
      "epoch": 0.0739540188486704,
      "grad_norm": 0.0058220489881932735,
      "learning_rate": 0.00015792238823327388,
      "loss": 11.5,
      "step": 462
    },
    {
      "epoch": 0.07411409248254197,
      "grad_norm": 0.002544667571783066,
      "learning_rate": 0.00015775038529013152,
      "loss": 11.5,
      "step": 463
    },
    {
      "epoch": 0.07427416611641355,
      "grad_norm": 0.0015520495362579823,
      "learning_rate": 0.0001575781256138485,
      "loss": 11.5,
      "step": 464
    },
    {
      "epoch": 0.07443423975028512,
      "grad_norm": 0.007448026444762945,
      "learning_rate": 0.00015740560997021648,
      "loss": 11.5,
      "step": 465
    },
    {
      "epoch": 0.07459431338415672,
      "grad_norm": 0.0019627476576715708,
      "learning_rate": 0.00015723283912616513,
      "loss": 11.5,
      "step": 466
    },
    {
      "epoch": 0.07475438701802829,
      "grad_norm": 0.004363425076007843,
      "learning_rate": 0.00015705981384975866,
      "loss": 11.5,
      "step": 467
    },
    {
      "epoch": 0.07491446065189987,
      "grad_norm": 0.0027951716911047697,
      "learning_rate": 0.0001568865349101923,
      "loss": 11.5,
      "step": 468
    },
    {
      "epoch": 0.07507453428577146,
      "grad_norm": 0.0024843004066497087,
      "learning_rate": 0.00015671300307778898,
      "loss": 11.5,
      "step": 469
    },
    {
      "epoch": 0.07523460791964304,
      "grad_norm": 0.002484348602592945,
      "learning_rate": 0.00015653921912399589,
      "loss": 11.5,
      "step": 470
    },
    {
      "epoch": 0.07539468155351461,
      "grad_norm": 0.0031197178177535534,
      "learning_rate": 0.00015636518382138107,
      "loss": 11.5,
      "step": 471
    },
    {
      "epoch": 0.07555475518738619,
      "grad_norm": 0.0030909341294318438,
      "learning_rate": 0.0001561908979436299,
      "loss": 11.5,
      "step": 472
    },
    {
      "epoch": 0.07571482882125778,
      "grad_norm": 0.0022891343105584383,
      "learning_rate": 0.00015601636226554168,
      "loss": 11.5,
      "step": 473
    },
    {
      "epoch": 0.07587490245512936,
      "grad_norm": 0.0042622266337275505,
      "learning_rate": 0.00015584157756302634,
      "loss": 11.5,
      "step": 474
    },
    {
      "epoch": 0.07603497608900094,
      "grad_norm": 0.005968592129647732,
      "learning_rate": 0.0001556665446131007,
      "loss": 11.5,
      "step": 475
    },
    {
      "epoch": 0.07619504972287253,
      "grad_norm": 0.00357861933298409,
      "learning_rate": 0.00015549126419388536,
      "loss": 11.5,
      "step": 476
    },
    {
      "epoch": 0.0763551233567441,
      "grad_norm": 0.004084532614797354,
      "learning_rate": 0.0001553157370846009,
      "loss": 11.5,
      "step": 477
    },
    {
      "epoch": 0.07651519699061568,
      "grad_norm": 0.005763905588537455,
      "learning_rate": 0.00015513996406556465,
      "loss": 11.5,
      "step": 478
    },
    {
      "epoch": 0.07667527062448726,
      "grad_norm": 0.0061824023723602295,
      "learning_rate": 0.00015496394591818716,
      "loss": 11.5,
      "step": 479
    },
    {
      "epoch": 0.07683534425835885,
      "grad_norm": 0.003055471694096923,
      "learning_rate": 0.0001547876834249687,
      "loss": 11.5,
      "step": 480
    },
    {
      "epoch": 0.07699541789223042,
      "grad_norm": 0.01285121776163578,
      "learning_rate": 0.00015461117736949577,
      "loss": 11.5,
      "step": 481
    },
    {
      "epoch": 0.077155491526102,
      "grad_norm": 0.0016418943414464593,
      "learning_rate": 0.00015443442853643762,
      "loss": 11.5,
      "step": 482
    },
    {
      "epoch": 0.07731556515997359,
      "grad_norm": 0.004201004281640053,
      "learning_rate": 0.00015425743771154294,
      "loss": 11.5,
      "step": 483
    },
    {
      "epoch": 0.07747563879384517,
      "grad_norm": 0.0018239245982840657,
      "learning_rate": 0.00015408020568163602,
      "loss": 11.5,
      "step": 484
    },
    {
      "epoch": 0.07763571242771675,
      "grad_norm": 0.003994714468717575,
      "learning_rate": 0.00015390273323461352,
      "loss": 11.5,
      "step": 485
    },
    {
      "epoch": 0.07779578606158834,
      "grad_norm": 0.002866100985556841,
      "learning_rate": 0.0001537250211594409,
      "loss": 11.5,
      "step": 486
    },
    {
      "epoch": 0.07795585969545991,
      "grad_norm": 0.004548719152808189,
      "learning_rate": 0.0001535470702461489,
      "loss": 11.5,
      "step": 487
    },
    {
      "epoch": 0.07811593332933149,
      "grad_norm": 0.0016402106266468763,
      "learning_rate": 0.00015336888128583,
      "loss": 11.5,
      "step": 488
    },
    {
      "epoch": 0.07827600696320307,
      "grad_norm": 0.0019115133909508586,
      "learning_rate": 0.000153190455070635,
      "loss": 11.5,
      "step": 489
    },
    {
      "epoch": 0.07843608059707466,
      "grad_norm": 0.0035825162194669247,
      "learning_rate": 0.00015301179239376938,
      "loss": 11.5,
      "step": 490
    },
    {
      "epoch": 0.07859615423094624,
      "grad_norm": 0.0030351774767041206,
      "learning_rate": 0.00015283289404948976,
      "loss": 11.5,
      "step": 491
    },
    {
      "epoch": 0.07875622786481781,
      "grad_norm": 0.004369048867374659,
      "learning_rate": 0.0001526537608331006,
      "loss": 11.5,
      "step": 492
    },
    {
      "epoch": 0.0789163014986894,
      "grad_norm": 0.023543706163764,
      "learning_rate": 0.00015247439354095041,
      "loss": 11.5,
      "step": 493
    },
    {
      "epoch": 0.07907637513256098,
      "grad_norm": 0.025057869032025337,
      "learning_rate": 0.00015229479297042823,
      "loss": 11.5,
      "step": 494
    },
    {
      "epoch": 0.07923644876643256,
      "grad_norm": 0.011006333865225315,
      "learning_rate": 0.00015211495991996027,
      "loss": 11.5,
      "step": 495
    },
    {
      "epoch": 0.07939652240030413,
      "grad_norm": 0.008471246808767319,
      "learning_rate": 0.0001519348951890062,
      "loss": 11.5,
      "step": 496
    },
    {
      "epoch": 0.07955659603417573,
      "grad_norm": 0.008287298493087292,
      "learning_rate": 0.0001517545995780556,
      "loss": 11.5,
      "step": 497
    },
    {
      "epoch": 0.0797166696680473,
      "grad_norm": 0.010373201221227646,
      "learning_rate": 0.00015157407388862452,
      "loss": 11.5,
      "step": 498
    },
    {
      "epoch": 0.07987674330191888,
      "grad_norm": 0.004067798610776663,
      "learning_rate": 0.00015139331892325179,
      "loss": 11.5,
      "step": 499
    },
    {
      "epoch": 0.08003681693579047,
      "grad_norm": 0.006332798860967159,
      "learning_rate": 0.0001512123354854955,
      "loss": 11.5,
      "step": 500
    },
    {
      "epoch": 0.08019689056966205,
      "grad_norm": 0.004832427483052015,
      "learning_rate": 0.0001510311243799295,
      "loss": 11.5,
      "step": 501
    },
    {
      "epoch": 0.08035696420353362,
      "grad_norm": 0.004609941504895687,
      "learning_rate": 0.00015084968641213958,
      "loss": 11.5,
      "step": 502
    },
    {
      "epoch": 0.0805170378374052,
      "grad_norm": 0.004927306901663542,
      "learning_rate": 0.00015066802238872023,
      "loss": 11.5,
      "step": 503
    },
    {
      "epoch": 0.08067711147127679,
      "grad_norm": 0.004748067818582058,
      "learning_rate": 0.0001504861331172709,
      "loss": 11.5,
      "step": 504
    },
    {
      "epoch": 0.08083718510514837,
      "grad_norm": 0.0033973974641412497,
      "learning_rate": 0.0001503040194063922,
      "loss": 11.5,
      "step": 505
    },
    {
      "epoch": 0.08099725873901994,
      "grad_norm": 0.0046224878169596195,
      "learning_rate": 0.00015012168206568268,
      "loss": 11.5,
      "step": 506
    },
    {
      "epoch": 0.08115733237289154,
      "grad_norm": 0.0028793816454708576,
      "learning_rate": 0.00014993912190573505,
      "loss": 11.5,
      "step": 507
    },
    {
      "epoch": 0.08131740600676311,
      "grad_norm": 0.008805199526250362,
      "learning_rate": 0.00014975633973813242,
      "loss": 11.5,
      "step": 508
    },
    {
      "epoch": 0.08147747964063469,
      "grad_norm": 0.002763141179457307,
      "learning_rate": 0.00014957333637544503,
      "loss": 11.5,
      "step": 509
    },
    {
      "epoch": 0.08163755327450627,
      "grad_norm": 0.002478482900187373,
      "learning_rate": 0.00014939011263122634,
      "loss": 11.5,
      "step": 510
    },
    {
      "epoch": 0.08179762690837786,
      "grad_norm": 0.0035602550487965345,
      "learning_rate": 0.0001492066693200096,
      "loss": 11.5,
      "step": 511
    },
    {
      "epoch": 0.08195770054224943,
      "grad_norm": 0.0036557859275490046,
      "learning_rate": 0.00014902300725730413,
      "loss": 11.5,
      "step": 512
    },
    {
      "epoch": 0.08211777417612101,
      "grad_norm": 0.00601975480094552,
      "learning_rate": 0.00014883912725959167,
      "loss": 11.5,
      "step": 513
    },
    {
      "epoch": 0.0822778478099926,
      "grad_norm": 0.0021397762466222048,
      "learning_rate": 0.00014865503014432292,
      "loss": 11.5,
      "step": 514
    },
    {
      "epoch": 0.08243792144386418,
      "grad_norm": 0.00622235843911767,
      "learning_rate": 0.00014847071672991367,
      "loss": 11.5,
      "step": 515
    },
    {
      "epoch": 0.08259799507773576,
      "grad_norm": 0.002387597691267729,
      "learning_rate": 0.0001482861878357414,
      "loss": 11.5,
      "step": 516
    },
    {
      "epoch": 0.08275806871160733,
      "grad_norm": 0.0027929286006838083,
      "learning_rate": 0.00014810144428214144,
      "loss": 11.5,
      "step": 517
    },
    {
      "epoch": 0.08291814234547892,
      "grad_norm": 0.0035270266234874725,
      "learning_rate": 0.0001479164868904034,
      "loss": 11.5,
      "step": 518
    },
    {
      "epoch": 0.0830782159793505,
      "grad_norm": 0.003800722537562251,
      "learning_rate": 0.00014773131648276758,
      "loss": 11.5,
      "step": 519
    },
    {
      "epoch": 0.08323828961322208,
      "grad_norm": 0.0030791962053626776,
      "learning_rate": 0.00014754593388242117,
      "loss": 11.5,
      "step": 520
    },
    {
      "epoch": 0.08339836324709367,
      "grad_norm": 0.0031547355465590954,
      "learning_rate": 0.0001473603399134948,
      "loss": 11.5,
      "step": 521
    },
    {
      "epoch": 0.08355843688096525,
      "grad_norm": 0.005380855407565832,
      "learning_rate": 0.0001471745354010586,
      "loss": 11.5,
      "step": 522
    },
    {
      "epoch": 0.08371851051483682,
      "grad_norm": 0.006104020401835442,
      "learning_rate": 0.00014698852117111884,
      "loss": 11.5,
      "step": 523
    },
    {
      "epoch": 0.08387858414870841,
      "grad_norm": 0.0033588106743991375,
      "learning_rate": 0.000146802298050614,
      "loss": 11.5,
      "step": 524
    },
    {
      "epoch": 0.08403865778257999,
      "grad_norm": 0.0022925056982785463,
      "learning_rate": 0.0001466158668674112,
      "loss": 11.5,
      "step": 525
    },
    {
      "epoch": 0.08419873141645157,
      "grad_norm": 0.0030669146217405796,
      "learning_rate": 0.00014642922845030257,
      "loss": 11.5,
      "step": 526
    },
    {
      "epoch": 0.08435880505032314,
      "grad_norm": 0.005041934084147215,
      "learning_rate": 0.0001462423836290015,
      "loss": 11.5,
      "step": 527
    },
    {
      "epoch": 0.08451887868419473,
      "grad_norm": 0.0041120778769254684,
      "learning_rate": 0.00014605533323413887,
      "loss": 11.5,
      "step": 528
    },
    {
      "epoch": 0.08467895231806631,
      "grad_norm": 0.0030222199857234955,
      "learning_rate": 0.00014586807809725962,
      "loss": 11.5,
      "step": 529
    },
    {
      "epoch": 0.08483902595193789,
      "grad_norm": 0.00622931495308876,
      "learning_rate": 0.00014568061905081875,
      "loss": 11.5,
      "step": 530
    },
    {
      "epoch": 0.08499909958580948,
      "grad_norm": 0.004166914615780115,
      "learning_rate": 0.00014549295692817778,
      "loss": 11.5,
      "step": 531
    },
    {
      "epoch": 0.08515917321968106,
      "grad_norm": 0.0025993871968239546,
      "learning_rate": 0.00014530509256360102,
      "loss": 11.5,
      "step": 532
    },
    {
      "epoch": 0.08531924685355263,
      "grad_norm": 0.002257340122014284,
      "learning_rate": 0.00014511702679225193,
      "loss": 11.5,
      "step": 533
    },
    {
      "epoch": 0.08547932048742421,
      "grad_norm": 0.002261596266180277,
      "learning_rate": 0.0001449287604501893,
      "loss": 11.5,
      "step": 534
    },
    {
      "epoch": 0.0856393941212958,
      "grad_norm": 0.0021795269567519426,
      "learning_rate": 0.00014474029437436348,
      "loss": 11.5,
      "step": 535
    },
    {
      "epoch": 0.08579946775516738,
      "grad_norm": 0.0021723133977502584,
      "learning_rate": 0.00014455162940261285,
      "loss": 11.5,
      "step": 536
    },
    {
      "epoch": 0.08595954138903895,
      "grad_norm": 0.0032066942658275366,
      "learning_rate": 0.0001443627663736599,
      "loss": 11.5,
      "step": 537
    },
    {
      "epoch": 0.08611961502291055,
      "grad_norm": 0.003661336377263069,
      "learning_rate": 0.00014417370612710778,
      "loss": 11.5,
      "step": 538
    },
    {
      "epoch": 0.08627968865678212,
      "grad_norm": 0.003820454701781273,
      "learning_rate": 0.00014398444950343623,
      "loss": 11.5,
      "step": 539
    },
    {
      "epoch": 0.0864397622906537,
      "grad_norm": 0.0021902904845774174,
      "learning_rate": 0.00014379499734399798,
      "loss": 11.5,
      "step": 540
    },
    {
      "epoch": 0.08659983592452528,
      "grad_norm": 0.002264488022774458,
      "learning_rate": 0.0001436053504910151,
      "loss": 11.5,
      "step": 541
    },
    {
      "epoch": 0.08675990955839687,
      "grad_norm": 0.0023856647312641144,
      "learning_rate": 0.0001434155097875752,
      "loss": 11.5,
      "step": 542
    },
    {
      "epoch": 0.08691998319226844,
      "grad_norm": 0.024050379171967506,
      "learning_rate": 0.00014322547607762762,
      "loss": 11.5,
      "step": 543
    },
    {
      "epoch": 0.08708005682614002,
      "grad_norm": 0.02946372516453266,
      "learning_rate": 0.0001430352502059797,
      "loss": 11.5,
      "step": 544
    },
    {
      "epoch": 0.08724013046001161,
      "grad_norm": 0.006515579763799906,
      "learning_rate": 0.0001428448330182931,
      "loss": 11.5,
      "step": 545
    },
    {
      "epoch": 0.08740020409388319,
      "grad_norm": 0.008249339647591114,
      "learning_rate": 0.00014265422536107993,
      "loss": 11.5,
      "step": 546
    },
    {
      "epoch": 0.08756027772775477,
      "grad_norm": 0.005820396356284618,
      "learning_rate": 0.00014246342808169914,
      "loss": 11.5,
      "step": 547
    },
    {
      "epoch": 0.08772035136162634,
      "grad_norm": 0.009405459277331829,
      "learning_rate": 0.00014227244202835257,
      "loss": 11.5,
      "step": 548
    },
    {
      "epoch": 0.08788042499549793,
      "grad_norm": 0.015361230820417404,
      "learning_rate": 0.0001420812680500813,
      "loss": 11.5,
      "step": 549
    },
    {
      "epoch": 0.08804049862936951,
      "grad_norm": 0.008878029882907867,
      "learning_rate": 0.00014188990699676184,
      "loss": 11.5,
      "step": 550
    },
    {
      "epoch": 0.08820057226324109,
      "grad_norm": 0.003142086323350668,
      "learning_rate": 0.00014169835971910238,
      "loss": 11.5,
      "step": 551
    },
    {
      "epoch": 0.08836064589711268,
      "grad_norm": 0.003109876997768879,
      "learning_rate": 0.0001415066270686389,
      "loss": 11.5,
      "step": 552
    },
    {
      "epoch": 0.08852071953098425,
      "grad_norm": 0.0030721533112227917,
      "learning_rate": 0.00014131470989773158,
      "loss": 11.5,
      "step": 553
    },
    {
      "epoch": 0.08868079316485583,
      "grad_norm": 0.004286905750632286,
      "learning_rate": 0.0001411226090595608,
      "loss": 11.5,
      "step": 554
    },
    {
      "epoch": 0.08884086679872741,
      "grad_norm": 0.005644454155117273,
      "learning_rate": 0.00014093032540812348,
      "loss": 11.5,
      "step": 555
    },
    {
      "epoch": 0.089000940432599,
      "grad_norm": 0.0034797850530594587,
      "learning_rate": 0.0001407378597982293,
      "loss": 11.5,
      "step": 556
    },
    {
      "epoch": 0.08916101406647058,
      "grad_norm": 0.0036068386398255825,
      "learning_rate": 0.00014054521308549673,
      "loss": 11.5,
      "step": 557
    },
    {
      "epoch": 0.08932108770034215,
      "grad_norm": 0.008354442194104195,
      "learning_rate": 0.0001403523861263495,
      "loss": 11.5,
      "step": 558
    },
    {
      "epoch": 0.08948116133421374,
      "grad_norm": 0.005250106565654278,
      "learning_rate": 0.00014015937977801256,
      "loss": 11.5,
      "step": 559
    },
    {
      "epoch": 0.08964123496808532,
      "grad_norm": 0.009559806436300278,
      "learning_rate": 0.00013996619489850822,
      "loss": 11.5,
      "step": 560
    },
    {
      "epoch": 0.0898013086019569,
      "grad_norm": 0.0026824921369552612,
      "learning_rate": 0.00013977283234665273,
      "loss": 11.5,
      "step": 561
    },
    {
      "epoch": 0.08996138223582847,
      "grad_norm": 0.0045197890140116215,
      "learning_rate": 0.00013957929298205195,
      "loss": 11.5,
      "step": 562
    },
    {
      "epoch": 0.09012145586970007,
      "grad_norm": 0.010180555284023285,
      "learning_rate": 0.00013938557766509792,
      "loss": 11.5,
      "step": 563
    },
    {
      "epoch": 0.09028152950357164,
      "grad_norm": 0.0024575996212661266,
      "learning_rate": 0.0001391916872569648,
      "loss": 11.5,
      "step": 564
    },
    {
      "epoch": 0.09044160313744322,
      "grad_norm": 0.0026084051933139563,
      "learning_rate": 0.00013899762261960518,
      "loss": 11.5,
      "step": 565
    },
    {
      "epoch": 0.09060167677131481,
      "grad_norm": 0.003972116392105818,
      "learning_rate": 0.0001388033846157462,
      "loss": 11.5,
      "step": 566
    },
    {
      "epoch": 0.09076175040518639,
      "grad_norm": 0.0028879332821816206,
      "learning_rate": 0.0001386089741088857,
      "loss": 11.5,
      "step": 567
    },
    {
      "epoch": 0.09092182403905796,
      "grad_norm": 0.0033570060040801764,
      "learning_rate": 0.00013841439196328836,
      "loss": 11.5,
      "step": 568
    },
    {
      "epoch": 0.09108189767292955,
      "grad_norm": 0.005385873839259148,
      "learning_rate": 0.00013821963904398193,
      "loss": 11.5,
      "step": 569
    },
    {
      "epoch": 0.09124197130680113,
      "grad_norm": 0.0034743715077638626,
      "learning_rate": 0.00013802471621675338,
      "loss": 11.5,
      "step": 570
    },
    {
      "epoch": 0.09140204494067271,
      "grad_norm": 0.0034987747203558683,
      "learning_rate": 0.00013782962434814492,
      "loss": 11.5,
      "step": 571
    },
    {
      "epoch": 0.09156211857454429,
      "grad_norm": 0.0045075914822518826,
      "learning_rate": 0.00013763436430545034,
      "loss": 11.5,
      "step": 572
    },
    {
      "epoch": 0.09172219220841588,
      "grad_norm": 0.0030352056492120028,
      "learning_rate": 0.00013743893695671096,
      "loss": 11.5,
      "step": 573
    },
    {
      "epoch": 0.09188226584228745,
      "grad_norm": 0.0051122200675308704,
      "learning_rate": 0.00013724334317071198,
      "loss": 11.5,
      "step": 574
    },
    {
      "epoch": 0.09204233947615903,
      "grad_norm": 0.003115959232673049,
      "learning_rate": 0.00013704758381697844,
      "loss": 11.5,
      "step": 575
    },
    {
      "epoch": 0.09220241311003062,
      "grad_norm": 0.0025965438690036535,
      "learning_rate": 0.00013685165976577146,
      "loss": 11.5,
      "step": 576
    },
    {
      "epoch": 0.0923624867439022,
      "grad_norm": 0.007525334600359201,
      "learning_rate": 0.0001366555718880843,
      "loss": 11.5,
      "step": 577
    },
    {
      "epoch": 0.09252256037777377,
      "grad_norm": 0.002660798141732812,
      "learning_rate": 0.00013645932105563844,
      "loss": 11.5,
      "step": 578
    },
    {
      "epoch": 0.09268263401164535,
      "grad_norm": 0.004253697581589222,
      "learning_rate": 0.00013626290814088005,
      "loss": 11.5,
      "step": 579
    },
    {
      "epoch": 0.09284270764551694,
      "grad_norm": 0.006708990782499313,
      "learning_rate": 0.00013606633401697557,
      "loss": 11.5,
      "step": 580
    },
    {
      "epoch": 0.09300278127938852,
      "grad_norm": 0.004029675852507353,
      "learning_rate": 0.00013586959955780824,
      "loss": 11.5,
      "step": 581
    },
    {
      "epoch": 0.0931628549132601,
      "grad_norm": 0.0039032679051160812,
      "learning_rate": 0.00013567270563797398,
      "loss": 11.5,
      "step": 582
    },
    {
      "epoch": 0.09332292854713169,
      "grad_norm": 0.0023409342393279076,
      "learning_rate": 0.00013547565313277776,
      "loss": 11.5,
      "step": 583
    },
    {
      "epoch": 0.09348300218100326,
      "grad_norm": 0.0038006675895303488,
      "learning_rate": 0.00013527844291822948,
      "loss": 11.5,
      "step": 584
    },
    {
      "epoch": 0.09364307581487484,
      "grad_norm": 0.0021854753140360117,
      "learning_rate": 0.0001350810758710401,
      "loss": 11.5,
      "step": 585
    },
    {
      "epoch": 0.09380314944874642,
      "grad_norm": 0.0035075852647423744,
      "learning_rate": 0.00013488355286861783,
      "loss": 11.5,
      "step": 586
    },
    {
      "epoch": 0.09396322308261801,
      "grad_norm": 0.0017785037634894252,
      "learning_rate": 0.0001346858747890642,
      "loss": 11.5,
      "step": 587
    },
    {
      "epoch": 0.09412329671648959,
      "grad_norm": 0.001983600901439786,
      "learning_rate": 0.00013448804251117003,
      "loss": 11.5,
      "step": 588
    },
    {
      "epoch": 0.09428337035036116,
      "grad_norm": 0.002027469454333186,
      "learning_rate": 0.0001342900569144119,
      "loss": 11.5,
      "step": 589
    },
    {
      "epoch": 0.09444344398423275,
      "grad_norm": 0.0018896710826084018,
      "learning_rate": 0.0001340919188789477,
      "loss": 11.5,
      "step": 590
    },
    {
      "epoch": 0.09460351761810433,
      "grad_norm": 0.004546232987195253,
      "learning_rate": 0.00013389362928561317,
      "loss": 11.5,
      "step": 591
    },
    {
      "epoch": 0.0947635912519759,
      "grad_norm": 0.0027206360828131437,
      "learning_rate": 0.00013369518901591772,
      "loss": 11.5,
      "step": 592
    },
    {
      "epoch": 0.09492366488584748,
      "grad_norm": 0.019346138462424278,
      "learning_rate": 0.00013349659895204067,
      "loss": 11.5,
      "step": 593
    },
    {
      "epoch": 0.09508373851971907,
      "grad_norm": 0.03009447269141674,
      "learning_rate": 0.0001332978599768272,
      "loss": 11.5,
      "step": 594
    },
    {
      "epoch": 0.09524381215359065,
      "grad_norm": 0.012629881501197815,
      "learning_rate": 0.00013309897297378455,
      "loss": 11.5,
      "step": 595
    },
    {
      "epoch": 0.09540388578746223,
      "grad_norm": 0.008901401422917843,
      "learning_rate": 0.00013289993882707797,
      "loss": 11.5,
      "step": 596
    },
    {
      "epoch": 0.09556395942133382,
      "grad_norm": 0.009643127210438251,
      "learning_rate": 0.00013270075842152678,
      "loss": 11.5,
      "step": 597
    },
    {
      "epoch": 0.0957240330552054,
      "grad_norm": 0.007158044725656509,
      "learning_rate": 0.00013250143264260074,
      "loss": 11.5,
      "step": 598
    },
    {
      "epoch": 0.09588410668907697,
      "grad_norm": 0.003668807679787278,
      "learning_rate": 0.0001323019623764156,
      "loss": 11.5,
      "step": 599
    },
    {
      "epoch": 0.09604418032294855,
      "grad_norm": 0.009742848575115204,
      "learning_rate": 0.00013210234850972964,
      "loss": 11.5,
      "step": 600
    },
    {
      "epoch": 0.09620425395682014,
      "grad_norm": 0.003795140888541937,
      "learning_rate": 0.0001319025919299394,
      "loss": 11.5,
      "step": 601
    },
    {
      "epoch": 0.09636432759069172,
      "grad_norm": 0.00407062191516161,
      "learning_rate": 0.00013170269352507597,
      "loss": 11.5,
      "step": 602
    },
    {
      "epoch": 0.0965244012245633,
      "grad_norm": 0.004391151014715433,
      "learning_rate": 0.0001315026541838008,
      "loss": 11.5,
      "step": 603
    },
    {
      "epoch": 0.09668447485843489,
      "grad_norm": 0.0037375150714069605,
      "learning_rate": 0.00013130247479540202,
      "loss": 11.5,
      "step": 604
    },
    {
      "epoch": 0.09684454849230646,
      "grad_norm": 0.0031808826606720686,
      "learning_rate": 0.00013110215624979025,
      "loss": 11.5,
      "step": 605
    },
    {
      "epoch": 0.09700462212617804,
      "grad_norm": 0.0024859332479536533,
      "learning_rate": 0.00013090169943749476,
      "loss": 11.5,
      "step": 606
    },
    {
      "epoch": 0.09716469576004963,
      "grad_norm": 0.002641265280544758,
      "learning_rate": 0.00013070110524965954,
      "loss": 11.5,
      "step": 607
    },
    {
      "epoch": 0.0973247693939212,
      "grad_norm": 0.0030788355506956577,
      "learning_rate": 0.00013050037457803924,
      "loss": 11.5,
      "step": 608
    },
    {
      "epoch": 0.09748484302779278,
      "grad_norm": 0.0029422331135720015,
      "learning_rate": 0.0001302995083149953,
      "loss": 11.5,
      "step": 609
    },
    {
      "epoch": 0.09764491666166436,
      "grad_norm": 0.004001727793365717,
      "learning_rate": 0.0001300985073534919,
      "loss": 11.5,
      "step": 610
    },
    {
      "epoch": 0.09780499029553595,
      "grad_norm": 0.0034312887582927942,
      "learning_rate": 0.00012989737258709203,
      "loss": 11.5,
      "step": 611
    },
    {
      "epoch": 0.09796506392940753,
      "grad_norm": 0.007745360489934683,
      "learning_rate": 0.00012969610490995358,
      "loss": 11.5,
      "step": 612
    },
    {
      "epoch": 0.0981251375632791,
      "grad_norm": 0.00498192897066474,
      "learning_rate": 0.00012949470521682528,
      "loss": 11.5,
      "step": 613
    },
    {
      "epoch": 0.0982852111971507,
      "grad_norm": 0.007052308414131403,
      "learning_rate": 0.0001292931744030427,
      "loss": 11.5,
      "step": 614
    },
    {
      "epoch": 0.09844528483102227,
      "grad_norm": 0.0052633509039878845,
      "learning_rate": 0.0001290915133645243,
      "loss": 11.5,
      "step": 615
    },
    {
      "epoch": 0.09860535846489385,
      "grad_norm": 0.003086523385718465,
      "learning_rate": 0.00012888972299776754,
      "loss": 11.5,
      "step": 616
    },
    {
      "epoch": 0.09876543209876543,
      "grad_norm": 0.004674405325204134,
      "learning_rate": 0.00012868780419984482,
      "loss": 11.5,
      "step": 617
    },
    {
      "epoch": 0.09892550573263702,
      "grad_norm": 0.0023710811510682106,
      "learning_rate": 0.00012848575786839943,
      "loss": 11.5,
      "step": 618
    },
    {
      "epoch": 0.0990855793665086,
      "grad_norm": 0.001511088339611888,
      "learning_rate": 0.0001282835849016416,
      "loss": 11.5,
      "step": 619
    },
    {
      "epoch": 0.09924565300038017,
      "grad_norm": 0.004471070598810911,
      "learning_rate": 0.00012808128619834461,
      "loss": 11.5,
      "step": 620
    },
    {
      "epoch": 0.09940572663425176,
      "grad_norm": 0.005246717482805252,
      "learning_rate": 0.0001278788626578407,
      "loss": 11.5,
      "step": 621
    },
    {
      "epoch": 0.09956580026812334,
      "grad_norm": 0.003558073192834854,
      "learning_rate": 0.00012767631518001698,
      "loss": 11.5,
      "step": 622
    },
    {
      "epoch": 0.09972587390199492,
      "grad_norm": 0.002810321282595396,
      "learning_rate": 0.00012747364466531163,
      "loss": 11.5,
      "step": 623
    },
    {
      "epoch": 0.09988594753586649,
      "grad_norm": 0.006239730399101973,
      "learning_rate": 0.00012727085201470973,
      "loss": 11.5,
      "step": 624
    },
    {
      "epoch": 0.10004602116973808,
      "grad_norm": 0.005400476977229118,
      "learning_rate": 0.00012706793812973941,
      "loss": 11.5,
      "step": 625
    },
    {
      "epoch": 0.10020609480360966,
      "grad_norm": 0.0026480902452021837,
      "learning_rate": 0.0001268649039124677,
      "loss": 11.5,
      "step": 626
    },
    {
      "epoch": 0.10036616843748124,
      "grad_norm": 0.003939393442124128,
      "learning_rate": 0.00012666175026549662,
      "loss": 11.5,
      "step": 627
    },
    {
      "epoch": 0.10052624207135283,
      "grad_norm": 0.006197950802743435,
      "learning_rate": 0.000126458478091959,
      "loss": 11.5,
      "step": 628
    },
    {
      "epoch": 0.1006863157052244,
      "grad_norm": 0.002940737409517169,
      "learning_rate": 0.00012625508829551473,
      "loss": 11.5,
      "step": 629
    },
    {
      "epoch": 0.10084638933909598,
      "grad_norm": 0.002679995261132717,
      "learning_rate": 0.00012605158178034654,
      "loss": 11.5,
      "step": 630
    },
    {
      "epoch": 0.10100646297296756,
      "grad_norm": 0.006134078837931156,
      "learning_rate": 0.00012584795945115603,
      "loss": 11.5,
      "step": 631
    },
    {
      "epoch": 0.10116653660683915,
      "grad_norm": 0.002539752284064889,
      "learning_rate": 0.0001256442222131597,
      "loss": 11.5,
      "step": 632
    },
    {
      "epoch": 0.10132661024071073,
      "grad_norm": 0.008050462231040001,
      "learning_rate": 0.0001254403709720848,
      "loss": 11.5,
      "step": 633
    },
    {
      "epoch": 0.1014866838745823,
      "grad_norm": 0.0033373672049492598,
      "learning_rate": 0.0001252364066341655,
      "loss": 11.5,
      "step": 634
    },
    {
      "epoch": 0.1016467575084539,
      "grad_norm": 0.00466674892231822,
      "learning_rate": 0.00012503233010613865,
      "loss": 11.5,
      "step": 635
    },
    {
      "epoch": 0.10180683114232547,
      "grad_norm": 0.0051351687870919704,
      "learning_rate": 0.00012482814229523997,
      "loss": 11.5,
      "step": 636
    },
    {
      "epoch": 0.10196690477619705,
      "grad_norm": 0.0030390277970582247,
      "learning_rate": 0.00012462384410919975,
      "loss": 11.5,
      "step": 637
    },
    {
      "epoch": 0.10212697841006863,
      "grad_norm": 0.001592720509506762,
      "learning_rate": 0.00012441943645623903,
      "loss": 11.5,
      "step": 638
    },
    {
      "epoch": 0.10228705204394022,
      "grad_norm": 0.002561771310865879,
      "learning_rate": 0.00012421492024506555,
      "loss": 11.5,
      "step": 639
    },
    {
      "epoch": 0.10244712567781179,
      "grad_norm": 0.0022247228771448135,
      "learning_rate": 0.00012401029638486953,
      "loss": 11.5,
      "step": 640
    },
    {
      "epoch": 0.10260719931168337,
      "grad_norm": 0.001579913543537259,
      "learning_rate": 0.0001238055657853198,
      "loss": 11.5,
      "step": 641
    },
    {
      "epoch": 0.10276727294555496,
      "grad_norm": 0.0026041765231639147,
      "learning_rate": 0.00012360072935655982,
      "loss": 11.5,
      "step": 642
    },
    {
      "epoch": 0.10292734657942654,
      "grad_norm": 0.028179192915558815,
      "learning_rate": 0.00012339578800920332,
      "loss": 11.5,
      "step": 643
    },
    {
      "epoch": 0.10308742021329811,
      "grad_norm": 0.04551348090171814,
      "learning_rate": 0.00012319074265433063,
      "loss": 11.5,
      "step": 644
    },
    {
      "epoch": 0.10324749384716969,
      "grad_norm": 0.006364596076309681,
      "learning_rate": 0.00012298559420348437,
      "loss": 11.5,
      "step": 645
    },
    {
      "epoch": 0.10340756748104128,
      "grad_norm": 0.006951974704861641,
      "learning_rate": 0.00012278034356866545,
      "loss": 11.5,
      "step": 646
    },
    {
      "epoch": 0.10356764111491286,
      "grad_norm": 0.009590779431164265,
      "learning_rate": 0.00012257499166232907,
      "loss": 11.5,
      "step": 647
    },
    {
      "epoch": 0.10372771474878444,
      "grad_norm": 0.0075108809396624565,
      "learning_rate": 0.0001223695393973807,
      "loss": 11.5,
      "step": 648
    },
    {
      "epoch": 0.10388778838265603,
      "grad_norm": 0.004725019447505474,
      "learning_rate": 0.0001221639876871719,
      "loss": 11.5,
      "step": 649
    },
    {
      "epoch": 0.1040478620165276,
      "grad_norm": 0.0038340131286531687,
      "learning_rate": 0.0001219583374454963,
      "loss": 11.5,
      "step": 650
    },
    {
      "epoch": 0.10420793565039918,
      "grad_norm": 0.002426853170618415,
      "learning_rate": 0.00012175258958658564,
      "loss": 11.5,
      "step": 651
    },
    {
      "epoch": 0.10436800928427077,
      "grad_norm": 0.0036109599750488997,
      "learning_rate": 0.00012154674502510555,
      "loss": 11.5,
      "step": 652
    },
    {
      "epoch": 0.10452808291814235,
      "grad_norm": 0.0047542257234454155,
      "learning_rate": 0.00012134080467615159,
      "loss": 11.5,
      "step": 653
    },
    {
      "epoch": 0.10468815655201393,
      "grad_norm": 0.00217737490311265,
      "learning_rate": 0.00012113476945524513,
      "loss": 11.5,
      "step": 654
    },
    {
      "epoch": 0.1048482301858855,
      "grad_norm": 0.0037848197389394045,
      "learning_rate": 0.00012092864027832933,
      "loss": 11.5,
      "step": 655
    },
    {
      "epoch": 0.10500830381975709,
      "grad_norm": 0.0030182679183781147,
      "learning_rate": 0.000120722418061765,
      "loss": 11.5,
      "step": 656
    },
    {
      "epoch": 0.10516837745362867,
      "grad_norm": 0.003835608484223485,
      "learning_rate": 0.0001205161037223266,
      "loss": 11.5,
      "step": 657
    },
    {
      "epoch": 0.10532845108750025,
      "grad_norm": 0.002789166523143649,
      "learning_rate": 0.00012030969817719808,
      "loss": 11.5,
      "step": 658
    },
    {
      "epoch": 0.10548852472137184,
      "grad_norm": 0.003341441275551915,
      "learning_rate": 0.00012010320234396894,
      "loss": 11.5,
      "step": 659
    },
    {
      "epoch": 0.10564859835524341,
      "grad_norm": 0.004561631940305233,
      "learning_rate": 0.00011989661714062999,
      "loss": 11.5,
      "step": 660
    },
    {
      "epoch": 0.10580867198911499,
      "grad_norm": 0.0027008343022316694,
      "learning_rate": 0.0001196899434855693,
      "loss": 11.5,
      "step": 661
    },
    {
      "epoch": 0.10596874562298657,
      "grad_norm": 0.002600240521132946,
      "learning_rate": 0.00011948318229756827,
      "loss": 11.5,
      "step": 662
    },
    {
      "epoch": 0.10612881925685816,
      "grad_norm": 0.0030214020516723394,
      "learning_rate": 0.00011927633449579735,
      "loss": 11.5,
      "step": 663
    },
    {
      "epoch": 0.10628889289072974,
      "grad_norm": 0.003407622454687953,
      "learning_rate": 0.0001190694009998121,
      "loss": 11.5,
      "step": 664
    },
    {
      "epoch": 0.10644896652460131,
      "grad_norm": 0.0038646699395030737,
      "learning_rate": 0.00011886238272954897,
      "loss": 11.5,
      "step": 665
    },
    {
      "epoch": 0.1066090401584729,
      "grad_norm": 0.004721539560705423,
      "learning_rate": 0.00011865528060532127,
      "loss": 11.5,
      "step": 666
    },
    {
      "epoch": 0.10676911379234448,
      "grad_norm": 0.002900291932746768,
      "learning_rate": 0.0001184480955478152,
      "loss": 11.5,
      "step": 667
    },
    {
      "epoch": 0.10692918742621606,
      "grad_norm": 0.00415404886007309,
      "learning_rate": 0.00011824082847808558,
      "loss": 11.5,
      "step": 668
    },
    {
      "epoch": 0.10708926106008763,
      "grad_norm": 0.00498941307887435,
      "learning_rate": 0.00011803348031755179,
      "loss": 11.5,
      "step": 669
    },
    {
      "epoch": 0.10724933469395923,
      "grad_norm": 0.0024129594676196575,
      "learning_rate": 0.0001178260519879937,
      "loss": 11.5,
      "step": 670
    },
    {
      "epoch": 0.1074094083278308,
      "grad_norm": 0.0036406132858246565,
      "learning_rate": 0.00011761854441154767,
      "loss": 11.5,
      "step": 671
    },
    {
      "epoch": 0.10756948196170238,
      "grad_norm": 0.0024914101231843233,
      "learning_rate": 0.00011741095851070228,
      "loss": 11.5,
      "step": 672
    },
    {
      "epoch": 0.10772955559557397,
      "grad_norm": 0.004080123733729124,
      "learning_rate": 0.00011720329520829429,
      "loss": 11.5,
      "step": 673
    },
    {
      "epoch": 0.10788962922944555,
      "grad_norm": 0.002997978124767542,
      "learning_rate": 0.0001169955554275046,
      "loss": 11.5,
      "step": 674
    },
    {
      "epoch": 0.10804970286331712,
      "grad_norm": 0.002375695388764143,
      "learning_rate": 0.0001167877400918541,
      "loss": 11.5,
      "step": 675
    },
    {
      "epoch": 0.1082097764971887,
      "grad_norm": 0.0030817838851362467,
      "learning_rate": 0.00011657985012519952,
      "loss": 11.5,
      "step": 676
    },
    {
      "epoch": 0.10836985013106029,
      "grad_norm": 0.002576148835942149,
      "learning_rate": 0.00011637188645172944,
      "loss": 11.5,
      "step": 677
    },
    {
      "epoch": 0.10852992376493187,
      "grad_norm": 0.0029999027028679848,
      "learning_rate": 0.00011616384999596006,
      "loss": 11.5,
      "step": 678
    },
    {
      "epoch": 0.10868999739880345,
      "grad_norm": 0.00403354037553072,
      "learning_rate": 0.00011595574168273111,
      "loss": 11.5,
      "step": 679
    },
    {
      "epoch": 0.10885007103267504,
      "grad_norm": 0.0036163197364658117,
      "learning_rate": 0.0001157475624372018,
      "loss": 11.5,
      "step": 680
    },
    {
      "epoch": 0.10901014466654661,
      "grad_norm": 0.002601220738142729,
      "learning_rate": 0.0001155393131848467,
      "loss": 11.5,
      "step": 681
    },
    {
      "epoch": 0.10917021830041819,
      "grad_norm": 0.003596701193600893,
      "learning_rate": 0.00011533099485145155,
      "loss": 11.5,
      "step": 682
    },
    {
      "epoch": 0.10933029193428977,
      "grad_norm": 0.004421269055455923,
      "learning_rate": 0.00011512260836310924,
      "loss": 11.5,
      "step": 683
    },
    {
      "epoch": 0.10949036556816136,
      "grad_norm": 0.0034209107980132103,
      "learning_rate": 0.00011491415464621562,
      "loss": 11.5,
      "step": 684
    },
    {
      "epoch": 0.10965043920203293,
      "grad_norm": 0.002435034839436412,
      "learning_rate": 0.00011470563462746541,
      "loss": 11.5,
      "step": 685
    },
    {
      "epoch": 0.10981051283590451,
      "grad_norm": 0.0034538828767836094,
      "learning_rate": 0.00011449704923384812,
      "loss": 11.5,
      "step": 686
    },
    {
      "epoch": 0.1099705864697761,
      "grad_norm": 0.0033488785848021507,
      "learning_rate": 0.00011428839939264382,
      "loss": 11.5,
      "step": 687
    },
    {
      "epoch": 0.11013066010364768,
      "grad_norm": 0.0019896412268280983,
      "learning_rate": 0.0001140796860314191,
      "loss": 11.5,
      "step": 688
    },
    {
      "epoch": 0.11029073373751926,
      "grad_norm": 0.0023177447728812695,
      "learning_rate": 0.00011387091007802297,
      "loss": 11.5,
      "step": 689
    },
    {
      "epoch": 0.11045080737139085,
      "grad_norm": 0.001368948258459568,
      "learning_rate": 0.0001136620724605827,
      "loss": 11.5,
      "step": 690
    },
    {
      "epoch": 0.11061088100526242,
      "grad_norm": 0.001555595314130187,
      "learning_rate": 0.00011345317410749964,
      "loss": 11.5,
      "step": 691
    },
    {
      "epoch": 0.110770954639134,
      "grad_norm": 0.015506098978221416,
      "learning_rate": 0.00011324421594744516,
      "loss": 11.5,
      "step": 692
    },
    {
      "epoch": 0.11093102827300558,
      "grad_norm": 0.03277930989861488,
      "learning_rate": 0.00011303519890935656,
      "loss": 11.5,
      "step": 693
    },
    {
      "epoch": 0.11109110190687717,
      "grad_norm": 0.009618213400244713,
      "learning_rate": 0.00011282612392243286,
      "loss": 11.5,
      "step": 694
    },
    {
      "epoch": 0.11125117554074875,
      "grad_norm": 0.005067052319645882,
      "learning_rate": 0.00011261699191613066,
      "loss": 11.5,
      "step": 695
    },
    {
      "epoch": 0.11141124917462032,
      "grad_norm": 0.008352566510438919,
      "learning_rate": 0.00011240780382016005,
      "loss": 11.5,
      "step": 696
    },
    {
      "epoch": 0.11157132280849191,
      "grad_norm": 0.007350957486778498,
      "learning_rate": 0.00011219856056448051,
      "loss": 11.5,
      "step": 697
    },
    {
      "epoch": 0.11173139644236349,
      "grad_norm": 0.0046548848040401936,
      "learning_rate": 0.00011198926307929664,
      "loss": 11.5,
      "step": 698
    },
    {
      "epoch": 0.11189147007623507,
      "grad_norm": 0.005582796409726143,
      "learning_rate": 0.00011177991229505431,
      "loss": 11.5,
      "step": 699
    },
    {
      "epoch": 0.11205154371010664,
      "grad_norm": 0.0065402076579630375,
      "learning_rate": 0.00011157050914243614,
      "loss": 11.5,
      "step": 700
    },
    {
      "epoch": 0.11221161734397823,
      "grad_norm": 0.005135664716362953,
      "learning_rate": 0.00011136105455235766,
      "loss": 11.5,
      "step": 701
    },
    {
      "epoch": 0.11237169097784981,
      "grad_norm": 0.0032237451523542404,
      "learning_rate": 0.00011115154945596305,
      "loss": 11.5,
      "step": 702
    },
    {
      "epoch": 0.11253176461172139,
      "grad_norm": 0.0020533131901174784,
      "learning_rate": 0.00011094199478462095,
      "loss": 11.5,
      "step": 703
    },
    {
      "epoch": 0.11269183824559298,
      "grad_norm": 0.00300579727627337,
      "learning_rate": 0.00011073239146992054,
      "loss": 11.5,
      "step": 704
    },
    {
      "epoch": 0.11285191187946456,
      "grad_norm": 0.004169172141700983,
      "learning_rate": 0.00011052274044366711,
      "loss": 11.5,
      "step": 705
    },
    {
      "epoch": 0.11301198551333613,
      "grad_norm": 0.0035526338033378124,
      "learning_rate": 0.00011031304263787812,
      "loss": 11.5,
      "step": 706
    },
    {
      "epoch": 0.11317205914720771,
      "grad_norm": 0.0021802200935781,
      "learning_rate": 0.00011010329898477891,
      "loss": 11.5,
      "step": 707
    },
    {
      "epoch": 0.1133321327810793,
      "grad_norm": 0.004188310820609331,
      "learning_rate": 0.0001098935104167988,
      "loss": 11.5,
      "step": 708
    },
    {
      "epoch": 0.11349220641495088,
      "grad_norm": 0.003933247644454241,
      "learning_rate": 0.00010968367786656663,
      "loss": 11.5,
      "step": 709
    },
    {
      "epoch": 0.11365228004882245,
      "grad_norm": 0.0032794740982353687,
      "learning_rate": 0.00010947380226690684,
      "loss": 11.5,
      "step": 710
    },
    {
      "epoch": 0.11381235368269405,
      "grad_norm": 0.008862190879881382,
      "learning_rate": 0.00010926388455083522,
      "loss": 11.5,
      "step": 711
    },
    {
      "epoch": 0.11397242731656562,
      "grad_norm": 0.0039347694255411625,
      "learning_rate": 0.00010905392565155477,
      "loss": 11.5,
      "step": 712
    },
    {
      "epoch": 0.1141325009504372,
      "grad_norm": 0.002004723297432065,
      "learning_rate": 0.00010884392650245165,
      "loss": 11.5,
      "step": 713
    },
    {
      "epoch": 0.11429257458430878,
      "grad_norm": 0.0030048920307308435,
      "learning_rate": 0.00010863388803709089,
      "loss": 11.5,
      "step": 714
    },
    {
      "epoch": 0.11445264821818037,
      "grad_norm": 0.0025611340533941984,
      "learning_rate": 0.00010842381118921232,
      "loss": 11.5,
      "step": 715
    },
    {
      "epoch": 0.11461272185205194,
      "grad_norm": 0.0017067677108570933,
      "learning_rate": 0.00010821369689272638,
      "loss": 11.5,
      "step": 716
    },
    {
      "epoch": 0.11477279548592352,
      "grad_norm": 0.0028548913542181253,
      "learning_rate": 0.00010800354608171003,
      "loss": 11.5,
      "step": 717
    },
    {
      "epoch": 0.11493286911979511,
      "grad_norm": 0.006282150745391846,
      "learning_rate": 0.00010779335969040252,
      "loss": 11.5,
      "step": 718
    },
    {
      "epoch": 0.11509294275366669,
      "grad_norm": 0.005091913044452667,
      "learning_rate": 0.00010758313865320134,
      "loss": 11.5,
      "step": 719
    },
    {
      "epoch": 0.11525301638753827,
      "grad_norm": 0.003715391969308257,
      "learning_rate": 0.00010737288390465792,
      "loss": 11.5,
      "step": 720
    },
    {
      "epoch": 0.11541309002140984,
      "grad_norm": 0.002450329950079322,
      "learning_rate": 0.00010716259637947357,
      "loss": 11.5,
      "step": 721
    },
    {
      "epoch": 0.11557316365528143,
      "grad_norm": 0.005074887536466122,
      "learning_rate": 0.00010695227701249537,
      "loss": 11.5,
      "step": 722
    },
    {
      "epoch": 0.11573323728915301,
      "grad_norm": 0.0037700904067605734,
      "learning_rate": 0.00010674192673871191,
      "loss": 11.5,
      "step": 723
    },
    {
      "epoch": 0.11589331092302459,
      "grad_norm": 0.006124040577560663,
      "learning_rate": 0.00010653154649324917,
      "loss": 11.5,
      "step": 724
    },
    {
      "epoch": 0.11605338455689618,
      "grad_norm": 0.0023981244303286076,
      "learning_rate": 0.00010632113721136636,
      "loss": 11.5,
      "step": 725
    },
    {
      "epoch": 0.11621345819076775,
      "grad_norm": 0.0017800109926611185,
      "learning_rate": 0.00010611069982845183,
      "loss": 11.5,
      "step": 726
    },
    {
      "epoch": 0.11637353182463933,
      "grad_norm": 0.0036051615606993437,
      "learning_rate": 0.00010590023528001884,
      "loss": 11.5,
      "step": 727
    },
    {
      "epoch": 0.11653360545851091,
      "grad_norm": 0.0050517236813902855,
      "learning_rate": 0.00010568974450170139,
      "loss": 11.5,
      "step": 728
    },
    {
      "epoch": 0.1166936790923825,
      "grad_norm": 0.0017460257513448596,
      "learning_rate": 0.00010547922842925008,
      "loss": 11.5,
      "step": 729
    },
    {
      "epoch": 0.11685375272625408,
      "grad_norm": 0.0035009782295674086,
      "learning_rate": 0.00010526868799852796,
      "loss": 11.5,
      "step": 730
    },
    {
      "epoch": 0.11701382636012565,
      "grad_norm": 0.0031001076567918062,
      "learning_rate": 0.0001050581241455064,
      "loss": 11.5,
      "step": 731
    },
    {
      "epoch": 0.11717389999399724,
      "grad_norm": 0.0017861028900370002,
      "learning_rate": 0.00010484753780626089,
      "loss": 11.5,
      "step": 732
    },
    {
      "epoch": 0.11733397362786882,
      "grad_norm": 0.004153000190854073,
      "learning_rate": 0.00010463692991696685,
      "loss": 11.5,
      "step": 733
    },
    {
      "epoch": 0.1174940472617404,
      "grad_norm": 0.0015630534617230296,
      "learning_rate": 0.00010442630141389549,
      "loss": 11.5,
      "step": 734
    },
    {
      "epoch": 0.11765412089561199,
      "grad_norm": 0.0030233291909098625,
      "learning_rate": 0.00010421565323340971,
      "loss": 11.5,
      "step": 735
    },
    {
      "epoch": 0.11781419452948357,
      "grad_norm": 0.0026963124983012676,
      "learning_rate": 0.00010400498631195992,
      "loss": 11.5,
      "step": 736
    },
    {
      "epoch": 0.11797426816335514,
      "grad_norm": 0.002723485231399536,
      "learning_rate": 0.00010379430158607975,
      "loss": 11.5,
      "step": 737
    },
    {
      "epoch": 0.11813434179722672,
      "grad_norm": 0.002175193512812257,
      "learning_rate": 0.000103583599992382,
      "loss": 11.5,
      "step": 738
    },
    {
      "epoch": 0.11829441543109831,
      "grad_norm": 0.00298920925706625,
      "learning_rate": 0.0001033728824675545,
      "loss": 11.5,
      "step": 739
    },
    {
      "epoch": 0.11845448906496989,
      "grad_norm": 0.003432897152379155,
      "learning_rate": 0.0001031621499483559,
      "loss": 11.5,
      "step": 740
    },
    {
      "epoch": 0.11861456269884146,
      "grad_norm": 0.0027004864532500505,
      "learning_rate": 0.00010295140337161146,
      "loss": 11.5,
      "step": 741
    },
    {
      "epoch": 0.11877463633271305,
      "grad_norm": 0.002965206978842616,
      "learning_rate": 0.00010274064367420897,
      "loss": 11.5,
      "step": 742
    },
    {
      "epoch": 0.11893470996658463,
      "grad_norm": 0.033101730048656464,
      "learning_rate": 0.00010252987179309459,
      "loss": 11.5,
      "step": 743
    },
    {
      "epoch": 0.11909478360045621,
      "grad_norm": 0.03653303533792496,
      "learning_rate": 0.00010231908866526851,
      "loss": 11.5,
      "step": 744
    },
    {
      "epoch": 0.11925485723432779,
      "grad_norm": 0.009369375184178352,
      "learning_rate": 0.00010210829522778111,
      "loss": 11.5,
      "step": 745
    },
    {
      "epoch": 0.11941493086819938,
      "grad_norm": 0.007763825356960297,
      "learning_rate": 0.00010189749241772844,
      "loss": 11.5,
      "step": 746
    },
    {
      "epoch": 0.11957500450207095,
      "grad_norm": 0.007243392523378134,
      "learning_rate": 0.00010168668117224825,
      "loss": 11.5,
      "step": 747
    },
    {
      "epoch": 0.11973507813594253,
      "grad_norm": 0.008052904158830643,
      "learning_rate": 0.00010147586242851585,
      "loss": 11.5,
      "step": 748
    },
    {
      "epoch": 0.11989515176981412,
      "grad_norm": 0.006970830261707306,
      "learning_rate": 0.00010126503712373982,
      "loss": 11.5,
      "step": 749
    },
    {
      "epoch": 0.1200552254036857,
      "grad_norm": 0.007045971695333719,
      "learning_rate": 0.00010105420619515798,
      "loss": 11.5,
      "step": 750
    },
    {
      "epoch": 0.1200552254036857,
      "eval_loss": 11.5,
      "eval_runtime": 59.3593,
      "eval_samples_per_second": 177.259,
      "eval_steps_per_second": 88.63,
      "step": 750
    },
    {
      "epoch": 0.12021529903755727,
      "grad_norm": 0.004605344031006098,
      "learning_rate": 0.00010084337058003303,
      "loss": 11.5,
      "step": 751
    },
    {
      "epoch": 0.12037537267142885,
      "grad_norm": 0.00208938866853714,
      "learning_rate": 0.00010063253121564868,
      "loss": 11.5,
      "step": 752
    },
    {
      "epoch": 0.12053544630530044,
      "grad_norm": 0.003603358054533601,
      "learning_rate": 0.00010042168903930514,
      "loss": 11.5,
      "step": 753
    },
    {
      "epoch": 0.12069551993917202,
      "grad_norm": 0.003973204176872969,
      "learning_rate": 0.00010021084498831522,
      "loss": 11.5,
      "step": 754
    },
    {
      "epoch": 0.1208555935730436,
      "grad_norm": 0.0031278200913220644,
      "learning_rate": 0.0001,
      "loss": 11.5,
      "step": 755
    },
    {
      "epoch": 0.12101566720691519,
      "grad_norm": 0.0022925559896975756,
      "learning_rate": 9.97891550116848e-05,
      "loss": 11.5,
      "step": 756
    },
    {
      "epoch": 0.12117574084078676,
      "grad_norm": 0.005406473763287067,
      "learning_rate": 9.957831096069488e-05,
      "loss": 11.5,
      "step": 757
    },
    {
      "epoch": 0.12133581447465834,
      "grad_norm": 0.00347351748496294,
      "learning_rate": 9.936746878435136e-05,
      "loss": 11.5,
      "step": 758
    },
    {
      "epoch": 0.12149588810852992,
      "grad_norm": 0.004065958317369223,
      "learning_rate": 9.915662941996699e-05,
      "loss": 11.5,
      "step": 759
    },
    {
      "epoch": 0.12165596174240151,
      "grad_norm": 0.003199318889528513,
      "learning_rate": 9.894579380484204e-05,
      "loss": 11.5,
      "step": 760
    },
    {
      "epoch": 0.12181603537627309,
      "grad_norm": 0.00647583557292819,
      "learning_rate": 9.873496287626019e-05,
      "loss": 11.5,
      "step": 761
    },
    {
      "epoch": 0.12197610901014466,
      "grad_norm": 0.004823511932045221,
      "learning_rate": 9.852413757148417e-05,
      "loss": 11.5,
      "step": 762
    },
    {
      "epoch": 0.12213618264401625,
      "grad_norm": 0.007430774159729481,
      "learning_rate": 9.831331882775178e-05,
      "loss": 11.5,
      "step": 763
    },
    {
      "epoch": 0.12229625627788783,
      "grad_norm": 0.004258694592863321,
      "learning_rate": 9.81025075822716e-05,
      "loss": 11.5,
      "step": 764
    },
    {
      "epoch": 0.1224563299117594,
      "grad_norm": 0.004434685222804546,
      "learning_rate": 9.789170477221891e-05,
      "loss": 11.5,
      "step": 765
    },
    {
      "epoch": 0.12261640354563098,
      "grad_norm": 0.005173634737730026,
      "learning_rate": 9.76809113347315e-05,
      "loss": 11.5,
      "step": 766
    },
    {
      "epoch": 0.12277647717950257,
      "grad_norm": 0.002468820894137025,
      "learning_rate": 9.747012820690543e-05,
      "loss": 11.5,
      "step": 767
    },
    {
      "epoch": 0.12293655081337415,
      "grad_norm": 0.0034938561730086803,
      "learning_rate": 9.725935632579104e-05,
      "loss": 11.5,
      "step": 768
    },
    {
      "epoch": 0.12309662444724573,
      "grad_norm": 0.002860844600945711,
      "learning_rate": 9.704859662838855e-05,
      "loss": 11.5,
      "step": 769
    },
    {
      "epoch": 0.12325669808111732,
      "grad_norm": 0.004840923007577658,
      "learning_rate": 9.683785005164411e-05,
      "loss": 11.5,
      "step": 770
    },
    {
      "epoch": 0.1234167717149889,
      "grad_norm": 0.003707458032295108,
      "learning_rate": 9.662711753244551e-05,
      "loss": 11.5,
      "step": 771
    },
    {
      "epoch": 0.12357684534886047,
      "grad_norm": 0.0015729664592072368,
      "learning_rate": 9.641640000761802e-05,
      "loss": 11.5,
      "step": 772
    },
    {
      "epoch": 0.12373691898273205,
      "grad_norm": 0.0019833880942314863,
      "learning_rate": 9.620569841392029e-05,
      "loss": 11.5,
      "step": 773
    },
    {
      "epoch": 0.12389699261660364,
      "grad_norm": 0.0022297173272818327,
      "learning_rate": 9.59950136880401e-05,
      "loss": 11.5,
      "step": 774
    },
    {
      "epoch": 0.12405706625047522,
      "grad_norm": 0.002988791326060891,
      "learning_rate": 9.57843467665903e-05,
      "loss": 11.5,
      "step": 775
    },
    {
      "epoch": 0.1242171398843468,
      "grad_norm": 0.0060118031688034534,
      "learning_rate": 9.557369858610453e-05,
      "loss": 11.5,
      "step": 776
    },
    {
      "epoch": 0.12437721351821839,
      "grad_norm": 0.005681234411895275,
      "learning_rate": 9.53630700830332e-05,
      "loss": 11.5,
      "step": 777
    },
    {
      "epoch": 0.12453728715208996,
      "grad_norm": 0.005485843401402235,
      "learning_rate": 9.51524621937391e-05,
      "loss": 11.5,
      "step": 778
    },
    {
      "epoch": 0.12469736078596154,
      "grad_norm": 0.002206267789006233,
      "learning_rate": 9.494187585449358e-05,
      "loss": 11.5,
      "step": 779
    },
    {
      "epoch": 0.12485743441983313,
      "grad_norm": 0.0023218491114676,
      "learning_rate": 9.473131200147205e-05,
      "loss": 11.5,
      "step": 780
    },
    {
      "epoch": 0.1250175080537047,
      "grad_norm": 0.001893517910502851,
      "learning_rate": 9.452077157074994e-05,
      "loss": 11.5,
      "step": 781
    },
    {
      "epoch": 0.12517758168757628,
      "grad_norm": 0.002466535195708275,
      "learning_rate": 9.431025549829862e-05,
      "loss": 11.5,
      "step": 782
    },
    {
      "epoch": 0.12533765532144786,
      "grad_norm": 0.00466206343844533,
      "learning_rate": 9.409976471998118e-05,
      "loss": 11.5,
      "step": 783
    },
    {
      "epoch": 0.12549772895531944,
      "grad_norm": 0.0027048313058912754,
      "learning_rate": 9.388930017154819e-05,
      "loss": 11.5,
      "step": 784
    },
    {
      "epoch": 0.12565780258919101,
      "grad_norm": 0.0014575169188901782,
      "learning_rate": 9.367886278863366e-05,
      "loss": 11.5,
      "step": 785
    },
    {
      "epoch": 0.12581787622306262,
      "grad_norm": 0.003571330104023218,
      "learning_rate": 9.346845350675088e-05,
      "loss": 11.5,
      "step": 786
    },
    {
      "epoch": 0.1259779498569342,
      "grad_norm": 0.0030264845117926598,
      "learning_rate": 9.325807326128814e-05,
      "loss": 11.5,
      "step": 787
    },
    {
      "epoch": 0.12613802349080577,
      "grad_norm": 0.002837283769622445,
      "learning_rate": 9.304772298750463e-05,
      "loss": 11.5,
      "step": 788
    },
    {
      "epoch": 0.12629809712467735,
      "grad_norm": 0.0029044628608971834,
      "learning_rate": 9.283740362052642e-05,
      "loss": 11.5,
      "step": 789
    },
    {
      "epoch": 0.12645817075854893,
      "grad_norm": 0.003517976263538003,
      "learning_rate": 9.26271160953421e-05,
      "loss": 11.5,
      "step": 790
    },
    {
      "epoch": 0.1266182443924205,
      "grad_norm": 0.00347169884480536,
      "learning_rate": 9.241686134679867e-05,
      "loss": 11.5,
      "step": 791
    },
    {
      "epoch": 0.12677831802629208,
      "grad_norm": 0.0036738195922225714,
      "learning_rate": 9.220664030959749e-05,
      "loss": 11.5,
      "step": 792
    },
    {
      "epoch": 0.12693839166016369,
      "grad_norm": 0.004519483540207148,
      "learning_rate": 9.199645391828999e-05,
      "loss": 11.5,
      "step": 793
    },
    {
      "epoch": 0.12709846529403526,
      "grad_norm": 0.013343493454158306,
      "learning_rate": 9.178630310727365e-05,
      "loss": 11.5,
      "step": 794
    },
    {
      "epoch": 0.12725853892790684,
      "grad_norm": 0.004912480246275663,
      "learning_rate": 9.157618881078772e-05,
      "loss": 11.5,
      "step": 795
    },
    {
      "epoch": 0.12741861256177842,
      "grad_norm": 0.0050626154989004135,
      "learning_rate": 9.136611196290915e-05,
      "loss": 11.5,
      "step": 796
    },
    {
      "epoch": 0.12757868619565,
      "grad_norm": 0.004409583285450935,
      "learning_rate": 9.115607349754834e-05,
      "loss": 11.5,
      "step": 797
    },
    {
      "epoch": 0.12773875982952157,
      "grad_norm": 0.00467911409214139,
      "learning_rate": 9.094607434844523e-05,
      "loss": 11.5,
      "step": 798
    },
    {
      "epoch": 0.12789883346339317,
      "grad_norm": 0.005608738400042057,
      "learning_rate": 9.07361154491648e-05,
      "loss": 11.5,
      "step": 799
    },
    {
      "epoch": 0.12805890709726475,
      "grad_norm": 0.008519706316292286,
      "learning_rate": 9.052619773309317e-05,
      "loss": 11.5,
      "step": 800
    },
    {
      "epoch": 0.12821898073113633,
      "grad_norm": 0.00285702059045434,
      "learning_rate": 9.031632213343339e-05,
      "loss": 11.5,
      "step": 801
    },
    {
      "epoch": 0.1283790543650079,
      "grad_norm": 0.005185387562960386,
      "learning_rate": 9.01064895832012e-05,
      "loss": 11.5,
      "step": 802
    },
    {
      "epoch": 0.12853912799887948,
      "grad_norm": 0.0026589753106236458,
      "learning_rate": 8.98967010152211e-05,
      "loss": 11.5,
      "step": 803
    },
    {
      "epoch": 0.12869920163275106,
      "grad_norm": 0.002747615799307823,
      "learning_rate": 8.968695736212193e-05,
      "loss": 11.5,
      "step": 804
    },
    {
      "epoch": 0.12885927526662264,
      "grad_norm": 0.006737317889928818,
      "learning_rate": 8.947725955633294e-05,
      "loss": 11.5,
      "step": 805
    },
    {
      "epoch": 0.12901934890049424,
      "grad_norm": 0.0025012276601046324,
      "learning_rate": 8.926760853007946e-05,
      "loss": 11.5,
      "step": 806
    },
    {
      "epoch": 0.12917942253436582,
      "grad_norm": 0.0026731386315077543,
      "learning_rate": 8.905800521537905e-05,
      "loss": 11.5,
      "step": 807
    },
    {
      "epoch": 0.1293394961682374,
      "grad_norm": 0.002023561391979456,
      "learning_rate": 8.884845054403699e-05,
      "loss": 11.5,
      "step": 808
    },
    {
      "epoch": 0.12949956980210897,
      "grad_norm": 0.0024797311052680016,
      "learning_rate": 8.863894544764236e-05,
      "loss": 11.5,
      "step": 809
    },
    {
      "epoch": 0.12965964343598055,
      "grad_norm": 0.0026280167512595654,
      "learning_rate": 8.84294908575639e-05,
      "loss": 11.5,
      "step": 810
    },
    {
      "epoch": 0.12981971706985213,
      "grad_norm": 0.0032640115823596716,
      "learning_rate": 8.822008770494572e-05,
      "loss": 11.5,
      "step": 811
    },
    {
      "epoch": 0.1299797907037237,
      "grad_norm": 0.0033080619759857655,
      "learning_rate": 8.801073692070337e-05,
      "loss": 11.5,
      "step": 812
    },
    {
      "epoch": 0.1301398643375953,
      "grad_norm": 0.002956042531877756,
      "learning_rate": 8.780143943551954e-05,
      "loss": 11.5,
      "step": 813
    },
    {
      "epoch": 0.13029993797146688,
      "grad_norm": 0.0029434009920805693,
      "learning_rate": 8.759219617983999e-05,
      "loss": 11.5,
      "step": 814
    },
    {
      "epoch": 0.13046001160533846,
      "grad_norm": 0.002761637791991234,
      "learning_rate": 8.738300808386935e-05,
      "loss": 11.5,
      "step": 815
    },
    {
      "epoch": 0.13062008523921004,
      "grad_norm": 0.0042599160224199295,
      "learning_rate": 8.717387607756713e-05,
      "loss": 11.5,
      "step": 816
    },
    {
      "epoch": 0.13078015887308161,
      "grad_norm": 0.001450824085623026,
      "learning_rate": 8.696480109064342e-05,
      "loss": 11.5,
      "step": 817
    },
    {
      "epoch": 0.1309402325069532,
      "grad_norm": 0.004807382822036743,
      "learning_rate": 8.675578405255485e-05,
      "loss": 11.5,
      "step": 818
    },
    {
      "epoch": 0.13110030614082477,
      "grad_norm": 0.0016322832088917494,
      "learning_rate": 8.654682589250038e-05,
      "loss": 11.5,
      "step": 819
    },
    {
      "epoch": 0.13126037977469637,
      "grad_norm": 0.003458446590229869,
      "learning_rate": 8.633792753941733e-05,
      "loss": 11.5,
      "step": 820
    },
    {
      "epoch": 0.13142045340856795,
      "grad_norm": 0.004010004922747612,
      "learning_rate": 8.612908992197705e-05,
      "loss": 11.5,
      "step": 821
    },
    {
      "epoch": 0.13158052704243953,
      "grad_norm": 0.0021207761019468307,
      "learning_rate": 8.592031396858093e-05,
      "loss": 11.5,
      "step": 822
    },
    {
      "epoch": 0.1317406006763111,
      "grad_norm": 0.0034215482883155346,
      "learning_rate": 8.571160060735624e-05,
      "loss": 11.5,
      "step": 823
    },
    {
      "epoch": 0.13190067431018268,
      "grad_norm": 0.004220360890030861,
      "learning_rate": 8.550295076615188e-05,
      "loss": 11.5,
      "step": 824
    },
    {
      "epoch": 0.13206074794405426,
      "grad_norm": 0.002873731777071953,
      "learning_rate": 8.529436537253458e-05,
      "loss": 11.5,
      "step": 825
    },
    {
      "epoch": 0.13222082157792583,
      "grad_norm": 0.002436636248603463,
      "learning_rate": 8.508584535378439e-05,
      "loss": 11.5,
      "step": 826
    },
    {
      "epoch": 0.13238089521179744,
      "grad_norm": 0.0023656017147004604,
      "learning_rate": 8.487739163689079e-05,
      "loss": 11.5,
      "step": 827
    },
    {
      "epoch": 0.13254096884566902,
      "grad_norm": 0.001973626669496298,
      "learning_rate": 8.466900514854847e-05,
      "loss": 11.5,
      "step": 828
    },
    {
      "epoch": 0.1327010424795406,
      "grad_norm": 0.008791493251919746,
      "learning_rate": 8.446068681515334e-05,
      "loss": 11.5,
      "step": 829
    },
    {
      "epoch": 0.13286111611341217,
      "grad_norm": 0.0036985352635383606,
      "learning_rate": 8.425243756279824e-05,
      "loss": 11.5,
      "step": 830
    },
    {
      "epoch": 0.13302118974728375,
      "grad_norm": 0.001968603115528822,
      "learning_rate": 8.404425831726894e-05,
      "loss": 11.5,
      "step": 831
    },
    {
      "epoch": 0.13318126338115532,
      "grad_norm": 0.004625111818313599,
      "learning_rate": 8.383615000404e-05,
      "loss": 11.5,
      "step": 832
    },
    {
      "epoch": 0.1333413370150269,
      "grad_norm": 0.003616010770201683,
      "learning_rate": 8.362811354827059e-05,
      "loss": 11.5,
      "step": 833
    },
    {
      "epoch": 0.1335014106488985,
      "grad_norm": 0.002287510549649596,
      "learning_rate": 8.342014987480047e-05,
      "loss": 11.5,
      "step": 834
    },
    {
      "epoch": 0.13366148428277008,
      "grad_norm": 0.0015214205486699939,
      "learning_rate": 8.321225990814591e-05,
      "loss": 11.5,
      "step": 835
    },
    {
      "epoch": 0.13382155791664166,
      "grad_norm": 0.003204460721462965,
      "learning_rate": 8.300444457249543e-05,
      "loss": 11.5,
      "step": 836
    },
    {
      "epoch": 0.13398163155051324,
      "grad_norm": 0.002396148629486561,
      "learning_rate": 8.279670479170573e-05,
      "loss": 11.5,
      "step": 837
    },
    {
      "epoch": 0.1341417051843848,
      "grad_norm": 0.002087327418848872,
      "learning_rate": 8.258904148929775e-05,
      "loss": 11.5,
      "step": 838
    },
    {
      "epoch": 0.1343017788182564,
      "grad_norm": 0.0035919982474297285,
      "learning_rate": 8.238145558845235e-05,
      "loss": 11.5,
      "step": 839
    },
    {
      "epoch": 0.13446185245212797,
      "grad_norm": 0.002836944768205285,
      "learning_rate": 8.217394801200631e-05,
      "loss": 11.5,
      "step": 840
    },
    {
      "epoch": 0.13462192608599957,
      "grad_norm": 0.0018963912734761834,
      "learning_rate": 8.196651968244826e-05,
      "loss": 11.5,
      "step": 841
    },
    {
      "epoch": 0.13478199971987115,
      "grad_norm": 0.002539979759603739,
      "learning_rate": 8.175917152191447e-05,
      "loss": 11.5,
      "step": 842
    },
    {
      "epoch": 0.13494207335374273,
      "grad_norm": 0.006271395366638899,
      "learning_rate": 8.15519044521848e-05,
      "loss": 11.5,
      "step": 843
    },
    {
      "epoch": 0.1351021469876143,
      "grad_norm": 0.029586156830191612,
      "learning_rate": 8.134471939467874e-05,
      "loss": 11.5,
      "step": 844
    },
    {
      "epoch": 0.13526222062148588,
      "grad_norm": 0.012032132595777512,
      "learning_rate": 8.113761727045105e-05,
      "loss": 11.5,
      "step": 845
    },
    {
      "epoch": 0.13542229425535746,
      "grad_norm": 0.005794750992208719,
      "learning_rate": 8.093059900018792e-05,
      "loss": 11.5,
      "step": 846
    },
    {
      "epoch": 0.13558236788922903,
      "grad_norm": 0.0057182288728654385,
      "learning_rate": 8.072366550420266e-05,
      "loss": 11.5,
      "step": 847
    },
    {
      "epoch": 0.13574244152310064,
      "grad_norm": 0.0050259060226380825,
      "learning_rate": 8.051681770243175e-05,
      "loss": 11.5,
      "step": 848
    },
    {
      "epoch": 0.13590251515697221,
      "grad_norm": 0.007448924705386162,
      "learning_rate": 8.031005651443073e-05,
      "loss": 11.5,
      "step": 849
    },
    {
      "epoch": 0.1360625887908438,
      "grad_norm": 0.0068702357821166515,
      "learning_rate": 8.010338285937006e-05,
      "loss": 11.5,
      "step": 850
    },
    {
      "epoch": 0.13622266242471537,
      "grad_norm": 0.003492058254778385,
      "learning_rate": 7.989679765603108e-05,
      "loss": 11.5,
      "step": 851
    },
    {
      "epoch": 0.13638273605858695,
      "grad_norm": 0.004327989183366299,
      "learning_rate": 7.969030182280192e-05,
      "loss": 11.5,
      "step": 852
    },
    {
      "epoch": 0.13654280969245852,
      "grad_norm": 0.003075564280152321,
      "learning_rate": 7.948389627767343e-05,
      "loss": 11.5,
      "step": 853
    },
    {
      "epoch": 0.1367028833263301,
      "grad_norm": 0.004318468272686005,
      "learning_rate": 7.927758193823501e-05,
      "loss": 11.5,
      "step": 854
    },
    {
      "epoch": 0.1368629569602017,
      "grad_norm": 0.002096753567457199,
      "learning_rate": 7.907135972167069e-05,
      "loss": 11.5,
      "step": 855
    },
    {
      "epoch": 0.13702303059407328,
      "grad_norm": 0.0069441478699445724,
      "learning_rate": 7.88652305447549e-05,
      "loss": 11.5,
      "step": 856
    },
    {
      "epoch": 0.13718310422794486,
      "grad_norm": 0.001898195710964501,
      "learning_rate": 7.865919532384844e-05,
      "loss": 11.5,
      "step": 857
    },
    {
      "epoch": 0.13734317786181643,
      "grad_norm": 0.002723297569900751,
      "learning_rate": 7.845325497489449e-05,
      "loss": 11.5,
      "step": 858
    },
    {
      "epoch": 0.137503251495688,
      "grad_norm": 0.0023458232171833515,
      "learning_rate": 7.82474104134144e-05,
      "loss": 11.5,
      "step": 859
    },
    {
      "epoch": 0.1376633251295596,
      "grad_norm": 0.002770025981590152,
      "learning_rate": 7.804166255450373e-05,
      "loss": 11.5,
      "step": 860
    },
    {
      "epoch": 0.13782339876343117,
      "grad_norm": 0.005793519783765078,
      "learning_rate": 7.783601231282812e-05,
      "loss": 11.5,
      "step": 861
    },
    {
      "epoch": 0.13798347239730277,
      "grad_norm": 0.004571247845888138,
      "learning_rate": 7.763046060261932e-05,
      "loss": 11.5,
      "step": 862
    },
    {
      "epoch": 0.13814354603117435,
      "grad_norm": 0.0017479181988164783,
      "learning_rate": 7.742500833767094e-05,
      "loss": 11.5,
      "step": 863
    },
    {
      "epoch": 0.13830361966504592,
      "grad_norm": 0.004760178737342358,
      "learning_rate": 7.721965643133458e-05,
      "loss": 11.5,
      "step": 864
    },
    {
      "epoch": 0.1384636932989175,
      "grad_norm": 0.0038296154234558344,
      "learning_rate": 7.701440579651566e-05,
      "loss": 11.5,
      "step": 865
    },
    {
      "epoch": 0.13862376693278908,
      "grad_norm": 0.0038842230569571257,
      "learning_rate": 7.680925734566937e-05,
      "loss": 11.5,
      "step": 866
    },
    {
      "epoch": 0.13878384056666065,
      "grad_norm": 0.0024726625997573137,
      "learning_rate": 7.660421199079669e-05,
      "loss": 11.5,
      "step": 867
    },
    {
      "epoch": 0.13894391420053223,
      "grad_norm": 0.003492773976176977,
      "learning_rate": 7.639927064344022e-05,
      "loss": 11.5,
      "step": 868
    },
    {
      "epoch": 0.13910398783440384,
      "grad_norm": 0.0036202871706336737,
      "learning_rate": 7.619443421468021e-05,
      "loss": 11.5,
      "step": 869
    },
    {
      "epoch": 0.1392640614682754,
      "grad_norm": 0.0031892103143036366,
      "learning_rate": 7.598970361513051e-05,
      "loss": 11.5,
      "step": 870
    },
    {
      "epoch": 0.139424135102147,
      "grad_norm": 0.0025079164188355207,
      "learning_rate": 7.578507975493448e-05,
      "loss": 11.5,
      "step": 871
    },
    {
      "epoch": 0.13958420873601857,
      "grad_norm": 0.0029436389449983835,
      "learning_rate": 7.558056354376098e-05,
      "loss": 11.5,
      "step": 872
    },
    {
      "epoch": 0.13974428236989014,
      "grad_norm": 0.0029052277095615864,
      "learning_rate": 7.537615589080027e-05,
      "loss": 11.5,
      "step": 873
    },
    {
      "epoch": 0.13990435600376172,
      "grad_norm": 0.0027314829640090466,
      "learning_rate": 7.517185770476006e-05,
      "loss": 11.5,
      "step": 874
    },
    {
      "epoch": 0.1400644296376333,
      "grad_norm": 0.002001818735152483,
      "learning_rate": 7.496766989386136e-05,
      "loss": 11.5,
      "step": 875
    },
    {
      "epoch": 0.1402245032715049,
      "grad_norm": 0.00688916863873601,
      "learning_rate": 7.476359336583454e-05,
      "loss": 11.5,
      "step": 876
    },
    {
      "epoch": 0.14038457690537648,
      "grad_norm": 0.00247250497341156,
      "learning_rate": 7.455962902791522e-05,
      "loss": 11.5,
      "step": 877
    },
    {
      "epoch": 0.14054465053924806,
      "grad_norm": 0.0036383874248713255,
      "learning_rate": 7.435577778684033e-05,
      "loss": 11.5,
      "step": 878
    },
    {
      "epoch": 0.14070472417311963,
      "grad_norm": 0.0035653701052069664,
      "learning_rate": 7.415204054884399e-05,
      "loss": 11.5,
      "step": 879
    },
    {
      "epoch": 0.1408647978069912,
      "grad_norm": 0.002242990303784609,
      "learning_rate": 7.394841821965345e-05,
      "loss": 11.5,
      "step": 880
    },
    {
      "epoch": 0.1410248714408628,
      "grad_norm": 0.00418773852288723,
      "learning_rate": 7.374491170448525e-05,
      "loss": 11.5,
      "step": 881
    },
    {
      "epoch": 0.1411849450747344,
      "grad_norm": 0.005222049541771412,
      "learning_rate": 7.3541521908041e-05,
      "loss": 11.5,
      "step": 882
    },
    {
      "epoch": 0.14134501870860597,
      "grad_norm": 0.003357511479407549,
      "learning_rate": 7.33382497345034e-05,
      "loss": 11.5,
      "step": 883
    },
    {
      "epoch": 0.14150509234247755,
      "grad_norm": 0.006907563656568527,
      "learning_rate": 7.313509608753231e-05,
      "loss": 11.5,
      "step": 884
    },
    {
      "epoch": 0.14166516597634912,
      "grad_norm": 0.0031600387301295996,
      "learning_rate": 7.293206187026061e-05,
      "loss": 11.5,
      "step": 885
    },
    {
      "epoch": 0.1418252396102207,
      "grad_norm": 0.001582453609444201,
      "learning_rate": 7.27291479852903e-05,
      "loss": 11.5,
      "step": 886
    },
    {
      "epoch": 0.14198531324409228,
      "grad_norm": 0.005144217517226934,
      "learning_rate": 7.252635533468843e-05,
      "loss": 11.5,
      "step": 887
    },
    {
      "epoch": 0.14214538687796385,
      "grad_norm": 0.0027015148662030697,
      "learning_rate": 7.232368481998309e-05,
      "loss": 11.5,
      "step": 888
    },
    {
      "epoch": 0.14230546051183546,
      "grad_norm": 0.0013171718455851078,
      "learning_rate": 7.212113734215932e-05,
      "loss": 11.5,
      "step": 889
    },
    {
      "epoch": 0.14246553414570703,
      "grad_norm": 0.003426140872761607,
      "learning_rate": 7.191871380165538e-05,
      "loss": 11.5,
      "step": 890
    },
    {
      "epoch": 0.1426256077795786,
      "grad_norm": 0.001977295847609639,
      "learning_rate": 7.17164150983584e-05,
      "loss": 11.5,
      "step": 891
    },
    {
      "epoch": 0.1427856814134502,
      "grad_norm": 0.002471641870215535,
      "learning_rate": 7.151424213160061e-05,
      "loss": 11.5,
      "step": 892
    },
    {
      "epoch": 0.14294575504732177,
      "grad_norm": 0.002837255597114563,
      "learning_rate": 7.131219580015521e-05,
      "loss": 11.5,
      "step": 893
    },
    {
      "epoch": 0.14310582868119334,
      "grad_norm": 0.021328654140233994,
      "learning_rate": 7.11102770022325e-05,
      "loss": 11.5,
      "step": 894
    },
    {
      "epoch": 0.14326590231506492,
      "grad_norm": 0.014724578708410263,
      "learning_rate": 7.090848663547574e-05,
      "loss": 11.5,
      "step": 895
    },
    {
      "epoch": 0.14342597594893652,
      "grad_norm": 0.006112853065133095,
      "learning_rate": 7.070682559695736e-05,
      "loss": 11.5,
      "step": 896
    },
    {
      "epoch": 0.1435860495828081,
      "grad_norm": 0.006259468849748373,
      "learning_rate": 7.050529478317476e-05,
      "loss": 11.5,
      "step": 897
    },
    {
      "epoch": 0.14374612321667968,
      "grad_norm": 0.008712248876690865,
      "learning_rate": 7.03038950900464e-05,
      "loss": 11.5,
      "step": 898
    },
    {
      "epoch": 0.14390619685055125,
      "grad_norm": 0.004674081690609455,
      "learning_rate": 7.010262741290798e-05,
      "loss": 11.5,
      "step": 899
    },
    {
      "epoch": 0.14406627048442283,
      "grad_norm": 0.006880821660161018,
      "learning_rate": 6.990149264650814e-05,
      "loss": 11.5,
      "step": 900
    },
    {
      "epoch": 0.1442263441182944,
      "grad_norm": 0.004148140083998442,
      "learning_rate": 6.970049168500474e-05,
      "loss": 11.5,
      "step": 901
    },
    {
      "epoch": 0.14438641775216599,
      "grad_norm": 0.004675343167036772,
      "learning_rate": 6.94996254219608e-05,
      "loss": 11.5,
      "step": 902
    },
    {
      "epoch": 0.1445464913860376,
      "grad_norm": 0.0016610127640888095,
      "learning_rate": 6.929889475034048e-05,
      "loss": 11.5,
      "step": 903
    },
    {
      "epoch": 0.14470656501990917,
      "grad_norm": 0.0041760834865272045,
      "learning_rate": 6.909830056250527e-05,
      "loss": 11.5,
      "step": 904
    },
    {
      "epoch": 0.14486663865378074,
      "grad_norm": 0.0026631078217178583,
      "learning_rate": 6.889784375020978e-05,
      "loss": 11.5,
      "step": 905
    },
    {
      "epoch": 0.14502671228765232,
      "grad_norm": 0.0032871628645807505,
      "learning_rate": 6.869752520459803e-05,
      "loss": 11.5,
      "step": 906
    },
    {
      "epoch": 0.1451867859215239,
      "grad_norm": 0.0025447129737585783,
      "learning_rate": 6.849734581619918e-05,
      "loss": 11.5,
      "step": 907
    },
    {
      "epoch": 0.14534685955539547,
      "grad_norm": 0.006851284299045801,
      "learning_rate": 6.829730647492404e-05,
      "loss": 11.5,
      "step": 908
    },
    {
      "epoch": 0.14550693318926705,
      "grad_norm": 0.0028427098877727985,
      "learning_rate": 6.80974080700606e-05,
      "loss": 11.5,
      "step": 909
    },
    {
      "epoch": 0.14566700682313866,
      "grad_norm": 0.002827555639669299,
      "learning_rate": 6.789765149027039e-05,
      "loss": 11.5,
      "step": 910
    },
    {
      "epoch": 0.14582708045701023,
      "grad_norm": 0.0024632092099636793,
      "learning_rate": 6.769803762358443e-05,
      "loss": 11.5,
      "step": 911
    },
    {
      "epoch": 0.1459871540908818,
      "grad_norm": 0.00859488733112812,
      "learning_rate": 6.749856735739928e-05,
      "loss": 11.5,
      "step": 912
    },
    {
      "epoch": 0.1461472277247534,
      "grad_norm": 0.0017382306978106499,
      "learning_rate": 6.729924157847323e-05,
      "loss": 11.5,
      "step": 913
    },
    {
      "epoch": 0.14630730135862496,
      "grad_norm": 0.0035761226899921894,
      "learning_rate": 6.710006117292209e-05,
      "loss": 11.5,
      "step": 914
    },
    {
      "epoch": 0.14646737499249654,
      "grad_norm": 0.006997543852776289,
      "learning_rate": 6.690102702621548e-05,
      "loss": 11.5,
      "step": 915
    },
    {
      "epoch": 0.14662744862636812,
      "grad_norm": 0.0035585190635174513,
      "learning_rate": 6.670214002317278e-05,
      "loss": 11.5,
      "step": 916
    },
    {
      "epoch": 0.14678752226023972,
      "grad_norm": 0.002104755025357008,
      "learning_rate": 6.650340104795932e-05,
      "loss": 11.5,
      "step": 917
    },
    {
      "epoch": 0.1469475958941113,
      "grad_norm": 0.004014565609395504,
      "learning_rate": 6.630481098408228e-05,
      "loss": 11.5,
      "step": 918
    },
    {
      "epoch": 0.14710766952798288,
      "grad_norm": 0.003054036758840084,
      "learning_rate": 6.610637071438686e-05,
      "loss": 11.5,
      "step": 919
    },
    {
      "epoch": 0.14726774316185445,
      "grad_norm": 0.0026505060959607363,
      "learning_rate": 6.590808112105232e-05,
      "loss": 11.5,
      "step": 920
    },
    {
      "epoch": 0.14742781679572603,
      "grad_norm": 0.0035651291254907846,
      "learning_rate": 6.570994308558812e-05,
      "loss": 11.5,
      "step": 921
    },
    {
      "epoch": 0.1475878904295976,
      "grad_norm": 0.0025029173120856285,
      "learning_rate": 6.551195748882997e-05,
      "loss": 11.5,
      "step": 922
    },
    {
      "epoch": 0.14774796406346918,
      "grad_norm": 0.0019678135868161917,
      "learning_rate": 6.531412521093586e-05,
      "loss": 11.5,
      "step": 923
    },
    {
      "epoch": 0.1479080376973408,
      "grad_norm": 0.0014836242189630866,
      "learning_rate": 6.51164471313822e-05,
      "loss": 11.5,
      "step": 924
    },
    {
      "epoch": 0.14806811133121237,
      "grad_norm": 0.0035512519534677267,
      "learning_rate": 6.491892412895995e-05,
      "loss": 11.5,
      "step": 925
    },
    {
      "epoch": 0.14822818496508394,
      "grad_norm": 0.004590109456330538,
      "learning_rate": 6.472155708177052e-05,
      "loss": 11.5,
      "step": 926
    },
    {
      "epoch": 0.14838825859895552,
      "grad_norm": 0.0026866630651056767,
      "learning_rate": 6.452434686722224e-05,
      "loss": 11.5,
      "step": 927
    },
    {
      "epoch": 0.1485483322328271,
      "grad_norm": 0.002269457560032606,
      "learning_rate": 6.432729436202604e-05,
      "loss": 11.5,
      "step": 928
    },
    {
      "epoch": 0.14870840586669867,
      "grad_norm": 0.007081727962940931,
      "learning_rate": 6.41304004421918e-05,
      "loss": 11.5,
      "step": 929
    },
    {
      "epoch": 0.14886847950057025,
      "grad_norm": 0.004162793979048729,
      "learning_rate": 6.393366598302446e-05,
      "loss": 11.5,
      "step": 930
    },
    {
      "epoch": 0.14902855313444185,
      "grad_norm": 0.00312308082357049,
      "learning_rate": 6.373709185911998e-05,
      "loss": 11.5,
      "step": 931
    },
    {
      "epoch": 0.14918862676831343,
      "grad_norm": 0.0027232684660702944,
      "learning_rate": 6.354067894436155e-05,
      "loss": 11.5,
      "step": 932
    },
    {
      "epoch": 0.149348700402185,
      "grad_norm": 0.0022869890090078115,
      "learning_rate": 6.334442811191576e-05,
      "loss": 11.5,
      "step": 933
    },
    {
      "epoch": 0.14950877403605659,
      "grad_norm": 0.002711891196668148,
      "learning_rate": 6.314834023422858e-05,
      "loss": 11.5,
      "step": 934
    },
    {
      "epoch": 0.14966884766992816,
      "grad_norm": 0.002728634513914585,
      "learning_rate": 6.295241618302156e-05,
      "loss": 11.5,
      "step": 935
    },
    {
      "epoch": 0.14982892130379974,
      "grad_norm": 0.001748494221828878,
      "learning_rate": 6.275665682928803e-05,
      "loss": 11.5,
      "step": 936
    },
    {
      "epoch": 0.14998899493767132,
      "grad_norm": 0.002195653971284628,
      "learning_rate": 6.256106304328905e-05,
      "loss": 11.5,
      "step": 937
    },
    {
      "epoch": 0.15014906857154292,
      "grad_norm": 0.003360239788889885,
      "learning_rate": 6.23656356945497e-05,
      "loss": 11.5,
      "step": 938
    },
    {
      "epoch": 0.1503091422054145,
      "grad_norm": 0.002994401380419731,
      "learning_rate": 6.21703756518551e-05,
      "loss": 11.5,
      "step": 939
    },
    {
      "epoch": 0.15046921583928607,
      "grad_norm": 0.003133380087092519,
      "learning_rate": 6.197528378324665e-05,
      "loss": 11.5,
      "step": 940
    },
    {
      "epoch": 0.15062928947315765,
      "grad_norm": 0.002414910588413477,
      "learning_rate": 6.17803609560181e-05,
      "loss": 11.5,
      "step": 941
    },
    {
      "epoch": 0.15078936310702923,
      "grad_norm": 0.003012928180396557,
      "learning_rate": 6.158560803671168e-05,
      "loss": 11.5,
      "step": 942
    },
    {
      "epoch": 0.1509494367409008,
      "grad_norm": 0.028782600536942482,
      "learning_rate": 6.139102589111435e-05,
      "loss": 11.5,
      "step": 943
    },
    {
      "epoch": 0.15110951037477238,
      "grad_norm": 0.01820896752178669,
      "learning_rate": 6.119661538425381e-05,
      "loss": 11.5,
      "step": 944
    },
    {
      "epoch": 0.151269584008644,
      "grad_norm": 0.007741926237940788,
      "learning_rate": 6.100237738039484e-05,
      "loss": 11.5,
      "step": 945
    },
    {
      "epoch": 0.15142965764251556,
      "grad_norm": 0.008180874399840832,
      "learning_rate": 6.0808312743035236e-05,
      "loss": 11.5,
      "step": 946
    },
    {
      "epoch": 0.15158973127638714,
      "grad_norm": 0.00771793769672513,
      "learning_rate": 6.061442233490211e-05,
      "loss": 11.5,
      "step": 947
    },
    {
      "epoch": 0.15174980491025872,
      "grad_norm": 0.005907203536480665,
      "learning_rate": 6.042070701794806e-05,
      "loss": 11.5,
      "step": 948
    },
    {
      "epoch": 0.1519098785441303,
      "grad_norm": 0.004037379287183285,
      "learning_rate": 6.0227167653347305e-05,
      "loss": 11.5,
      "step": 949
    },
    {
      "epoch": 0.15206995217800187,
      "grad_norm": 0.007686179596930742,
      "learning_rate": 6.0033805101491794e-05,
      "loss": 11.5,
      "step": 950
    },
    {
      "epoch": 0.15223002581187345,
      "grad_norm": 0.0024342418182641268,
      "learning_rate": 5.98406202219875e-05,
      "loss": 11.5,
      "step": 951
    },
    {
      "epoch": 0.15239009944574505,
      "grad_norm": 0.004280984867364168,
      "learning_rate": 5.964761387365052e-05,
      "loss": 11.5,
      "step": 952
    },
    {
      "epoch": 0.15255017307961663,
      "grad_norm": 0.008481497876346111,
      "learning_rate": 5.9454786914503255e-05,
      "loss": 11.5,
      "step": 953
    },
    {
      "epoch": 0.1527102467134882,
      "grad_norm": 0.0027742015663534403,
      "learning_rate": 5.926214020177074e-05,
      "loss": 11.5,
      "step": 954
    },
    {
      "epoch": 0.15287032034735978,
      "grad_norm": 0.002178241265937686,
      "learning_rate": 5.9069674591876534e-05,
      "loss": 11.5,
      "step": 955
    },
    {
      "epoch": 0.15303039398123136,
      "grad_norm": 0.005778731778264046,
      "learning_rate": 5.887739094043923e-05,
      "loss": 11.5,
      "step": 956
    },
    {
      "epoch": 0.15319046761510294,
      "grad_norm": 0.00665576197206974,
      "learning_rate": 5.868529010226845e-05,
      "loss": 11.5,
      "step": 957
    },
    {
      "epoch": 0.15335054124897451,
      "grad_norm": 0.0027354592457413673,
      "learning_rate": 5.849337293136112e-05,
      "loss": 11.5,
      "step": 958
    },
    {
      "epoch": 0.15351061488284612,
      "grad_norm": 0.0025162161327898502,
      "learning_rate": 5.830164028089766e-05,
      "loss": 11.5,
      "step": 959
    },
    {
      "epoch": 0.1536706885167177,
      "grad_norm": 0.002760212868452072,
      "learning_rate": 5.811009300323818e-05,
      "loss": 11.5,
      "step": 960
    },
    {
      "epoch": 0.15383076215058927,
      "grad_norm": 0.0020097976084798574,
      "learning_rate": 5.791873194991872e-05,
      "loss": 11.5,
      "step": 961
    },
    {
      "epoch": 0.15399083578446085,
      "grad_norm": 0.004208473023027182,
      "learning_rate": 5.7727557971647427e-05,
      "loss": 11.5,
      "step": 962
    },
    {
      "epoch": 0.15415090941833243,
      "grad_norm": 0.004097167402505875,
      "learning_rate": 5.7536571918300864e-05,
      "loss": 11.5,
      "step": 963
    },
    {
      "epoch": 0.154310983052204,
      "grad_norm": 0.005076675675809383,
      "learning_rate": 5.734577463892008e-05,
      "loss": 11.5,
      "step": 964
    },
    {
      "epoch": 0.1544710566860756,
      "grad_norm": 0.0037228353321552277,
      "learning_rate": 5.7155166981706956e-05,
      "loss": 11.5,
      "step": 965
    },
    {
      "epoch": 0.15463113031994719,
      "grad_norm": 0.0034742022398859262,
      "learning_rate": 5.6964749794020354e-05,
      "loss": 11.5,
      "step": 966
    },
    {
      "epoch": 0.15479120395381876,
      "grad_norm": 0.0034378506243228912,
      "learning_rate": 5.6774523922372394e-05,
      "loss": 11.5,
      "step": 967
    },
    {
      "epoch": 0.15495127758769034,
      "grad_norm": 0.0036437655799090862,
      "learning_rate": 5.6584490212424804e-05,
      "loss": 11.5,
      "step": 968
    },
    {
      "epoch": 0.15511135122156192,
      "grad_norm": 0.002839934080839157,
      "learning_rate": 5.639464950898491e-05,
      "loss": 11.5,
      "step": 969
    },
    {
      "epoch": 0.1552714248554335,
      "grad_norm": 0.0028093059081584215,
      "learning_rate": 5.620500265600206e-05,
      "loss": 11.5,
      "step": 970
    },
    {
      "epoch": 0.15543149848930507,
      "grad_norm": 0.003821355989202857,
      "learning_rate": 5.601555049656382e-05,
      "loss": 11.5,
      "step": 971
    },
    {
      "epoch": 0.15559157212317667,
      "grad_norm": 0.0015342511469498277,
      "learning_rate": 5.58262938728922e-05,
      "loss": 11.5,
      "step": 972
    },
    {
      "epoch": 0.15575164575704825,
      "grad_norm": 0.004114728886634111,
      "learning_rate": 5.563723362634008e-05,
      "loss": 11.5,
      "step": 973
    },
    {
      "epoch": 0.15591171939091983,
      "grad_norm": 0.005461197812110186,
      "learning_rate": 5.544837059738719e-05,
      "loss": 11.5,
      "step": 974
    },
    {
      "epoch": 0.1560717930247914,
      "grad_norm": 0.003300967626273632,
      "learning_rate": 5.525970562563656e-05,
      "loss": 11.5,
      "step": 975
    },
    {
      "epoch": 0.15623186665866298,
      "grad_norm": 0.0038338752929121256,
      "learning_rate": 5.507123954981073e-05,
      "loss": 11.5,
      "step": 976
    },
    {
      "epoch": 0.15639194029253456,
      "grad_norm": 0.0038169529289007187,
      "learning_rate": 5.488297320774807e-05,
      "loss": 11.5,
      "step": 977
    },
    {
      "epoch": 0.15655201392640614,
      "grad_norm": 0.002384957391768694,
      "learning_rate": 5.4694907436399e-05,
      "loss": 11.5,
      "step": 978
    },
    {
      "epoch": 0.15671208756027774,
      "grad_norm": 0.002962407423183322,
      "learning_rate": 5.4507043071822284e-05,
      "loss": 11.5,
      "step": 979
    },
    {
      "epoch": 0.15687216119414932,
      "grad_norm": 0.002982159610837698,
      "learning_rate": 5.431938094918132e-05,
      "loss": 11.5,
      "step": 980
    },
    {
      "epoch": 0.1570322348280209,
      "grad_norm": 0.006295071914792061,
      "learning_rate": 5.41319219027404e-05,
      "loss": 11.5,
      "step": 981
    },
    {
      "epoch": 0.15719230846189247,
      "grad_norm": 0.0028039771132171154,
      "learning_rate": 5.394466676586114e-05,
      "loss": 11.5,
      "step": 982
    },
    {
      "epoch": 0.15735238209576405,
      "grad_norm": 0.0031086585950106382,
      "learning_rate": 5.375761637099854e-05,
      "loss": 11.5,
      "step": 983
    },
    {
      "epoch": 0.15751245572963563,
      "grad_norm": 0.0018480699509382248,
      "learning_rate": 5.357077154969742e-05,
      "loss": 11.5,
      "step": 984
    },
    {
      "epoch": 0.1576725293635072,
      "grad_norm": 0.005203374195843935,
      "learning_rate": 5.3384133132588784e-05,
      "loss": 11.5,
      "step": 985
    },
    {
      "epoch": 0.1578326029973788,
      "grad_norm": 0.0016744579188525677,
      "learning_rate": 5.3197701949386e-05,
      "loss": 11.5,
      "step": 986
    },
    {
      "epoch": 0.15799267663125038,
      "grad_norm": 0.003136550774797797,
      "learning_rate": 5.301147882888116e-05,
      "loss": 11.5,
      "step": 987
    },
    {
      "epoch": 0.15815275026512196,
      "grad_norm": 0.002736804075539112,
      "learning_rate": 5.28254645989414e-05,
      "loss": 11.5,
      "step": 988
    },
    {
      "epoch": 0.15831282389899354,
      "grad_norm": 0.00409566517919302,
      "learning_rate": 5.2639660086505226e-05,
      "loss": 11.5,
      "step": 989
    },
    {
      "epoch": 0.15847289753286511,
      "grad_norm": 0.002291615353897214,
      "learning_rate": 5.2454066117578815e-05,
      "loss": 11.5,
      "step": 990
    },
    {
      "epoch": 0.1586329711667367,
      "grad_norm": 0.002706994069740176,
      "learning_rate": 5.226868351723244e-05,
      "loss": 11.5,
      "step": 991
    },
    {
      "epoch": 0.15879304480060827,
      "grad_norm": 0.01766454055905342,
      "learning_rate": 5.2083513109596616e-05,
      "loss": 11.5,
      "step": 992
    },
    {
      "epoch": 0.15895311843447987,
      "grad_norm": 0.04807746782898903,
      "learning_rate": 5.189855571785859e-05,
      "loss": 11.5,
      "step": 993
    },
    {
      "epoch": 0.15911319206835145,
      "grad_norm": 0.012728063389658928,
      "learning_rate": 5.171381216425863e-05,
      "loss": 11.5,
      "step": 994
    },
    {
      "epoch": 0.15927326570222303,
      "grad_norm": 0.005414154846221209,
      "learning_rate": 5.152928327008635e-05,
      "loss": 11.5,
      "step": 995
    },
    {
      "epoch": 0.1594333393360946,
      "grad_norm": 0.008969629183411598,
      "learning_rate": 5.134496985567714e-05,
      "loss": 11.5,
      "step": 996
    },
    {
      "epoch": 0.15959341296996618,
      "grad_norm": 0.0029964849818497896,
      "learning_rate": 5.116087274040837e-05,
      "loss": 11.5,
      "step": 997
    },
    {
      "epoch": 0.15975348660383776,
      "grad_norm": 0.004791382234543562,
      "learning_rate": 5.0976992742695925e-05,
      "loss": 11.5,
      "step": 998
    },
    {
      "epoch": 0.15991356023770933,
      "grad_norm": 0.008166301995515823,
      "learning_rate": 5.07933306799904e-05,
      "loss": 11.5,
      "step": 999
    },
    {
      "epoch": 0.16007363387158094,
      "grad_norm": 0.007119907531887293,
      "learning_rate": 5.060988736877366e-05,
      "loss": 11.5,
      "step": 1000
    },
    {
      "epoch": 0.16023370750545252,
      "grad_norm": 0.002332548378035426,
      "learning_rate": 5.042666362455498e-05,
      "loss": 11.5,
      "step": 1001
    },
    {
      "epoch": 0.1603937811393241,
      "grad_norm": 0.004620182793587446,
      "learning_rate": 5.024366026186755e-05,
      "loss": 11.5,
      "step": 1002
    },
    {
      "epoch": 0.16055385477319567,
      "grad_norm": 0.0025372488889843225,
      "learning_rate": 5.006087809426496e-05,
      "loss": 11.5,
      "step": 1003
    },
    {
      "epoch": 0.16071392840706725,
      "grad_norm": 0.003780423430725932,
      "learning_rate": 4.987831793431731e-05,
      "loss": 11.5,
      "step": 1004
    },
    {
      "epoch": 0.16087400204093882,
      "grad_norm": 0.0023666718043386936,
      "learning_rate": 4.9695980593607817e-05,
      "loss": 11.5,
      "step": 1005
    },
    {
      "epoch": 0.1610340756748104,
      "grad_norm": 0.010241476818919182,
      "learning_rate": 4.9513866882729146e-05,
      "loss": 11.5,
      "step": 1006
    },
    {
      "epoch": 0.161194149308682,
      "grad_norm": 0.0031386364717036486,
      "learning_rate": 4.9331977611279777e-05,
      "loss": 11.5,
      "step": 1007
    },
    {
      "epoch": 0.16135422294255358,
      "grad_norm": 0.002919365419074893,
      "learning_rate": 4.9150313587860433e-05,
      "loss": 11.5,
      "step": 1008
    },
    {
      "epoch": 0.16151429657642516,
      "grad_norm": 0.0030420273542404175,
      "learning_rate": 4.896887562007054e-05,
      "loss": 11.5,
      "step": 1009
    },
    {
      "epoch": 0.16167437021029674,
      "grad_norm": 0.0041471365839242935,
      "learning_rate": 4.8787664514504504e-05,
      "loss": 11.5,
      "step": 1010
    },
    {
      "epoch": 0.1618344438441683,
      "grad_norm": 0.001787104643881321,
      "learning_rate": 4.860668107674823e-05,
      "loss": 11.5,
      "step": 1011
    },
    {
      "epoch": 0.1619945174780399,
      "grad_norm": 0.001213782117702067,
      "learning_rate": 4.8425926111375506e-05,
      "loss": 11.5,
      "step": 1012
    },
    {
      "epoch": 0.16215459111191147,
      "grad_norm": 0.007795942481607199,
      "learning_rate": 4.824540042194443e-05,
      "loss": 11.5,
      "step": 1013
    },
    {
      "epoch": 0.16231466474578307,
      "grad_norm": 0.001617083209566772,
      "learning_rate": 4.8065104810993856e-05,
      "loss": 11.5,
      "step": 1014
    },
    {
      "epoch": 0.16247473837965465,
      "grad_norm": 0.004990368615835905,
      "learning_rate": 4.788504008003978e-05,
      "loss": 11.5,
      "step": 1015
    },
    {
      "epoch": 0.16263481201352623,
      "grad_norm": 0.0036852476187050343,
      "learning_rate": 4.770520702957182e-05,
      "loss": 11.5,
      "step": 1016
    },
    {
      "epoch": 0.1627948856473978,
      "grad_norm": 0.003142181783914566,
      "learning_rate": 4.752560645904962e-05,
      "loss": 11.5,
      "step": 1017
    },
    {
      "epoch": 0.16295495928126938,
      "grad_norm": 0.0034457738511264324,
      "learning_rate": 4.734623916689941e-05,
      "loss": 11.5,
      "step": 1018
    },
    {
      "epoch": 0.16311503291514096,
      "grad_norm": 0.003712948877364397,
      "learning_rate": 4.716710595051022e-05,
      "loss": 11.5,
      "step": 1019
    },
    {
      "epoch": 0.16327510654901253,
      "grad_norm": 0.0032946669962257147,
      "learning_rate": 4.698820760623064e-05,
      "loss": 11.5,
      "step": 1020
    },
    {
      "epoch": 0.16343518018288414,
      "grad_norm": 0.003699769964441657,
      "learning_rate": 4.6809544929365004e-05,
      "loss": 11.5,
      "step": 1021
    },
    {
      "epoch": 0.16359525381675571,
      "grad_norm": 0.003574742702767253,
      "learning_rate": 4.663111871417e-05,
      "loss": 11.5,
      "step": 1022
    },
    {
      "epoch": 0.1637553274506273,
      "grad_norm": 0.003883534809574485,
      "learning_rate": 4.645292975385111e-05,
      "loss": 11.5,
      "step": 1023
    },
    {
      "epoch": 0.16391540108449887,
      "grad_norm": 0.0018628851976245642,
      "learning_rate": 4.627497884055912e-05,
      "loss": 11.5,
      "step": 1024
    },
    {
      "epoch": 0.16407547471837045,
      "grad_norm": 0.0017472283216193318,
      "learning_rate": 4.609726676538652e-05,
      "loss": 11.5,
      "step": 1025
    },
    {
      "epoch": 0.16423554835224202,
      "grad_norm": 0.002226916141808033,
      "learning_rate": 4.591979431836402e-05,
      "loss": 11.5,
      "step": 1026
    },
    {
      "epoch": 0.1643956219861136,
      "grad_norm": 0.0016684042057022452,
      "learning_rate": 4.574256228845706e-05,
      "loss": 11.5,
      "step": 1027
    },
    {
      "epoch": 0.1645556956199852,
      "grad_norm": 0.0028440216556191444,
      "learning_rate": 4.5565571463562365e-05,
      "loss": 11.5,
      "step": 1028
    },
    {
      "epoch": 0.16471576925385678,
      "grad_norm": 0.003003405872732401,
      "learning_rate": 4.5388822630504256e-05,
      "loss": 11.5,
      "step": 1029
    },
    {
      "epoch": 0.16487584288772836,
      "grad_norm": 0.003543854458257556,
      "learning_rate": 4.521231657503132e-05,
      "loss": 11.5,
      "step": 1030
    },
    {
      "epoch": 0.16503591652159993,
      "grad_norm": 0.0027281527873128653,
      "learning_rate": 4.503605408181286e-05,
      "loss": 11.5,
      "step": 1031
    },
    {
      "epoch": 0.1651959901554715,
      "grad_norm": 0.00303859100677073,
      "learning_rate": 4.486003593443537e-05,
      "loss": 11.5,
      "step": 1032
    },
    {
      "epoch": 0.1653560637893431,
      "grad_norm": 0.0027469072956591845,
      "learning_rate": 4.468426291539914e-05,
      "loss": 11.5,
      "step": 1033
    },
    {
      "epoch": 0.16551613742321467,
      "grad_norm": 0.0026546616572886705,
      "learning_rate": 4.4508735806114654e-05,
      "loss": 11.5,
      "step": 1034
    },
    {
      "epoch": 0.16567621105708627,
      "grad_norm": 0.0019743379671126604,
      "learning_rate": 4.433345538689929e-05,
      "loss": 11.5,
      "step": 1035
    },
    {
      "epoch": 0.16583628469095785,
      "grad_norm": 0.0032873465679585934,
      "learning_rate": 4.415842243697369e-05,
      "loss": 11.5,
      "step": 1036
    },
    {
      "epoch": 0.16599635832482942,
      "grad_norm": 0.0029843030497431755,
      "learning_rate": 4.39836377344583e-05,
      "loss": 11.5,
      "step": 1037
    },
    {
      "epoch": 0.166156431958701,
      "grad_norm": 0.0018177861347794533,
      "learning_rate": 4.380910205637012e-05,
      "loss": 11.5,
      "step": 1038
    },
    {
      "epoch": 0.16631650559257258,
      "grad_norm": 0.0035029847640544176,
      "learning_rate": 4.363481617861893e-05,
      "loss": 11.5,
      "step": 1039
    },
    {
      "epoch": 0.16647657922644415,
      "grad_norm": 0.0028489981777966022,
      "learning_rate": 4.346078087600412e-05,
      "loss": 11.5,
      "step": 1040
    },
    {
      "epoch": 0.16663665286031573,
      "grad_norm": 0.002027237555012107,
      "learning_rate": 4.3286996922211034e-05,
      "loss": 11.5,
      "step": 1041
    },
    {
      "epoch": 0.16679672649418734,
      "grad_norm": 0.0029321161564439535,
      "learning_rate": 4.311346508980772e-05,
      "loss": 11.5,
      "step": 1042
    },
    {
      "epoch": 0.1669568001280589,
      "grad_norm": 0.027388835325837135,
      "learning_rate": 4.2940186150241365e-05,
      "loss": 11.5,
      "step": 1043
    },
    {
      "epoch": 0.1671168737619305,
      "grad_norm": 0.018330741673707962,
      "learning_rate": 4.27671608738349e-05,
      "loss": 11.5,
      "step": 1044
    },
    {
      "epoch": 0.16727694739580207,
      "grad_norm": 0.005885079503059387,
      "learning_rate": 4.2594390029783534e-05,
      "loss": 11.5,
      "step": 1045
    },
    {
      "epoch": 0.16743702102967364,
      "grad_norm": 0.008645343594253063,
      "learning_rate": 4.242187438615153e-05,
      "loss": 11.5,
      "step": 1046
    },
    {
      "epoch": 0.16759709466354522,
      "grad_norm": 0.008244359865784645,
      "learning_rate": 4.224961470986849e-05,
      "loss": 11.5,
      "step": 1047
    },
    {
      "epoch": 0.16775716829741683,
      "grad_norm": 0.013236356899142265,
      "learning_rate": 4.207761176672614e-05,
      "loss": 11.5,
      "step": 1048
    },
    {
      "epoch": 0.1679172419312884,
      "grad_norm": 0.005240214988589287,
      "learning_rate": 4.190586632137491e-05,
      "loss": 11.5,
      "step": 1049
    },
    {
      "epoch": 0.16807731556515998,
      "grad_norm": 0.004562577232718468,
      "learning_rate": 4.173437913732048e-05,
      "loss": 11.5,
      "step": 1050
    },
    {
      "epoch": 0.16823738919903156,
      "grad_norm": 0.0016778433928266168,
      "learning_rate": 4.156315097692037e-05,
      "loss": 11.5,
      "step": 1051
    },
    {
      "epoch": 0.16839746283290313,
      "grad_norm": 0.0028510624542832375,
      "learning_rate": 4.139218260138074e-05,
      "loss": 11.5,
      "step": 1052
    },
    {
      "epoch": 0.1685575364667747,
      "grad_norm": 0.0021169029641896486,
      "learning_rate": 4.12214747707527e-05,
      "loss": 11.5,
      "step": 1053
    },
    {
      "epoch": 0.1687176101006463,
      "grad_norm": 0.007874535396695137,
      "learning_rate": 4.1051028243929125e-05,
      "loss": 11.5,
      "step": 1054
    },
    {
      "epoch": 0.1688776837345179,
      "grad_norm": 0.0024130248930305243,
      "learning_rate": 4.088084377864135e-05,
      "loss": 11.5,
      "step": 1055
    },
    {
      "epoch": 0.16903775736838947,
      "grad_norm": 0.0034061060287058353,
      "learning_rate": 4.07109221314556e-05,
      "loss": 11.5,
      "step": 1056
    },
    {
      "epoch": 0.16919783100226105,
      "grad_norm": 0.002557958010584116,
      "learning_rate": 4.054126405776971e-05,
      "loss": 11.5,
      "step": 1057
    },
    {
      "epoch": 0.16935790463613262,
      "grad_norm": 0.005619005765765905,
      "learning_rate": 4.037187031180985e-05,
      "loss": 11.5,
      "step": 1058
    },
    {
      "epoch": 0.1695179782700042,
      "grad_norm": 0.0027678897604346275,
      "learning_rate": 4.020274164662707e-05,
      "loss": 11.5,
      "step": 1059
    },
    {
      "epoch": 0.16967805190387578,
      "grad_norm": 0.0022015825379639864,
      "learning_rate": 4.003387881409397e-05,
      "loss": 11.5,
      "step": 1060
    },
    {
      "epoch": 0.16983812553774735,
      "grad_norm": 0.0033486501779407263,
      "learning_rate": 3.986528256490141e-05,
      "loss": 11.5,
      "step": 1061
    },
    {
      "epoch": 0.16999819917161896,
      "grad_norm": 0.004518864210695028,
      "learning_rate": 3.969695364855511e-05,
      "loss": 11.5,
      "step": 1062
    },
    {
      "epoch": 0.17015827280549053,
      "grad_norm": 0.0028971631545573473,
      "learning_rate": 3.952889281337235e-05,
      "loss": 11.5,
      "step": 1063
    },
    {
      "epoch": 0.1703183464393621,
      "grad_norm": 0.0033338076900690794,
      "learning_rate": 3.93611008064786e-05,
      "loss": 11.5,
      "step": 1064
    },
    {
      "epoch": 0.1704784200732337,
      "grad_norm": 0.003616789123043418,
      "learning_rate": 3.9193578373804364e-05,
      "loss": 11.5,
      "step": 1065
    },
    {
      "epoch": 0.17063849370710527,
      "grad_norm": 0.00574265094473958,
      "learning_rate": 3.90263262600816e-05,
      "loss": 11.5,
      "step": 1066
    },
    {
      "epoch": 0.17079856734097684,
      "grad_norm": 0.003767479443922639,
      "learning_rate": 3.88593452088406e-05,
      "loss": 11.5,
      "step": 1067
    },
    {
      "epoch": 0.17095864097484842,
      "grad_norm": 0.003367670811712742,
      "learning_rate": 3.869263596240661e-05,
      "loss": 11.5,
      "step": 1068
    },
    {
      "epoch": 0.17111871460872002,
      "grad_norm": 0.0025567905977368355,
      "learning_rate": 3.8526199261896544e-05,
      "loss": 11.5,
      "step": 1069
    },
    {
      "epoch": 0.1712787882425916,
      "grad_norm": 0.0031548349652439356,
      "learning_rate": 3.836003584721577e-05,
      "loss": 11.5,
      "step": 1070
    },
    {
      "epoch": 0.17143886187646318,
      "grad_norm": 0.0037915778812021017,
      "learning_rate": 3.8194146457054655e-05,
      "loss": 11.5,
      "step": 1071
    },
    {
      "epoch": 0.17159893551033475,
      "grad_norm": 0.0023920745588839054,
      "learning_rate": 3.802853182888543e-05,
      "loss": 11.5,
      "step": 1072
    },
    {
      "epoch": 0.17175900914420633,
      "grad_norm": 0.0018295550253242254,
      "learning_rate": 3.786319269895877e-05,
      "loss": 11.5,
      "step": 1073
    },
    {
      "epoch": 0.1719190827780779,
      "grad_norm": 0.003734629601240158,
      "learning_rate": 3.769812980230074e-05,
      "loss": 11.5,
      "step": 1074
    },
    {
      "epoch": 0.17207915641194949,
      "grad_norm": 0.0012786126462742686,
      "learning_rate": 3.7533343872709294e-05,
      "loss": 11.5,
      "step": 1075
    },
    {
      "epoch": 0.1722392300458211,
      "grad_norm": 0.0034075656440109015,
      "learning_rate": 3.736883564275112e-05,
      "loss": 11.5,
      "step": 1076
    },
    {
      "epoch": 0.17239930367969267,
      "grad_norm": 0.0026780336629599333,
      "learning_rate": 3.7204605843758386e-05,
      "loss": 11.5,
      "step": 1077
    },
    {
      "epoch": 0.17255937731356424,
      "grad_norm": 0.0032900224905461073,
      "learning_rate": 3.704065520582549e-05,
      "loss": 11.5,
      "step": 1078
    },
    {
      "epoch": 0.17271945094743582,
      "grad_norm": 0.00796513445675373,
      "learning_rate": 3.6876984457805786e-05,
      "loss": 11.5,
      "step": 1079
    },
    {
      "epoch": 0.1728795245813074,
      "grad_norm": 0.006374262273311615,
      "learning_rate": 3.671359432730834e-05,
      "loss": 11.5,
      "step": 1080
    },
    {
      "epoch": 0.17303959821517897,
      "grad_norm": 0.0018266340484842658,
      "learning_rate": 3.655048554069478e-05,
      "loss": 11.5,
      "step": 1081
    },
    {
      "epoch": 0.17319967184905055,
      "grad_norm": 0.003731823293492198,
      "learning_rate": 3.638765882307589e-05,
      "loss": 11.5,
      "step": 1082
    },
    {
      "epoch": 0.17335974548292216,
      "grad_norm": 0.003684400813654065,
      "learning_rate": 3.6225114898308634e-05,
      "loss": 11.5,
      "step": 1083
    },
    {
      "epoch": 0.17351981911679373,
      "grad_norm": 0.003037514863535762,
      "learning_rate": 3.6062854488992714e-05,
      "loss": 11.5,
      "step": 1084
    },
    {
      "epoch": 0.1736798927506653,
      "grad_norm": 0.0035959489177912474,
      "learning_rate": 3.5900878316467454e-05,
      "loss": 11.5,
      "step": 1085
    },
    {
      "epoch": 0.1738399663845369,
      "grad_norm": 0.0020323735661804676,
      "learning_rate": 3.573918710080857e-05,
      "loss": 11.5,
      "step": 1086
    },
    {
      "epoch": 0.17400004001840846,
      "grad_norm": 0.002366045257076621,
      "learning_rate": 3.5577781560825066e-05,
      "loss": 11.5,
      "step": 1087
    },
    {
      "epoch": 0.17416011365228004,
      "grad_norm": 0.00238984078168869,
      "learning_rate": 3.541666241405588e-05,
      "loss": 11.5,
      "step": 1088
    },
    {
      "epoch": 0.17432018728615162,
      "grad_norm": 0.003099254798144102,
      "learning_rate": 3.5255830376766764e-05,
      "loss": 11.5,
      "step": 1089
    },
    {
      "epoch": 0.17448026092002322,
      "grad_norm": 0.0018906679470092058,
      "learning_rate": 3.509528616394716e-05,
      "loss": 11.5,
      "step": 1090
    },
    {
      "epoch": 0.1746403345538948,
      "grad_norm": 0.00253277737647295,
      "learning_rate": 3.4935030489306883e-05,
      "loss": 11.5,
      "step": 1091
    },
    {
      "epoch": 0.17480040818776638,
      "grad_norm": 0.0018051579827442765,
      "learning_rate": 3.4775064065273165e-05,
      "loss": 11.5,
      "step": 1092
    },
    {
      "epoch": 0.17496048182163795,
      "grad_norm": 0.002916001481935382,
      "learning_rate": 3.4615387602987236e-05,
      "loss": 11.5,
      "step": 1093
    },
    {
      "epoch": 0.17512055545550953,
      "grad_norm": 0.02692965790629387,
      "learning_rate": 3.445600181230134e-05,
      "loss": 11.5,
      "step": 1094
    },
    {
      "epoch": 0.1752806290893811,
      "grad_norm": 0.016595419496297836,
      "learning_rate": 3.429690740177549e-05,
      "loss": 11.5,
      "step": 1095
    },
    {
      "epoch": 0.17544070272325268,
      "grad_norm": 0.006738040596246719,
      "learning_rate": 3.413810507867436e-05,
      "loss": 11.5,
      "step": 1096
    },
    {
      "epoch": 0.1756007763571243,
      "grad_norm": 0.0090487627312541,
      "learning_rate": 3.397959554896415e-05,
      "loss": 11.5,
      "step": 1097
    },
    {
      "epoch": 0.17576084999099587,
      "grad_norm": 0.005679016001522541,
      "learning_rate": 3.3821379517309405e-05,
      "loss": 11.5,
      "step": 1098
    },
    {
      "epoch": 0.17592092362486744,
      "grad_norm": 0.009319297969341278,
      "learning_rate": 3.3663457687069924e-05,
      "loss": 11.5,
      "step": 1099
    },
    {
      "epoch": 0.17608099725873902,
      "grad_norm": 0.006360052153468132,
      "learning_rate": 3.350583076029754e-05,
      "loss": 11.5,
      "step": 1100
    },
    {
      "epoch": 0.1762410708926106,
      "grad_norm": 0.003051420673727989,
      "learning_rate": 3.334849943773323e-05,
      "loss": 11.5,
      "step": 1101
    },
    {
      "epoch": 0.17640114452648217,
      "grad_norm": 0.0020076115615665913,
      "learning_rate": 3.319146441880371e-05,
      "loss": 11.5,
      "step": 1102
    },
    {
      "epoch": 0.17656121816035375,
      "grad_norm": 0.0069500780664384365,
      "learning_rate": 3.3034726401618444e-05,
      "loss": 11.5,
      "step": 1103
    },
    {
      "epoch": 0.17672129179422535,
      "grad_norm": 0.003077697241678834,
      "learning_rate": 3.28782860829667e-05,
      "loss": 11.5,
      "step": 1104
    },
    {
      "epoch": 0.17688136542809693,
      "grad_norm": 0.00394095666706562,
      "learning_rate": 3.272214415831418e-05,
      "loss": 11.5,
      "step": 1105
    },
    {
      "epoch": 0.1770414390619685,
      "grad_norm": 0.0034618766512721777,
      "learning_rate": 3.2566301321800085e-05,
      "loss": 11.5,
      "step": 1106
    },
    {
      "epoch": 0.17720151269584009,
      "grad_norm": 0.0036810175515711308,
      "learning_rate": 3.241075826623401e-05,
      "loss": 11.5,
      "step": 1107
    },
    {
      "epoch": 0.17736158632971166,
      "grad_norm": 0.003591476008296013,
      "learning_rate": 3.225551568309284e-05,
      "loss": 11.5,
      "step": 1108
    },
    {
      "epoch": 0.17752165996358324,
      "grad_norm": 0.003950075246393681,
      "learning_rate": 3.210057426251773e-05,
      "loss": 11.5,
      "step": 1109
    },
    {
      "epoch": 0.17768173359745482,
      "grad_norm": 0.003185428213328123,
      "learning_rate": 3.1945934693310896e-05,
      "loss": 11.5,
      "step": 1110
    },
    {
      "epoch": 0.17784180723132642,
      "grad_norm": 0.005893181078135967,
      "learning_rate": 3.179159766293282e-05,
      "loss": 11.5,
      "step": 1111
    },
    {
      "epoch": 0.178001880865198,
      "grad_norm": 0.0025459767784923315,
      "learning_rate": 3.163756385749889e-05,
      "loss": 11.5,
      "step": 1112
    },
    {
      "epoch": 0.17816195449906957,
      "grad_norm": 0.005068275611847639,
      "learning_rate": 3.148383396177653e-05,
      "loss": 11.5,
      "step": 1113
    },
    {
      "epoch": 0.17832202813294115,
      "grad_norm": 0.001588392653502524,
      "learning_rate": 3.133040865918213e-05,
      "loss": 11.5,
      "step": 1114
    },
    {
      "epoch": 0.17848210176681273,
      "grad_norm": 0.0036992819514125586,
      "learning_rate": 3.117728863177796e-05,
      "loss": 11.5,
      "step": 1115
    },
    {
      "epoch": 0.1786421754006843,
      "grad_norm": 0.005307686049491167,
      "learning_rate": 3.102447456026919e-05,
      "loss": 11.5,
      "step": 1116
    },
    {
      "epoch": 0.17880224903455588,
      "grad_norm": 0.0036241610068827868,
      "learning_rate": 3.0871967124000834e-05,
      "loss": 11.5,
      "step": 1117
    },
    {
      "epoch": 0.1789623226684275,
      "grad_norm": 0.0023742804769426584,
      "learning_rate": 3.0719767000954714e-05,
      "loss": 11.5,
      "step": 1118
    },
    {
      "epoch": 0.17912239630229906,
      "grad_norm": 0.0025262085255235434,
      "learning_rate": 3.056787486774656e-05,
      "loss": 11.5,
      "step": 1119
    },
    {
      "epoch": 0.17928246993617064,
      "grad_norm": 0.003705256851390004,
      "learning_rate": 3.041629139962283e-05,
      "loss": 11.5,
      "step": 1120
    },
    {
      "epoch": 0.17944254357004222,
      "grad_norm": 0.004558330401778221,
      "learning_rate": 3.0265017270457775e-05,
      "loss": 11.5,
      "step": 1121
    },
    {
      "epoch": 0.1796026172039138,
      "grad_norm": 0.002673452254384756,
      "learning_rate": 3.0114053152750556e-05,
      "loss": 11.5,
      "step": 1122
    },
    {
      "epoch": 0.17976269083778537,
      "grad_norm": 0.0027173180133104324,
      "learning_rate": 2.9963399717622077e-05,
      "loss": 11.5,
      "step": 1123
    },
    {
      "epoch": 0.17992276447165695,
      "grad_norm": 0.004756426904350519,
      "learning_rate": 2.98130576348121e-05,
      "loss": 11.5,
      "step": 1124
    },
    {
      "epoch": 0.18008283810552855,
      "grad_norm": 0.0025077357422560453,
      "learning_rate": 2.966302757267625e-05,
      "loss": 11.5,
      "step": 1125
    },
    {
      "epoch": 0.18008283810552855,
      "eval_loss": 11.5,
      "eval_runtime": 59.0614,
      "eval_samples_per_second": 178.154,
      "eval_steps_per_second": 89.077,
      "step": 1125
    },
    {
      "epoch": 0.18024291173940013,
      "grad_norm": 0.005658532027155161,
      "learning_rate": 2.9513310198183065e-05,
      "loss": 11.5,
      "step": 1126
    },
    {
      "epoch": 0.1804029853732717,
      "grad_norm": 0.00252137565985322,
      "learning_rate": 2.936390617691097e-05,
      "loss": 11.5,
      "step": 1127
    },
    {
      "epoch": 0.18056305900714328,
      "grad_norm": 0.003056940156966448,
      "learning_rate": 2.9214816173045356e-05,
      "loss": 11.5,
      "step": 1128
    },
    {
      "epoch": 0.18072313264101486,
      "grad_norm": 0.004033989738672972,
      "learning_rate": 2.906604084937572e-05,
      "loss": 11.5,
      "step": 1129
    },
    {
      "epoch": 0.18088320627488644,
      "grad_norm": 0.003481508931145072,
      "learning_rate": 2.8917580867292526e-05,
      "loss": 11.5,
      "step": 1130
    },
    {
      "epoch": 0.18104327990875804,
      "grad_norm": 0.0026058063376694918,
      "learning_rate": 2.8769436886784408e-05,
      "loss": 11.5,
      "step": 1131
    },
    {
      "epoch": 0.18120335354262962,
      "grad_norm": 0.004511214327067137,
      "learning_rate": 2.862160956643517e-05,
      "loss": 11.5,
      "step": 1132
    },
    {
      "epoch": 0.1813634271765012,
      "grad_norm": 0.001757331076078117,
      "learning_rate": 2.847409956342092e-05,
      "loss": 11.5,
      "step": 1133
    },
    {
      "epoch": 0.18152350081037277,
      "grad_norm": 0.0030876228120177984,
      "learning_rate": 2.8326907533507074e-05,
      "loss": 11.5,
      "step": 1134
    },
    {
      "epoch": 0.18168357444424435,
      "grad_norm": 0.0035963074769824743,
      "learning_rate": 2.8180034131045464e-05,
      "loss": 11.5,
      "step": 1135
    },
    {
      "epoch": 0.18184364807811593,
      "grad_norm": 0.0022286765743047,
      "learning_rate": 2.8033480008971546e-05,
      "loss": 11.5,
      "step": 1136
    },
    {
      "epoch": 0.1820037217119875,
      "grad_norm": 0.003147304290905595,
      "learning_rate": 2.7887245818801277e-05,
      "loss": 11.5,
      "step": 1137
    },
    {
      "epoch": 0.1821637953458591,
      "grad_norm": 0.0016215078067034483,
      "learning_rate": 2.7741332210628345e-05,
      "loss": 11.5,
      "step": 1138
    },
    {
      "epoch": 0.18232386897973069,
      "grad_norm": 0.0019105322426185012,
      "learning_rate": 2.759573983312138e-05,
      "loss": 11.5,
      "step": 1139
    },
    {
      "epoch": 0.18248394261360226,
      "grad_norm": 0.0024289079010486603,
      "learning_rate": 2.7450469333520855e-05,
      "loss": 11.5,
      "step": 1140
    },
    {
      "epoch": 0.18264401624747384,
      "grad_norm": 0.0027825345750898123,
      "learning_rate": 2.730552135763632e-05,
      "loss": 11.5,
      "step": 1141
    },
    {
      "epoch": 0.18280408988134542,
      "grad_norm": 0.0023849569261074066,
      "learning_rate": 2.7160896549843562e-05,
      "loss": 11.5,
      "step": 1142
    },
    {
      "epoch": 0.182964163515217,
      "grad_norm": 0.00504154060035944,
      "learning_rate": 2.701659555308169e-05,
      "loss": 11.5,
      "step": 1143
    },
    {
      "epoch": 0.18312423714908857,
      "grad_norm": 0.01448983047157526,
      "learning_rate": 2.6872619008850274e-05,
      "loss": 11.5,
      "step": 1144
    },
    {
      "epoch": 0.18328431078296017,
      "grad_norm": 0.008898479864001274,
      "learning_rate": 2.672896755720654e-05,
      "loss": 11.5,
      "step": 1145
    },
    {
      "epoch": 0.18344438441683175,
      "grad_norm": 0.008333215489983559,
      "learning_rate": 2.6585641836762433e-05,
      "loss": 11.5,
      "step": 1146
    },
    {
      "epoch": 0.18360445805070333,
      "grad_norm": 0.007012536749243736,
      "learning_rate": 2.6442642484681944e-05,
      "loss": 11.5,
      "step": 1147
    },
    {
      "epoch": 0.1837645316845749,
      "grad_norm": 0.006049661431461573,
      "learning_rate": 2.6299970136678077e-05,
      "loss": 11.5,
      "step": 1148
    },
    {
      "epoch": 0.18392460531844648,
      "grad_norm": 0.007174614351242781,
      "learning_rate": 2.6157625427010156e-05,
      "loss": 11.5,
      "step": 1149
    },
    {
      "epoch": 0.18408467895231806,
      "grad_norm": 0.008941186591982841,
      "learning_rate": 2.6015608988480955e-05,
      "loss": 11.5,
      "step": 1150
    },
    {
      "epoch": 0.18424475258618964,
      "grad_norm": 0.007125054951757193,
      "learning_rate": 2.5873921452433915e-05,
      "loss": 11.5,
      "step": 1151
    },
    {
      "epoch": 0.18440482622006124,
      "grad_norm": 0.003278638469055295,
      "learning_rate": 2.57325634487503e-05,
      "loss": 11.5,
      "step": 1152
    },
    {
      "epoch": 0.18456489985393282,
      "grad_norm": 0.002455964218825102,
      "learning_rate": 2.5591535605846383e-05,
      "loss": 11.5,
      "step": 1153
    },
    {
      "epoch": 0.1847249734878044,
      "grad_norm": 0.0021103492472320795,
      "learning_rate": 2.5450838550670808e-05,
      "loss": 11.5,
      "step": 1154
    },
    {
      "epoch": 0.18488504712167597,
      "grad_norm": 0.0031680497340857983,
      "learning_rate": 2.5310472908701555e-05,
      "loss": 11.5,
      "step": 1155
    },
    {
      "epoch": 0.18504512075554755,
      "grad_norm": 0.002496772212907672,
      "learning_rate": 2.5170439303943294e-05,
      "loss": 11.5,
      "step": 1156
    },
    {
      "epoch": 0.18520519438941913,
      "grad_norm": 0.005319160409271717,
      "learning_rate": 2.503073835892471e-05,
      "loss": 11.5,
      "step": 1157
    },
    {
      "epoch": 0.1853652680232907,
      "grad_norm": 0.0033112477976828814,
      "learning_rate": 2.4891370694695517e-05,
      "loss": 11.5,
      "step": 1158
    },
    {
      "epoch": 0.1855253416571623,
      "grad_norm": 0.004564044065773487,
      "learning_rate": 2.4752336930823837e-05,
      "loss": 11.5,
      "step": 1159
    },
    {
      "epoch": 0.18568541529103388,
      "grad_norm": 0.0019350239308550954,
      "learning_rate": 2.4613637685393432e-05,
      "loss": 11.5,
      "step": 1160
    },
    {
      "epoch": 0.18584548892490546,
      "grad_norm": 0.0016299671260640025,
      "learning_rate": 2.4475273575000936e-05,
      "loss": 11.5,
      "step": 1161
    },
    {
      "epoch": 0.18600556255877704,
      "grad_norm": 0.007100170943886042,
      "learning_rate": 2.4337245214753103e-05,
      "loss": 11.5,
      "step": 1162
    },
    {
      "epoch": 0.18616563619264861,
      "grad_norm": 0.006092431955039501,
      "learning_rate": 2.4199553218264093e-05,
      "loss": 11.5,
      "step": 1163
    },
    {
      "epoch": 0.1863257098265202,
      "grad_norm": 0.005135354120284319,
      "learning_rate": 2.4062198197652752e-05,
      "loss": 11.5,
      "step": 1164
    },
    {
      "epoch": 0.18648578346039177,
      "grad_norm": 0.001947320532053709,
      "learning_rate": 2.3925180763539844e-05,
      "loss": 11.5,
      "step": 1165
    },
    {
      "epoch": 0.18664585709426337,
      "grad_norm": 0.006747918203473091,
      "learning_rate": 2.3788501525045438e-05,
      "loss": 11.5,
      "step": 1166
    },
    {
      "epoch": 0.18680593072813495,
      "grad_norm": 0.003693880746141076,
      "learning_rate": 2.3652161089786086e-05,
      "loss": 11.5,
      "step": 1167
    },
    {
      "epoch": 0.18696600436200653,
      "grad_norm": 0.003128764918074012,
      "learning_rate": 2.351616006387214e-05,
      "loss": 11.5,
      "step": 1168
    },
    {
      "epoch": 0.1871260779958781,
      "grad_norm": 0.0047349585220217705,
      "learning_rate": 2.3380499051905137e-05,
      "loss": 11.5,
      "step": 1169
    },
    {
      "epoch": 0.18728615162974968,
      "grad_norm": 0.0024411038029938936,
      "learning_rate": 2.324517865697501e-05,
      "loss": 11.5,
      "step": 1170
    },
    {
      "epoch": 0.18744622526362126,
      "grad_norm": 0.0044091069139540195,
      "learning_rate": 2.3110199480657525e-05,
      "loss": 11.5,
      "step": 1171
    },
    {
      "epoch": 0.18760629889749283,
      "grad_norm": 0.002249740995466709,
      "learning_rate": 2.2975562123011495e-05,
      "loss": 11.5,
      "step": 1172
    },
    {
      "epoch": 0.18776637253136444,
      "grad_norm": 0.00363954552449286,
      "learning_rate": 2.2841267182576143e-05,
      "loss": 11.5,
      "step": 1173
    },
    {
      "epoch": 0.18792644616523602,
      "grad_norm": 0.0034002477768808603,
      "learning_rate": 2.2707315256368433e-05,
      "loss": 11.5,
      "step": 1174
    },
    {
      "epoch": 0.1880865197991076,
      "grad_norm": 0.0022575787734240294,
      "learning_rate": 2.2573706939880555e-05,
      "loss": 11.5,
      "step": 1175
    },
    {
      "epoch": 0.18824659343297917,
      "grad_norm": 0.001263917307369411,
      "learning_rate": 2.2440442827077045e-05,
      "loss": 11.5,
      "step": 1176
    },
    {
      "epoch": 0.18840666706685075,
      "grad_norm": 0.0030521685257554054,
      "learning_rate": 2.230752351039228e-05,
      "loss": 11.5,
      "step": 1177
    },
    {
      "epoch": 0.18856674070072232,
      "grad_norm": 0.003970122896134853,
      "learning_rate": 2.2174949580727832e-05,
      "loss": 11.5,
      "step": 1178
    },
    {
      "epoch": 0.1887268143345939,
      "grad_norm": 0.0015181894414126873,
      "learning_rate": 2.2042721627449846e-05,
      "loss": 11.5,
      "step": 1179
    },
    {
      "epoch": 0.1888868879684655,
      "grad_norm": 0.0021002558059990406,
      "learning_rate": 2.1910840238386398e-05,
      "loss": 11.5,
      "step": 1180
    },
    {
      "epoch": 0.18904696160233708,
      "grad_norm": 0.005683374125510454,
      "learning_rate": 2.1779305999824884e-05,
      "loss": 11.5,
      "step": 1181
    },
    {
      "epoch": 0.18920703523620866,
      "grad_norm": 0.00221837917342782,
      "learning_rate": 2.164811949650942e-05,
      "loss": 11.5,
      "step": 1182
    },
    {
      "epoch": 0.18936710887008024,
      "grad_norm": 0.003727814182639122,
      "learning_rate": 2.1517281311638217e-05,
      "loss": 11.5,
      "step": 1183
    },
    {
      "epoch": 0.1895271825039518,
      "grad_norm": 0.0021374353673309088,
      "learning_rate": 2.1386792026861103e-05,
      "loss": 11.5,
      "step": 1184
    },
    {
      "epoch": 0.1896872561378234,
      "grad_norm": 0.004052836447954178,
      "learning_rate": 2.125665222227675e-05,
      "loss": 11.5,
      "step": 1185
    },
    {
      "epoch": 0.18984732977169497,
      "grad_norm": 0.003397738328203559,
      "learning_rate": 2.112686247643024e-05,
      "loss": 11.5,
      "step": 1186
    },
    {
      "epoch": 0.19000740340556657,
      "grad_norm": 0.006731340661644936,
      "learning_rate": 2.09974233663104e-05,
      "loss": 11.5,
      "step": 1187
    },
    {
      "epoch": 0.19016747703943815,
      "grad_norm": 0.002169186482205987,
      "learning_rate": 2.0868335467347366e-05,
      "loss": 11.5,
      "step": 1188
    },
    {
      "epoch": 0.19032755067330973,
      "grad_norm": 0.0017040061065927148,
      "learning_rate": 2.073959935340988e-05,
      "loss": 11.5,
      "step": 1189
    },
    {
      "epoch": 0.1904876243071813,
      "grad_norm": 0.0026962559204548597,
      "learning_rate": 2.06112155968028e-05,
      "loss": 11.5,
      "step": 1190
    },
    {
      "epoch": 0.19064769794105288,
      "grad_norm": 0.0027670180425047874,
      "learning_rate": 2.0483184768264596e-05,
      "loss": 11.5,
      "step": 1191
    },
    {
      "epoch": 0.19080777157492446,
      "grad_norm": 0.006315532140433788,
      "learning_rate": 2.035550743696468e-05,
      "loss": 11.5,
      "step": 1192
    },
    {
      "epoch": 0.19096784520879603,
      "grad_norm": 0.03239641711115837,
      "learning_rate": 2.022818417050113e-05,
      "loss": 11.5,
      "step": 1193
    },
    {
      "epoch": 0.19112791884266764,
      "grad_norm": 0.03733925521373749,
      "learning_rate": 2.0101215534897855e-05,
      "loss": 11.5,
      "step": 1194
    },
    {
      "epoch": 0.19128799247653921,
      "grad_norm": 0.007610844448208809,
      "learning_rate": 1.99746020946023e-05,
      "loss": 11.5,
      "step": 1195
    },
    {
      "epoch": 0.1914480661104108,
      "grad_norm": 0.007685669232159853,
      "learning_rate": 1.9848344412482854e-05,
      "loss": 11.5,
      "step": 1196
    },
    {
      "epoch": 0.19160813974428237,
      "grad_norm": 0.006268704775720835,
      "learning_rate": 1.9722443049826344e-05,
      "loss": 11.5,
      "step": 1197
    },
    {
      "epoch": 0.19176821337815395,
      "grad_norm": 0.010067098774015903,
      "learning_rate": 1.9596898566335576e-05,
      "loss": 11.5,
      "step": 1198
    },
    {
      "epoch": 0.19192828701202552,
      "grad_norm": 0.008086483925580978,
      "learning_rate": 1.9471711520126824e-05,
      "loss": 11.5,
      "step": 1199
    },
    {
      "epoch": 0.1920883606458971,
      "grad_norm": 0.005931831896305084,
      "learning_rate": 1.9346882467727325e-05,
      "loss": 11.5,
      "step": 1200
    },
    {
      "epoch": 0.1922484342797687,
      "grad_norm": 0.0045110140927135944,
      "learning_rate": 1.9222411964072884e-05,
      "loss": 11.5,
      "step": 1201
    },
    {
      "epoch": 0.19240850791364028,
      "grad_norm": 0.004013038240373135,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 11.5,
      "step": 1202
    },
    {
      "epoch": 0.19256858154751186,
      "grad_norm": 0.0038493447937071323,
      "learning_rate": 1.8974548814769944e-05,
      "loss": 11.5,
      "step": 1203
    },
    {
      "epoch": 0.19272865518138343,
      "grad_norm": 0.0025400177109986544,
      "learning_rate": 1.8851157271013442e-05,
      "loss": 11.5,
      "step": 1204
    },
    {
      "epoch": 0.192888728815255,
      "grad_norm": 0.003950870595872402,
      "learning_rate": 1.872812647978095e-05,
      "loss": 11.5,
      "step": 1205
    },
    {
      "epoch": 0.1930488024491266,
      "grad_norm": 0.0026451449375599623,
      "learning_rate": 1.8605456988014015e-05,
      "loss": 11.5,
      "step": 1206
    },
    {
      "epoch": 0.19320887608299817,
      "grad_norm": 0.003675904357805848,
      "learning_rate": 1.8483149341047923e-05,
      "loss": 11.5,
      "step": 1207
    },
    {
      "epoch": 0.19336894971686977,
      "grad_norm": 0.0031217210926115513,
      "learning_rate": 1.8361204082609352e-05,
      "loss": 11.5,
      "step": 1208
    },
    {
      "epoch": 0.19352902335074135,
      "grad_norm": 0.006900711916387081,
      "learning_rate": 1.8239621754813995e-05,
      "loss": 11.5,
      "step": 1209
    },
    {
      "epoch": 0.19368909698461292,
      "grad_norm": 0.0030354110058397055,
      "learning_rate": 1.811840289816409e-05,
      "loss": 11.5,
      "step": 1210
    },
    {
      "epoch": 0.1938491706184845,
      "grad_norm": 0.007082383614033461,
      "learning_rate": 1.799754805154603e-05,
      "loss": 11.5,
      "step": 1211
    },
    {
      "epoch": 0.19400924425235608,
      "grad_norm": 0.00182642275467515,
      "learning_rate": 1.787705775222802e-05,
      "loss": 11.5,
      "step": 1212
    },
    {
      "epoch": 0.19416931788622765,
      "grad_norm": 0.006203912664204836,
      "learning_rate": 1.775693253585763e-05,
      "loss": 11.5,
      "step": 1213
    },
    {
      "epoch": 0.19432939152009926,
      "grad_norm": 0.006941215600818396,
      "learning_rate": 1.763717293645939e-05,
      "loss": 11.5,
      "step": 1214
    },
    {
      "epoch": 0.19448946515397084,
      "grad_norm": 0.0022211989853531122,
      "learning_rate": 1.7517779486432495e-05,
      "loss": 11.5,
      "step": 1215
    },
    {
      "epoch": 0.1946495387878424,
      "grad_norm": 0.002246611751616001,
      "learning_rate": 1.7398752716548395e-05,
      "loss": 11.5,
      "step": 1216
    },
    {
      "epoch": 0.194809612421714,
      "grad_norm": 0.004534405190497637,
      "learning_rate": 1.728009315594843e-05,
      "loss": 11.5,
      "step": 1217
    },
    {
      "epoch": 0.19496968605558557,
      "grad_norm": 0.0020383824594318867,
      "learning_rate": 1.716180133214149e-05,
      "loss": 11.5,
      "step": 1218
    },
    {
      "epoch": 0.19512975968945714,
      "grad_norm": 0.002337103709578514,
      "learning_rate": 1.704387777100165e-05,
      "loss": 11.5,
      "step": 1219
    },
    {
      "epoch": 0.19528983332332872,
      "grad_norm": 0.0020474609918892384,
      "learning_rate": 1.6926322996765897e-05,
      "loss": 11.5,
      "step": 1220
    },
    {
      "epoch": 0.19544990695720033,
      "grad_norm": 0.003318862058222294,
      "learning_rate": 1.6809137532031704e-05,
      "loss": 11.5,
      "step": 1221
    },
    {
      "epoch": 0.1956099805910719,
      "grad_norm": 0.0028475450817495584,
      "learning_rate": 1.6692321897754758e-05,
      "loss": 11.5,
      "step": 1222
    },
    {
      "epoch": 0.19577005422494348,
      "grad_norm": 0.0026161440182477236,
      "learning_rate": 1.65758766132467e-05,
      "loss": 11.5,
      "step": 1223
    },
    {
      "epoch": 0.19593012785881506,
      "grad_norm": 0.0018915985710918903,
      "learning_rate": 1.6459802196172668e-05,
      "loss": 11.5,
      "step": 1224
    },
    {
      "epoch": 0.19609020149268663,
      "grad_norm": 0.004110247828066349,
      "learning_rate": 1.634409916254914e-05,
      "loss": 11.5,
      "step": 1225
    },
    {
      "epoch": 0.1962502751265582,
      "grad_norm": 0.0049685207195580006,
      "learning_rate": 1.622876802674158e-05,
      "loss": 11.5,
      "step": 1226
    },
    {
      "epoch": 0.1964103487604298,
      "grad_norm": 0.006498860195279121,
      "learning_rate": 1.6113809301462125e-05,
      "loss": 11.5,
      "step": 1227
    },
    {
      "epoch": 0.1965704223943014,
      "grad_norm": 0.0023258759174495935,
      "learning_rate": 1.599922349776738e-05,
      "loss": 11.5,
      "step": 1228
    },
    {
      "epoch": 0.19673049602817297,
      "grad_norm": 0.0038611027412116528,
      "learning_rate": 1.5885011125056047e-05,
      "loss": 11.5,
      "step": 1229
    },
    {
      "epoch": 0.19689056966204455,
      "grad_norm": 0.0019651183392852545,
      "learning_rate": 1.5771172691066794e-05,
      "loss": 11.5,
      "step": 1230
    },
    {
      "epoch": 0.19705064329591612,
      "grad_norm": 0.0020572866778820753,
      "learning_rate": 1.565770870187585e-05,
      "loss": 11.5,
      "step": 1231
    },
    {
      "epoch": 0.1972107169297877,
      "grad_norm": 0.0029037913773208857,
      "learning_rate": 1.5544619661894864e-05,
      "loss": 11.5,
      "step": 1232
    },
    {
      "epoch": 0.19737079056365928,
      "grad_norm": 0.0018012108048424125,
      "learning_rate": 1.543190607386861e-05,
      "loss": 11.5,
      "step": 1233
    },
    {
      "epoch": 0.19753086419753085,
      "grad_norm": 0.00278186216019094,
      "learning_rate": 1.5319568438872745e-05,
      "loss": 11.5,
      "step": 1234
    },
    {
      "epoch": 0.19769093783140246,
      "grad_norm": 0.0022009124513715506,
      "learning_rate": 1.520760725631164e-05,
      "loss": 11.5,
      "step": 1235
    },
    {
      "epoch": 0.19785101146527403,
      "grad_norm": 0.003694578306749463,
      "learning_rate": 1.5096023023916094e-05,
      "loss": 11.5,
      "step": 1236
    },
    {
      "epoch": 0.1980110850991456,
      "grad_norm": 0.0024776288773864508,
      "learning_rate": 1.498481623774115e-05,
      "loss": 11.5,
      "step": 1237
    },
    {
      "epoch": 0.1981711587330172,
      "grad_norm": 0.002001419896259904,
      "learning_rate": 1.4873987392163947e-05,
      "loss": 11.5,
      "step": 1238
    },
    {
      "epoch": 0.19833123236688877,
      "grad_norm": 0.002676997799426317,
      "learning_rate": 1.4763536979881354e-05,
      "loss": 11.5,
      "step": 1239
    },
    {
      "epoch": 0.19849130600076034,
      "grad_norm": 0.001963255926966667,
      "learning_rate": 1.4653465491908003e-05,
      "loss": 11.5,
      "step": 1240
    },
    {
      "epoch": 0.19865137963463192,
      "grad_norm": 0.007502126041799784,
      "learning_rate": 1.4543773417573925e-05,
      "loss": 11.5,
      "step": 1241
    },
    {
      "epoch": 0.19881145326850352,
      "grad_norm": 0.002534693107008934,
      "learning_rate": 1.4434461244522458e-05,
      "loss": 11.5,
      "step": 1242
    },
    {
      "epoch": 0.1989715269023751,
      "grad_norm": 0.016475824639201164,
      "learning_rate": 1.4325529458708065e-05,
      "loss": 11.5,
      "step": 1243
    },
    {
      "epoch": 0.19913160053624668,
      "grad_norm": 0.021618835628032684,
      "learning_rate": 1.4216978544394177e-05,
      "loss": 11.5,
      "step": 1244
    },
    {
      "epoch": 0.19929167417011825,
      "grad_norm": 0.006856040563434362,
      "learning_rate": 1.4108808984151023e-05,
      "loss": 11.5,
      "step": 1245
    },
    {
      "epoch": 0.19945174780398983,
      "grad_norm": 0.007929130457341671,
      "learning_rate": 1.4001021258853509e-05,
      "loss": 11.5,
      "step": 1246
    },
    {
      "epoch": 0.1996118214378614,
      "grad_norm": 0.009471126832067966,
      "learning_rate": 1.3893615847679065e-05,
      "loss": 11.5,
      "step": 1247
    },
    {
      "epoch": 0.19977189507173299,
      "grad_norm": 0.009036310948431492,
      "learning_rate": 1.3786593228105494e-05,
      "loss": 11.5,
      "step": 1248
    },
    {
      "epoch": 0.1999319687056046,
      "grad_norm": 0.003158113220706582,
      "learning_rate": 1.3679953875908957e-05,
      "loss": 11.5,
      "step": 1249
    },
    {
      "epoch": 0.20009204233947617,
      "grad_norm": 0.003167160553857684,
      "learning_rate": 1.3573698265161683e-05,
      "loss": 11.5,
      "step": 1250
    },
    {
      "epoch": 0.20025211597334774,
      "grad_norm": 0.0021652011200785637,
      "learning_rate": 1.3467826868229994e-05,
      "loss": 11.5,
      "step": 1251
    },
    {
      "epoch": 0.20041218960721932,
      "grad_norm": 0.0020336401648819447,
      "learning_rate": 1.3362340155772146e-05,
      "loss": 11.5,
      "step": 1252
    },
    {
      "epoch": 0.2005722632410909,
      "grad_norm": 0.0040008570067584515,
      "learning_rate": 1.3257238596736266e-05,
      "loss": 11.5,
      "step": 1253
    },
    {
      "epoch": 0.20073233687496247,
      "grad_norm": 0.0027364951092749834,
      "learning_rate": 1.3152522658358245e-05,
      "loss": 11.5,
      "step": 1254
    },
    {
      "epoch": 0.20089241050883405,
      "grad_norm": 0.006214543711394072,
      "learning_rate": 1.3048192806159721e-05,
      "loss": 11.5,
      "step": 1255
    },
    {
      "epoch": 0.20105248414270566,
      "grad_norm": 0.003878774121403694,
      "learning_rate": 1.2944249503945894e-05,
      "loss": 11.5,
      "step": 1256
    },
    {
      "epoch": 0.20121255777657723,
      "grad_norm": 0.0018572742119431496,
      "learning_rate": 1.2840693213803545e-05,
      "loss": 11.5,
      "step": 1257
    },
    {
      "epoch": 0.2013726314104488,
      "grad_norm": 0.0018154243007302284,
      "learning_rate": 1.2737524396099032e-05,
      "loss": 11.5,
      "step": 1258
    },
    {
      "epoch": 0.2015327050443204,
      "grad_norm": 0.004351686220616102,
      "learning_rate": 1.2634743509476088e-05,
      "loss": 11.5,
      "step": 1259
    },
    {
      "epoch": 0.20169277867819196,
      "grad_norm": 0.0021896574180573225,
      "learning_rate": 1.2532351010853916e-05,
      "loss": 11.5,
      "step": 1260
    },
    {
      "epoch": 0.20185285231206354,
      "grad_norm": 0.0019422408659011126,
      "learning_rate": 1.243034735542512e-05,
      "loss": 11.5,
      "step": 1261
    },
    {
      "epoch": 0.20201292594593512,
      "grad_norm": 0.003049128921702504,
      "learning_rate": 1.2328732996653669e-05,
      "loss": 11.5,
      "step": 1262
    },
    {
      "epoch": 0.20217299957980672,
      "grad_norm": 0.0054011475294828415,
      "learning_rate": 1.2227508386272878e-05,
      "loss": 11.5,
      "step": 1263
    },
    {
      "epoch": 0.2023330732136783,
      "grad_norm": 0.0034021218307316303,
      "learning_rate": 1.212667397428342e-05,
      "loss": 11.5,
      "step": 1264
    },
    {
      "epoch": 0.20249314684754988,
      "grad_norm": 0.0020046860445290804,
      "learning_rate": 1.2026230208951306e-05,
      "loss": 11.5,
      "step": 1265
    },
    {
      "epoch": 0.20265322048142145,
      "grad_norm": 0.004417333286255598,
      "learning_rate": 1.1926177536805905e-05,
      "loss": 11.5,
      "step": 1266
    },
    {
      "epoch": 0.20281329411529303,
      "grad_norm": 0.00409281998872757,
      "learning_rate": 1.1826516402637989e-05,
      "loss": 11.5,
      "step": 1267
    },
    {
      "epoch": 0.2029733677491646,
      "grad_norm": 0.00307185435667634,
      "learning_rate": 1.1727247249497685e-05,
      "loss": 11.5,
      "step": 1268
    },
    {
      "epoch": 0.20313344138303618,
      "grad_norm": 0.0031310522463172674,
      "learning_rate": 1.1628370518692533e-05,
      "loss": 11.5,
      "step": 1269
    },
    {
      "epoch": 0.2032935150169078,
      "grad_norm": 0.002145033562555909,
      "learning_rate": 1.152988664978556e-05,
      "loss": 11.5,
      "step": 1270
    },
    {
      "epoch": 0.20345358865077937,
      "grad_norm": 0.0024463185109198093,
      "learning_rate": 1.1431796080593283e-05,
      "loss": 11.5,
      "step": 1271
    },
    {
      "epoch": 0.20361366228465094,
      "grad_norm": 0.0031195313204079866,
      "learning_rate": 1.1334099247183783e-05,
      "loss": 11.5,
      "step": 1272
    },
    {
      "epoch": 0.20377373591852252,
      "grad_norm": 0.002764628268778324,
      "learning_rate": 1.1236796583874787e-05,
      "loss": 11.5,
      "step": 1273
    },
    {
      "epoch": 0.2039338095523941,
      "grad_norm": 0.0031756593380123377,
      "learning_rate": 1.1139888523231678e-05,
      "loss": 11.5,
      "step": 1274
    },
    {
      "epoch": 0.20409388318626567,
      "grad_norm": 0.004369680769741535,
      "learning_rate": 1.1043375496065611e-05,
      "loss": 11.5,
      "step": 1275
    },
    {
      "epoch": 0.20425395682013725,
      "grad_norm": 0.002668399829417467,
      "learning_rate": 1.0947257931431642e-05,
      "loss": 11.5,
      "step": 1276
    },
    {
      "epoch": 0.20441403045400885,
      "grad_norm": 0.0035743105690926313,
      "learning_rate": 1.0851536256626705e-05,
      "loss": 11.5,
      "step": 1277
    },
    {
      "epoch": 0.20457410408788043,
      "grad_norm": 0.00246331631205976,
      "learning_rate": 1.0756210897187812e-05,
      "loss": 11.5,
      "step": 1278
    },
    {
      "epoch": 0.204734177721752,
      "grad_norm": 0.0032423525117337704,
      "learning_rate": 1.0661282276890127e-05,
      "loss": 11.5,
      "step": 1279
    },
    {
      "epoch": 0.20489425135562359,
      "grad_norm": 0.002488408237695694,
      "learning_rate": 1.0566750817745074e-05,
      "loss": 11.5,
      "step": 1280
    },
    {
      "epoch": 0.20505432498949516,
      "grad_norm": 0.0026047928258776665,
      "learning_rate": 1.0472616939998492e-05,
      "loss": 11.5,
      "step": 1281
    },
    {
      "epoch": 0.20521439862336674,
      "grad_norm": 0.0033032901119440794,
      "learning_rate": 1.0378881062128731e-05,
      "loss": 11.5,
      "step": 1282
    },
    {
      "epoch": 0.20537447225723832,
      "grad_norm": 0.001280674827285111,
      "learning_rate": 1.0285543600844804e-05,
      "loss": 11.5,
      "step": 1283
    },
    {
      "epoch": 0.20553454589110992,
      "grad_norm": 0.002876692684367299,
      "learning_rate": 1.019260497108453e-05,
      "loss": 11.5,
      "step": 1284
    },
    {
      "epoch": 0.2056946195249815,
      "grad_norm": 0.0041841669008135796,
      "learning_rate": 1.010006558601274e-05,
      "loss": 11.5,
      "step": 1285
    },
    {
      "epoch": 0.20585469315885307,
      "grad_norm": 0.0027890666387975216,
      "learning_rate": 1.000792585701934e-05,
      "loss": 11.5,
      "step": 1286
    },
    {
      "epoch": 0.20601476679272465,
      "grad_norm": 0.0034449852537363768,
      "learning_rate": 9.91618619371757e-06,
      "loss": 11.5,
      "step": 1287
    },
    {
      "epoch": 0.20617484042659623,
      "grad_norm": 0.002114606788381934,
      "learning_rate": 9.82484700394215e-06,
      "loss": 11.5,
      "step": 1288
    },
    {
      "epoch": 0.2063349140604678,
      "grad_norm": 0.0029558613896369934,
      "learning_rate": 9.73390869374743e-06,
      "loss": 11.5,
      "step": 1289
    },
    {
      "epoch": 0.20649498769433938,
      "grad_norm": 0.002221523318439722,
      "learning_rate": 9.643371667405698e-06,
      "loss": 11.5,
      "step": 1290
    },
    {
      "epoch": 0.206655061328211,
      "grad_norm": 0.0027887243777513504,
      "learning_rate": 9.553236327405246e-06,
      "loss": 11.5,
      "step": 1291
    },
    {
      "epoch": 0.20681513496208256,
      "grad_norm": 0.002874466124922037,
      "learning_rate": 9.463503074448677e-06,
      "loss": 11.5,
      "step": 1292
    },
    {
      "epoch": 0.20697520859595414,
      "grad_norm": 0.02173793874680996,
      "learning_rate": 9.374172307451068e-06,
      "loss": 11.5,
      "step": 1293
    },
    {
      "epoch": 0.20713528222982572,
      "grad_norm": 0.02837592177093029,
      "learning_rate": 9.285244423538197e-06,
      "loss": 11.5,
      "step": 1294
    },
    {
      "epoch": 0.2072953558636973,
      "grad_norm": 0.008146749809384346,
      "learning_rate": 9.196719818044886e-06,
      "loss": 11.5,
      "step": 1295
    },
    {
      "epoch": 0.20745542949756887,
      "grad_norm": 0.007599384989589453,
      "learning_rate": 9.108598884513053e-06,
      "loss": 11.5,
      "step": 1296
    },
    {
      "epoch": 0.20761550313144048,
      "grad_norm": 0.00396609166637063,
      "learning_rate": 9.020882014690136e-06,
      "loss": 11.5,
      "step": 1297
    },
    {
      "epoch": 0.20777557676531205,
      "grad_norm": 0.010361498221755028,
      "learning_rate": 8.933569598527247e-06,
      "loss": 11.5,
      "step": 1298
    },
    {
      "epoch": 0.20793565039918363,
      "grad_norm": 0.004484712146222591,
      "learning_rate": 8.846662024177477e-06,
      "loss": 11.5,
      "step": 1299
    },
    {
      "epoch": 0.2080957240330552,
      "grad_norm": 0.0075991349294781685,
      "learning_rate": 8.760159677994172e-06,
      "loss": 11.5,
      "step": 1300
    },
    {
      "epoch": 0.20825579766692678,
      "grad_norm": 0.005831120070070028,
      "learning_rate": 8.674062944529216e-06,
      "loss": 11.5,
      "step": 1301
    },
    {
      "epoch": 0.20841587130079836,
      "grad_norm": 0.004072224721312523,
      "learning_rate": 8.588372206531292e-06,
      "loss": 11.5,
      "step": 1302
    },
    {
      "epoch": 0.20857594493466994,
      "grad_norm": 0.007944183424115181,
      "learning_rate": 8.503087844944213e-06,
      "loss": 11.5,
      "step": 1303
    },
    {
      "epoch": 0.20873601856854154,
      "grad_norm": 0.0033783623948693275,
      "learning_rate": 8.418210238905256e-06,
      "loss": 11.5,
      "step": 1304
    },
    {
      "epoch": 0.20889609220241312,
      "grad_norm": 0.0015663927188143134,
      "learning_rate": 8.333739765743398e-06,
      "loss": 11.5,
      "step": 1305
    },
    {
      "epoch": 0.2090561658362847,
      "grad_norm": 0.002214925130829215,
      "learning_rate": 8.249676800977658e-06,
      "loss": 11.5,
      "step": 1306
    },
    {
      "epoch": 0.20921623947015627,
      "grad_norm": 0.005886987783014774,
      "learning_rate": 8.16602171831553e-06,
      "loss": 11.5,
      "step": 1307
    },
    {
      "epoch": 0.20937631310402785,
      "grad_norm": 0.002971059875562787,
      "learning_rate": 8.082774889651168e-06,
      "loss": 11.5,
      "step": 1308
    },
    {
      "epoch": 0.20953638673789943,
      "grad_norm": 0.006135622970759869,
      "learning_rate": 7.999936685063835e-06,
      "loss": 11.5,
      "step": 1309
    },
    {
      "epoch": 0.209696460371771,
      "grad_norm": 0.004374191164970398,
      "learning_rate": 7.91750747281621e-06,
      "loss": 11.5,
      "step": 1310
    },
    {
      "epoch": 0.2098565340056426,
      "grad_norm": 0.003110012039542198,
      "learning_rate": 7.835487619352811e-06,
      "loss": 11.5,
      "step": 1311
    },
    {
      "epoch": 0.21001660763951419,
      "grad_norm": 0.0021591773256659508,
      "learning_rate": 7.753877489298244e-06,
      "loss": 11.5,
      "step": 1312
    },
    {
      "epoch": 0.21017668127338576,
      "grad_norm": 0.0021113939583301544,
      "learning_rate": 7.67267744545579e-06,
      "loss": 11.5,
      "step": 1313
    },
    {
      "epoch": 0.21033675490725734,
      "grad_norm": 0.005725356284528971,
      "learning_rate": 7.591887848805545e-06,
      "loss": 11.5,
      "step": 1314
    },
    {
      "epoch": 0.21049682854112892,
      "grad_norm": 0.005279659293591976,
      "learning_rate": 7.5115090585029966e-06,
      "loss": 11.5,
      "step": 1315
    },
    {
      "epoch": 0.2106569021750005,
      "grad_norm": 0.0030834954231977463,
      "learning_rate": 7.431541431877342e-06,
      "loss": 11.5,
      "step": 1316
    },
    {
      "epoch": 0.21081697580887207,
      "grad_norm": 0.002347782487049699,
      "learning_rate": 7.351985324429933e-06,
      "loss": 11.5,
      "step": 1317
    },
    {
      "epoch": 0.21097704944274367,
      "grad_norm": 0.004431252367794514,
      "learning_rate": 7.272841089832694e-06,
      "loss": 11.5,
      "step": 1318
    },
    {
      "epoch": 0.21113712307661525,
      "grad_norm": 0.0027293700259178877,
      "learning_rate": 7.194109079926514e-06,
      "loss": 11.5,
      "step": 1319
    },
    {
      "epoch": 0.21129719671048683,
      "grad_norm": 0.004201219882816076,
      "learning_rate": 7.115789644719728e-06,
      "loss": 11.5,
      "step": 1320
    },
    {
      "epoch": 0.2114572703443584,
      "grad_norm": 0.0030196679290384054,
      "learning_rate": 7.037883132386547e-06,
      "loss": 11.5,
      "step": 1321
    },
    {
      "epoch": 0.21161734397822998,
      "grad_norm": 0.0053329793736338615,
      "learning_rate": 6.960389889265517e-06,
      "loss": 11.5,
      "step": 1322
    },
    {
      "epoch": 0.21177741761210156,
      "grad_norm": 0.003384033450856805,
      "learning_rate": 6.883310259857944e-06,
      "loss": 11.5,
      "step": 1323
    },
    {
      "epoch": 0.21193749124597314,
      "grad_norm": 0.006386162247508764,
      "learning_rate": 6.806644586826383e-06,
      "loss": 11.5,
      "step": 1324
    },
    {
      "epoch": 0.21209756487984474,
      "grad_norm": 0.0023678236175328493,
      "learning_rate": 6.730393210993147e-06,
      "loss": 11.5,
      "step": 1325
    },
    {
      "epoch": 0.21225763851371632,
      "grad_norm": 0.00347006693482399,
      "learning_rate": 6.654556471338746e-06,
      "loss": 11.5,
      "step": 1326
    },
    {
      "epoch": 0.2124177121475879,
      "grad_norm": 0.003324992023408413,
      "learning_rate": 6.579134705000412e-06,
      "loss": 11.5,
      "step": 1327
    },
    {
      "epoch": 0.21257778578145947,
      "grad_norm": 0.0045981016010046005,
      "learning_rate": 6.504128247270546e-06,
      "loss": 11.5,
      "step": 1328
    },
    {
      "epoch": 0.21273785941533105,
      "grad_norm": 0.007631343323737383,
      "learning_rate": 6.429537431595312e-06,
      "loss": 11.5,
      "step": 1329
    },
    {
      "epoch": 0.21289793304920263,
      "grad_norm": 0.0033836376387625933,
      "learning_rate": 6.355362589573077e-06,
      "loss": 11.5,
      "step": 1330
    },
    {
      "epoch": 0.2130580066830742,
      "grad_norm": 0.002158495131880045,
      "learning_rate": 6.2816040509530165e-06,
      "loss": 11.5,
      "step": 1331
    },
    {
      "epoch": 0.2132180803169458,
      "grad_norm": 0.00441649928689003,
      "learning_rate": 6.2082621436335475e-06,
      "loss": 11.5,
      "step": 1332
    },
    {
      "epoch": 0.21337815395081738,
      "grad_norm": 0.001920916954986751,
      "learning_rate": 6.135337193660962e-06,
      "loss": 11.5,
      "step": 1333
    },
    {
      "epoch": 0.21353822758468896,
      "grad_norm": 0.0026197098195552826,
      "learning_rate": 6.062829525227909e-06,
      "loss": 11.5,
      "step": 1334
    },
    {
      "epoch": 0.21369830121856054,
      "grad_norm": 0.004652595147490501,
      "learning_rate": 5.990739460672024e-06,
      "loss": 11.5,
      "step": 1335
    },
    {
      "epoch": 0.21385837485243211,
      "grad_norm": 0.002218803623691201,
      "learning_rate": 5.9190673204744255e-06,
      "loss": 11.5,
      "step": 1336
    },
    {
      "epoch": 0.2140184484863037,
      "grad_norm": 0.0023052345495671034,
      "learning_rate": 5.84781342325833e-06,
      "loss": 11.5,
      "step": 1337
    },
    {
      "epoch": 0.21417852212017527,
      "grad_norm": 0.003893439657986164,
      "learning_rate": 5.77697808578761e-06,
      "loss": 11.5,
      "step": 1338
    },
    {
      "epoch": 0.21433859575404687,
      "grad_norm": 0.0009984035277739167,
      "learning_rate": 5.706561622965467e-06,
      "loss": 11.5,
      "step": 1339
    },
    {
      "epoch": 0.21449866938791845,
      "grad_norm": 0.0017044260166585445,
      "learning_rate": 5.636564347832907e-06,
      "loss": 11.5,
      "step": 1340
    },
    {
      "epoch": 0.21465874302179003,
      "grad_norm": 0.010417683981359005,
      "learning_rate": 5.566986571567401e-06,
      "loss": 11.5,
      "step": 1341
    },
    {
      "epoch": 0.2148188166556616,
      "grad_norm": 0.01849612034857273,
      "learning_rate": 5.497828603481569e-06,
      "loss": 11.5,
      "step": 1342
    },
    {
      "epoch": 0.21497889028953318,
      "grad_norm": 0.02373194694519043,
      "learning_rate": 5.429090751021704e-06,
      "loss": 11.5,
      "step": 1343
    },
    {
      "epoch": 0.21513896392340476,
      "grad_norm": 0.011821235530078411,
      "learning_rate": 5.3607733197664436e-06,
      "loss": 11.5,
      "step": 1344
    },
    {
      "epoch": 0.21529903755727633,
      "grad_norm": 0.0074108256958425045,
      "learning_rate": 5.2928766134254345e-06,
      "loss": 11.5,
      "step": 1345
    },
    {
      "epoch": 0.21545911119114794,
      "grad_norm": 0.005540342070162296,
      "learning_rate": 5.225400933837954e-06,
      "loss": 11.5,
      "step": 1346
    },
    {
      "epoch": 0.21561918482501952,
      "grad_norm": 0.009911085478961468,
      "learning_rate": 5.158346580971573e-06,
      "loss": 11.5,
      "step": 1347
    },
    {
      "epoch": 0.2157792584588911,
      "grad_norm": 0.007390840910375118,
      "learning_rate": 5.091713852920854e-06,
      "loss": 11.5,
      "step": 1348
    },
    {
      "epoch": 0.21593933209276267,
      "grad_norm": 0.0036254727747291327,
      "learning_rate": 5.025503045905933e-06,
      "loss": 11.5,
      "step": 1349
    },
    {
      "epoch": 0.21609940572663425,
      "grad_norm": 0.00854894146323204,
      "learning_rate": 4.959714454271369e-06,
      "loss": 11.5,
      "step": 1350
    },
    {
      "epoch": 0.21625947936050582,
      "grad_norm": 0.0033870558254420757,
      "learning_rate": 4.8943483704846475e-06,
      "loss": 11.5,
      "step": 1351
    },
    {
      "epoch": 0.2164195529943774,
      "grad_norm": 0.005913980770856142,
      "learning_rate": 4.829405085134997e-06,
      "loss": 11.5,
      "step": 1352
    },
    {
      "epoch": 0.216579626628249,
      "grad_norm": 0.007272148970514536,
      "learning_rate": 4.764884886932086e-06,
      "loss": 11.5,
      "step": 1353
    },
    {
      "epoch": 0.21673970026212058,
      "grad_norm": 0.0030211645644158125,
      "learning_rate": 4.700788062704687e-06,
      "loss": 11.5,
      "step": 1354
    },
    {
      "epoch": 0.21689977389599216,
      "grad_norm": 0.0013875890290364623,
      "learning_rate": 4.6371148973994525e-06,
      "loss": 11.5,
      "step": 1355
    },
    {
      "epoch": 0.21705984752986374,
      "grad_norm": 0.002735642483457923,
      "learning_rate": 4.573865674079625e-06,
      "loss": 11.5,
      "step": 1356
    },
    {
      "epoch": 0.2172199211637353,
      "grad_norm": 0.002947563771158457,
      "learning_rate": 4.511040673923828e-06,
      "loss": 11.5,
      "step": 1357
    },
    {
      "epoch": 0.2173799947976069,
      "grad_norm": 0.0042845094576478004,
      "learning_rate": 4.448640176224694e-06,
      "loss": 11.5,
      "step": 1358
    },
    {
      "epoch": 0.21754006843147847,
      "grad_norm": 0.004379062447696924,
      "learning_rate": 4.386664458387779e-06,
      "loss": 11.5,
      "step": 1359
    },
    {
      "epoch": 0.21770014206535007,
      "grad_norm": 0.0032331692054867744,
      "learning_rate": 4.325113795930203e-06,
      "loss": 11.5,
      "step": 1360
    },
    {
      "epoch": 0.21786021569922165,
      "grad_norm": 0.0020940350368618965,
      "learning_rate": 4.263988462479484e-06,
      "loss": 11.5,
      "step": 1361
    },
    {
      "epoch": 0.21802028933309323,
      "grad_norm": 0.0035770414397120476,
      "learning_rate": 4.203288729772326e-06,
      "loss": 11.5,
      "step": 1362
    },
    {
      "epoch": 0.2181803629669648,
      "grad_norm": 0.0055876183323562145,
      "learning_rate": 4.143014867653383e-06,
      "loss": 11.5,
      "step": 1363
    },
    {
      "epoch": 0.21834043660083638,
      "grad_norm": 0.005337531678378582,
      "learning_rate": 4.083167144074073e-06,
      "loss": 11.5,
      "step": 1364
    },
    {
      "epoch": 0.21850051023470796,
      "grad_norm": 0.004889491945505142,
      "learning_rate": 4.023745825091407e-06,
      "loss": 11.5,
      "step": 1365
    },
    {
      "epoch": 0.21866058386857953,
      "grad_norm": 0.002258522668853402,
      "learning_rate": 3.964751174866765e-06,
      "loss": 11.5,
      "step": 1366
    },
    {
      "epoch": 0.21882065750245114,
      "grad_norm": 0.0026405653916299343,
      "learning_rate": 3.906183455664725e-06,
      "loss": 11.5,
      "step": 1367
    },
    {
      "epoch": 0.21898073113632271,
      "grad_norm": 0.006426986772567034,
      "learning_rate": 3.84804292785198e-06,
      "loss": 11.5,
      "step": 1368
    },
    {
      "epoch": 0.2191408047701943,
      "grad_norm": 0.003312518587335944,
      "learning_rate": 3.7903298498960572e-06,
      "loss": 11.5,
      "step": 1369
    },
    {
      "epoch": 0.21930087840406587,
      "grad_norm": 0.0017910711467266083,
      "learning_rate": 3.7330444783642338e-06,
      "loss": 11.5,
      "step": 1370
    },
    {
      "epoch": 0.21946095203793745,
      "grad_norm": 0.0036273514851927757,
      "learning_rate": 3.676187067922421e-06,
      "loss": 11.5,
      "step": 1371
    },
    {
      "epoch": 0.21962102567180902,
      "grad_norm": 0.0038595781661570072,
      "learning_rate": 3.619757871333973e-06,
      "loss": 11.5,
      "step": 1372
    },
    {
      "epoch": 0.2197810993056806,
      "grad_norm": 0.00221767439506948,
      "learning_rate": 3.563757139458579e-06,
      "loss": 11.5,
      "step": 1373
    },
    {
      "epoch": 0.2199411729395522,
      "grad_norm": 0.0021597740706056356,
      "learning_rate": 3.5081851212512175e-06,
      "loss": 11.5,
      "step": 1374
    },
    {
      "epoch": 0.22010124657342378,
      "grad_norm": 0.002075984375551343,
      "learning_rate": 3.4530420637609363e-06,
      "loss": 11.5,
      "step": 1375
    },
    {
      "epoch": 0.22026132020729536,
      "grad_norm": 0.0031187019776552916,
      "learning_rate": 3.3983282121298086e-06,
      "loss": 11.5,
      "step": 1376
    },
    {
      "epoch": 0.22042139384116693,
      "grad_norm": 0.004900255240499973,
      "learning_rate": 3.3440438095919126e-06,
      "loss": 11.5,
      "step": 1377
    },
    {
      "epoch": 0.2205814674750385,
      "grad_norm": 0.002167940605431795,
      "learning_rate": 3.290189097472096e-06,
      "loss": 11.5,
      "step": 1378
    },
    {
      "epoch": 0.2207415411089101,
      "grad_norm": 0.002858449937775731,
      "learning_rate": 3.236764315185037e-06,
      "loss": 11.5,
      "step": 1379
    },
    {
      "epoch": 0.2209016147427817,
      "grad_norm": 0.0024274946190416813,
      "learning_rate": 3.1837697002341293e-06,
      "loss": 11.5,
      "step": 1380
    },
    {
      "epoch": 0.22106168837665327,
      "grad_norm": 0.0024481264408677816,
      "learning_rate": 3.131205488210409e-06,
      "loss": 11.5,
      "step": 1381
    },
    {
      "epoch": 0.22122176201052485,
      "grad_norm": 0.0024221399798989296,
      "learning_rate": 3.0790719127915646e-06,
      "loss": 11.5,
      "step": 1382
    },
    {
      "epoch": 0.22138183564439642,
      "grad_norm": 0.0027032729703933,
      "learning_rate": 3.0273692057408265e-06,
      "loss": 11.5,
      "step": 1383
    },
    {
      "epoch": 0.221541909278268,
      "grad_norm": 0.003121505258604884,
      "learning_rate": 2.976097596905969e-06,
      "loss": 11.5,
      "step": 1384
    },
    {
      "epoch": 0.22170198291213958,
      "grad_norm": 0.002749078441411257,
      "learning_rate": 2.9252573142183326e-06,
      "loss": 11.5,
      "step": 1385
    },
    {
      "epoch": 0.22186205654601115,
      "grad_norm": 0.005534507799893618,
      "learning_rate": 2.874848583691714e-06,
      "loss": 11.5,
      "step": 1386
    },
    {
      "epoch": 0.22202213017988276,
      "grad_norm": 0.002249793615192175,
      "learning_rate": 2.8248716294214774e-06,
      "loss": 11.5,
      "step": 1387
    },
    {
      "epoch": 0.22218220381375434,
      "grad_norm": 0.002555784536525607,
      "learning_rate": 2.7753266735834338e-06,
      "loss": 11.5,
      "step": 1388
    },
    {
      "epoch": 0.2223422774476259,
      "grad_norm": 0.005491218063980341,
      "learning_rate": 2.7262139364329643e-06,
      "loss": 11.5,
      "step": 1389
    },
    {
      "epoch": 0.2225023510814975,
      "grad_norm": 0.003095269901677966,
      "learning_rate": 2.677533636303964e-06,
      "loss": 11.5,
      "step": 1390
    },
    {
      "epoch": 0.22266242471536907,
      "grad_norm": 0.0014023822732269764,
      "learning_rate": 2.6292859896079213e-06,
      "loss": 11.5,
      "step": 1391
    },
    {
      "epoch": 0.22282249834924064,
      "grad_norm": 0.004665051586925983,
      "learning_rate": 2.581471210832931e-06,
      "loss": 11.5,
      "step": 1392
    },
    {
      "epoch": 0.22298257198311222,
      "grad_norm": 0.02085324190557003,
      "learning_rate": 2.5340895125427364e-06,
      "loss": 11.5,
      "step": 1393
    },
    {
      "epoch": 0.22314264561698383,
      "grad_norm": 0.0144707215949893,
      "learning_rate": 2.4871411053757898e-06,
      "loss": 11.5,
      "step": 1394
    },
    {
      "epoch": 0.2233027192508554,
      "grad_norm": 0.01008582953363657,
      "learning_rate": 2.440626198044327e-06,
      "loss": 11.5,
      "step": 1395
    },
    {
      "epoch": 0.22346279288472698,
      "grad_norm": 0.006674542091786861,
      "learning_rate": 2.394544997333437e-06,
      "loss": 11.5,
      "step": 1396
    },
    {
      "epoch": 0.22362286651859856,
      "grad_norm": 0.007663378491997719,
      "learning_rate": 2.3488977081001394e-06,
      "loss": 11.5,
      "step": 1397
    },
    {
      "epoch": 0.22378294015247013,
      "grad_norm": 0.0060112918727099895,
      "learning_rate": 2.3036845332724543e-06,
      "loss": 11.5,
      "step": 1398
    },
    {
      "epoch": 0.2239430137863417,
      "grad_norm": 0.00751597760245204,
      "learning_rate": 2.2589056738485324e-06,
      "loss": 11.5,
      "step": 1399
    },
    {
      "epoch": 0.2241030874202133,
      "grad_norm": 0.005465385038405657,
      "learning_rate": 2.2145613288957478e-06,
      "loss": 11.5,
      "step": 1400
    },
    {
      "epoch": 0.2242631610540849,
      "grad_norm": 0.002556642983108759,
      "learning_rate": 2.170651695549786e-06,
      "loss": 11.5,
      "step": 1401
    },
    {
      "epoch": 0.22442323468795647,
      "grad_norm": 0.0032211760990321636,
      "learning_rate": 2.1271769690138332e-06,
      "loss": 11.5,
      "step": 1402
    },
    {
      "epoch": 0.22458330832182805,
      "grad_norm": 0.0027126914355903864,
      "learning_rate": 2.084137342557646e-06,
      "loss": 11.5,
      "step": 1403
    },
    {
      "epoch": 0.22474338195569962,
      "grad_norm": 0.0041853804141283035,
      "learning_rate": 2.0415330075166937e-06,
      "loss": 11.5,
      "step": 1404
    },
    {
      "epoch": 0.2249034555895712,
      "grad_norm": 0.00473013473674655,
      "learning_rate": 1.9993641532913833e-06,
      "loss": 11.5,
      "step": 1405
    },
    {
      "epoch": 0.22506352922344278,
      "grad_norm": 0.0026882446836680174,
      "learning_rate": 1.9576309673461357e-06,
      "loss": 11.5,
      "step": 1406
    },
    {
      "epoch": 0.22522360285731435,
      "grad_norm": 0.0030842390842735767,
      "learning_rate": 1.916333635208556e-06,
      "loss": 11.5,
      "step": 1407
    },
    {
      "epoch": 0.22538367649118596,
      "grad_norm": 0.0039044285658746958,
      "learning_rate": 1.8754723404686425e-06,
      "loss": 11.5,
      "step": 1408
    },
    {
      "epoch": 0.22554375012505753,
      "grad_norm": 0.0030411509796977043,
      "learning_rate": 1.8350472647780116e-06,
      "loss": 11.5,
      "step": 1409
    },
    {
      "epoch": 0.2257038237589291,
      "grad_norm": 0.0017332800198346376,
      "learning_rate": 1.7950585878489856e-06,
      "loss": 11.5,
      "step": 1410
    },
    {
      "epoch": 0.2258638973928007,
      "grad_norm": 0.0015333160990849137,
      "learning_rate": 1.7555064874538397e-06,
      "loss": 11.5,
      "step": 1411
    },
    {
      "epoch": 0.22602397102667227,
      "grad_norm": 0.0015668972628191113,
      "learning_rate": 1.7163911394240672e-06,
      "loss": 11.5,
      "step": 1412
    },
    {
      "epoch": 0.22618404466054384,
      "grad_norm": 0.0032179735135287046,
      "learning_rate": 1.6777127176495043e-06,
      "loss": 11.5,
      "step": 1413
    },
    {
      "epoch": 0.22634411829441542,
      "grad_norm": 0.002921745413914323,
      "learning_rate": 1.6394713940776296e-06,
      "loss": 11.5,
      "step": 1414
    },
    {
      "epoch": 0.22650419192828702,
      "grad_norm": 0.0026629441417753696,
      "learning_rate": 1.6016673387127646e-06,
      "loss": 11.5,
      "step": 1415
    },
    {
      "epoch": 0.2266642655621586,
      "grad_norm": 0.003579874522984028,
      "learning_rate": 1.5643007196153302e-06,
      "loss": 11.5,
      "step": 1416
    },
    {
      "epoch": 0.22682433919603018,
      "grad_norm": 0.0021721201483160257,
      "learning_rate": 1.5273717029010925e-06,
      "loss": 11.5,
      "step": 1417
    },
    {
      "epoch": 0.22698441282990175,
      "grad_norm": 0.002987267915159464,
      "learning_rate": 1.4908804527404286e-06,
      "loss": 11.5,
      "step": 1418
    },
    {
      "epoch": 0.22714448646377333,
      "grad_norm": 0.005420115776360035,
      "learning_rate": 1.4548271313575835e-06,
      "loss": 11.5,
      "step": 1419
    },
    {
      "epoch": 0.2273045600976449,
      "grad_norm": 0.005826962646096945,
      "learning_rate": 1.4192118990299707e-06,
      "loss": 11.5,
      "step": 1420
    },
    {
      "epoch": 0.22746463373151649,
      "grad_norm": 0.002273988677188754,
      "learning_rate": 1.3840349140874619e-06,
      "loss": 11.5,
      "step": 1421
    },
    {
      "epoch": 0.2276247073653881,
      "grad_norm": 0.0024928958155214787,
      "learning_rate": 1.3492963329116537e-06,
      "loss": 11.5,
      "step": 1422
    },
    {
      "epoch": 0.22778478099925967,
      "grad_norm": 0.004260348156094551,
      "learning_rate": 1.3149963099352014e-06,
      "loss": 11.5,
      "step": 1423
    },
    {
      "epoch": 0.22794485463313124,
      "grad_norm": 0.004182882606983185,
      "learning_rate": 1.2811349976411202e-06,
      "loss": 11.5,
      "step": 1424
    },
    {
      "epoch": 0.22810492826700282,
      "grad_norm": 0.003101121634244919,
      "learning_rate": 1.2477125465620853e-06,
      "loss": 11.5,
      "step": 1425
    },
    {
      "epoch": 0.2282650019008744,
      "grad_norm": 0.00238924752920866,
      "learning_rate": 1.2147291052798216e-06,
      "loss": 11.5,
      "step": 1426
    },
    {
      "epoch": 0.22842507553474597,
      "grad_norm": 0.0032329584937542677,
      "learning_rate": 1.1821848204243814e-06,
      "loss": 11.5,
      "step": 1427
    },
    {
      "epoch": 0.22858514916861755,
      "grad_norm": 0.005832359194755554,
      "learning_rate": 1.1500798366735233e-06,
      "loss": 11.5,
      "step": 1428
    },
    {
      "epoch": 0.22874522280248916,
      "grad_norm": 0.004793527070432901,
      "learning_rate": 1.1184142967520794e-06,
      "loss": 11.5,
      "step": 1429
    },
    {
      "epoch": 0.22890529643636073,
      "grad_norm": 0.002899014391005039,
      "learning_rate": 1.0871883414312777e-06,
      "loss": 11.5,
      "step": 1430
    },
    {
      "epoch": 0.2290653700702323,
      "grad_norm": 0.0017772008432075381,
      "learning_rate": 1.0564021095281652e-06,
      "loss": 11.5,
      "step": 1431
    },
    {
      "epoch": 0.2292254437041039,
      "grad_norm": 0.004641447216272354,
      "learning_rate": 1.0260557379049519e-06,
      "loss": 11.5,
      "step": 1432
    },
    {
      "epoch": 0.22938551733797546,
      "grad_norm": 0.004166705533862114,
      "learning_rate": 9.96149361468457e-07,
      "loss": 11.5,
      "step": 1433
    },
    {
      "epoch": 0.22954559097184704,
      "grad_norm": 0.0020900522358715534,
      "learning_rate": 9.66683113169431e-07,
      "loss": 11.5,
      "step": 1434
    },
    {
      "epoch": 0.22970566460571862,
      "grad_norm": 0.002282516798004508,
      "learning_rate": 9.376571240020227e-07,
      "loss": 11.5,
      "step": 1435
    },
    {
      "epoch": 0.22986573823959022,
      "grad_norm": 0.003710755379870534,
      "learning_rate": 9.090715230031688e-07,
      "loss": 11.5,
      "step": 1436
    },
    {
      "epoch": 0.2300258118734618,
      "grad_norm": 0.0018695998005568981,
      "learning_rate": 8.809264372520609e-07,
      "loss": 11.5,
      "step": 1437
    },
    {
      "epoch": 0.23018588550733338,
      "grad_norm": 0.0014625567710027099,
      "learning_rate": 8.532219918695128e-07,
      "loss": 11.5,
      "step": 1438
    },
    {
      "epoch": 0.23034595914120495,
      "grad_norm": 0.004999818746000528,
      "learning_rate": 8.259583100174606e-07,
      "loss": 11.5,
      "step": 1439
    },
    {
      "epoch": 0.23050603277507653,
      "grad_norm": 0.002869151532649994,
      "learning_rate": 7.991355128984079e-07,
      "loss": 11.5,
      "step": 1440
    },
    {
      "epoch": 0.2306661064089481,
      "grad_norm": 0.0023289695382118225,
      "learning_rate": 7.727537197548707e-07,
      "loss": 11.5,
      "step": 1441
    },
    {
      "epoch": 0.23082618004281968,
      "grad_norm": 0.0017972509376704693,
      "learning_rate": 7.468130478688218e-07,
      "loss": 11.5,
      "step": 1442
    },
    {
      "epoch": 0.2309862536766913,
      "grad_norm": 0.044782571494579315,
      "learning_rate": 7.213136125612586e-07,
      "loss": 11.5,
      "step": 1443
    },
    {
      "epoch": 0.23114632731056287,
      "grad_norm": 0.011034706607460976,
      "learning_rate": 6.962555271915805e-07,
      "loss": 11.5,
      "step": 1444
    },
    {
      "epoch": 0.23130640094443444,
      "grad_norm": 0.005636899266391993,
      "learning_rate": 6.716389031571568e-07,
      "loss": 11.5,
      "step": 1445
    },
    {
      "epoch": 0.23146647457830602,
      "grad_norm": 0.012516049668192863,
      "learning_rate": 6.474638498928265e-07,
      "loss": 11.5,
      "step": 1446
    },
    {
      "epoch": 0.2316265482121776,
      "grad_norm": 0.006045103073120117,
      "learning_rate": 6.237304748703543e-07,
      "loss": 11.5,
      "step": 1447
    },
    {
      "epoch": 0.23178662184604917,
      "grad_norm": 0.00623738020658493,
      "learning_rate": 6.004388835980423e-07,
      "loss": 11.5,
      "step": 1448
    },
    {
      "epoch": 0.23194669547992075,
      "grad_norm": 0.0038605364970862865,
      "learning_rate": 5.77589179620186e-07,
      "loss": 11.5,
      "step": 1449
    },
    {
      "epoch": 0.23210676911379235,
      "grad_norm": 0.007407385855913162,
      "learning_rate": 5.55181464516652e-07,
      "loss": 11.5,
      "step": 1450
    },
    {
      "epoch": 0.23226684274766393,
      "grad_norm": 0.0058756982907652855,
      "learning_rate": 5.332158379024122e-07,
      "loss": 11.5,
      "step": 1451
    },
    {
      "epoch": 0.2324269163815355,
      "grad_norm": 0.003662444418296218,
      "learning_rate": 5.116923974270993e-07,
      "loss": 11.5,
      "step": 1452
    },
    {
      "epoch": 0.23258699001540709,
      "grad_norm": 0.0038290186785161495,
      "learning_rate": 4.906112387745965e-07,
      "loss": 11.5,
      "step": 1453
    },
    {
      "epoch": 0.23274706364927866,
      "grad_norm": 0.003163130022585392,
      "learning_rate": 4.6997245566257064e-07,
      "loss": 11.5,
      "step": 1454
    },
    {
      "epoch": 0.23290713728315024,
      "grad_norm": 0.002829737262800336,
      "learning_rate": 4.497761398421063e-07,
      "loss": 11.5,
      "step": 1455
    },
    {
      "epoch": 0.23306721091702182,
      "grad_norm": 0.00424747820943594,
      "learning_rate": 4.3002238109723927e-07,
      "loss": 11.5,
      "step": 1456
    },
    {
      "epoch": 0.23322728455089342,
      "grad_norm": 0.003198497463017702,
      "learning_rate": 4.107112672446123e-07,
      "loss": 11.5,
      "step": 1457
    },
    {
      "epoch": 0.233387358184765,
      "grad_norm": 0.003337225643917918,
      "learning_rate": 3.9184288413306456e-07,
      "loss": 11.5,
      "step": 1458
    },
    {
      "epoch": 0.23354743181863657,
      "grad_norm": 0.004608597606420517,
      "learning_rate": 3.734173156432208e-07,
      "loss": 11.5,
      "step": 1459
    },
    {
      "epoch": 0.23370750545250815,
      "grad_norm": 0.005992481950670481,
      "learning_rate": 3.554346436871581e-07,
      "loss": 11.5,
      "step": 1460
    },
    {
      "epoch": 0.23386757908637973,
      "grad_norm": 0.0015647500986233354,
      "learning_rate": 3.3789494820803957e-07,
      "loss": 11.5,
      "step": 1461
    },
    {
      "epoch": 0.2340276527202513,
      "grad_norm": 0.00299090170301497,
      "learning_rate": 3.2079830717972606e-07,
      "loss": 11.5,
      "step": 1462
    },
    {
      "epoch": 0.2341877263541229,
      "grad_norm": 0.0028376192785799503,
      "learning_rate": 3.041447966064648e-07,
      "loss": 11.5,
      "step": 1463
    },
    {
      "epoch": 0.2343477999879945,
      "grad_norm": 0.002717991592362523,
      "learning_rate": 2.8793449052254563e-07,
      "loss": 11.5,
      "step": 1464
    },
    {
      "epoch": 0.23450787362186606,
      "grad_norm": 0.0019830649252980947,
      "learning_rate": 2.721674609919345e-07,
      "loss": 11.5,
      "step": 1465
    },
    {
      "epoch": 0.23466794725573764,
      "grad_norm": 0.0012316781794652343,
      "learning_rate": 2.568437781080069e-07,
      "loss": 11.5,
      "step": 1466
    },
    {
      "epoch": 0.23482802088960922,
      "grad_norm": 0.0031201078090816736,
      "learning_rate": 2.4196350999320384e-07,
      "loss": 11.5,
      "step": 1467
    },
    {
      "epoch": 0.2349880945234808,
      "grad_norm": 0.003829725319519639,
      "learning_rate": 2.275267227987321e-07,
      "loss": 11.5,
      "step": 1468
    },
    {
      "epoch": 0.23514816815735237,
      "grad_norm": 0.003428353928029537,
      "learning_rate": 2.135334807042866e-07,
      "loss": 11.5,
      "step": 1469
    },
    {
      "epoch": 0.23530824179122398,
      "grad_norm": 0.0036914802622050047,
      "learning_rate": 1.9998384591773944e-07,
      "loss": 11.5,
      "step": 1470
    },
    {
      "epoch": 0.23546831542509555,
      "grad_norm": 0.0024707193952053785,
      "learning_rate": 1.8687787867489592e-07,
      "loss": 11.5,
      "step": 1471
    },
    {
      "epoch": 0.23562838905896713,
      "grad_norm": 0.002799749607220292,
      "learning_rate": 1.7421563723919454e-07,
      "loss": 11.5,
      "step": 1472
    },
    {
      "epoch": 0.2357884626928387,
      "grad_norm": 0.003706209594383836,
      "learning_rate": 1.6199717790145174e-07,
      "loss": 11.5,
      "step": 1473
    },
    {
      "epoch": 0.23594853632671028,
      "grad_norm": 0.0034813382662832737,
      "learning_rate": 1.5022255497962879e-07,
      "loss": 11.5,
      "step": 1474
    },
    {
      "epoch": 0.23610860996058186,
      "grad_norm": 0.002922093030065298,
      "learning_rate": 1.3889182081860962e-07,
      "loss": 11.5,
      "step": 1475
    },
    {
      "epoch": 0.23626868359445344,
      "grad_norm": 0.0027509492356330156,
      "learning_rate": 1.2800502578991235e-07,
      "loss": 11.5,
      "step": 1476
    },
    {
      "epoch": 0.23642875722832504,
      "grad_norm": 0.0042092702351510525,
      "learning_rate": 1.1756221829148928e-07,
      "loss": 11.5,
      "step": 1477
    },
    {
      "epoch": 0.23658883086219662,
      "grad_norm": 0.0024770977906882763,
      "learning_rate": 1.0756344474753821e-07,
      "loss": 11.5,
      "step": 1478
    },
    {
      "epoch": 0.2367489044960682,
      "grad_norm": 0.0025954816956073046,
      "learning_rate": 9.800874960826933e-08,
      "loss": 11.5,
      "step": 1479
    },
    {
      "epoch": 0.23690897812993977,
      "grad_norm": 0.001979141030460596,
      "learning_rate": 8.889817534969425e-08,
      "loss": 11.5,
      "step": 1480
    },
    {
      "epoch": 0.23706905176381135,
      "grad_norm": 0.005076172295957804,
      "learning_rate": 8.023176247348163e-08,
      "loss": 11.5,
      "step": 1481
    },
    {
      "epoch": 0.23722912539768293,
      "grad_norm": 0.0013020483311265707,
      "learning_rate": 7.200954950673522e-08,
      "loss": 11.5,
      "step": 1482
    },
    {
      "epoch": 0.2373891990315545,
      "grad_norm": 0.002819462912157178,
      "learning_rate": 6.423157300184946e-08,
      "loss": 11.5,
      "step": 1483
    },
    {
      "epoch": 0.2375492726654261,
      "grad_norm": 0.004089313559234142,
      "learning_rate": 5.6897867536331864e-08,
      "loss": 11.5,
      "step": 1484
    },
    {
      "epoch": 0.23770934629929769,
      "grad_norm": 0.0035174668300896883,
      "learning_rate": 5.000846571264761e-08,
      "loss": 11.5,
      "step": 1485
    },
    {
      "epoch": 0.23786941993316926,
      "grad_norm": 0.0026240090373903513,
      "learning_rate": 4.35633981580974e-08,
      "loss": 11.5,
      "step": 1486
    },
    {
      "epoch": 0.23802949356704084,
      "grad_norm": 0.0020378362387418747,
      "learning_rate": 3.756269352462871e-08,
      "loss": 11.5,
      "step": 1487
    },
    {
      "epoch": 0.23818956720091242,
      "grad_norm": 0.002563786692917347,
      "learning_rate": 3.20063784888025e-08,
      "loss": 11.5,
      "step": 1488
    },
    {
      "epoch": 0.238349640834784,
      "grad_norm": 0.0014814968453720212,
      "learning_rate": 2.6894477751548964e-08,
      "loss": 11.5,
      "step": 1489
    },
    {
      "epoch": 0.23850971446865557,
      "grad_norm": 0.0015452037332579494,
      "learning_rate": 2.222701403818972e-08,
      "loss": 11.5,
      "step": 1490
    },
    {
      "epoch": 0.23866978810252718,
      "grad_norm": 0.004355388227850199,
      "learning_rate": 1.8004008098226887e-08,
      "loss": 11.5,
      "step": 1491
    },
    {
      "epoch": 0.23882986173639875,
      "grad_norm": 0.001626601442694664,
      "learning_rate": 1.4225478705309769e-08,
      "loss": 11.5,
      "step": 1492
    },
    {
      "epoch": 0.23898993537027033,
      "grad_norm": 0.028113674372434616,
      "learning_rate": 1.0891442657134932e-08,
      "loss": 11.5,
      "step": 1493
    },
    {
      "epoch": 0.2391500090041419,
      "grad_norm": 0.026651926338672638,
      "learning_rate": 8.001914775401798e-09,
      "loss": 11.5,
      "step": 1494
    },
    {
      "epoch": 0.23931008263801348,
      "grad_norm": 0.006215530447661877,
      "learning_rate": 5.5569079056794206e-09,
      "loss": 11.5,
      "step": 1495
    },
    {
      "epoch": 0.23947015627188506,
      "grad_norm": 0.005797202233225107,
      "learning_rate": 3.5564329174064824e-09,
      "loss": 11.5,
      "step": 1496
    },
    {
      "epoch": 0.23963022990575664,
      "grad_norm": 0.006935896351933479,
      "learning_rate": 2.0004987038246824e-09,
      "loss": 11.5,
      "step": 1497
    },
    {
      "epoch": 0.23979030353962824,
      "grad_norm": 0.005933831911534071,
      "learning_rate": 8.891121819565306e-10,
      "loss": 11.5,
      "step": 1498
    },
    {
      "epoch": 0.23995037717349982,
      "grad_norm": 0.007345221936702728,
      "learning_rate": 2.2227829252763344e-10,
      "loss": 11.5,
      "step": 1499
    },
    {
      "epoch": 0.2401104508073714,
      "grad_norm": 0.003996612038463354,
      "learning_rate": 0.0,
      "loss": 11.5,
      "step": 1500
    },
    {
      "epoch": 0.2401104508073714,
      "eval_loss": 11.5,
      "eval_runtime": 59.2015,
      "eval_samples_per_second": 177.732,
      "eval_steps_per_second": 88.866,
      "step": 1500
    }
  ],
  "logging_steps": 1,
  "max_steps": 1500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 375,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 239204040278016.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}