{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 15.0,
  "eval_steps": 500,
  "global_step": 12255,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0012239902080783353,
      "grad_norm": 0.18583729171749852,
      "learning_rate": 4.999999917854624e-06,
      "loss": 0.7751,
      "step": 1
    },
    {
      "epoch": 0.0024479804161566705,
      "grad_norm": 0.20406357721085652,
      "learning_rate": 4.999999671418499e-06,
      "loss": 0.8304,
      "step": 2
    },
    {
      "epoch": 0.0036719706242350062,
      "grad_norm": 0.21291625994602287,
      "learning_rate": 4.999999260691644e-06,
      "loss": 1.0053,
      "step": 3
    },
    {
      "epoch": 0.004895960832313341,
      "grad_norm": 0.2877744427598735,
      "learning_rate": 4.999998685674084e-06,
      "loss": 0.7683,
      "step": 4
    },
    {
      "epoch": 0.006119951040391677,
      "grad_norm": 0.22350743109660962,
      "learning_rate": 4.999997946365856e-06,
      "loss": 0.9978,
      "step": 5
    },
    {
      "epoch": 0.0073439412484700125,
      "grad_norm": 0.2840586328444629,
      "learning_rate": 4.999997042767011e-06,
      "loss": 1.6078,
      "step": 6
    },
    {
      "epoch": 0.008567931456548347,
      "grad_norm": 0.23613497057256475,
      "learning_rate": 4.999995974877606e-06,
      "loss": 0.719,
      "step": 7
    },
    {
      "epoch": 0.009791921664626682,
      "grad_norm": 0.2970787719278101,
      "learning_rate": 4.999994742697714e-06,
      "loss": 0.9873,
      "step": 8
    },
    {
      "epoch": 0.011015911872705019,
      "grad_norm": 0.2070582341188991,
      "learning_rate": 4.999993346227413e-06,
      "loss": 0.6846,
      "step": 9
    },
    {
      "epoch": 0.012239902080783354,
      "grad_norm": 0.30932752611485503,
      "learning_rate": 4.999991785466797e-06,
      "loss": 1.3271,
      "step": 10
    },
    {
      "epoch": 0.01346389228886169,
      "grad_norm": 0.24978429465929539,
      "learning_rate": 4.999990060415967e-06,
      "loss": 0.6926,
      "step": 11
    },
    {
      "epoch": 0.014687882496940025,
      "grad_norm": 0.19147231300046869,
      "learning_rate": 4.999988171075038e-06,
      "loss": 0.6701,
      "step": 12
    },
    {
      "epoch": 0.01591187270501836,
      "grad_norm": 0.30225290731864235,
      "learning_rate": 4.999986117444132e-06,
      "loss": 0.7883,
      "step": 13
    },
    {
      "epoch": 0.017135862913096694,
      "grad_norm": 0.34854122308275026,
      "learning_rate": 4.9999838995233864e-06,
      "loss": 1.0646,
      "step": 14
    },
    {
      "epoch": 0.01835985312117503,
      "grad_norm": 0.35994514446338466,
      "learning_rate": 4.999981517312945e-06,
      "loss": 0.917,
      "step": 15
    },
    {
      "epoch": 0.019583843329253364,
      "grad_norm": 0.19253165865850944,
      "learning_rate": 4.9999789708129654e-06,
      "loss": 0.7728,
      "step": 16
    },
    {
      "epoch": 0.0208078335373317,
      "grad_norm": 0.25014201537622865,
      "learning_rate": 4.999976260023615e-06,
      "loss": 0.7767,
      "step": 17
    },
    {
      "epoch": 0.022031823745410038,
      "grad_norm": 0.1739379338522603,
      "learning_rate": 4.999973384945071e-06,
      "loss": 0.653,
      "step": 18
    },
    {
      "epoch": 0.023255813953488372,
      "grad_norm": 0.3584818482270971,
      "learning_rate": 4.999970345577521e-06,
      "loss": 1.2892,
      "step": 19
    },
    {
      "epoch": 0.02447980416156671,
      "grad_norm": 0.3080763209165389,
      "learning_rate": 4.999967141921169e-06,
      "loss": 1.2724,
      "step": 20
    },
    {
      "epoch": 0.025703794369645042,
      "grad_norm": 0.3078190263976052,
      "learning_rate": 4.999963773976223e-06,
      "loss": 1.5101,
      "step": 21
    },
    {
      "epoch": 0.02692778457772338,
      "grad_norm": 0.35734267049928353,
      "learning_rate": 4.999960241742904e-06,
      "loss": 1.0969,
      "step": 22
    },
    {
      "epoch": 0.028151774785801713,
      "grad_norm": 0.29870690242698555,
      "learning_rate": 4.9999565452214455e-06,
      "loss": 0.957,
      "step": 23
    },
    {
      "epoch": 0.02937576499388005,
      "grad_norm": 0.31183960915376496,
      "learning_rate": 4.999952684412089e-06,
      "loss": 0.8774,
      "step": 24
    },
    {
      "epoch": 0.030599755201958383,
      "grad_norm": 0.26729570980176937,
      "learning_rate": 4.999948659315088e-06,
      "loss": 0.9801,
      "step": 25
    },
    {
      "epoch": 0.03182374541003672,
      "grad_norm": 0.3298521691537802,
      "learning_rate": 4.999944469930708e-06,
      "loss": 1.4352,
      "step": 26
    },
    {
      "epoch": 0.033047735618115054,
      "grad_norm": 0.30511614902908596,
      "learning_rate": 4.9999401162592256e-06,
      "loss": 0.9802,
      "step": 27
    },
    {
      "epoch": 0.03427172582619339,
      "grad_norm": 0.26480266153469656,
      "learning_rate": 4.999935598300924e-06,
      "loss": 0.6735,
      "step": 28
    },
    {
      "epoch": 0.03549571603427173,
      "grad_norm": 0.26529159306625855,
      "learning_rate": 4.999930916056102e-06,
      "loss": 0.9023,
      "step": 29
    },
    {
      "epoch": 0.03671970624235006,
      "grad_norm": 0.3579345710211942,
      "learning_rate": 4.999926069525066e-06,
      "loss": 0.934,
      "step": 30
    },
    {
      "epoch": 0.037943696450428395,
      "grad_norm": 0.1866932821938979,
      "learning_rate": 4.999921058708136e-06,
      "loss": 0.7769,
      "step": 31
    },
    {
      "epoch": 0.03916768665850673,
      "grad_norm": 0.3184281967758577,
      "learning_rate": 4.999915883605641e-06,
      "loss": 0.683,
      "step": 32
    },
    {
      "epoch": 0.04039167686658507,
      "grad_norm": 0.27541521490140086,
      "learning_rate": 4.99991054421792e-06,
      "loss": 1.3672,
      "step": 33
    },
    {
      "epoch": 0.0416156670746634,
      "grad_norm": 0.2721316520895182,
      "learning_rate": 4.999905040545325e-06,
      "loss": 1.0381,
      "step": 34
    },
    {
      "epoch": 0.042839657282741736,
      "grad_norm": 0.20120575079725292,
      "learning_rate": 4.999899372588217e-06,
      "loss": 0.7746,
      "step": 35
    },
    {
      "epoch": 0.044063647490820076,
      "grad_norm": 0.27818574153822795,
      "learning_rate": 4.9998935403469685e-06,
      "loss": 0.8716,
      "step": 36
    },
    {
      "epoch": 0.04528763769889841,
      "grad_norm": 0.40361412171293226,
      "learning_rate": 4.999887543821963e-06,
      "loss": 1.0859,
      "step": 37
    },
    {
      "epoch": 0.046511627906976744,
      "grad_norm": 0.2456651991051225,
      "learning_rate": 4.999881383013595e-06,
      "loss": 1.0212,
      "step": 38
    },
    {
      "epoch": 0.04773561811505508,
      "grad_norm": 0.2060748267108159,
      "learning_rate": 4.999875057922269e-06,
      "loss": 0.8636,
      "step": 39
    },
    {
      "epoch": 0.04895960832313342,
      "grad_norm": 0.41509728031167886,
      "learning_rate": 4.9998685685484e-06,
      "loss": 0.911,
      "step": 40
    },
    {
      "epoch": 0.05018359853121175,
      "grad_norm": 0.21505365801947904,
      "learning_rate": 4.999861914892415e-06,
      "loss": 0.7291,
      "step": 41
    },
    {
      "epoch": 0.051407588739290085,
      "grad_norm": 0.3544708832311817,
      "learning_rate": 4.999855096954752e-06,
      "loss": 0.7691,
      "step": 42
    },
    {
      "epoch": 0.05263157894736842,
      "grad_norm": 0.32765260031528076,
      "learning_rate": 4.999848114735858e-06,
      "loss": 0.8236,
      "step": 43
    },
    {
      "epoch": 0.05385556915544676,
      "grad_norm": 0.3303817423046128,
      "learning_rate": 4.999840968236192e-06,
      "loss": 0.8613,
      "step": 44
    },
    {
      "epoch": 0.05507955936352509,
      "grad_norm": 0.2572617478071188,
      "learning_rate": 4.999833657456224e-06,
      "loss": 0.874,
      "step": 45
    },
    {
      "epoch": 0.056303549571603426,
      "grad_norm": 0.31006055814937555,
      "learning_rate": 4.999826182396435e-06,
      "loss": 0.8179,
      "step": 46
    },
    {
      "epoch": 0.05752753977968176,
      "grad_norm": 0.2656685858221822,
      "learning_rate": 4.9998185430573145e-06,
      "loss": 0.767,
      "step": 47
    },
    {
      "epoch": 0.0587515299877601,
      "grad_norm": 0.36661459170923244,
      "learning_rate": 4.999810739439365e-06,
      "loss": 0.8963,
      "step": 48
    },
    {
      "epoch": 0.05997552019583843,
      "grad_norm": 0.26792611459222554,
      "learning_rate": 4.999802771543101e-06,
      "loss": 0.89,
      "step": 49
    },
    {
      "epoch": 0.06119951040391677,
      "grad_norm": 0.48121373293033187,
      "learning_rate": 4.999794639369045e-06,
      "loss": 1.3291,
      "step": 50
    },
    {
      "epoch": 0.06242350061199511,
      "grad_norm": 0.26272074677420965,
      "learning_rate": 4.99978634291773e-06,
      "loss": 0.6825,
      "step": 51
    },
    {
      "epoch": 0.06364749082007344,
      "grad_norm": 0.2598096821654224,
      "learning_rate": 4.999777882189705e-06,
      "loss": 0.6748,
      "step": 52
    },
    {
      "epoch": 0.06487148102815178,
      "grad_norm": 0.2668709897768842,
      "learning_rate": 4.999769257185522e-06,
      "loss": 0.9461,
      "step": 53
    },
    {
      "epoch": 0.06609547123623011,
      "grad_norm": 0.3375475064373023,
      "learning_rate": 4.99976046790575e-06,
      "loss": 0.8876,
      "step": 54
    },
    {
      "epoch": 0.06731946144430845,
      "grad_norm": 0.4198840161163024,
      "learning_rate": 4.999751514350966e-06,
      "loss": 1.6459,
      "step": 55
    },
    {
      "epoch": 0.06854345165238677,
      "grad_norm": 0.34369233553235295,
      "learning_rate": 4.999742396521758e-06,
      "loss": 0.7366,
      "step": 56
    },
    {
      "epoch": 0.06976744186046512,
      "grad_norm": 0.40703261331120943,
      "learning_rate": 4.999733114418726e-06,
      "loss": 1.3698,
      "step": 57
    },
    {
      "epoch": 0.07099143206854346,
      "grad_norm": 0.22293367791017168,
      "learning_rate": 4.999723668042479e-06,
      "loss": 0.4068,
      "step": 58
    },
    {
      "epoch": 0.07221542227662178,
      "grad_norm": 0.4008775770379318,
      "learning_rate": 4.9997140573936385e-06,
      "loss": 0.7147,
      "step": 59
    },
    {
      "epoch": 0.07343941248470012,
      "grad_norm": 0.4194534781140063,
      "learning_rate": 4.9997042824728365e-06,
      "loss": 0.9667,
      "step": 60
    },
    {
      "epoch": 0.07466340269277846,
      "grad_norm": 0.42035518786031906,
      "learning_rate": 4.9996943432807144e-06,
      "loss": 0.9308,
      "step": 61
    },
    {
      "epoch": 0.07588739290085679,
      "grad_norm": 0.29454843150959004,
      "learning_rate": 4.999684239817925e-06,
      "loss": 0.9713,
      "step": 62
    },
    {
      "epoch": 0.07711138310893513,
      "grad_norm": 0.33329763298008036,
      "learning_rate": 4.999673972085135e-06,
      "loss": 1.021,
      "step": 63
    },
    {
      "epoch": 0.07833537331701346,
      "grad_norm": 0.29752155694550203,
      "learning_rate": 4.999663540083016e-06,
      "loss": 0.6485,
      "step": 64
    },
    {
      "epoch": 0.0795593635250918,
      "grad_norm": 0.3947136376006826,
      "learning_rate": 4.999652943812255e-06,
      "loss": 1.7957,
      "step": 65
    },
    {
      "epoch": 0.08078335373317014,
      "grad_norm": 0.20583114475708844,
      "learning_rate": 4.999642183273549e-06,
      "loss": 0.6997,
      "step": 66
    },
    {
      "epoch": 0.08200734394124846,
      "grad_norm": 0.18289509535433704,
      "learning_rate": 4.999631258467603e-06,
      "loss": 0.4703,
      "step": 67
    },
    {
      "epoch": 0.0832313341493268,
      "grad_norm": 0.2274688855191698,
      "learning_rate": 4.999620169395136e-06,
      "loss": 0.7131,
      "step": 68
    },
    {
      "epoch": 0.08445532435740515,
      "grad_norm": 0.35792525848819423,
      "learning_rate": 4.9996089160568775e-06,
      "loss": 0.8018,
      "step": 69
    },
    {
      "epoch": 0.08567931456548347,
      "grad_norm": 0.3448967587172274,
      "learning_rate": 4.999597498453567e-06,
      "loss": 0.8188,
      "step": 70
    },
    {
      "epoch": 0.08690330477356181,
      "grad_norm": 0.3715378107688424,
      "learning_rate": 4.9995859165859525e-06,
      "loss": 0.7089,
      "step": 71
    },
    {
      "epoch": 0.08812729498164015,
      "grad_norm": 0.2314946710165429,
      "learning_rate": 4.999574170454799e-06,
      "loss": 0.7165,
      "step": 72
    },
    {
      "epoch": 0.08935128518971848,
      "grad_norm": 0.5062107082444923,
      "learning_rate": 4.999562260060875e-06,
      "loss": 0.7468,
      "step": 73
    },
    {
      "epoch": 0.09057527539779682,
      "grad_norm": 0.3613440311794418,
      "learning_rate": 4.999550185404965e-06,
      "loss": 1.1806,
      "step": 74
    },
    {
      "epoch": 0.09179926560587515,
      "grad_norm": 0.36850743951287973,
      "learning_rate": 4.999537946487862e-06,
      "loss": 1.3959,
      "step": 75
    },
    {
      "epoch": 0.09302325581395349,
      "grad_norm": 0.2735739225496828,
      "learning_rate": 4.999525543310369e-06,
      "loss": 1.0208,
      "step": 76
    },
    {
      "epoch": 0.09424724602203183,
      "grad_norm": 0.31992463657244324,
      "learning_rate": 4.999512975873305e-06,
      "loss": 0.6857,
      "step": 77
    },
    {
      "epoch": 0.09547123623011015,
      "grad_norm": 0.19353299905849947,
      "learning_rate": 4.999500244177491e-06,
      "loss": 0.5956,
      "step": 78
    },
    {
      "epoch": 0.0966952264381885,
      "grad_norm": 0.3590515193305031,
      "learning_rate": 4.999487348223767e-06,
      "loss": 1.8687,
      "step": 79
    },
    {
      "epoch": 0.09791921664626684,
      "grad_norm": 0.30964026487168866,
      "learning_rate": 4.99947428801298e-06,
      "loss": 0.7706,
      "step": 80
    },
    {
      "epoch": 0.09914320685434516,
      "grad_norm": 0.5067456087223579,
      "learning_rate": 4.9994610635459865e-06,
      "loss": 0.9452,
      "step": 81
    },
    {
      "epoch": 0.1003671970624235,
      "grad_norm": 0.40985466873086296,
      "learning_rate": 4.999447674823657e-06,
      "loss": 0.8859,
      "step": 82
    },
    {
      "epoch": 0.10159118727050184,
      "grad_norm": 0.39190374257953436,
      "learning_rate": 4.9994341218468715e-06,
      "loss": 1.3009,
      "step": 83
    },
    {
      "epoch": 0.10281517747858017,
      "grad_norm": 0.28923539054173913,
      "learning_rate": 4.99942040461652e-06,
      "loss": 0.7321,
      "step": 84
    },
    {
      "epoch": 0.10403916768665851,
      "grad_norm": 0.2722920220496856,
      "learning_rate": 4.999406523133504e-06,
      "loss": 0.7567,
      "step": 85
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 0.3277663148624042,
      "learning_rate": 4.999392477398737e-06,
      "loss": 0.8629,
      "step": 86
    },
    {
      "epoch": 0.10648714810281518,
      "grad_norm": 0.5195962241294879,
      "learning_rate": 4.99937826741314e-06,
      "loss": 1.257,
      "step": 87
    },
    {
      "epoch": 0.10771113831089352,
      "grad_norm": 0.3903200576204231,
      "learning_rate": 4.999363893177648e-06,
      "loss": 1.4092,
      "step": 88
    },
    {
      "epoch": 0.10893512851897184,
      "grad_norm": 0.4906747473421015,
      "learning_rate": 4.999349354693206e-06,
      "loss": 1.6772,
      "step": 89
    },
    {
      "epoch": 0.11015911872705018,
      "grad_norm": 0.3121759660794242,
      "learning_rate": 4.9993346519607685e-06,
      "loss": 1.8001,
      "step": 90
    },
    {
      "epoch": 0.11138310893512852,
      "grad_norm": 0.2892084262986656,
      "learning_rate": 4.999319784981302e-06,
      "loss": 0.6667,
      "step": 91
    },
    {
      "epoch": 0.11260709914320685,
      "grad_norm": 0.26490607637177427,
      "learning_rate": 4.999304753755784e-06,
      "loss": 0.8763,
      "step": 92
    },
    {
      "epoch": 0.11383108935128519,
      "grad_norm": 0.18640634686040375,
      "learning_rate": 4.9992895582852e-06,
      "loss": 0.5593,
      "step": 93
    },
    {
      "epoch": 0.11505507955936352,
      "grad_norm": 0.2748875373849093,
      "learning_rate": 4.9992741985705524e-06,
      "loss": 0.5294,
      "step": 94
    },
    {
      "epoch": 0.11627906976744186,
      "grad_norm": 0.6441254262036021,
      "learning_rate": 4.999258674612849e-06,
      "loss": 0.8691,
      "step": 95
    },
    {
      "epoch": 0.1175030599755202,
      "grad_norm": 0.5235182704075505,
      "learning_rate": 4.999242986413108e-06,
      "loss": 1.1176,
      "step": 96
    },
    {
      "epoch": 0.11872705018359853,
      "grad_norm": 0.3556262696346673,
      "learning_rate": 4.999227133972364e-06,
      "loss": 0.6873,
      "step": 97
    },
    {
      "epoch": 0.11995104039167687,
      "grad_norm": 0.34559006466503145,
      "learning_rate": 4.999211117291655e-06,
      "loss": 0.6915,
      "step": 98
    },
    {
      "epoch": 0.1211750305997552,
      "grad_norm": 0.4455493985241837,
      "learning_rate": 4.999194936372036e-06,
      "loss": 0.7893,
      "step": 99
    },
    {
      "epoch": 0.12239902080783353,
      "grad_norm": 0.3422702470959658,
      "learning_rate": 4.99917859121457e-06,
      "loss": 0.7701,
      "step": 100
    },
    {
      "epoch": 0.12362301101591187,
      "grad_norm": 0.4425943797862132,
      "learning_rate": 4.99916208182033e-06,
      "loss": 0.7172,
      "step": 101
    },
    {
      "epoch": 0.12484700122399021,
      "grad_norm": 0.40506374911245696,
      "learning_rate": 4.9991454081904025e-06,
      "loss": 1.2405,
      "step": 102
    },
    {
      "epoch": 0.12607099143206854,
      "grad_norm": 0.28721291352960154,
      "learning_rate": 4.999128570325882e-06,
      "loss": 0.7086,
      "step": 103
    },
    {
      "epoch": 0.12729498164014688,
      "grad_norm": 0.42411263793572984,
      "learning_rate": 4.999111568227875e-06,
      "loss": 0.9167,
      "step": 104
    },
    {
      "epoch": 0.12851897184822522,
      "grad_norm": 0.46092441192090267,
      "learning_rate": 4.9990944018975e-06,
      "loss": 0.7406,
      "step": 105
    },
    {
      "epoch": 0.12974296205630356,
      "grad_norm": 0.36830448393020426,
      "learning_rate": 4.999077071335885e-06,
      "loss": 0.8869,
      "step": 106
    },
    {
      "epoch": 0.13096695226438188,
      "grad_norm": 0.44533454781577375,
      "learning_rate": 4.999059576544167e-06,
      "loss": 0.807,
      "step": 107
    },
    {
      "epoch": 0.13219094247246022,
      "grad_norm": 0.28515020625555076,
      "learning_rate": 4.999041917523497e-06,
      "loss": 0.7689,
      "step": 108
    },
    {
      "epoch": 0.13341493268053856,
      "grad_norm": 0.634481287771021,
      "learning_rate": 4.999024094275037e-06,
      "loss": 0.7283,
      "step": 109
    },
    {
      "epoch": 0.1346389228886169,
      "grad_norm": 0.39843886467027073,
      "learning_rate": 4.9990061067999555e-06,
      "loss": 0.8053,
      "step": 110
    },
    {
      "epoch": 0.13586291309669524,
      "grad_norm": 0.4253410945943641,
      "learning_rate": 4.998987955099436e-06,
      "loss": 0.7736,
      "step": 111
    },
    {
      "epoch": 0.13708690330477355,
      "grad_norm": 0.36722373631265515,
      "learning_rate": 4.998969639174671e-06,
      "loss": 0.9464,
      "step": 112
    },
    {
      "epoch": 0.1383108935128519,
      "grad_norm": 0.2732427884721167,
      "learning_rate": 4.998951159026866e-06,
      "loss": 0.833,
      "step": 113
    },
    {
      "epoch": 0.13953488372093023,
      "grad_norm": 0.34663595616055065,
      "learning_rate": 4.998932514657232e-06,
      "loss": 0.7257,
      "step": 114
    },
    {
      "epoch": 0.14075887392900857,
      "grad_norm": 0.5998368754726062,
      "learning_rate": 4.998913706066997e-06,
      "loss": 0.7323,
      "step": 115
    },
    {
      "epoch": 0.1419828641370869,
      "grad_norm": 0.4473463614973645,
      "learning_rate": 4.9988947332573955e-06,
      "loss": 1.2615,
      "step": 116
    },
    {
      "epoch": 0.14320685434516525,
      "grad_norm": 0.4486765780437808,
      "learning_rate": 4.998875596229675e-06,
      "loss": 1.5905,
      "step": 117
    },
    {
      "epoch": 0.14443084455324356,
      "grad_norm": 0.4219723228963629,
      "learning_rate": 4.998856294985093e-06,
      "loss": 0.8601,
      "step": 118
    },
    {
      "epoch": 0.1456548347613219,
      "grad_norm": 0.2506786547797268,
      "learning_rate": 4.998836829524918e-06,
      "loss": 0.6768,
      "step": 119
    },
    {
      "epoch": 0.14687882496940025,
      "grad_norm": 0.4501037316082423,
      "learning_rate": 4.99881719985043e-06,
      "loss": 0.8671,
      "step": 120
    },
    {
      "epoch": 0.1481028151774786,
      "grad_norm": 0.44614537792632397,
      "learning_rate": 4.9987974059629165e-06,
      "loss": 0.6588,
      "step": 121
    },
    {
      "epoch": 0.14932680538555693,
      "grad_norm": 0.2563130623514102,
      "learning_rate": 4.998777447863681e-06,
      "loss": 0.5567,
      "step": 122
    },
    {
      "epoch": 0.15055079559363524,
      "grad_norm": 0.4199130939198221,
      "learning_rate": 4.998757325554034e-06,
      "loss": 0.9347,
      "step": 123
    },
    {
      "epoch": 0.15177478580171358,
      "grad_norm": 0.3361919551388072,
      "learning_rate": 4.998737039035297e-06,
      "loss": 1.129,
      "step": 124
    },
    {
      "epoch": 0.15299877600979192,
      "grad_norm": 0.2960923862167825,
      "learning_rate": 4.998716588308805e-06,
      "loss": 0.5396,
      "step": 125
    },
    {
      "epoch": 0.15422276621787026,
      "grad_norm": 0.37670334524566873,
      "learning_rate": 4.9986959733759e-06,
      "loss": 0.7446,
      "step": 126
    },
    {
      "epoch": 0.1554467564259486,
      "grad_norm": 0.37407387501653966,
      "learning_rate": 4.998675194237939e-06,
      "loss": 1.2103,
      "step": 127
    },
    {
      "epoch": 0.15667074663402691,
      "grad_norm": 0.34935983601064924,
      "learning_rate": 4.9986542508962845e-06,
      "loss": 0.7557,
      "step": 128
    },
    {
      "epoch": 0.15789473684210525,
      "grad_norm": 0.2707127564841584,
      "learning_rate": 4.998633143352315e-06,
      "loss": 0.7545,
      "step": 129
    },
    {
      "epoch": 0.1591187270501836,
      "grad_norm": 0.24472003358768396,
      "learning_rate": 4.998611871607418e-06,
      "loss": 0.6538,
      "step": 130
    },
    {
      "epoch": 0.16034271725826194,
      "grad_norm": 0.347989151027408,
      "learning_rate": 4.99859043566299e-06,
      "loss": 0.6373,
      "step": 131
    },
    {
      "epoch": 0.16156670746634028,
      "grad_norm": 0.3410028328075718,
      "learning_rate": 4.9985688355204395e-06,
      "loss": 0.6982,
      "step": 132
    },
    {
      "epoch": 0.16279069767441862,
      "grad_norm": 0.4622243517751799,
      "learning_rate": 4.998547071181188e-06,
      "loss": 0.7655,
      "step": 133
    },
    {
      "epoch": 0.16401468788249693,
      "grad_norm": 0.24536179750121562,
      "learning_rate": 4.9985251426466635e-06,
      "loss": 0.6953,
      "step": 134
    },
    {
      "epoch": 0.16523867809057527,
      "grad_norm": 0.2593736638478067,
      "learning_rate": 4.998503049918308e-06,
      "loss": 0.6774,
      "step": 135
    },
    {
      "epoch": 0.1664626682986536,
      "grad_norm": 0.3122150544885731,
      "learning_rate": 4.998480792997574e-06,
      "loss": 0.6568,
      "step": 136
    },
    {
      "epoch": 0.16768665850673195,
      "grad_norm": 0.49403594015867824,
      "learning_rate": 4.998458371885924e-06,
      "loss": 1.3702,
      "step": 137
    },
    {
      "epoch": 0.1689106487148103,
      "grad_norm": 0.49563883047758617,
      "learning_rate": 4.99843578658483e-06,
      "loss": 1.5568,
      "step": 138
    },
    {
      "epoch": 0.1701346389228886,
      "grad_norm": 0.5328672596622324,
      "learning_rate": 4.9984130370957774e-06,
      "loss": 0.7061,
      "step": 139
    },
    {
      "epoch": 0.17135862913096694,
      "grad_norm": 0.4616635577958037,
      "learning_rate": 4.998390123420261e-06,
      "loss": 1.3856,
      "step": 140
    },
    {
      "epoch": 0.17258261933904528,
      "grad_norm": 0.42410787586749155,
      "learning_rate": 4.998367045559786e-06,
      "loss": 0.8944,
      "step": 141
    },
    {
      "epoch": 0.17380660954712362,
      "grad_norm": 0.3157157425050123,
      "learning_rate": 4.998343803515871e-06,
      "loss": 0.8737,
      "step": 142
    },
    {
      "epoch": 0.17503059975520197,
      "grad_norm": 0.39421935061211766,
      "learning_rate": 4.9983203972900404e-06,
      "loss": 0.6572,
      "step": 143
    },
    {
      "epoch": 0.1762545899632803,
      "grad_norm": 0.36976427689507774,
      "learning_rate": 4.998296826883835e-06,
      "loss": 0.8748,
      "step": 144
    },
    {
      "epoch": 0.17747858017135862,
      "grad_norm": 0.37479242639243654,
      "learning_rate": 4.998273092298802e-06,
      "loss": 0.5115,
      "step": 145
    },
    {
      "epoch": 0.17870257037943696,
      "grad_norm": 0.4514158745969038,
      "learning_rate": 4.998249193536502e-06,
      "loss": 0.7835,
      "step": 146
    },
    {
      "epoch": 0.1799265605875153,
      "grad_norm": 0.4951291207657622,
      "learning_rate": 4.998225130598506e-06,
      "loss": 0.6473,
      "step": 147
    },
    {
      "epoch": 0.18115055079559364,
      "grad_norm": 0.3386131828475296,
      "learning_rate": 4.9982009034863945e-06,
      "loss": 0.8011,
      "step": 148
    },
    {
      "epoch": 0.18237454100367198,
      "grad_norm": 0.3228627462553721,
      "learning_rate": 4.998176512201759e-06,
      "loss": 0.8726,
      "step": 149
    },
    {
      "epoch": 0.1835985312117503,
      "grad_norm": 0.5854932496433183,
      "learning_rate": 4.998151956746204e-06,
      "loss": 0.8107,
      "step": 150
    },
    {
      "epoch": 0.18482252141982863,
      "grad_norm": 0.4399488624127159,
      "learning_rate": 4.998127237121343e-06,
      "loss": 1.3297,
      "step": 151
    },
    {
      "epoch": 0.18604651162790697,
      "grad_norm": 0.7223478344080757,
      "learning_rate": 4.998102353328799e-06,
      "loss": 0.7384,
      "step": 152
    },
    {
      "epoch": 0.18727050183598531,
      "grad_norm": 0.4119591602064238,
      "learning_rate": 4.998077305370208e-06,
      "loss": 0.7381,
      "step": 153
    },
    {
      "epoch": 0.18849449204406366,
      "grad_norm": 0.3857796813404751,
      "learning_rate": 4.998052093247216e-06,
      "loss": 1.1831,
      "step": 154
    },
    {
      "epoch": 0.189718482252142,
      "grad_norm": 0.6318475758628039,
      "learning_rate": 4.998026716961482e-06,
      "loss": 0.6631,
      "step": 155
    },
    {
      "epoch": 0.1909424724602203,
      "grad_norm": 0.32050742437596175,
      "learning_rate": 4.998001176514669e-06,
      "loss": 0.6584,
      "step": 156
    },
    {
      "epoch": 0.19216646266829865,
      "grad_norm": 0.3915227502604706,
      "learning_rate": 4.9979754719084605e-06,
      "loss": 1.106,
      "step": 157
    },
    {
      "epoch": 0.193390452876377,
      "grad_norm": 0.7184424755359442,
      "learning_rate": 4.997949603144543e-06,
      "loss": 0.6375,
      "step": 158
    },
    {
      "epoch": 0.19461444308445533,
      "grad_norm": 0.5318068215351058,
      "learning_rate": 4.997923570224616e-06,
      "loss": 1.2265,
      "step": 159
    },
    {
      "epoch": 0.19583843329253367,
      "grad_norm": 0.5495236392930165,
      "learning_rate": 4.997897373150391e-06,
      "loss": 1.068,
      "step": 160
    },
    {
      "epoch": 0.19706242350061198,
      "grad_norm": 0.4896808464872852,
      "learning_rate": 4.99787101192359e-06,
      "loss": 1.107,
      "step": 161
    },
    {
      "epoch": 0.19828641370869032,
      "grad_norm": 0.4592888365279043,
      "learning_rate": 4.997844486545945e-06,
      "loss": 0.5706,
      "step": 162
    },
    {
      "epoch": 0.19951040391676866,
      "grad_norm": 0.37159977396263055,
      "learning_rate": 4.9978177970192e-06,
      "loss": 0.9603,
      "step": 163
    },
    {
      "epoch": 0.200734394124847,
      "grad_norm": 0.3919580648883766,
      "learning_rate": 4.997790943345108e-06,
      "loss": 1.1347,
      "step": 164
    },
    {
      "epoch": 0.20195838433292534,
      "grad_norm": 0.4848483989999201,
      "learning_rate": 4.997763925525433e-06,
      "loss": 0.9064,
      "step": 165
    },
    {
      "epoch": 0.20318237454100369,
      "grad_norm": 0.2445361529142901,
      "learning_rate": 4.997736743561953e-06,
      "loss": 0.4401,
      "step": 166
    },
    {
      "epoch": 0.204406364749082,
      "grad_norm": 0.49908539255042084,
      "learning_rate": 4.9977093974564505e-06,
      "loss": 0.8189,
      "step": 167
    },
    {
      "epoch": 0.20563035495716034,
      "grad_norm": 0.30279213170087316,
      "learning_rate": 4.997681887210726e-06,
      "loss": 0.9745,
      "step": 168
    },
    {
      "epoch": 0.20685434516523868,
      "grad_norm": 0.5715880439135204,
      "learning_rate": 4.997654212826586e-06,
      "loss": 0.6207,
      "step": 169
    },
    {
      "epoch": 0.20807833537331702,
      "grad_norm": 0.27475549346668576,
      "learning_rate": 4.99762637430585e-06,
      "loss": 0.6313,
      "step": 170
    },
    {
      "epoch": 0.20930232558139536,
      "grad_norm": 0.8468479347937827,
      "learning_rate": 4.997598371650346e-06,
      "loss": 1.3491,
      "step": 171
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 0.30388117983313656,
      "learning_rate": 4.9975702048619155e-06,
      "loss": 0.638,
      "step": 172
    },
    {
      "epoch": 0.211750305997552,
      "grad_norm": 0.5767394139051092,
      "learning_rate": 4.997541873942409e-06,
      "loss": 0.8063,
      "step": 173
    },
    {
      "epoch": 0.21297429620563035,
      "grad_norm": 0.5787388882265994,
      "learning_rate": 4.9975133788936876e-06,
      "loss": 1.3126,
      "step": 174
    },
    {
      "epoch": 0.2141982864137087,
      "grad_norm": 0.38625235096294347,
      "learning_rate": 4.997484719717624e-06,
      "loss": 0.8975,
      "step": 175
    },
    {
      "epoch": 0.21542227662178703,
      "grad_norm": 0.3826102878704471,
      "learning_rate": 4.997455896416104e-06,
      "loss": 0.4821,
      "step": 176
    },
    {
      "epoch": 0.21664626682986537,
      "grad_norm": 0.5100533216743708,
      "learning_rate": 4.997426908991019e-06,
      "loss": 0.6693,
      "step": 177
    },
    {
      "epoch": 0.2178702570379437,
      "grad_norm": 0.3240272144167246,
      "learning_rate": 4.997397757444275e-06,
      "loss": 0.75,
      "step": 178
    },
    {
      "epoch": 0.21909424724602203,
      "grad_norm": 0.5071827270438135,
      "learning_rate": 4.997368441777788e-06,
      "loss": 1.3654,
      "step": 179
    },
    {
      "epoch": 0.22031823745410037,
      "grad_norm": 0.28815113962935535,
      "learning_rate": 4.997338961993483e-06,
      "loss": 0.6788,
      "step": 180
    },
    {
      "epoch": 0.2215422276621787,
      "grad_norm": 0.49956327183259586,
      "learning_rate": 4.9973093180932995e-06,
      "loss": 0.7429,
      "step": 181
    },
    {
      "epoch": 0.22276621787025705,
      "grad_norm": 0.2905320526365547,
      "learning_rate": 4.997279510079184e-06,
      "loss": 0.6445,
      "step": 182
    },
    {
      "epoch": 0.22399020807833536,
      "grad_norm": 0.5239616159135687,
      "learning_rate": 4.997249537953096e-06,
      "loss": 0.5992,
      "step": 183
    },
    {
      "epoch": 0.2252141982864137,
      "grad_norm": 0.4648600472659148,
      "learning_rate": 4.997219401717005e-06,
      "loss": 0.7933,
      "step": 184
    },
    {
      "epoch": 0.22643818849449204,
      "grad_norm": 0.47836539467822914,
      "learning_rate": 4.997189101372892e-06,
      "loss": 0.78,
      "step": 185
    },
    {
      "epoch": 0.22766217870257038,
      "grad_norm": 0.5689519798620852,
      "learning_rate": 4.997158636922748e-06,
      "loss": 1.8561,
      "step": 186
    },
    {
      "epoch": 0.22888616891064872,
      "grad_norm": 0.3249181825098272,
      "learning_rate": 4.997128008368574e-06,
      "loss": 0.5759,
      "step": 187
    },
    {
      "epoch": 0.23011015911872704,
      "grad_norm": 0.43628931796956344,
      "learning_rate": 4.997097215712383e-06,
      "loss": 1.2084,
      "step": 188
    },
    {
      "epoch": 0.23133414932680538,
      "grad_norm": 0.5661250013530805,
      "learning_rate": 4.9970662589562004e-06,
      "loss": 1.0399,
      "step": 189
    },
    {
      "epoch": 0.23255813953488372,
      "grad_norm": 0.5227652020785303,
      "learning_rate": 4.997035138102057e-06,
      "loss": 1.7628,
      "step": 190
    },
    {
      "epoch": 0.23378212974296206,
      "grad_norm": 0.3039968524820034,
      "learning_rate": 4.997003853152003e-06,
      "loss": 0.7608,
      "step": 191
    },
    {
      "epoch": 0.2350061199510404,
      "grad_norm": 0.5004286280906974,
      "learning_rate": 4.99697240410809e-06,
      "loss": 0.7315,
      "step": 192
    },
    {
      "epoch": 0.23623011015911874,
      "grad_norm": 0.5471560096499389,
      "learning_rate": 4.9969407909723865e-06,
      "loss": 0.6934,
      "step": 193
    },
    {
      "epoch": 0.23745410036719705,
      "grad_norm": 0.41295993902503664,
      "learning_rate": 4.99690901374697e-06,
      "loss": 0.7179,
      "step": 194
    },
    {
      "epoch": 0.2386780905752754,
      "grad_norm": 0.3894074261579104,
      "learning_rate": 4.9968770724339284e-06,
      "loss": 1.2983,
      "step": 195
    },
    {
      "epoch": 0.23990208078335373,
      "grad_norm": 0.33281237615234205,
      "learning_rate": 4.996844967035361e-06,
      "loss": 0.7737,
      "step": 196
    },
    {
      "epoch": 0.24112607099143207,
      "grad_norm": 0.6370346789944287,
      "learning_rate": 4.996812697553378e-06,
      "loss": 1.5354,
      "step": 197
    },
    {
      "epoch": 0.2423500611995104,
      "grad_norm": 0.40147533713216516,
      "learning_rate": 4.996780263990099e-06,
      "loss": 1.4154,
      "step": 198
    },
    {
      "epoch": 0.24357405140758873,
      "grad_norm": 0.4736800171228452,
      "learning_rate": 4.996747666347655e-06,
      "loss": 0.8009,
      "step": 199
    },
    {
      "epoch": 0.24479804161566707,
      "grad_norm": 0.3341724444131733,
      "learning_rate": 4.996714904628191e-06,
      "loss": 0.7838,
      "step": 200
    },
    {
      "epoch": 0.2460220318237454,
      "grad_norm": 0.5437138362384825,
      "learning_rate": 4.996681978833857e-06,
      "loss": 0.9131,
      "step": 201
    },
    {
      "epoch": 0.24724602203182375,
      "grad_norm": 0.4218633779837018,
      "learning_rate": 4.996648888966819e-06,
      "loss": 0.7337,
      "step": 202
    },
    {
      "epoch": 0.2484700122399021,
      "grad_norm": 0.284889446000958,
      "learning_rate": 4.9966156350292495e-06,
      "loss": 0.7668,
      "step": 203
    },
    {
      "epoch": 0.24969400244798043,
      "grad_norm": 0.4141797542372401,
      "learning_rate": 4.996582217023336e-06,
      "loss": 0.9985,
      "step": 204
    },
    {
      "epoch": 0.25091799265605874,
      "grad_norm": 0.5678675624576675,
      "learning_rate": 4.996548634951272e-06,
      "loss": 0.7437,
      "step": 205
    },
    {
      "epoch": 0.2521419828641371,
      "grad_norm": 0.5162332751942637,
      "learning_rate": 4.996514888815267e-06,
      "loss": 1.2859,
      "step": 206
    },
    {
      "epoch": 0.2533659730722154,
      "grad_norm": 0.4395001801952133,
      "learning_rate": 4.996480978617536e-06,
      "loss": 0.7203,
      "step": 207
    },
    {
      "epoch": 0.25458996328029376,
      "grad_norm": 0.533497126973168,
      "learning_rate": 4.99644690436031e-06,
      "loss": 0.7629,
      "step": 208
    },
    {
      "epoch": 0.2558139534883721,
      "grad_norm": 0.36556328854877673,
      "learning_rate": 4.996412666045827e-06,
      "loss": 0.792,
      "step": 209
    },
    {
      "epoch": 0.25703794369645044,
      "grad_norm": 0.5027566620987959,
      "learning_rate": 4.996378263676337e-06,
      "loss": 0.7591,
      "step": 210
    },
    {
      "epoch": 0.2582619339045288,
      "grad_norm": 0.4690908461976835,
      "learning_rate": 4.996343697254101e-06,
      "loss": 1.4626,
      "step": 211
    },
    {
      "epoch": 0.2594859241126071,
      "grad_norm": 0.485475298306494,
      "learning_rate": 4.996308966781391e-06,
      "loss": 1.2105,
      "step": 212
    },
    {
      "epoch": 0.2607099143206854,
      "grad_norm": 0.4049909223288257,
      "learning_rate": 4.996274072260489e-06,
      "loss": 0.6471,
      "step": 213
    },
    {
      "epoch": 0.26193390452876375,
      "grad_norm": 0.43623035400909826,
      "learning_rate": 4.9962390136936875e-06,
      "loss": 0.7631,
      "step": 214
    },
    {
      "epoch": 0.2631578947368421,
      "grad_norm": 0.4328970092945313,
      "learning_rate": 4.996203791083291e-06,
      "loss": 0.8536,
      "step": 215
    },
    {
      "epoch": 0.26438188494492043,
      "grad_norm": 0.4129059493539081,
      "learning_rate": 4.996168404431614e-06,
      "loss": 1.1369,
      "step": 216
    },
    {
      "epoch": 0.26560587515299877,
      "grad_norm": 0.5512147023001712,
      "learning_rate": 4.996132853740983e-06,
      "loss": 0.603,
      "step": 217
    },
    {
      "epoch": 0.2668298653610771,
      "grad_norm": 0.32062319796630956,
      "learning_rate": 4.996097139013732e-06,
      "loss": 0.5787,
      "step": 218
    },
    {
      "epoch": 0.26805385556915545,
      "grad_norm": 0.5885131760628513,
      "learning_rate": 4.996061260252211e-06,
      "loss": 0.9231,
      "step": 219
    },
    {
      "epoch": 0.2692778457772338,
      "grad_norm": 0.44250343208886816,
      "learning_rate": 4.9960252174587755e-06,
      "loss": 0.869,
      "step": 220
    },
    {
      "epoch": 0.27050183598531213,
      "grad_norm": 0.6770444474548993,
      "learning_rate": 4.995989010635795e-06,
      "loss": 0.7859,
      "step": 221
    },
    {
      "epoch": 0.2717258261933905,
      "grad_norm": 0.41483569214928795,
      "learning_rate": 4.995952639785648e-06,
      "loss": 0.7092,
      "step": 222
    },
    {
      "epoch": 0.2729498164014688,
      "grad_norm": 0.45009232931162785,
      "learning_rate": 4.995916104910726e-06,
      "loss": 0.5354,
      "step": 223
    },
    {
      "epoch": 0.2741738066095471,
      "grad_norm": 0.43389200474036227,
      "learning_rate": 4.99587940601343e-06,
      "loss": 0.6763,
      "step": 224
    },
    {
      "epoch": 0.27539779681762544,
      "grad_norm": 0.3495561501216181,
      "learning_rate": 4.995842543096171e-06,
      "loss": 0.9284,
      "step": 225
    },
    {
      "epoch": 0.2766217870257038,
      "grad_norm": 0.6919048432012822,
      "learning_rate": 4.9958055161613706e-06,
      "loss": 0.7265,
      "step": 226
    },
    {
      "epoch": 0.2778457772337821,
      "grad_norm": 0.34441159988406594,
      "learning_rate": 4.995768325211463e-06,
      "loss": 0.5909,
      "step": 227
    },
    {
      "epoch": 0.27906976744186046,
      "grad_norm": 0.47574339841571883,
      "learning_rate": 4.995730970248893e-06,
      "loss": 0.606,
      "step": 228
    },
    {
      "epoch": 0.2802937576499388,
      "grad_norm": 0.4941533686083936,
      "learning_rate": 4.9956934512761145e-06,
      "loss": 1.5199,
      "step": 229
    },
    {
      "epoch": 0.28151774785801714,
      "grad_norm": 0.3742334713270821,
      "learning_rate": 4.995655768295593e-06,
      "loss": 0.7398,
      "step": 230
    },
    {
      "epoch": 0.2827417380660955,
      "grad_norm": 0.45036144161461145,
      "learning_rate": 4.995617921309806e-06,
      "loss": 0.6636,
      "step": 231
    },
    {
      "epoch": 0.2839657282741738,
      "grad_norm": 0.40031611547530066,
      "learning_rate": 4.995579910321239e-06,
      "loss": 0.7104,
      "step": 232
    },
    {
      "epoch": 0.28518971848225216,
      "grad_norm": 0.374237680233108,
      "learning_rate": 4.995541735332391e-06,
      "loss": 0.9785,
      "step": 233
    },
    {
      "epoch": 0.2864137086903305,
      "grad_norm": 0.509762012418007,
      "learning_rate": 4.995503396345771e-06,
      "loss": 0.7464,
      "step": 234
    },
    {
      "epoch": 0.2876376988984088,
      "grad_norm": 0.5496727138740468,
      "learning_rate": 4.9954648933638975e-06,
      "loss": 0.7652,
      "step": 235
    },
    {
      "epoch": 0.28886168910648713,
      "grad_norm": 0.4188285414593592,
      "learning_rate": 4.995426226389301e-06,
      "loss": 0.8009,
      "step": 236
    },
    {
      "epoch": 0.29008567931456547,
      "grad_norm": 0.32169460293596425,
      "learning_rate": 4.995387395424523e-06,
      "loss": 0.7628,
      "step": 237
    },
    {
      "epoch": 0.2913096695226438,
      "grad_norm": 0.464132936363322,
      "learning_rate": 4.9953484004721155e-06,
      "loss": 0.6802,
      "step": 238
    },
    {
      "epoch": 0.29253365973072215,
      "grad_norm": 0.3539045069204836,
      "learning_rate": 4.995309241534641e-06,
      "loss": 0.7423,
      "step": 239
    },
    {
      "epoch": 0.2937576499388005,
      "grad_norm": 0.5825407156686288,
      "learning_rate": 4.995269918614672e-06,
      "loss": 0.5653,
      "step": 240
    },
    {
      "epoch": 0.29498164014687883,
      "grad_norm": 0.4627979340308164,
      "learning_rate": 4.995230431714794e-06,
      "loss": 0.7908,
      "step": 241
    },
    {
      "epoch": 0.2962056303549572,
      "grad_norm": 0.5766648265598003,
      "learning_rate": 4.995190780837601e-06,
      "loss": 0.9683,
      "step": 242
    },
    {
      "epoch": 0.2974296205630355,
      "grad_norm": 0.45446441684394123,
      "learning_rate": 4.995150965985699e-06,
      "loss": 0.8602,
      "step": 243
    },
    {
      "epoch": 0.29865361077111385,
      "grad_norm": 0.6837464625447014,
      "learning_rate": 4.9951109871617045e-06,
      "loss": 0.5667,
      "step": 244
    },
    {
      "epoch": 0.2998776009791922,
      "grad_norm": 0.41141226748470483,
      "learning_rate": 4.9950708443682446e-06,
      "loss": 0.6613,
      "step": 245
    },
    {
      "epoch": 0.3011015911872705,
      "grad_norm": 0.43729198998748986,
      "learning_rate": 4.995030537607957e-06,
      "loss": 0.7126,
      "step": 246
    },
    {
      "epoch": 0.3023255813953488,
      "grad_norm": 0.6110751343503649,
      "learning_rate": 4.9949900668834916e-06,
      "loss": 0.7024,
      "step": 247
    },
    {
      "epoch": 0.30354957160342716,
      "grad_norm": 0.5053383323646585,
      "learning_rate": 4.994949432197507e-06,
      "loss": 0.6215,
      "step": 248
    },
    {
      "epoch": 0.3047735618115055,
      "grad_norm": 0.5390976496088581,
      "learning_rate": 4.994908633552674e-06,
      "loss": 1.1544,
      "step": 249
    },
    {
      "epoch": 0.30599755201958384,
      "grad_norm": 0.5491463550819378,
      "learning_rate": 4.994867670951674e-06,
      "loss": 0.4781,
      "step": 250
    },
    {
      "epoch": 0.3072215422276622,
      "grad_norm": 0.6831890373805769,
      "learning_rate": 4.994826544397198e-06,
      "loss": 0.9995,
      "step": 251
    },
    {
      "epoch": 0.3084455324357405,
      "grad_norm": 0.6019644145564317,
      "learning_rate": 4.994785253891949e-06,
      "loss": 1.3746,
      "step": 252
    },
    {
      "epoch": 0.30966952264381886,
      "grad_norm": 0.464487943349067,
      "learning_rate": 4.994743799438641e-06,
      "loss": 0.6875,
      "step": 253
    },
    {
      "epoch": 0.3108935128518972,
      "grad_norm": 0.5497408102914363,
      "learning_rate": 4.9947021810399986e-06,
      "loss": 0.6446,
      "step": 254
    },
    {
      "epoch": 0.31211750305997554,
      "grad_norm": 0.829023176350708,
      "learning_rate": 4.994660398698755e-06,
      "loss": 1.0661,
      "step": 255
    },
    {
      "epoch": 0.31334149326805383,
      "grad_norm": 0.3986199249741592,
      "learning_rate": 4.9946184524176576e-06,
      "loss": 0.6443,
      "step": 256
    },
    {
      "epoch": 0.31456548347613217,
      "grad_norm": 0.4633947127041751,
      "learning_rate": 4.994576342199462e-06,
      "loss": 0.6902,
      "step": 257
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 0.47760380163452454,
      "learning_rate": 4.994534068046936e-06,
      "loss": 0.644,
      "step": 258
    },
    {
      "epoch": 0.31701346389228885,
      "grad_norm": 1.0075439480350235,
      "learning_rate": 4.994491629962859e-06,
      "loss": 0.6522,
      "step": 259
    },
    {
      "epoch": 0.3182374541003672,
      "grad_norm": 0.4689984187228857,
      "learning_rate": 4.994449027950018e-06,
      "loss": 0.7934,
      "step": 260
    },
    {
      "epoch": 0.31946144430844553,
      "grad_norm": 0.5307467615288092,
      "learning_rate": 4.994406262011213e-06,
      "loss": 1.3388,
      "step": 261
    },
    {
      "epoch": 0.32068543451652387,
      "grad_norm": 0.47202037017977416,
      "learning_rate": 4.994363332149254e-06,
      "loss": 0.7637,
      "step": 262
    },
    {
      "epoch": 0.3219094247246022,
      "grad_norm": 0.35722945212055746,
      "learning_rate": 4.994320238366963e-06,
      "loss": 0.6222,
      "step": 263
    },
    {
      "epoch": 0.32313341493268055,
      "grad_norm": 0.5247557166547449,
      "learning_rate": 4.994276980667173e-06,
      "loss": 1.2315,
      "step": 264
    },
    {
      "epoch": 0.3243574051407589,
      "grad_norm": 0.6510691266406681,
      "learning_rate": 4.994233559052724e-06,
      "loss": 0.7935,
      "step": 265
    },
    {
      "epoch": 0.32558139534883723,
      "grad_norm": 0.5808808864794651,
      "learning_rate": 4.994189973526472e-06,
      "loss": 0.694,
      "step": 266
    },
    {
      "epoch": 0.3268053855569155,
      "grad_norm": 0.342859550230117,
      "learning_rate": 4.99414622409128e-06,
      "loss": 0.6828,
      "step": 267
    },
    {
      "epoch": 0.32802937576499386,
      "grad_norm": 0.43612585685740884,
      "learning_rate": 4.994102310750023e-06,
      "loss": 0.6533,
      "step": 268
    },
    {
      "epoch": 0.3292533659730722,
      "grad_norm": 0.5436454281415156,
      "learning_rate": 4.994058233505588e-06,
      "loss": 1.6295,
      "step": 269
    },
    {
      "epoch": 0.33047735618115054,
      "grad_norm": 0.43775295110254103,
      "learning_rate": 4.99401399236087e-06,
      "loss": 0.7311,
      "step": 270
    },
    {
      "epoch": 0.3317013463892289,
      "grad_norm": 0.5170552538273601,
      "learning_rate": 4.993969587318777e-06,
      "loss": 1.167,
      "step": 271
    },
    {
      "epoch": 0.3329253365973072,
      "grad_norm": 0.5440519486265794,
      "learning_rate": 4.993925018382228e-06,
      "loss": 0.5029,
      "step": 272
    },
    {
      "epoch": 0.33414932680538556,
      "grad_norm": 0.35139266911721506,
      "learning_rate": 4.993880285554151e-06,
      "loss": 0.682,
      "step": 273
    },
    {
      "epoch": 0.3353733170134639,
      "grad_norm": 0.5468861395513627,
      "learning_rate": 4.9938353888374864e-06,
      "loss": 0.6786,
      "step": 274
    },
    {
      "epoch": 0.33659730722154224,
      "grad_norm": 0.4382209259198201,
      "learning_rate": 4.993790328235184e-06,
      "loss": 0.6551,
      "step": 275
    },
    {
      "epoch": 0.3378212974296206,
      "grad_norm": 0.7234003051270653,
      "learning_rate": 4.9937451037502035e-06,
      "loss": 0.729,
      "step": 276
    },
    {
      "epoch": 0.3390452876376989,
      "grad_norm": 0.4397032702863513,
      "learning_rate": 4.99369971538552e-06,
      "loss": 1.1771,
      "step": 277
    },
    {
      "epoch": 0.3402692778457772,
      "grad_norm": 0.7131330442490159,
      "learning_rate": 4.993654163144115e-06,
      "loss": 0.6182,
      "step": 278
    },
    {
      "epoch": 0.34149326805385555,
      "grad_norm": 0.624111336386168,
      "learning_rate": 4.993608447028981e-06,
      "loss": 0.7858,
      "step": 279
    },
    {
      "epoch": 0.3427172582619339,
      "grad_norm": 0.46245621252372643,
      "learning_rate": 4.9935625670431235e-06,
      "loss": 0.6259,
      "step": 280
    },
    {
      "epoch": 0.34394124847001223,
      "grad_norm": 0.7632145813638226,
      "learning_rate": 4.993516523189556e-06,
      "loss": 0.9955,
      "step": 281
    },
    {
      "epoch": 0.34516523867809057,
      "grad_norm": 0.6707122084619282,
      "learning_rate": 4.9934703154713075e-06,
      "loss": 1.2381,
      "step": 282
    },
    {
      "epoch": 0.3463892288861689,
      "grad_norm": 0.4520286005651424,
      "learning_rate": 4.99342394389141e-06,
      "loss": 0.7872,
      "step": 283
    },
    {
      "epoch": 0.34761321909424725,
      "grad_norm": 0.3916751700619931,
      "learning_rate": 4.993377408452916e-06,
      "loss": 0.8928,
      "step": 284
    },
    {
      "epoch": 0.3488372093023256,
      "grad_norm": 0.6570141277578657,
      "learning_rate": 4.993330709158879e-06,
      "loss": 0.4832,
      "step": 285
    },
    {
      "epoch": 0.35006119951040393,
      "grad_norm": 0.4003850438158674,
      "learning_rate": 4.993283846012371e-06,
      "loss": 0.7022,
      "step": 286
    },
    {
      "epoch": 0.35128518971848227,
      "grad_norm": 0.3036266746080802,
      "learning_rate": 4.993236819016471e-06,
      "loss": 0.4886,
      "step": 287
    },
    {
      "epoch": 0.3525091799265606,
      "grad_norm": 0.3622207827400109,
      "learning_rate": 4.9931896281742685e-06,
      "loss": 0.6166,
      "step": 288
    },
    {
      "epoch": 0.3537331701346389,
      "grad_norm": 0.3880251816279412,
      "learning_rate": 4.993142273488866e-06,
      "loss": 0.6824,
      "step": 289
    },
    {
      "epoch": 0.35495716034271724,
      "grad_norm": 0.5738188001442078,
      "learning_rate": 4.9930947549633745e-06,
      "loss": 1.1688,
      "step": 290
    },
    {
      "epoch": 0.3561811505507956,
      "grad_norm": 0.4362385965173308,
      "learning_rate": 4.9930470726009165e-06,
      "loss": 0.7815,
      "step": 291
    },
    {
      "epoch": 0.3574051407588739,
      "grad_norm": 0.5442023426597746,
      "learning_rate": 4.992999226404627e-06,
      "loss": 1.7166,
      "step": 292
    },
    {
      "epoch": 0.35862913096695226,
      "grad_norm": 0.5710824241256561,
      "learning_rate": 4.992951216377648e-06,
      "loss": 0.6388,
      "step": 293
    },
    {
      "epoch": 0.3598531211750306,
      "grad_norm": 0.6390812083022306,
      "learning_rate": 4.992903042523137e-06,
      "loss": 0.7242,
      "step": 294
    },
    {
      "epoch": 0.36107711138310894,
      "grad_norm": 0.8321065324171557,
      "learning_rate": 4.992854704844257e-06,
      "loss": 1.121,
      "step": 295
    },
    {
      "epoch": 0.3623011015911873,
      "grad_norm": 0.48345521754380033,
      "learning_rate": 4.992806203344188e-06,
      "loss": 0.6511,
      "step": 296
    },
    {
      "epoch": 0.3635250917992656,
      "grad_norm": 0.442311019177579,
      "learning_rate": 4.992757538026115e-06,
      "loss": 1.1869,
      "step": 297
    },
    {
      "epoch": 0.36474908200734396,
      "grad_norm": 0.36259486182817524,
      "learning_rate": 4.992708708893237e-06,
      "loss": 0.7094,
      "step": 298
    },
    {
      "epoch": 0.3659730722154223,
      "grad_norm": 0.5544932696828196,
      "learning_rate": 4.992659715948763e-06,
      "loss": 1.5548,
      "step": 299
    },
    {
      "epoch": 0.3671970624235006,
      "grad_norm": 0.5607043561772156,
      "learning_rate": 4.99261055919591e-06,
      "loss": 1.5498,
      "step": 300
    },
    {
      "epoch": 0.3684210526315789,
      "grad_norm": 0.4106643603096727,
      "learning_rate": 4.992561238637912e-06,
      "loss": 0.6165,
      "step": 301
    },
    {
      "epoch": 0.36964504283965727,
      "grad_norm": 0.3431100429385,
      "learning_rate": 4.992511754278009e-06,
      "loss": 0.7248,
      "step": 302
    },
    {
      "epoch": 0.3708690330477356,
      "grad_norm": 0.48499515589131414,
      "learning_rate": 4.9924621061194524e-06,
      "loss": 0.6237,
      "step": 303
    },
    {
      "epoch": 0.37209302325581395,
      "grad_norm": 0.5995655457611933,
      "learning_rate": 4.992412294165505e-06,
      "loss": 0.5666,
      "step": 304
    },
    {
      "epoch": 0.3733170134638923,
      "grad_norm": 0.4765058018222547,
      "learning_rate": 4.99236231841944e-06,
      "loss": 1.0014,
      "step": 305
    },
    {
      "epoch": 0.37454100367197063,
      "grad_norm": 0.5353333723369841,
      "learning_rate": 4.992312178884543e-06,
      "loss": 1.0534,
      "step": 306
    },
    {
      "epoch": 0.37576499388004897,
      "grad_norm": 0.3624605466616526,
      "learning_rate": 4.992261875564108e-06,
      "loss": 0.7108,
      "step": 307
    },
    {
      "epoch": 0.3769889840881273,
      "grad_norm": 0.35899087384306405,
      "learning_rate": 4.99221140846144e-06,
      "loss": 0.5408,
      "step": 308
    },
    {
      "epoch": 0.37821297429620565,
      "grad_norm": 0.487077037642005,
      "learning_rate": 4.992160777579857e-06,
      "loss": 0.8632,
      "step": 309
    },
    {
      "epoch": 0.379436964504284,
      "grad_norm": 1.0139106930067168,
      "learning_rate": 4.992109982922685e-06,
      "loss": 0.5317,
      "step": 310
    },
    {
      "epoch": 0.3806609547123623,
      "grad_norm": 0.4696845443377511,
      "learning_rate": 4.992059024493263e-06,
      "loss": 0.5028,
      "step": 311
    },
    {
      "epoch": 0.3818849449204406,
      "grad_norm": 0.4499122175232395,
      "learning_rate": 4.992007902294939e-06,
      "loss": 0.9648,
      "step": 312
    },
    {
      "epoch": 0.38310893512851896,
      "grad_norm": 0.5356240915106696,
      "learning_rate": 4.991956616331074e-06,
      "loss": 0.7667,
      "step": 313
    },
    {
      "epoch": 0.3843329253365973,
      "grad_norm": 0.7173577965714655,
      "learning_rate": 4.9919051666050355e-06,
      "loss": 2.0693,
      "step": 314
    },
    {
      "epoch": 0.38555691554467564,
      "grad_norm": 0.6139317501138394,
      "learning_rate": 4.9918535531202075e-06,
      "loss": 1.7767,
      "step": 315
    },
    {
      "epoch": 0.386780905752754,
      "grad_norm": 0.3593650422494968,
      "learning_rate": 4.991801775879981e-06,
      "loss": 0.6836,
      "step": 316
    },
    {
      "epoch": 0.3880048959608323,
      "grad_norm": 0.7812116567688177,
      "learning_rate": 4.991749834887758e-06,
      "loss": 0.8054,
      "step": 317
    },
    {
      "epoch": 0.38922888616891066,
      "grad_norm": 0.7470535859823948,
      "learning_rate": 4.991697730146952e-06,
      "loss": 0.6041,
      "step": 318
    },
    {
      "epoch": 0.390452876376989,
      "grad_norm": 0.6404465963351149,
      "learning_rate": 4.9916454616609875e-06,
      "loss": 1.1557,
      "step": 319
    },
    {
      "epoch": 0.39167686658506734,
      "grad_norm": 0.3153155888607163,
      "learning_rate": 4.9915930294332995e-06,
      "loss": 0.5525,
      "step": 320
    },
    {
      "epoch": 0.3929008567931457,
      "grad_norm": 0.6655474444195767,
      "learning_rate": 4.991540433467333e-06,
      "loss": 0.74,
      "step": 321
    },
    {
      "epoch": 0.39412484700122397,
      "grad_norm": 0.4130331472288031,
      "learning_rate": 4.991487673766544e-06,
      "loss": 0.6494,
      "step": 322
    },
    {
      "epoch": 0.3953488372093023,
      "grad_norm": 0.9084530571943739,
      "learning_rate": 4.991434750334402e-06,
      "loss": 0.7103,
      "step": 323
    },
    {
      "epoch": 0.39657282741738065,
      "grad_norm": 0.4806525902993342,
      "learning_rate": 4.991381663174382e-06,
      "loss": 1.005,
      "step": 324
    },
    {
      "epoch": 0.397796817625459,
      "grad_norm": 0.39113371583719014,
      "learning_rate": 4.991328412289975e-06,
      "loss": 0.4884,
      "step": 325
    },
    {
      "epoch": 0.3990208078335373,
      "grad_norm": 0.5011128516738553,
      "learning_rate": 4.991274997684679e-06,
      "loss": 0.7204,
      "step": 326
    },
    {
      "epoch": 0.40024479804161567,
      "grad_norm": 0.5277401592985527,
      "learning_rate": 4.991221419362005e-06,
      "loss": 1.0627,
      "step": 327
    },
    {
      "epoch": 0.401468788249694,
      "grad_norm": 0.5599662265938297,
      "learning_rate": 4.991167677325473e-06,
      "loss": 1.1175,
      "step": 328
    },
    {
      "epoch": 0.40269277845777235,
      "grad_norm": 0.36090832619308766,
      "learning_rate": 4.991113771578617e-06,
      "loss": 0.832,
      "step": 329
    },
    {
      "epoch": 0.4039167686658507,
      "grad_norm": 0.6014499449533965,
      "learning_rate": 4.991059702124976e-06,
      "loss": 0.6821,
      "step": 330
    },
    {
      "epoch": 0.40514075887392903,
      "grad_norm": 0.6380961030988447,
      "learning_rate": 4.991005468968108e-06,
      "loss": 0.6623,
      "step": 331
    },
    {
      "epoch": 0.40636474908200737,
      "grad_norm": 0.5566817907514316,
      "learning_rate": 4.990951072111572e-06,
      "loss": 0.8899,
      "step": 332
    },
    {
      "epoch": 0.40758873929008566,
      "grad_norm": 0.3452644881067129,
      "learning_rate": 4.990896511558945e-06,
      "loss": 0.7251,
      "step": 333
    },
    {
      "epoch": 0.408812729498164,
      "grad_norm": 0.44533300591607666,
      "learning_rate": 4.990841787313813e-06,
      "loss": 0.3994,
      "step": 334
    },
    {
      "epoch": 0.41003671970624234,
      "grad_norm": 0.7012291935157753,
      "learning_rate": 4.990786899379771e-06,
      "loss": 1.2105,
      "step": 335
    },
    {
      "epoch": 0.4112607099143207,
      "grad_norm": 0.7699730158235925,
      "learning_rate": 4.9907318477604276e-06,
      "loss": 0.7512,
      "step": 336
    },
    {
      "epoch": 0.412484700122399,
      "grad_norm": 0.5844222721735775,
      "learning_rate": 4.990676632459399e-06,
      "loss": 0.741,
      "step": 337
    },
    {
      "epoch": 0.41370869033047736,
      "grad_norm": 0.9357690522851926,
      "learning_rate": 4.990621253480315e-06,
      "loss": 0.7072,
      "step": 338
    },
    {
      "epoch": 0.4149326805385557,
      "grad_norm": 0.27407095469032267,
      "learning_rate": 4.990565710826815e-06,
      "loss": 0.4287,
      "step": 339
    },
    {
      "epoch": 0.41615667074663404,
      "grad_norm": 0.5085447392685897,
      "learning_rate": 4.9905100045025465e-06,
      "loss": 1.422,
      "step": 340
    },
    {
      "epoch": 0.4173806609547124,
      "grad_norm": 0.7415706065836827,
      "learning_rate": 4.990454134511174e-06,
      "loss": 1.4713,
      "step": 341
    },
    {
      "epoch": 0.4186046511627907,
      "grad_norm": 0.4549838538864066,
      "learning_rate": 4.990398100856367e-06,
      "loss": 0.7243,
      "step": 342
    },
    {
      "epoch": 0.41982864137086906,
      "grad_norm": 0.4512242161507683,
      "learning_rate": 4.9903419035418075e-06,
      "loss": 0.8547,
      "step": 343
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 0.534164923256363,
      "learning_rate": 4.9902855425711905e-06,
      "loss": 0.7069,
      "step": 344
    },
    {
      "epoch": 0.4222766217870257,
      "grad_norm": 0.469234775417887,
      "learning_rate": 4.990229017948217e-06,
      "loss": 1.3196,
      "step": 345
    },
    {
      "epoch": 0.423500611995104,
      "grad_norm": 0.4842586594577994,
      "learning_rate": 4.990172329676605e-06,
      "loss": 0.6216,
      "step": 346
    },
    {
      "epoch": 0.42472460220318237,
      "grad_norm": 0.7143824981640142,
      "learning_rate": 4.9901154777600766e-06,
      "loss": 0.6318,
      "step": 347
    },
    {
      "epoch": 0.4259485924112607,
      "grad_norm": 0.4957859191910802,
      "learning_rate": 4.99005846220237e-06,
      "loss": 0.814,
      "step": 348
    },
    {
      "epoch": 0.42717258261933905,
      "grad_norm": 0.4420966835734801,
      "learning_rate": 4.9900012830072305e-06,
      "loss": 0.736,
      "step": 349
    },
    {
      "epoch": 0.4283965728274174,
      "grad_norm": 0.6145427384004296,
      "learning_rate": 4.9899439401784165e-06,
      "loss": 1.2759,
      "step": 350
    },
    {
      "epoch": 0.42962056303549573,
      "grad_norm": 0.36005809544555506,
      "learning_rate": 4.989886433719697e-06,
      "loss": 0.7074,
      "step": 351
    },
    {
      "epoch": 0.43084455324357407,
      "grad_norm": 0.5530317148312526,
      "learning_rate": 4.989828763634851e-06,
      "loss": 0.7222,
      "step": 352
    },
    {
      "epoch": 0.4320685434516524,
      "grad_norm": 0.5903230053277813,
      "learning_rate": 4.989770929927668e-06,
      "loss": 0.999,
      "step": 353
    },
    {
      "epoch": 0.43329253365973075,
      "grad_norm": 0.5235973098449931,
      "learning_rate": 4.989712932601947e-06,
      "loss": 0.623,
      "step": 354
    },
    {
      "epoch": 0.43451652386780903,
      "grad_norm": 0.8238368780573823,
      "learning_rate": 4.989654771661502e-06,
      "loss": 0.6532,
      "step": 355
    },
    {
      "epoch": 0.4357405140758874,
      "grad_norm": 0.3961066520599492,
      "learning_rate": 4.989596447110154e-06,
      "loss": 0.7365,
      "step": 356
    },
    {
      "epoch": 0.4369645042839657,
      "grad_norm": 0.4602146080601334,
      "learning_rate": 4.989537958951736e-06,
      "loss": 0.6349,
      "step": 357
    },
    {
      "epoch": 0.43818849449204406,
      "grad_norm": 0.38742547392598503,
      "learning_rate": 4.98947930719009e-06,
      "loss": 0.6247,
      "step": 358
    },
    {
      "epoch": 0.4394124847001224,
      "grad_norm": 0.42036788302585487,
      "learning_rate": 4.989420491829074e-06,
      "loss": 0.5722,
      "step": 359
    },
    {
      "epoch": 0.44063647490820074,
      "grad_norm": 0.7959798845821433,
      "learning_rate": 4.98936151287255e-06,
      "loss": 1.0968,
      "step": 360
    },
    {
      "epoch": 0.4418604651162791,
      "grad_norm": 0.738986970951876,
      "learning_rate": 4.989302370324395e-06,
      "loss": 1.0574,
      "step": 361
    },
    {
      "epoch": 0.4430844553243574,
      "grad_norm": 0.2604864666489375,
      "learning_rate": 4.989243064188495e-06,
      "loss": 0.4387,
      "step": 362
    },
    {
      "epoch": 0.44430844553243576,
      "grad_norm": 0.5476739282324891,
      "learning_rate": 4.989183594468748e-06,
      "loss": 0.5895,
      "step": 363
    },
    {
      "epoch": 0.4455324357405141,
      "grad_norm": 0.8000170519937692,
      "learning_rate": 4.989123961169064e-06,
      "loss": 0.5301,
      "step": 364
    },
    {
      "epoch": 0.4467564259485924,
      "grad_norm": 0.5883493340829177,
      "learning_rate": 4.989064164293357e-06,
      "loss": 1.2867,
      "step": 365
    },
    {
      "epoch": 0.4479804161566707,
      "grad_norm": 0.4351096501408344,
      "learning_rate": 4.989004203845561e-06,
      "loss": 0.5018,
      "step": 366
    },
    {
      "epoch": 0.44920440636474906,
      "grad_norm": 0.4826233005913043,
      "learning_rate": 4.9889440798296155e-06,
      "loss": 0.655,
      "step": 367
    },
    {
      "epoch": 0.4504283965728274,
      "grad_norm": 0.4503050011088164,
      "learning_rate": 4.988883792249469e-06,
      "loss": 0.6846,
      "step": 368
    },
    {
      "epoch": 0.45165238678090575,
      "grad_norm": 0.7312499673085558,
      "learning_rate": 4.988823341109087e-06,
      "loss": 0.5229,
      "step": 369
    },
    {
      "epoch": 0.4528763769889841,
      "grad_norm": 0.8825890360993164,
      "learning_rate": 4.988762726412441e-06,
      "loss": 0.8082,
      "step": 370
    },
    {
      "epoch": 0.4541003671970624,
      "grad_norm": 0.584121075932636,
      "learning_rate": 4.988701948163512e-06,
      "loss": 0.5286,
      "step": 371
    },
    {
      "epoch": 0.45532435740514077,
      "grad_norm": 0.7633584400222349,
      "learning_rate": 4.988641006366298e-06,
      "loss": 0.6654,
      "step": 372
    },
    {
      "epoch": 0.4565483476132191,
      "grad_norm": 0.6532785362101784,
      "learning_rate": 4.988579901024801e-06,
      "loss": 1.3154,
      "step": 373
    },
    {
      "epoch": 0.45777233782129745,
      "grad_norm": 0.5291382291351212,
      "learning_rate": 4.988518632143038e-06,
      "loss": 0.5225,
      "step": 374
    },
    {
      "epoch": 0.4589963280293758,
      "grad_norm": 0.6468474741863817,
      "learning_rate": 4.988457199725034e-06,
      "loss": 0.6772,
      "step": 375
    },
    {
      "epoch": 0.4602203182374541,
      "grad_norm": 0.4335137465584569,
      "learning_rate": 4.988395603774827e-06,
      "loss": 0.6934,
      "step": 376
    },
    {
      "epoch": 0.4614443084455324,
      "grad_norm": 0.9198073708896262,
      "learning_rate": 4.9883338442964655e-06,
      "loss": 0.9603,
      "step": 377
    },
    {
      "epoch": 0.46266829865361075,
      "grad_norm": 0.5760645355202023,
      "learning_rate": 4.988271921294008e-06,
      "loss": 1.2621,
      "step": 378
    },
    {
      "epoch": 0.4638922888616891,
      "grad_norm": 0.7885276885999555,
      "learning_rate": 4.988209834771522e-06,
      "loss": 0.5408,
      "step": 379
    },
    {
      "epoch": 0.46511627906976744,
      "grad_norm": 0.8238883625693336,
      "learning_rate": 4.98814758473309e-06,
      "loss": 0.4811,
      "step": 380
    },
    {
      "epoch": 0.4663402692778458,
      "grad_norm": 0.5576815178557181,
      "learning_rate": 4.988085171182802e-06,
      "loss": 0.6299,
      "step": 381
    },
    {
      "epoch": 0.4675642594859241,
      "grad_norm": 0.5123263432878964,
      "learning_rate": 4.988022594124758e-06,
      "loss": 0.7589,
      "step": 382
    },
    {
      "epoch": 0.46878824969400246,
      "grad_norm": 0.6220823653637169,
      "learning_rate": 4.987959853563072e-06,
      "loss": 0.9005,
      "step": 383
    },
    {
      "epoch": 0.4700122399020808,
      "grad_norm": 0.6290503139592389,
      "learning_rate": 4.987896949501867e-06,
      "loss": 0.9105,
      "step": 384
    },
    {
      "epoch": 0.47123623011015914,
      "grad_norm": 0.9401316401593696,
      "learning_rate": 4.987833881945277e-06,
      "loss": 0.7446,
      "step": 385
    },
    {
      "epoch": 0.4724602203182375,
      "grad_norm": 0.9051181878739166,
      "learning_rate": 4.987770650897446e-06,
      "loss": 0.7699,
      "step": 386
    },
    {
      "epoch": 0.47368421052631576,
      "grad_norm": 0.35399677965369547,
      "learning_rate": 4.987707256362529e-06,
      "loss": 0.6705,
      "step": 387
    },
    {
      "epoch": 0.4749082007343941,
      "grad_norm": 0.4664473022740672,
      "learning_rate": 4.987643698344693e-06,
      "loss": 0.7531,
      "step": 388
    },
    {
      "epoch": 0.47613219094247244,
      "grad_norm": 0.6268223032680224,
      "learning_rate": 4.9875799768481134e-06,
      "loss": 0.8883,
      "step": 389
    },
    {
      "epoch": 0.4773561811505508,
      "grad_norm": 0.7930645555375703,
      "learning_rate": 4.987516091876979e-06,
      "loss": 0.6697,
      "step": 390
    },
    {
      "epoch": 0.4785801713586291,
      "grad_norm": 0.39005757476249786,
      "learning_rate": 4.987452043435489e-06,
      "loss": 0.6306,
      "step": 391
    },
    {
      "epoch": 0.47980416156670747,
      "grad_norm": 0.33717536882333426,
      "learning_rate": 4.98738783152785e-06,
      "loss": 0.5174,
      "step": 392
    },
    {
      "epoch": 0.4810281517747858,
      "grad_norm": 0.5096723567856368,
      "learning_rate": 4.9873234561582826e-06,
      "loss": 0.7911,
      "step": 393
    },
    {
      "epoch": 0.48225214198286415,
      "grad_norm": 0.5771711738205841,
      "learning_rate": 4.987258917331018e-06,
      "loss": 0.4946,
      "step": 394
    },
    {
      "epoch": 0.4834761321909425,
      "grad_norm": 0.6184236099986771,
      "learning_rate": 4.987194215050297e-06,
      "loss": 1.6691,
      "step": 395
    },
    {
      "epoch": 0.4847001223990208,
      "grad_norm": 0.7192296461953666,
      "learning_rate": 4.987129349320372e-06,
      "loss": 0.6894,
      "step": 396
    },
    {
      "epoch": 0.48592411260709917,
      "grad_norm": 0.5298906853294455,
      "learning_rate": 4.987064320145506e-06,
      "loss": 0.7028,
      "step": 397
    },
    {
      "epoch": 0.48714810281517745,
      "grad_norm": 0.40377559904858246,
      "learning_rate": 4.986999127529971e-06,
      "loss": 0.6399,
      "step": 398
    },
    {
      "epoch": 0.4883720930232558,
      "grad_norm": 0.455341828053919,
      "learning_rate": 4.986933771478052e-06,
      "loss": 0.6907,
      "step": 399
    },
    {
      "epoch": 0.48959608323133413,
      "grad_norm": 0.6681580290457239,
      "learning_rate": 4.986868251994045e-06,
      "loss": 1.3346,
      "step": 400
    },
    {
      "epoch": 0.4908200734394125,
      "grad_norm": 0.5057058961872322,
      "learning_rate": 4.986802569082254e-06,
      "loss": 0.6,
      "step": 401
    },
    {
      "epoch": 0.4920440636474908,
      "grad_norm": 0.4720055776923247,
      "learning_rate": 4.986736722746997e-06,
      "loss": 0.4718,
      "step": 402
    },
    {
      "epoch": 0.49326805385556916,
      "grad_norm": 0.7991662775483814,
      "learning_rate": 4.986670712992599e-06,
      "loss": 0.552,
      "step": 403
    },
    {
      "epoch": 0.4944920440636475,
      "grad_norm": 0.6535808127274191,
      "learning_rate": 4.986604539823399e-06,
      "loss": 0.7138,
      "step": 404
    },
    {
      "epoch": 0.49571603427172584,
      "grad_norm": 0.5658656474191684,
      "learning_rate": 4.986538203243748e-06,
      "loss": 0.7007,
      "step": 405
    },
    {
      "epoch": 0.4969400244798042,
      "grad_norm": 0.455049664118394,
      "learning_rate": 4.986471703258002e-06,
      "loss": 0.8734,
      "step": 406
    },
    {
      "epoch": 0.4981640146878825,
      "grad_norm": 0.5990395698316427,
      "learning_rate": 4.9864050398705334e-06,
      "loss": 0.7472,
      "step": 407
    },
    {
      "epoch": 0.49938800489596086,
      "grad_norm": 0.5479802574372877,
      "learning_rate": 4.986338213085721e-06,
      "loss": 1.2621,
      "step": 408
    },
    {
      "epoch": 0.5006119951040392,
      "grad_norm": 0.6532806167863214,
      "learning_rate": 4.986271222907958e-06,
      "loss": 0.6825,
      "step": 409
    },
    {
      "epoch": 0.5018359853121175,
      "grad_norm": 0.6120126735894845,
      "learning_rate": 4.9862040693416466e-06,
      "loss": 1.2561,
      "step": 410
    },
    {
      "epoch": 0.5030599755201959,
      "grad_norm": 0.5032771691644705,
      "learning_rate": 4.9861367523911995e-06,
      "loss": 0.8265,
      "step": 411
    },
    {
      "epoch": 0.5042839657282742,
      "grad_norm": 0.5928601683665099,
      "learning_rate": 4.9860692720610405e-06,
      "loss": 0.6969,
      "step": 412
    },
    {
      "epoch": 0.5055079559363526,
      "grad_norm": 0.3991212570382625,
      "learning_rate": 4.9860016283556045e-06,
      "loss": 0.706,
      "step": 413
    },
    {
      "epoch": 0.5067319461444308,
      "grad_norm": 0.44493232390900905,
      "learning_rate": 4.985933821279336e-06,
      "loss": 1.0315,
      "step": 414
    },
    {
      "epoch": 0.5079559363525091,
      "grad_norm": 0.5326504993434127,
      "learning_rate": 4.985865850836693e-06,
      "loss": 0.6428,
      "step": 415
    },
    {
      "epoch": 0.5091799265605875,
      "grad_norm": 0.75296526985409,
      "learning_rate": 4.985797717032139e-06,
      "loss": 0.7676,
      "step": 416
    },
    {
      "epoch": 0.5104039167686658,
      "grad_norm": 0.6071213565299739,
      "learning_rate": 4.985729419870154e-06,
      "loss": 0.7815,
      "step": 417
    },
    {
      "epoch": 0.5116279069767442,
      "grad_norm": 0.7313162025393259,
      "learning_rate": 4.985660959355226e-06,
      "loss": 0.6697,
      "step": 418
    },
    {
      "epoch": 0.5128518971848225,
      "grad_norm": 1.0084870827345973,
      "learning_rate": 4.985592335491854e-06,
      "loss": 0.6198,
      "step": 419
    },
    {
      "epoch": 0.5140758873929009,
      "grad_norm": 0.6003035571504491,
      "learning_rate": 4.9855235482845466e-06,
      "loss": 1.3909,
      "step": 420
    },
    {
      "epoch": 0.5152998776009792,
      "grad_norm": 0.6579874918589667,
      "learning_rate": 4.985454597737825e-06,
      "loss": 0.7469,
      "step": 421
    },
    {
      "epoch": 0.5165238678090576,
      "grad_norm": 0.6969597178767786,
      "learning_rate": 4.9853854838562196e-06,
      "loss": 1.5377,
      "step": 422
    },
    {
      "epoch": 0.5177478580171359,
      "grad_norm": 0.8000478783932281,
      "learning_rate": 4.9853162066442745e-06,
      "loss": 0.6222,
      "step": 423
    },
    {
      "epoch": 0.5189718482252142,
      "grad_norm": 0.7666703968425487,
      "learning_rate": 4.9852467661065395e-06,
      "loss": 1.2538,
      "step": 424
    },
    {
      "epoch": 0.5201958384332925,
      "grad_norm": 0.8465989219790139,
      "learning_rate": 4.98517716224758e-06,
      "loss": 0.6529,
      "step": 425
    },
    {
      "epoch": 0.5214198286413708,
      "grad_norm": 0.9688003399358155,
      "learning_rate": 4.98510739507197e-06,
      "loss": 0.5989,
      "step": 426
    },
    {
      "epoch": 0.5226438188494492,
      "grad_norm": 0.4637067568173058,
      "learning_rate": 4.985037464584293e-06,
      "loss": 0.4672,
      "step": 427
    },
    {
      "epoch": 0.5238678090575275,
      "grad_norm": 0.4574913334463103,
      "learning_rate": 4.984967370789146e-06,
      "loss": 0.5859,
      "step": 428
    },
    {
      "epoch": 0.5250917992656059,
      "grad_norm": 0.5114573213837407,
      "learning_rate": 4.9848971136911335e-06,
      "loss": 0.4764,
      "step": 429
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 0.6809683006620858,
      "learning_rate": 4.9848266932948745e-06,
      "loss": 0.5473,
      "step": 430
    },
    {
      "epoch": 0.5275397796817626,
      "grad_norm": 0.6463385524260794,
      "learning_rate": 4.984756109604996e-06,
      "loss": 1.3814,
      "step": 431
    },
    {
      "epoch": 0.5287637698898409,
      "grad_norm": 0.40419008330189565,
      "learning_rate": 4.984685362626135e-06,
      "loss": 0.7705,
      "step": 432
    },
    {
      "epoch": 0.5299877600979193,
      "grad_norm": 0.4939438732325167,
      "learning_rate": 4.9846144523629435e-06,
      "loss": 0.7647,
      "step": 433
    },
    {
      "epoch": 0.5312117503059975,
      "grad_norm": 0.9281581150066616,
      "learning_rate": 4.98454337882008e-06,
      "loss": 0.5892,
      "step": 434
    },
    {
      "epoch": 0.5324357405140759,
      "grad_norm": 0.48154020714924345,
      "learning_rate": 4.984472142002216e-06,
      "loss": 0.6756,
      "step": 435
    },
    {
      "epoch": 0.5336597307221542,
      "grad_norm": 0.5285488763420756,
      "learning_rate": 4.984400741914032e-06,
      "loss": 0.4437,
      "step": 436
    },
    {
      "epoch": 0.5348837209302325,
      "grad_norm": 0.5774225748447701,
      "learning_rate": 4.98432917856022e-06,
      "loss": 0.8708,
      "step": 437
    },
    {
      "epoch": 0.5361077111383109,
      "grad_norm": 0.4693383447942286,
      "learning_rate": 4.9842574519454825e-06,
      "loss": 0.8761,
      "step": 438
    },
    {
      "epoch": 0.5373317013463892,
      "grad_norm": 0.4550208973916101,
      "learning_rate": 4.984185562074535e-06,
      "loss": 0.5698,
      "step": 439
    },
    {
      "epoch": 0.5385556915544676,
      "grad_norm": 0.6150252036978977,
      "learning_rate": 4.9841135089521e-06,
      "loss": 0.9911,
      "step": 440
    },
    {
      "epoch": 0.5397796817625459,
      "grad_norm": 0.5628510230051749,
      "learning_rate": 4.984041292582914e-06,
      "loss": 0.6923,
      "step": 441
    },
    {
      "epoch": 0.5410036719706243,
      "grad_norm": 1.1405020898484086,
      "learning_rate": 4.983968912971722e-06,
      "loss": 1.0072,
      "step": 442
    },
    {
      "epoch": 0.5422276621787026,
      "grad_norm": 0.6941681422603433,
      "learning_rate": 4.98389637012328e-06,
      "loss": 0.5952,
      "step": 443
    },
    {
      "epoch": 0.543451652386781,
      "grad_norm": 0.5840750201714761,
      "learning_rate": 4.983823664042357e-06,
      "loss": 0.9915,
      "step": 444
    },
    {
      "epoch": 0.5446756425948592,
      "grad_norm": 0.35238863025625233,
      "learning_rate": 4.983750794733728e-06,
      "loss": 0.4847,
      "step": 445
    },
    {
      "epoch": 0.5458996328029376,
      "grad_norm": 0.9765951182060455,
      "learning_rate": 4.983677762202185e-06,
      "loss": 0.7767,
      "step": 446
    },
    {
      "epoch": 0.5471236230110159,
      "grad_norm": 0.5762381466614729,
      "learning_rate": 4.9836045664525255e-06,
      "loss": 0.7466,
      "step": 447
    },
    {
      "epoch": 0.5483476132190942,
      "grad_norm": 0.3986092045277627,
      "learning_rate": 4.9835312074895605e-06,
      "loss": 0.6021,
      "step": 448
    },
    {
      "epoch": 0.5495716034271726,
      "grad_norm": 0.7268919269682568,
      "learning_rate": 4.9834576853181095e-06,
      "loss": 0.8209,
      "step": 449
    },
    {
      "epoch": 0.5507955936352509,
      "grad_norm": 0.6555639447350133,
      "learning_rate": 4.9833839999430065e-06,
      "loss": 0.7777,
      "step": 450
    },
    {
      "epoch": 0.5520195838433293,
      "grad_norm": 0.66561695542955,
      "learning_rate": 4.9833101513690915e-06,
      "loss": 0.742,
      "step": 451
    },
    {
      "epoch": 0.5532435740514076,
      "grad_norm": 0.494781765119413,
      "learning_rate": 4.983236139601219e-06,
      "loss": 1.0711,
      "step": 452
    },
    {
      "epoch": 0.554467564259486,
      "grad_norm": 0.721069858553008,
      "learning_rate": 4.983161964644251e-06,
      "loss": 1.1872,
      "step": 453
    },
    {
      "epoch": 0.5556915544675642,
      "grad_norm": 0.72919590776234,
      "learning_rate": 4.983087626503065e-06,
      "loss": 0.6297,
      "step": 454
    },
    {
      "epoch": 0.5569155446756426,
      "grad_norm": 0.7450867470397293,
      "learning_rate": 4.983013125182543e-06,
      "loss": 0.7224,
      "step": 455
    },
    {
      "epoch": 0.5581395348837209,
      "grad_norm": 0.4084870021019476,
      "learning_rate": 4.982938460687583e-06,
      "loss": 0.6538,
      "step": 456
    },
    {
      "epoch": 0.5593635250917993,
      "grad_norm": 0.7509680628872014,
      "learning_rate": 4.982863633023092e-06,
      "loss": 0.3776,
      "step": 457
    },
    {
      "epoch": 0.5605875152998776,
      "grad_norm": 1.087997625617814,
      "learning_rate": 4.982788642193985e-06,
      "loss": 0.6623,
      "step": 458
    },
    {
      "epoch": 0.5618115055079559,
      "grad_norm": 0.48339687799418907,
      "learning_rate": 4.9827134882051924e-06,
      "loss": 0.5704,
      "step": 459
    },
    {
      "epoch": 0.5630354957160343,
      "grad_norm": 0.5534075806710526,
      "learning_rate": 4.982638171061652e-06,
      "loss": 0.7673,
      "step": 460
    },
    {
      "epoch": 0.5642594859241126,
      "grad_norm": 0.40986544335244157,
      "learning_rate": 4.982562690768314e-06,
      "loss": 0.646,
      "step": 461
    },
    {
      "epoch": 0.565483476132191,
      "grad_norm": 0.6381486235919178,
      "learning_rate": 4.982487047330139e-06,
      "loss": 1.066,
      "step": 462
    },
    {
      "epoch": 0.5667074663402693,
      "grad_norm": 0.7125070778783265,
      "learning_rate": 4.982411240752097e-06,
      "loss": 1.3099,
      "step": 463
    },
    {
      "epoch": 0.5679314565483476,
      "grad_norm": 0.5959287446038983,
      "learning_rate": 4.982335271039169e-06,
      "loss": 0.6067,
      "step": 464
    },
    {
      "epoch": 0.5691554467564259,
      "grad_norm": 0.6086512098470025,
      "learning_rate": 4.982259138196349e-06,
      "loss": 0.7151,
      "step": 465
    },
    {
      "epoch": 0.5703794369645043,
      "grad_norm": 0.46145738752682813,
      "learning_rate": 4.98218284222864e-06,
      "loss": 0.6642,
      "step": 466
    },
    {
      "epoch": 0.5716034271725826,
      "grad_norm": 0.5922458645503862,
      "learning_rate": 4.982106383141056e-06,
      "loss": 1.2248,
      "step": 467
    },
    {
      "epoch": 0.572827417380661,
      "grad_norm": 0.5554179979490513,
      "learning_rate": 4.9820297609386195e-06,
      "loss": 1.4382,
      "step": 468
    },
    {
      "epoch": 0.5740514075887393,
      "grad_norm": 0.3869879843858493,
      "learning_rate": 4.981952975626368e-06,
      "loss": 0.6222,
      "step": 469
    },
    {
      "epoch": 0.5752753977968176,
      "grad_norm": 0.7165812862697734,
      "learning_rate": 4.981876027209348e-06,
      "loss": 1.062,
      "step": 470
    },
    {
      "epoch": 0.576499388004896,
      "grad_norm": 0.3587533524927841,
      "learning_rate": 4.981798915692615e-06,
      "loss": 0.611,
      "step": 471
    },
    {
      "epoch": 0.5777233782129743,
      "grad_norm": 0.4620040648647825,
      "learning_rate": 4.9817216410812365e-06,
      "loss": 0.6235,
      "step": 472
    },
    {
      "epoch": 0.5789473684210527,
      "grad_norm": 0.6847007914054791,
      "learning_rate": 4.981644203380292e-06,
      "loss": 0.7786,
      "step": 473
    },
    {
      "epoch": 0.5801713586291309,
      "grad_norm": 0.8634056289002575,
      "learning_rate": 4.981566602594868e-06,
      "loss": 1.3832,
      "step": 474
    },
    {
      "epoch": 0.5813953488372093,
      "grad_norm": 0.7668323236402386,
      "learning_rate": 4.981488838730066e-06,
      "loss": 1.5351,
      "step": 475
    },
    {
      "epoch": 0.5826193390452876,
      "grad_norm": 0.5460972502140486,
      "learning_rate": 4.981410911790996e-06,
      "loss": 0.6451,
      "step": 476
    },
    {
      "epoch": 0.583843329253366,
      "grad_norm": 0.4847622397163458,
      "learning_rate": 4.981332821782779e-06,
      "loss": 0.5821,
      "step": 477
    },
    {
      "epoch": 0.5850673194614443,
      "grad_norm": 0.667526872565768,
      "learning_rate": 4.9812545687105476e-06,
      "loss": 0.6665,
      "step": 478
    },
    {
      "epoch": 0.5862913096695227,
      "grad_norm": 0.7171543323825912,
      "learning_rate": 4.981176152579442e-06,
      "loss": 0.5822,
      "step": 479
    },
    {
      "epoch": 0.587515299877601,
      "grad_norm": 0.8667584694004078,
      "learning_rate": 4.981097573394618e-06,
      "loss": 0.6365,
      "step": 480
    },
    {
      "epoch": 0.5887392900856793,
      "grad_norm": 0.7365407819701043,
      "learning_rate": 4.981018831161238e-06,
      "loss": 0.7338,
      "step": 481
    },
    {
      "epoch": 0.5899632802937577,
      "grad_norm": 0.6997802356459719,
      "learning_rate": 4.980939925884477e-06,
      "loss": 0.8013,
      "step": 482
    },
    {
      "epoch": 0.591187270501836,
      "grad_norm": 0.49963032895103354,
      "learning_rate": 4.98086085756952e-06,
      "loss": 0.7466,
      "step": 483
    },
    {
      "epoch": 0.5924112607099143,
      "grad_norm": 0.8920943812354478,
      "learning_rate": 4.980781626221565e-06,
      "loss": 0.7156,
      "step": 484
    },
    {
      "epoch": 0.5936352509179926,
      "grad_norm": 0.6661321240021298,
      "learning_rate": 4.980702231845816e-06,
      "loss": 0.7356,
      "step": 485
    },
    {
      "epoch": 0.594859241126071,
      "grad_norm": 0.4170324798701388,
      "learning_rate": 4.980622674447493e-06,
      "loss": 0.6441,
      "step": 486
    },
    {
      "epoch": 0.5960832313341493,
      "grad_norm": 0.8502814114814677,
      "learning_rate": 4.980542954031822e-06,
      "loss": 0.5912,
      "step": 487
    },
    {
      "epoch": 0.5973072215422277,
      "grad_norm": 0.5836125655231348,
      "learning_rate": 4.980463070604042e-06,
      "loss": 0.8118,
      "step": 488
    },
    {
      "epoch": 0.598531211750306,
      "grad_norm": 0.8087845455291846,
      "learning_rate": 4.980383024169405e-06,
      "loss": 0.7508,
      "step": 489
    },
    {
      "epoch": 0.5997552019583844,
      "grad_norm": 0.8757690894135692,
      "learning_rate": 4.98030281473317e-06,
      "loss": 1.2787,
      "step": 490
    },
    {
      "epoch": 0.6009791921664627,
      "grad_norm": 0.7872153042853589,
      "learning_rate": 4.980222442300608e-06,
      "loss": 1.1471,
      "step": 491
    },
    {
      "epoch": 0.602203182374541,
      "grad_norm": 0.5914271112228281,
      "learning_rate": 4.980141906877001e-06,
      "loss": 0.6121,
      "step": 492
    },
    {
      "epoch": 0.6034271725826194,
      "grad_norm": 0.9602661555495826,
      "learning_rate": 4.98006120846764e-06,
      "loss": 1.3528,
      "step": 493
    },
    {
      "epoch": 0.6046511627906976,
      "grad_norm": 0.8457421366249023,
      "learning_rate": 4.979980347077831e-06,
      "loss": 0.8904,
      "step": 494
    },
    {
      "epoch": 0.605875152998776,
      "grad_norm": 0.7803270149784354,
      "learning_rate": 4.979899322712885e-06,
      "loss": 0.8407,
      "step": 495
    },
    {
      "epoch": 0.6070991432068543,
      "grad_norm": 0.9003791171851305,
      "learning_rate": 4.979818135378129e-06,
      "loss": 0.6624,
      "step": 496
    },
    {
      "epoch": 0.6083231334149327,
      "grad_norm": 0.38048113984728554,
      "learning_rate": 4.979736785078897e-06,
      "loss": 0.4568,
      "step": 497
    },
    {
      "epoch": 0.609547123623011,
      "grad_norm": 0.5460484270490091,
      "learning_rate": 4.979655271820535e-06,
      "loss": 0.8807,
      "step": 498
    },
    {
      "epoch": 0.6107711138310894,
      "grad_norm": 0.8607446671430177,
      "learning_rate": 4.9795735956084015e-06,
      "loss": 2.4248,
      "step": 499
    },
    {
      "epoch": 0.6119951040391677,
      "grad_norm": 0.8657540161264972,
      "learning_rate": 4.9794917564478616e-06,
      "loss": 1.0421,
      "step": 500
    },
    {
      "epoch": 0.6132190942472461,
      "grad_norm": 0.786380900764491,
      "learning_rate": 4.979409754344294e-06,
      "loss": 1.1132,
      "step": 501
    },
    {
      "epoch": 0.6144430844553244,
      "grad_norm": 0.5541101339567411,
      "learning_rate": 4.979327589303089e-06,
      "loss": 0.6173,
      "step": 502
    },
    {
      "epoch": 0.6156670746634026,
      "grad_norm": 0.7338735579088909,
      "learning_rate": 4.979245261329645e-06,
      "loss": 0.9687,
      "step": 503
    },
    {
      "epoch": 0.616891064871481,
      "grad_norm": 0.6086756002681308,
      "learning_rate": 4.979162770429372e-06,
      "loss": 0.5935,
      "step": 504
    },
    {
      "epoch": 0.6181150550795593,
      "grad_norm": 0.7954198181003124,
      "learning_rate": 4.979080116607692e-06,
      "loss": 0.5992,
      "step": 505
    },
    {
      "epoch": 0.6193390452876377,
      "grad_norm": 0.5815447401847034,
      "learning_rate": 4.978997299870036e-06,
      "loss": 0.8095,
      "step": 506
    },
    {
      "epoch": 0.620563035495716,
      "grad_norm": 0.6904210968325463,
      "learning_rate": 4.978914320221847e-06,
      "loss": 1.0425,
      "step": 507
    },
    {
      "epoch": 0.6217870257037944,
      "grad_norm": 0.6509154661965433,
      "learning_rate": 4.978831177668577e-06,
      "loss": 0.6879,
      "step": 508
    },
    {
      "epoch": 0.6230110159118727,
      "grad_norm": 0.32876693792041606,
      "learning_rate": 4.978747872215691e-06,
      "loss": 0.2984,
      "step": 509
    },
    {
      "epoch": 0.6242350061199511,
      "grad_norm": 0.7048658689041563,
      "learning_rate": 4.978664403868664e-06,
      "loss": 0.7401,
      "step": 510
    },
    {
      "epoch": 0.6254589963280294,
      "grad_norm": 0.6802817915400121,
      "learning_rate": 4.978580772632979e-06,
      "loss": 1.2961,
      "step": 511
    },
    {
      "epoch": 0.6266829865361077,
      "grad_norm": 0.9447830556444728,
      "learning_rate": 4.9784969785141334e-06,
      "loss": 0.6635,
      "step": 512
    },
    {
      "epoch": 0.627906976744186,
      "grad_norm": 0.605067025896805,
      "learning_rate": 4.978413021517634e-06,
      "loss": 0.6016,
      "step": 513
    },
    {
      "epoch": 0.6291309669522643,
      "grad_norm": 0.6973401878576624,
      "learning_rate": 4.9783289016489975e-06,
      "loss": 1.7901,
      "step": 514
    },
    {
      "epoch": 0.6303549571603427,
      "grad_norm": 0.5743499259562451,
      "learning_rate": 4.978244618913754e-06,
      "loss": 0.6964,
      "step": 515
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 0.5765570989674189,
      "learning_rate": 4.978160173317439e-06,
      "loss": 1.2022,
      "step": 516
    },
    {
      "epoch": 0.6328029375764994,
      "grad_norm": 0.676210526239049,
      "learning_rate": 4.978075564865603e-06,
      "loss": 0.5173,
      "step": 517
    },
    {
      "epoch": 0.6340269277845777,
      "grad_norm": 0.6101515429201264,
      "learning_rate": 4.9779907935638076e-06,
      "loss": 0.7999,
      "step": 518
    },
    {
      "epoch": 0.6352509179926561,
      "grad_norm": 0.625029551653209,
      "learning_rate": 4.977905859417622e-06,
      "loss": 1.0577,
      "step": 519
    },
    {
      "epoch": 0.6364749082007344,
      "grad_norm": 0.7728096552634098,
      "learning_rate": 4.9778207624326294e-06,
      "loss": 0.5477,
      "step": 520
    },
    {
      "epoch": 0.6376988984088128,
      "grad_norm": 0.43009825366727467,
      "learning_rate": 4.97773550261442e-06,
      "loss": 0.6058,
      "step": 521
    },
    {
      "epoch": 0.6389228886168911,
      "grad_norm": 0.8387121402128179,
      "learning_rate": 4.977650079968599e-06,
      "loss": 0.8564,
      "step": 522
    },
    {
      "epoch": 0.6401468788249693,
      "grad_norm": 0.5950306634511766,
      "learning_rate": 4.977564494500778e-06,
      "loss": 1.1089,
      "step": 523
    },
    {
      "epoch": 0.6413708690330477,
      "grad_norm": 0.5943989281202401,
      "learning_rate": 4.977478746216583e-06,
      "loss": 0.6134,
      "step": 524
    },
    {
      "epoch": 0.642594859241126,
      "grad_norm": 0.729651571226729,
      "learning_rate": 4.977392835121649e-06,
      "loss": 0.7043,
      "step": 525
    },
    {
      "epoch": 0.6438188494492044,
      "grad_norm": 0.9178273079178908,
      "learning_rate": 4.9773067612216194e-06,
      "loss": 0.7013,
      "step": 526
    },
    {
      "epoch": 0.6450428396572827,
      "grad_norm": 0.7462894848550342,
      "learning_rate": 4.9772205245221525e-06,
      "loss": 1.0939,
      "step": 527
    },
    {
      "epoch": 0.6462668298653611,
      "grad_norm": 0.7004286005767373,
      "learning_rate": 4.977134125028916e-06,
      "loss": 0.5541,
      "step": 528
    },
    {
      "epoch": 0.6474908200734394,
      "grad_norm": 0.9715965342324924,
      "learning_rate": 4.977047562747588e-06,
      "loss": 0.59,
      "step": 529
    },
    {
      "epoch": 0.6487148102815178,
      "grad_norm": 0.744286503821514,
      "learning_rate": 4.9769608376838555e-06,
      "loss": 1.1145,
      "step": 530
    },
    {
      "epoch": 0.6499388004895961,
      "grad_norm": 0.9508178288411085,
      "learning_rate": 4.976873949843417e-06,
      "loss": 0.7731,
      "step": 531
    },
    {
      "epoch": 0.6511627906976745,
      "grad_norm": 0.780669900476306,
      "learning_rate": 4.976786899231986e-06,
      "loss": 0.7806,
      "step": 532
    },
    {
      "epoch": 0.6523867809057528,
      "grad_norm": 0.5254012370160568,
      "learning_rate": 4.976699685855279e-06,
      "loss": 0.6822,
      "step": 533
    },
    {
      "epoch": 0.653610771113831,
      "grad_norm": 0.6146151754181215,
      "learning_rate": 4.976612309719031e-06,
      "loss": 0.8659,
      "step": 534
    },
    {
      "epoch": 0.6548347613219094,
      "grad_norm": 0.9731637117312572,
      "learning_rate": 4.976524770828981e-06,
      "loss": 0.9978,
      "step": 535
    },
    {
      "epoch": 0.6560587515299877,
      "grad_norm": 0.80898410376378,
      "learning_rate": 4.976437069190884e-06,
      "loss": 1.9593,
      "step": 536
    },
    {
      "epoch": 0.6572827417380661,
      "grad_norm": 0.49631139959251924,
      "learning_rate": 4.976349204810502e-06,
      "loss": 0.6042,
      "step": 537
    },
    {
      "epoch": 0.6585067319461444,
      "grad_norm": 0.5877262837299358,
      "learning_rate": 4.97626117769361e-06,
      "loss": 0.7233,
      "step": 538
    },
    {
      "epoch": 0.6597307221542228,
      "grad_norm": 0.5006659854179215,
      "learning_rate": 4.976172987845992e-06,
      "loss": 0.4859,
      "step": 539
    },
    {
      "epoch": 0.6609547123623011,
      "grad_norm": 0.7243688811014526,
      "learning_rate": 4.976084635273445e-06,
      "loss": 1.4443,
      "step": 540
    },
    {
      "epoch": 0.6621787025703795,
      "grad_norm": 0.5788182162667203,
      "learning_rate": 4.975996119981773e-06,
      "loss": 1.2865,
      "step": 541
    },
    {
      "epoch": 0.6634026927784578,
      "grad_norm": 0.7618498816492038,
      "learning_rate": 4.975907441976795e-06,
      "loss": 0.6311,
      "step": 542
    },
    {
      "epoch": 0.6646266829865362,
      "grad_norm": 0.9105223655593707,
      "learning_rate": 4.975818601264337e-06,
      "loss": 0.8695,
      "step": 543
    },
    {
      "epoch": 0.6658506731946144,
      "grad_norm": 0.5709587329130438,
      "learning_rate": 4.975729597850238e-06,
      "loss": 0.7136,
      "step": 544
    },
    {
      "epoch": 0.6670746634026927,
      "grad_norm": 0.5709572171217463,
      "learning_rate": 4.9756404317403475e-06,
      "loss": 1.4518,
      "step": 545
    },
    {
      "epoch": 0.6682986536107711,
      "grad_norm": 0.7137365348802489,
      "learning_rate": 4.9755511029405235e-06,
      "loss": 0.6105,
      "step": 546
    },
    {
      "epoch": 0.6695226438188494,
      "grad_norm": 0.9472825614416733,
      "learning_rate": 4.975461611456639e-06,
      "loss": 1.0258,
      "step": 547
    },
    {
      "epoch": 0.6707466340269278,
      "grad_norm": 0.5948426326943427,
      "learning_rate": 4.975371957294572e-06,
      "loss": 0.7631,
      "step": 548
    },
    {
      "epoch": 0.6719706242350061,
      "grad_norm": 0.49442036024015285,
      "learning_rate": 4.975282140460216e-06,
      "loss": 0.5656,
      "step": 549
    },
    {
      "epoch": 0.6731946144430845,
      "grad_norm": 0.5141243670686299,
      "learning_rate": 4.975192160959474e-06,
      "loss": 0.6187,
      "step": 550
    },
    {
      "epoch": 0.6744186046511628,
      "grad_norm": 0.6447537114880092,
      "learning_rate": 4.975102018798258e-06,
      "loss": 0.6707,
      "step": 551
    },
    {
      "epoch": 0.6756425948592412,
      "grad_norm": 1.0498450919663247,
      "learning_rate": 4.9750117139824916e-06,
      "loss": 0.8024,
      "step": 552
    },
    {
      "epoch": 0.6768665850673194,
      "grad_norm": 0.3967421027806449,
      "learning_rate": 4.9749212465181095e-06,
      "loss": 0.4971,
      "step": 553
    },
    {
      "epoch": 0.6780905752753978,
      "grad_norm": 0.4999352613405446,
      "learning_rate": 4.974830616411059e-06,
      "loss": 0.6739,
      "step": 554
    },
    {
      "epoch": 0.6793145654834761,
      "grad_norm": 0.5726588192115494,
      "learning_rate": 4.974739823667292e-06,
      "loss": 0.978,
      "step": 555
    },
    {
      "epoch": 0.6805385556915544,
      "grad_norm": 0.723750119249683,
      "learning_rate": 4.974648868292778e-06,
      "loss": 0.5329,
      "step": 556
    },
    {
      "epoch": 0.6817625458996328,
      "grad_norm": 0.5514615518789243,
      "learning_rate": 4.9745577502934936e-06,
      "loss": 1.1273,
      "step": 557
    },
    {
      "epoch": 0.6829865361077111,
      "grad_norm": 0.6237740221028041,
      "learning_rate": 4.974466469675427e-06,
      "loss": 0.7259,
      "step": 558
    },
    {
      "epoch": 0.6842105263157895,
      "grad_norm": 0.441421519775364,
      "learning_rate": 4.974375026444575e-06,
      "loss": 0.6798,
      "step": 559
    },
    {
      "epoch": 0.6854345165238678,
      "grad_norm": 0.9338845353458881,
      "learning_rate": 4.97428342060695e-06,
      "loss": 0.7226,
      "step": 560
    },
    {
      "epoch": 0.6866585067319462,
      "grad_norm": 0.6267436130602022,
      "learning_rate": 4.974191652168569e-06,
      "loss": 0.6096,
      "step": 561
    },
    {
      "epoch": 0.6878824969400245,
      "grad_norm": 0.8211413779042164,
      "learning_rate": 4.974099721135466e-06,
      "loss": 1.5565,
      "step": 562
    },
    {
      "epoch": 0.6891064871481029,
      "grad_norm": 0.6910690891205508,
      "learning_rate": 4.974007627513678e-06,
      "loss": 1.2202,
      "step": 563
    },
    {
      "epoch": 0.6903304773561811,
      "grad_norm": 0.5917342611493596,
      "learning_rate": 4.973915371309262e-06,
      "loss": 0.9193,
      "step": 564
    },
    {
      "epoch": 0.6915544675642595,
      "grad_norm": 1.0435096061682623,
      "learning_rate": 4.973822952528276e-06,
      "loss": 1.5729,
      "step": 565
    },
    {
      "epoch": 0.6927784577723378,
      "grad_norm": 0.6072455238693985,
      "learning_rate": 4.973730371176796e-06,
      "loss": 0.6981,
      "step": 566
    },
    {
      "epoch": 0.6940024479804161,
      "grad_norm": 0.7833292444100382,
      "learning_rate": 4.973637627260907e-06,
      "loss": 0.6399,
      "step": 567
    },
    {
      "epoch": 0.6952264381884945,
      "grad_norm": 0.5908805712931489,
      "learning_rate": 4.973544720786701e-06,
      "loss": 0.9483,
      "step": 568
    },
    {
      "epoch": 0.6964504283965728,
      "grad_norm": 0.9507882820554344,
      "learning_rate": 4.973451651760286e-06,
      "loss": 1.1759,
      "step": 569
    },
    {
      "epoch": 0.6976744186046512,
      "grad_norm": 0.5395558365491222,
      "learning_rate": 4.973358420187776e-06,
      "loss": 0.6891,
      "step": 570
    },
    {
      "epoch": 0.6988984088127295,
      "grad_norm": 0.8226985692636989,
      "learning_rate": 4.973265026075299e-06,
      "loss": 0.9661,
      "step": 571
    },
    {
      "epoch": 0.7001223990208079,
      "grad_norm": 0.8020154739677174,
      "learning_rate": 4.973171469428993e-06,
      "loss": 0.6196,
      "step": 572
    },
    {
      "epoch": 0.7013463892288861,
      "grad_norm": 0.4430199949633676,
      "learning_rate": 4.973077750255006e-06,
      "loss": 0.6245,
      "step": 573
    },
    {
      "epoch": 0.7025703794369645,
      "grad_norm": 0.8217842159188796,
      "learning_rate": 4.972983868559497e-06,
      "loss": 1.086,
      "step": 574
    },
    {
      "epoch": 0.7037943696450428,
      "grad_norm": 0.6507059382623177,
      "learning_rate": 4.9728898243486335e-06,
      "loss": 0.5705,
      "step": 575
    },
    {
      "epoch": 0.7050183598531212,
      "grad_norm": 0.5685690088846183,
      "learning_rate": 4.972795617628598e-06,
      "loss": 0.7152,
      "step": 576
    },
    {
      "epoch": 0.7062423500611995,
      "grad_norm": 0.5979174639210477,
      "learning_rate": 4.972701248405581e-06,
      "loss": 0.988,
      "step": 577
    },
    {
      "epoch": 0.7074663402692778,
      "grad_norm": 0.6342720690962881,
      "learning_rate": 4.972606716685784e-06,
      "loss": 1.0455,
      "step": 578
    },
    {
      "epoch": 0.7086903304773562,
      "grad_norm": 0.5674760213224617,
      "learning_rate": 4.9725120224754184e-06,
      "loss": 1.0505,
      "step": 579
    },
    {
      "epoch": 0.7099143206854345,
      "grad_norm": 0.5223678278906756,
      "learning_rate": 4.972417165780709e-06,
      "loss": 0.7916,
      "step": 580
    },
    {
      "epoch": 0.7111383108935129,
      "grad_norm": 0.7098437120298289,
      "learning_rate": 4.972322146607887e-06,
      "loss": 0.7439,
      "step": 581
    },
    {
      "epoch": 0.7123623011015912,
      "grad_norm": 0.5554196168253557,
      "learning_rate": 4.972226964963198e-06,
      "loss": 0.7848,
      "step": 582
    },
    {
      "epoch": 0.7135862913096696,
      "grad_norm": 0.8028778572167714,
      "learning_rate": 4.972131620852898e-06,
      "loss": 1.0801,
      "step": 583
    },
    {
      "epoch": 0.7148102815177478,
      "grad_norm": 0.8788064974366373,
      "learning_rate": 4.9720361142832516e-06,
      "loss": 1.1014,
      "step": 584
    },
    {
      "epoch": 0.7160342717258262,
      "grad_norm": 1.0507662863719238,
      "learning_rate": 4.9719404452605345e-06,
      "loss": 0.5231,
      "step": 585
    },
    {
      "epoch": 0.7172582619339045,
      "grad_norm": 0.8519649680014808,
      "learning_rate": 4.971844613791035e-06,
      "loss": 0.7638,
      "step": 586
    },
    {
      "epoch": 0.7184822521419829,
      "grad_norm": 0.8676498937975138,
      "learning_rate": 4.97174861988105e-06,
      "loss": 0.4428,
      "step": 587
    },
    {
      "epoch": 0.7197062423500612,
      "grad_norm": 0.39356823885394826,
      "learning_rate": 4.971652463536888e-06,
      "loss": 0.4902,
      "step": 588
    },
    {
      "epoch": 0.7209302325581395,
      "grad_norm": 0.7042322941656949,
      "learning_rate": 4.971556144764868e-06,
      "loss": 1.658,
      "step": 589
    },
    {
      "epoch": 0.7221542227662179,
      "grad_norm": 1.1854771058805402,
      "learning_rate": 4.97145966357132e-06,
      "loss": 0.8904,
      "step": 590
    },
    {
      "epoch": 0.7233782129742962,
      "grad_norm": 0.5708757192214396,
      "learning_rate": 4.971363019962584e-06,
      "loss": 0.7227,
      "step": 591
    },
    {
      "epoch": 0.7246022031823746,
      "grad_norm": 0.6772516152524453,
      "learning_rate": 4.971266213945012e-06,
      "loss": 0.4588,
      "step": 592
    },
    {
      "epoch": 0.7258261933904528,
      "grad_norm": 1.0012987474314943,
      "learning_rate": 4.971169245524964e-06,
      "loss": 0.6288,
      "step": 593
    },
    {
      "epoch": 0.7270501835985312,
      "grad_norm": 0.9484340869296382,
      "learning_rate": 4.971072114708813e-06,
      "loss": 0.581,
      "step": 594
    },
    {
      "epoch": 0.7282741738066095,
      "grad_norm": 0.615108701886359,
      "learning_rate": 4.970974821502944e-06,
      "loss": 0.6183,
      "step": 595
    },
    {
      "epoch": 0.7294981640146879,
      "grad_norm": 0.7012882258838486,
      "learning_rate": 4.970877365913749e-06,
      "loss": 0.5819,
      "step": 596
    },
    {
      "epoch": 0.7307221542227662,
      "grad_norm": 0.47208445383336833,
      "learning_rate": 4.970779747947632e-06,
      "loss": 0.5465,
      "step": 597
    },
    {
      "epoch": 0.7319461444308446,
      "grad_norm": 0.513142195829669,
      "learning_rate": 4.970681967611009e-06,
      "loss": 0.4998,
      "step": 598
    },
    {
      "epoch": 0.7331701346389229,
      "grad_norm": 0.4816548410982356,
      "learning_rate": 4.9705840249103054e-06,
      "loss": 0.6552,
      "step": 599
    },
    {
      "epoch": 0.7343941248470012,
      "grad_norm": 0.6431509174434921,
      "learning_rate": 4.970485919851958e-06,
      "loss": 1.3001,
      "step": 600
    },
    {
      "epoch": 0.7356181150550796,
      "grad_norm": 0.712352793323645,
      "learning_rate": 4.970387652442413e-06,
      "loss": 1.4916,
      "step": 601
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 1.0396576730053728,
      "learning_rate": 4.970289222688129e-06,
      "loss": 1.1944,
      "step": 602
    },
    {
      "epoch": 0.7380660954712362,
      "grad_norm": 0.7592000215963804,
      "learning_rate": 4.970190630595575e-06,
      "loss": 0.8748,
      "step": 603
    },
    {
      "epoch": 0.7392900856793145,
      "grad_norm": 0.9156707634423166,
      "learning_rate": 4.970091876171229e-06,
      "loss": 0.594,
      "step": 604
    },
    {
      "epoch": 0.7405140758873929,
      "grad_norm": 0.6451404632893534,
      "learning_rate": 4.969992959421581e-06,
      "loss": 0.6855,
      "step": 605
    },
    {
      "epoch": 0.7417380660954712,
      "grad_norm": 0.46130620716085075,
      "learning_rate": 4.969893880353132e-06,
      "loss": 0.6904,
      "step": 606
    },
    {
      "epoch": 0.7429620563035496,
      "grad_norm": 1.0270600784677892,
      "learning_rate": 4.969794638972392e-06,
      "loss": 0.7618,
      "step": 607
    },
    {
      "epoch": 0.7441860465116279,
      "grad_norm": 0.7836370519805019,
      "learning_rate": 4.9696952352858844e-06,
      "loss": 0.705,
      "step": 608
    },
    {
      "epoch": 0.7454100367197063,
      "grad_norm": 0.39752796690613,
      "learning_rate": 4.9695956693001394e-06,
      "loss": 0.5664,
      "step": 609
    },
    {
      "epoch": 0.7466340269277846,
      "grad_norm": 0.7742064722344367,
      "learning_rate": 4.969495941021702e-06,
      "loss": 0.6021,
      "step": 610
    },
    {
      "epoch": 0.7478580171358629,
      "grad_norm": 0.9269461702870032,
      "learning_rate": 4.969396050457126e-06,
      "loss": 1.3448,
      "step": 611
    },
    {
      "epoch": 0.7490820073439413,
      "grad_norm": 0.5962969508176195,
      "learning_rate": 4.969295997612974e-06,
      "loss": 0.5883,
      "step": 612
    },
    {
      "epoch": 0.7503059975520195,
      "grad_norm": 0.998690953858965,
      "learning_rate": 4.969195782495823e-06,
      "loss": 0.6252,
      "step": 613
    },
    {
      "epoch": 0.7515299877600979,
      "grad_norm": 0.6369734465150525,
      "learning_rate": 4.969095405112259e-06,
      "loss": 0.5243,
      "step": 614
    },
    {
      "epoch": 0.7527539779681762,
      "grad_norm": 0.5883419096865912,
      "learning_rate": 4.968994865468876e-06,
      "loss": 0.9644,
      "step": 615
    },
    {
      "epoch": 0.7539779681762546,
      "grad_norm": 0.6266375168065322,
      "learning_rate": 4.9688941635722835e-06,
      "loss": 0.6935,
      "step": 616
    },
    {
      "epoch": 0.7552019583843329,
      "grad_norm": 0.6529873979701071,
      "learning_rate": 4.968793299429098e-06,
      "loss": 1.3436,
      "step": 617
    },
    {
      "epoch": 0.7564259485924113,
      "grad_norm": 0.46924977448314387,
      "learning_rate": 4.968692273045949e-06,
      "loss": 0.779,
      "step": 618
    },
    {
      "epoch": 0.7576499388004896,
      "grad_norm": 0.4417891405151123,
      "learning_rate": 4.968591084429474e-06,
      "loss": 0.7529,
      "step": 619
    },
    {
      "epoch": 0.758873929008568,
      "grad_norm": 0.5595233063808529,
      "learning_rate": 4.968489733586324e-06,
      "loss": 0.5163,
      "step": 620
    },
    {
      "epoch": 0.7600979192166463,
      "grad_norm": 0.687497280904428,
      "learning_rate": 4.968388220523158e-06,
      "loss": 0.849,
      "step": 621
    },
    {
      "epoch": 0.7613219094247246,
      "grad_norm": 0.8279073121259894,
      "learning_rate": 4.96828654524665e-06,
      "loss": 0.6802,
      "step": 622
    },
    {
      "epoch": 0.762545899632803,
      "grad_norm": 0.4282792593944984,
      "learning_rate": 4.968184707763478e-06,
      "loss": 0.5985,
      "step": 623
    },
    {
      "epoch": 0.7637698898408812,
      "grad_norm": 0.7291351662089038,
      "learning_rate": 4.968082708080337e-06,
      "loss": 0.7812,
      "step": 624
    },
    {
      "epoch": 0.7649938800489596,
      "grad_norm": 1.1061164288585412,
      "learning_rate": 4.96798054620393e-06,
      "loss": 1.3159,
      "step": 625
    },
    {
      "epoch": 0.7662178702570379,
      "grad_norm": 1.2197055829355696,
      "learning_rate": 4.967878222140969e-06,
      "loss": 1.1424,
      "step": 626
    },
    {
      "epoch": 0.7674418604651163,
      "grad_norm": 0.6475891215948089,
      "learning_rate": 4.967775735898179e-06,
      "loss": 0.8342,
      "step": 627
    },
    {
      "epoch": 0.7686658506731946,
      "grad_norm": 1.0532057669413615,
      "learning_rate": 4.9676730874822965e-06,
      "loss": 0.6239,
      "step": 628
    },
    {
      "epoch": 0.769889840881273,
      "grad_norm": 0.809246997189568,
      "learning_rate": 4.967570276900065e-06,
      "loss": 0.6994,
      "step": 629
    },
    {
      "epoch": 0.7711138310893513,
      "grad_norm": 0.7892053930779026,
      "learning_rate": 4.967467304158242e-06,
      "loss": 0.9058,
      "step": 630
    },
    {
      "epoch": 0.7723378212974297,
      "grad_norm": 1.0314193793563253,
      "learning_rate": 4.967364169263594e-06,
      "loss": 1.1671,
      "step": 631
    },
    {
      "epoch": 0.773561811505508,
      "grad_norm": 0.9123532849570993,
      "learning_rate": 4.9672608722228995e-06,
      "loss": 0.525,
      "step": 632
    },
    {
      "epoch": 0.7747858017135862,
      "grad_norm": 0.8480696357200321,
      "learning_rate": 4.967157413042946e-06,
      "loss": 1.5879,
      "step": 633
    },
    {
      "epoch": 0.7760097919216646,
      "grad_norm": 0.8596720181825599,
      "learning_rate": 4.967053791730533e-06,
      "loss": 0.5371,
      "step": 634
    },
    {
      "epoch": 0.7772337821297429,
      "grad_norm": 0.6211059219747544,
      "learning_rate": 4.96695000829247e-06,
      "loss": 0.4747,
      "step": 635
    },
    {
      "epoch": 0.7784577723378213,
      "grad_norm": 0.4629490378032919,
      "learning_rate": 4.966846062735576e-06,
      "loss": 0.5684,
      "step": 636
    },
    {
      "epoch": 0.7796817625458996,
      "grad_norm": 0.806056164140906,
      "learning_rate": 4.966741955066684e-06,
      "loss": 0.6234,
      "step": 637
    },
    {
      "epoch": 0.780905752753978,
      "grad_norm": 0.7714680181902983,
      "learning_rate": 4.966637685292635e-06,
      "loss": 0.5484,
      "step": 638
    },
    {
      "epoch": 0.7821297429620563,
      "grad_norm": 0.9269158255118782,
      "learning_rate": 4.966533253420279e-06,
      "loss": 0.643,
      "step": 639
    },
    {
      "epoch": 0.7833537331701347,
      "grad_norm": 0.39727557307009653,
      "learning_rate": 4.966428659456482e-06,
      "loss": 0.5241,
      "step": 640
    },
    {
      "epoch": 0.784577723378213,
      "grad_norm": 0.5498888346071843,
      "learning_rate": 4.966323903408116e-06,
      "loss": 0.6624,
      "step": 641
    },
    {
      "epoch": 0.7858017135862914,
      "grad_norm": 0.9582460661242013,
      "learning_rate": 4.966218985282065e-06,
      "loss": 1.102,
      "step": 642
    },
    {
      "epoch": 0.7870257037943696,
      "grad_norm": 0.6924637960627974,
      "learning_rate": 4.966113905085225e-06,
      "loss": 1.4242,
      "step": 643
    },
    {
      "epoch": 0.7882496940024479,
      "grad_norm": 0.6435657697670896,
      "learning_rate": 4.9660086628245e-06,
      "loss": 0.7195,
      "step": 644
    },
    {
      "epoch": 0.7894736842105263,
      "grad_norm": 0.8642952500877826,
      "learning_rate": 4.965903258506806e-06,
      "loss": 1.5262,
      "step": 645
    },
    {
      "epoch": 0.7906976744186046,
      "grad_norm": 1.0681830827384629,
      "learning_rate": 4.9657976921390714e-06,
      "loss": 0.6774,
      "step": 646
    },
    {
      "epoch": 0.791921664626683,
      "grad_norm": 1.0524664740300034,
      "learning_rate": 4.965691963728233e-06,
      "loss": 1.0384,
      "step": 647
    },
    {
      "epoch": 0.7931456548347613,
      "grad_norm": 0.6254499843442785,
      "learning_rate": 4.965586073281239e-06,
      "loss": 0.7753,
      "step": 648
    },
    {
      "epoch": 0.7943696450428397,
      "grad_norm": 0.6386283497926499,
      "learning_rate": 4.965480020805047e-06,
      "loss": 0.6537,
      "step": 649
    },
    {
      "epoch": 0.795593635250918,
      "grad_norm": 0.5542214681334583,
      "learning_rate": 4.9653738063066285e-06,
      "loss": 0.4985,
      "step": 650
    },
    {
      "epoch": 0.7968176254589964,
      "grad_norm": 0.7385661588704525,
      "learning_rate": 4.965267429792961e-06,
      "loss": 0.7591,
      "step": 651
    },
    {
      "epoch": 0.7980416156670747,
      "grad_norm": 0.7988023399399428,
      "learning_rate": 4.965160891271037e-06,
      "loss": 1.4882,
      "step": 652
    },
    {
      "epoch": 0.799265605875153,
      "grad_norm": 0.6219625214624916,
      "learning_rate": 4.9650541907478566e-06,
      "loss": 0.5383,
      "step": 653
    },
    {
      "epoch": 0.8004895960832313,
      "grad_norm": 0.6750578805333489,
      "learning_rate": 4.964947328230433e-06,
      "loss": 0.7228,
      "step": 654
    },
    {
      "epoch": 0.8017135862913096,
      "grad_norm": 0.611696364581251,
      "learning_rate": 4.964840303725788e-06,
      "loss": 0.736,
      "step": 655
    },
    {
      "epoch": 0.802937576499388,
      "grad_norm": 0.8365574692074819,
      "learning_rate": 4.964733117240955e-06,
      "loss": 0.7894,
      "step": 656
    },
    {
      "epoch": 0.8041615667074663,
      "grad_norm": 0.37500351164507584,
      "learning_rate": 4.964625768782978e-06,
      "loss": 0.4694,
      "step": 657
    },
    {
      "epoch": 0.8053855569155447,
      "grad_norm": 0.8599796209464186,
      "learning_rate": 4.964518258358912e-06,
      "loss": 0.8291,
      "step": 658
    },
    {
      "epoch": 0.806609547123623,
      "grad_norm": 0.7475307753112096,
      "learning_rate": 4.964410585975821e-06,
      "loss": 1.5502,
      "step": 659
    },
    {
      "epoch": 0.8078335373317014,
      "grad_norm": 0.8015958302816727,
      "learning_rate": 4.964302751640782e-06,
      "loss": 1.326,
      "step": 660
    },
    {
      "epoch": 0.8090575275397797,
      "grad_norm": 1.0528153516904397,
      "learning_rate": 4.9641947553608805e-06,
      "loss": 1.0994,
      "step": 661
    },
    {
      "epoch": 0.8102815177478581,
      "grad_norm": 0.6757406701884192,
      "learning_rate": 4.964086597143214e-06,
      "loss": 0.816,
      "step": 662
    },
    {
      "epoch": 0.8115055079559363,
      "grad_norm": 0.6082119601165287,
      "learning_rate": 4.963978276994891e-06,
      "loss": 0.8588,
      "step": 663
    },
    {
      "epoch": 0.8127294981640147,
      "grad_norm": 0.42903926514342355,
      "learning_rate": 4.963869794923029e-06,
      "loss": 0.6817,
      "step": 664
    },
    {
      "epoch": 0.813953488372093,
      "grad_norm": 0.8183416215602448,
      "learning_rate": 4.963761150934757e-06,
      "loss": 0.8746,
      "step": 665
    },
    {
      "epoch": 0.8151774785801713,
      "grad_norm": 0.5790604514629629,
      "learning_rate": 4.963652345037215e-06,
      "loss": 0.5978,
      "step": 666
    },
    {
      "epoch": 0.8164014687882497,
      "grad_norm": 0.9841913474099444,
      "learning_rate": 4.963543377237554e-06,
      "loss": 0.8679,
      "step": 667
    },
    {
      "epoch": 0.817625458996328,
      "grad_norm": 0.5242588053675499,
      "learning_rate": 4.963434247542934e-06,
      "loss": 0.6791,
      "step": 668
    },
    {
      "epoch": 0.8188494492044064,
      "grad_norm": 0.6524840579850808,
      "learning_rate": 4.9633249559605264e-06,
      "loss": 0.9256,
      "step": 669
    },
    {
      "epoch": 0.8200734394124847,
      "grad_norm": 0.5638342011037669,
      "learning_rate": 4.963215502497514e-06,
      "loss": 0.9675,
      "step": 670
    },
    {
      "epoch": 0.8212974296205631,
      "grad_norm": 0.5582692750055275,
      "learning_rate": 4.96310588716109e-06,
      "loss": 0.8604,
      "step": 671
    },
    {
      "epoch": 0.8225214198286414,
      "grad_norm": 0.7894170377790417,
      "learning_rate": 4.9629961099584575e-06,
      "loss": 1.3202,
      "step": 672
    },
    {
      "epoch": 0.8237454100367197,
      "grad_norm": 0.768432594533552,
      "learning_rate": 4.96288617089683e-06,
      "loss": 0.4463,
      "step": 673
    },
    {
      "epoch": 0.824969400244798,
      "grad_norm": 0.5474461322631754,
      "learning_rate": 4.962776069983434e-06,
      "loss": 1.0516,
      "step": 674
    },
    {
      "epoch": 0.8261933904528764,
      "grad_norm": 0.9023793086012241,
      "learning_rate": 4.962665807225503e-06,
      "loss": 0.7088,
      "step": 675
    },
    {
      "epoch": 0.8274173806609547,
      "grad_norm": 0.8770622743672146,
      "learning_rate": 4.962555382630284e-06,
      "loss": 0.5896,
      "step": 676
    },
    {
      "epoch": 0.828641370869033,
      "grad_norm": 0.40190443269097986,
      "learning_rate": 4.962444796205034e-06,
      "loss": 0.4439,
      "step": 677
    },
    {
      "epoch": 0.8298653610771114,
      "grad_norm": 0.9237737974360988,
      "learning_rate": 4.96233404795702e-06,
      "loss": 0.6128,
      "step": 678
    },
    {
      "epoch": 0.8310893512851897,
      "grad_norm": 0.6992319859751572,
      "learning_rate": 4.962223137893518e-06,
      "loss": 1.3304,
      "step": 679
    },
    {
      "epoch": 0.8323133414932681,
      "grad_norm": 0.6258347473825795,
      "learning_rate": 4.96211206602182e-06,
      "loss": 0.5668,
      "step": 680
    },
    {
      "epoch": 0.8335373317013464,
      "grad_norm": 0.8470335439349213,
      "learning_rate": 4.962000832349224e-06,
      "loss": 0.522,
      "step": 681
    },
    {
      "epoch": 0.8347613219094248,
      "grad_norm": 1.1234994612112508,
      "learning_rate": 4.9618894368830396e-06,
      "loss": 0.4874,
      "step": 682
    },
    {
      "epoch": 0.835985312117503,
      "grad_norm": 1.0489500450141993,
      "learning_rate": 4.961777879630587e-06,
      "loss": 0.6158,
      "step": 683
    },
    {
      "epoch": 0.8372093023255814,
      "grad_norm": 0.6241119450309481,
      "learning_rate": 4.961666160599198e-06,
      "loss": 0.8178,
      "step": 684
    },
    {
      "epoch": 0.8384332925336597,
      "grad_norm": 1.0682003904914388,
      "learning_rate": 4.961554279796214e-06,
      "loss": 0.5875,
      "step": 685
    },
    {
      "epoch": 0.8396572827417381,
      "grad_norm": 0.6472953076353286,
      "learning_rate": 4.961442237228987e-06,
      "loss": 0.6897,
      "step": 686
    },
    {
      "epoch": 0.8408812729498164,
      "grad_norm": 0.485126674032805,
      "learning_rate": 4.961330032904881e-06,
      "loss": 0.9089,
      "step": 687
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.939181228389836,
      "learning_rate": 4.961217666831268e-06,
      "loss": 0.5114,
      "step": 688
    },
    {
      "epoch": 0.8433292533659731,
      "grad_norm": 0.7543209755588531,
      "learning_rate": 4.961105139015534e-06,
      "loss": 0.5583,
      "step": 689
    },
    {
      "epoch": 0.8445532435740514,
      "grad_norm": 0.4744987718552312,
      "learning_rate": 4.960992449465073e-06,
      "loss": 0.6766,
      "step": 690
    },
    {
      "epoch": 0.8457772337821298,
      "grad_norm": 0.8144338828937318,
      "learning_rate": 4.960879598187292e-06,
      "loss": 0.6966,
      "step": 691
    },
    {
      "epoch": 0.847001223990208,
      "grad_norm": 1.1445769205634282,
      "learning_rate": 4.960766585189604e-06,
      "loss": 0.5002,
      "step": 692
    },
    {
      "epoch": 0.8482252141982864,
      "grad_norm": 0.7189221539713921,
      "learning_rate": 4.960653410479439e-06,
      "loss": 0.586,
      "step": 693
    },
    {
      "epoch": 0.8494492044063647,
      "grad_norm": 0.48895865285309326,
      "learning_rate": 4.9605400740642336e-06,
      "loss": 0.667,
      "step": 694
    },
    {
      "epoch": 0.8506731946144431,
      "grad_norm": 0.777080565568276,
      "learning_rate": 4.960426575951434e-06,
      "loss": 1.2394,
      "step": 695
    },
    {
      "epoch": 0.8518971848225214,
      "grad_norm": 0.8822613054732716,
      "learning_rate": 4.960312916148502e-06,
      "loss": 0.5631,
      "step": 696
    },
    {
      "epoch": 0.8531211750305998,
      "grad_norm": 0.5824593370286995,
      "learning_rate": 4.960199094662904e-06,
      "loss": 0.424,
      "step": 697
    },
    {
      "epoch": 0.8543451652386781,
      "grad_norm": 0.6520130072079502,
      "learning_rate": 4.960085111502122e-06,
      "loss": 1.197,
      "step": 698
    },
    {
      "epoch": 0.8555691554467564,
      "grad_norm": 0.7655563199471871,
      "learning_rate": 4.959970966673645e-06,
      "loss": 0.5878,
      "step": 699
    },
    {
      "epoch": 0.8567931456548348,
      "grad_norm": 1.1527368782096508,
      "learning_rate": 4.9598566601849755e-06,
      "loss": 1.1473,
      "step": 700
    },
    {
      "epoch": 0.8580171358629131,
      "grad_norm": 0.8243418319838043,
      "learning_rate": 4.959742192043624e-06,
      "loss": 1.287,
      "step": 701
    },
    {
      "epoch": 0.8592411260709915,
      "grad_norm": 0.44765985850280393,
      "learning_rate": 4.959627562257113e-06,
      "loss": 0.7051,
      "step": 702
    },
    {
      "epoch": 0.8604651162790697,
      "grad_norm": 0.6765836162428761,
      "learning_rate": 4.959512770832977e-06,
      "loss": 0.608,
      "step": 703
    },
    {
      "epoch": 0.8616891064871481,
      "grad_norm": 0.7104304635049636,
      "learning_rate": 4.959397817778759e-06,
      "loss": 1.3051,
      "step": 704
    },
    {
      "epoch": 0.8629130966952264,
      "grad_norm": 0.7503986020799235,
      "learning_rate": 4.959282703102013e-06,
      "loss": 1.3645,
      "step": 705
    },
    {
      "epoch": 0.8641370869033048,
      "grad_norm": 1.0291781668201516,
      "learning_rate": 4.959167426810304e-06,
      "loss": 0.9358,
      "step": 706
    },
    {
      "epoch": 0.8653610771113831,
      "grad_norm": 1.236303290364792,
      "learning_rate": 4.959051988911208e-06,
      "loss": 0.7589,
      "step": 707
    },
    {
      "epoch": 0.8665850673194615,
      "grad_norm": 0.5529621778113976,
      "learning_rate": 4.95893638941231e-06,
      "loss": 0.3963,
      "step": 708
    },
    {
      "epoch": 0.8678090575275398,
      "grad_norm": 0.5655819564631934,
      "learning_rate": 4.958820628321207e-06,
      "loss": 0.7413,
      "step": 709
    },
    {
      "epoch": 0.8690330477356181,
      "grad_norm": 0.6326436098735134,
      "learning_rate": 4.958704705645508e-06,
      "loss": 0.6838,
      "step": 710
    },
    {
      "epoch": 0.8702570379436965,
      "grad_norm": 0.5610922186390335,
      "learning_rate": 4.958588621392829e-06,
      "loss": 0.6588,
      "step": 711
    },
    {
      "epoch": 0.8714810281517748,
      "grad_norm": 1.0853802862978965,
      "learning_rate": 4.9584723755708e-06,
      "loss": 1.0627,
      "step": 712
    },
    {
      "epoch": 0.8727050183598531,
      "grad_norm": 0.4181265139862361,
      "learning_rate": 4.95835596818706e-06,
      "loss": 0.4741,
      "step": 713
    },
    {
      "epoch": 0.8739290085679314,
      "grad_norm": 0.5697305138603498,
      "learning_rate": 4.958239399249257e-06,
      "loss": 0.4528,
      "step": 714
    },
    {
      "epoch": 0.8751529987760098,
      "grad_norm": 0.5409908090439541,
      "learning_rate": 4.958122668765055e-06,
      "loss": 0.7722,
      "step": 715
    },
    {
      "epoch": 0.8763769889840881,
      "grad_norm": 0.6114565244788789,
      "learning_rate": 4.9580057767421216e-06,
      "loss": 0.6247,
      "step": 716
    },
    {
      "epoch": 0.8776009791921665,
      "grad_norm": 0.5492951193229206,
      "learning_rate": 4.9578887231881415e-06,
      "loss": 0.8009,
      "step": 717
    },
    {
      "epoch": 0.8788249694002448,
      "grad_norm": 0.8082722482767791,
      "learning_rate": 4.957771508110804e-06,
      "loss": 0.536,
      "step": 718
    },
    {
      "epoch": 0.8800489596083231,
      "grad_norm": 1.1912454600581321,
      "learning_rate": 4.9576541315178155e-06,
      "loss": 0.8341,
      "step": 719
    },
    {
      "epoch": 0.8812729498164015,
      "grad_norm": 0.6627422431388482,
      "learning_rate": 4.957536593416887e-06,
      "loss": 0.5589,
      "step": 720
    },
    {
      "epoch": 0.8824969400244798,
      "grad_norm": 0.8281215538387402,
      "learning_rate": 4.957418893815743e-06,
      "loss": 0.8272,
      "step": 721
    },
    {
      "epoch": 0.8837209302325582,
      "grad_norm": 0.2322118039477947,
      "learning_rate": 4.95730103272212e-06,
      "loss": 0.1485,
      "step": 722
    },
    {
      "epoch": 0.8849449204406364,
      "grad_norm": 0.771347945813748,
      "learning_rate": 4.95718301014376e-06,
      "loss": 1.3325,
      "step": 723
    },
    {
      "epoch": 0.8861689106487148,
      "grad_norm": 1.0376997785252853,
      "learning_rate": 4.957064826088422e-06,
      "loss": 0.3694,
      "step": 724
    },
    {
      "epoch": 0.8873929008567931,
      "grad_norm": 0.45775974748006704,
      "learning_rate": 4.956946480563872e-06,
      "loss": 0.6373,
      "step": 725
    },
    {
      "epoch": 0.8886168910648715,
      "grad_norm": 0.8508035796461484,
      "learning_rate": 4.956827973577887e-06,
      "loss": 0.9275,
      "step": 726
    },
    {
      "epoch": 0.8898408812729498,
      "grad_norm": 0.46320961532413946,
      "learning_rate": 4.956709305138254e-06,
      "loss": 0.6821,
      "step": 727
    },
    {
      "epoch": 0.8910648714810282,
      "grad_norm": 0.9018089617689261,
      "learning_rate": 4.956590475252773e-06,
      "loss": 1.1907,
      "step": 728
    },
    {
      "epoch": 0.8922888616891065,
      "grad_norm": 0.503750120577611,
      "learning_rate": 4.956471483929252e-06,
      "loss": 0.6648,
      "step": 729
    },
    {
      "epoch": 0.8935128518971848,
      "grad_norm": 0.8408772095120547,
      "learning_rate": 4.95635233117551e-06,
      "loss": 1.0152,
      "step": 730
    },
    {
      "epoch": 0.8947368421052632,
      "grad_norm": 0.5631755390842799,
      "learning_rate": 4.956233016999379e-06,
      "loss": 0.7104,
      "step": 731
    },
    {
      "epoch": 0.8959608323133414,
      "grad_norm": 0.8712703886908723,
      "learning_rate": 4.956113541408699e-06,
      "loss": 1.0763,
      "step": 732
    },
    {
      "epoch": 0.8971848225214198,
      "grad_norm": 0.512408541132229,
      "learning_rate": 4.955993904411322e-06,
      "loss": 0.7247,
      "step": 733
    },
    {
      "epoch": 0.8984088127294981,
      "grad_norm": 0.6824974015011425,
      "learning_rate": 4.9558741060151095e-06,
      "loss": 0.9624,
      "step": 734
    },
    {
      "epoch": 0.8996328029375765,
      "grad_norm": 0.5183321760850187,
      "learning_rate": 4.955754146227934e-06,
      "loss": 0.6987,
      "step": 735
    },
    {
      "epoch": 0.9008567931456548,
      "grad_norm": 1.0467640809491838,
      "learning_rate": 4.955634025057679e-06,
      "loss": 0.5277,
      "step": 736
    },
    {
      "epoch": 0.9020807833537332,
      "grad_norm": 0.5811866370147754,
      "learning_rate": 4.95551374251224e-06,
      "loss": 1.169,
      "step": 737
    },
    {
      "epoch": 0.9033047735618115,
      "grad_norm": 0.6467966511771853,
      "learning_rate": 4.955393298599519e-06,
      "loss": 0.6356,
      "step": 738
    },
    {
      "epoch": 0.9045287637698899,
      "grad_norm": 0.705212707899457,
      "learning_rate": 4.955272693327432e-06,
      "loss": 0.801,
      "step": 739
    },
    {
      "epoch": 0.9057527539779682,
      "grad_norm": 0.628522507726457,
      "learning_rate": 4.955151926703906e-06,
      "loss": 1.006,
      "step": 740
    },
    {
      "epoch": 0.9069767441860465,
      "grad_norm": 0.6229832723062967,
      "learning_rate": 4.955030998736876e-06,
      "loss": 0.3779,
      "step": 741
    },
    {
      "epoch": 0.9082007343941249,
      "grad_norm": 0.7466982422594095,
      "learning_rate": 4.95490990943429e-06,
      "loss": 0.5944,
      "step": 742
    },
    {
      "epoch": 0.9094247246022031,
      "grad_norm": 0.6324108113406297,
      "learning_rate": 4.954788658804104e-06,
      "loss": 0.7041,
      "step": 743
    },
    {
      "epoch": 0.9106487148102815,
      "grad_norm": 0.7062328351787057,
      "learning_rate": 4.954667246854287e-06,
      "loss": 1.171,
      "step": 744
    },
    {
      "epoch": 0.9118727050183598,
      "grad_norm": 0.6043655151870077,
      "learning_rate": 4.954545673592818e-06,
      "loss": 0.7547,
      "step": 745
    },
    {
      "epoch": 0.9130966952264382,
      "grad_norm": 0.6288704008008857,
      "learning_rate": 4.954423939027685e-06,
      "loss": 0.6992,
      "step": 746
    },
    {
      "epoch": 0.9143206854345165,
      "grad_norm": 0.56448098498115,
      "learning_rate": 4.95430204316689e-06,
      "loss": 0.5258,
      "step": 747
    },
    {
      "epoch": 0.9155446756425949,
      "grad_norm": 0.953467737764038,
      "learning_rate": 4.954179986018442e-06,
      "loss": 0.6948,
      "step": 748
    },
    {
      "epoch": 0.9167686658506732,
      "grad_norm": 0.6567046061576766,
      "learning_rate": 4.954057767590364e-06,
      "loss": 0.6245,
      "step": 749
    },
    {
      "epoch": 0.9179926560587516,
      "grad_norm": 1.003548014952242,
      "learning_rate": 4.9539353878906855e-06,
      "loss": 0.657,
      "step": 750
    },
    {
      "epoch": 0.9192166462668299,
      "grad_norm": 0.7329053091410532,
      "learning_rate": 4.9538128469274495e-06,
      "loss": 0.406,
      "step": 751
    },
    {
      "epoch": 0.9204406364749081,
      "grad_norm": 0.9746605148531934,
      "learning_rate": 4.9536901447087094e-06,
      "loss": 0.7816,
      "step": 752
    },
    {
      "epoch": 0.9216646266829865,
      "grad_norm": 0.7637650437816885,
      "learning_rate": 4.95356728124253e-06,
      "loss": 0.6554,
      "step": 753
    },
    {
      "epoch": 0.9228886168910648,
      "grad_norm": 0.6660208266325064,
      "learning_rate": 4.953444256536982e-06,
      "loss": 0.6415,
      "step": 754
    },
    {
      "epoch": 0.9241126070991432,
      "grad_norm": 0.9819341653189343,
      "learning_rate": 4.9533210706001526e-06,
      "loss": 1.5814,
      "step": 755
    },
    {
      "epoch": 0.9253365973072215,
      "grad_norm": 0.9377834858093372,
      "learning_rate": 4.9531977234401376e-06,
      "loss": 0.7036,
      "step": 756
    },
    {
      "epoch": 0.9265605875152999,
      "grad_norm": 0.7616575323489688,
      "learning_rate": 4.9530742150650405e-06,
      "loss": 0.4911,
      "step": 757
    },
    {
      "epoch": 0.9277845777233782,
      "grad_norm": 0.6953791207814416,
      "learning_rate": 4.95295054548298e-06,
      "loss": 1.65,
      "step": 758
    },
    {
      "epoch": 0.9290085679314566,
      "grad_norm": 1.1387378973646702,
      "learning_rate": 4.952826714702082e-06,
      "loss": 0.5804,
      "step": 759
    },
    {
      "epoch": 0.9302325581395349,
      "grad_norm": 0.8861422220151547,
      "learning_rate": 4.952702722730486e-06,
      "loss": 0.5614,
      "step": 760
    },
    {
      "epoch": 0.9314565483476133,
      "grad_norm": 0.7418298836843789,
      "learning_rate": 4.9525785695763385e-06,
      "loss": 0.7323,
      "step": 761
    },
    {
      "epoch": 0.9326805385556916,
      "grad_norm": 1.0368621356180565,
      "learning_rate": 4.952454255247798e-06,
      "loss": 1.2654,
      "step": 762
    },
    {
      "epoch": 0.9339045287637698,
      "grad_norm": 0.744680815090098,
      "learning_rate": 4.952329779753036e-06,
      "loss": 0.5789,
      "step": 763
    },
    {
      "epoch": 0.9351285189718482,
      "grad_norm": 0.7435190315239415,
      "learning_rate": 4.952205143100231e-06,
      "loss": 0.6035,
      "step": 764
    },
    {
      "epoch": 0.9363525091799265,
      "grad_norm": 0.9180800699512491,
      "learning_rate": 4.952080345297573e-06,
      "loss": 0.8417,
      "step": 765
    },
    {
      "epoch": 0.9375764993880049,
      "grad_norm": 0.5401721647431306,
      "learning_rate": 4.951955386353265e-06,
      "loss": 0.824,
      "step": 766
    },
    {
      "epoch": 0.9388004895960832,
      "grad_norm": 0.8520554366017021,
      "learning_rate": 4.951830266275519e-06,
      "loss": 0.746,
      "step": 767
    },
    {
      "epoch": 0.9400244798041616,
      "grad_norm": 0.41268144712817595,
      "learning_rate": 4.951704985072556e-06,
      "loss": 0.5081,
      "step": 768
    },
    {
      "epoch": 0.9412484700122399,
      "grad_norm": 0.7399246329327385,
      "learning_rate": 4.95157954275261e-06,
      "loss": 0.8574,
      "step": 769
    },
    {
      "epoch": 0.9424724602203183,
      "grad_norm": 0.6699501069245549,
      "learning_rate": 4.951453939323924e-06,
      "loss": 0.7014,
      "step": 770
    },
    {
      "epoch": 0.9436964504283966,
      "grad_norm": 0.7953659085232874,
      "learning_rate": 4.951328174794752e-06,
      "loss": 0.7177,
      "step": 771
    },
    {
      "epoch": 0.944920440636475,
      "grad_norm": 1.4365935603083932,
      "learning_rate": 4.951202249173359e-06,
      "loss": 0.9069,
      "step": 772
    },
    {
      "epoch": 0.9461444308445532,
      "grad_norm": 0.9958209988883617,
      "learning_rate": 4.951076162468022e-06,
      "loss": 1.154,
      "step": 773
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 0.8172148241879345,
      "learning_rate": 4.950949914687024e-06,
      "loss": 1.0645,
      "step": 774
    },
    {
      "epoch": 0.9485924112607099,
      "grad_norm": 0.6654613847396502,
      "learning_rate": 4.9508235058386635e-06,
      "loss": 0.5749,
      "step": 775
    },
    {
      "epoch": 0.9498164014687882,
      "grad_norm": 0.349047639859916,
      "learning_rate": 4.950696935931247e-06,
      "loss": 0.2597,
      "step": 776
    },
    {
      "epoch": 0.9510403916768666,
      "grad_norm": 0.6259195946879196,
      "learning_rate": 4.950570204973092e-06,
      "loss": 0.9118,
      "step": 777
    },
    {
      "epoch": 0.9522643818849449,
      "grad_norm": 0.7726247437629674,
      "learning_rate": 4.950443312972528e-06,
      "loss": 1.4145,
      "step": 778
    },
    {
      "epoch": 0.9534883720930233,
      "grad_norm": 1.2905346863679434,
      "learning_rate": 4.950316259937893e-06,
      "loss": 0.5826,
      "step": 779
    },
    {
      "epoch": 0.9547123623011016,
      "grad_norm": 0.9103500823309562,
      "learning_rate": 4.950189045877536e-06,
      "loss": 0.5736,
      "step": 780
    },
    {
      "epoch": 0.95593635250918,
      "grad_norm": 0.6626792535247903,
      "learning_rate": 4.950061670799818e-06,
      "loss": 0.5445,
      "step": 781
    },
    {
      "epoch": 0.9571603427172583,
      "grad_norm": 1.0176661356963794,
      "learning_rate": 4.949934134713109e-06,
      "loss": 1.144,
      "step": 782
    },
    {
      "epoch": 0.9583843329253366,
      "grad_norm": 1.0613860820373697,
      "learning_rate": 4.9498064376257905e-06,
      "loss": 1.1083,
      "step": 783
    },
    {
      "epoch": 0.9596083231334149,
      "grad_norm": 0.8661539175464111,
      "learning_rate": 4.949678579546254e-06,
      "loss": 0.8527,
      "step": 784
    },
    {
      "epoch": 0.9608323133414932,
      "grad_norm": 1.0089892450035782,
      "learning_rate": 4.949550560482902e-06,
      "loss": 1.2164,
      "step": 785
    },
    {
      "epoch": 0.9620563035495716,
      "grad_norm": 0.6471811639787152,
      "learning_rate": 4.949422380444147e-06,
      "loss": 0.67,
      "step": 786
    },
    {
      "epoch": 0.9632802937576499,
      "grad_norm": 1.3305504669122221,
      "learning_rate": 4.9492940394384135e-06,
      "loss": 0.7003,
      "step": 787
    },
    {
      "epoch": 0.9645042839657283,
      "grad_norm": 0.6822602927488968,
      "learning_rate": 4.949165537474134e-06,
      "loss": 0.5909,
      "step": 788
    },
    {
      "epoch": 0.9657282741738066,
      "grad_norm": 1.265269610791293,
      "learning_rate": 4.949036874559755e-06,
      "loss": 0.5821,
      "step": 789
    },
    {
      "epoch": 0.966952264381885,
      "grad_norm": 0.5048014197446955,
      "learning_rate": 4.9489080507037314e-06,
      "loss": 0.4543,
      "step": 790
    },
    {
      "epoch": 0.9681762545899633,
      "grad_norm": 0.5077967861665366,
      "learning_rate": 4.9487790659145285e-06,
      "loss": 0.5157,
      "step": 791
    },
    {
      "epoch": 0.9694002447980417,
      "grad_norm": 1.1738487056815181,
      "learning_rate": 4.9486499202006225e-06,
      "loss": 0.5897,
      "step": 792
    },
    {
      "epoch": 0.9706242350061199,
      "grad_norm": 0.8593748606891259,
      "learning_rate": 4.9485206135705e-06,
      "loss": 0.659,
      "step": 793
    },
    {
      "epoch": 0.9718482252141983,
      "grad_norm": 0.9106326648832945,
      "learning_rate": 4.94839114603266e-06,
      "loss": 0.5134,
      "step": 794
    },
    {
      "epoch": 0.9730722154222766,
      "grad_norm": 1.1073536618977897,
      "learning_rate": 4.94826151759561e-06,
      "loss": 0.5198,
      "step": 795
    },
    {
      "epoch": 0.9742962056303549,
      "grad_norm": 0.774719335187564,
      "learning_rate": 4.948131728267868e-06,
      "loss": 0.6796,
      "step": 796
    },
    {
      "epoch": 0.9755201958384333,
      "grad_norm": 0.5551126390132856,
      "learning_rate": 4.948001778057964e-06,
      "loss": 0.7378,
      "step": 797
    },
    {
      "epoch": 0.9767441860465116,
      "grad_norm": 0.9278909177767704,
      "learning_rate": 4.947871666974438e-06,
      "loss": 0.527,
      "step": 798
    },
    {
      "epoch": 0.97796817625459,
      "grad_norm": 0.6459144722406741,
      "learning_rate": 4.94774139502584e-06,
      "loss": 0.6721,
      "step": 799
    },
    {
      "epoch": 0.9791921664626683,
      "grad_norm": 0.7560365405148642,
      "learning_rate": 4.9476109622207305e-06,
      "loss": 0.5276,
      "step": 800
    },
    {
      "epoch": 0.9804161566707467,
      "grad_norm": 0.9382628125879249,
      "learning_rate": 4.947480368567682e-06,
      "loss": 0.462,
      "step": 801
    },
    {
      "epoch": 0.981640146878825,
      "grad_norm": 0.7358258340675429,
      "learning_rate": 4.9473496140752765e-06,
      "loss": 1.6796,
      "step": 802
    },
    {
      "epoch": 0.9828641370869033,
      "grad_norm": 0.8665619231728571,
      "learning_rate": 4.9472186987521065e-06,
      "loss": 1.6072,
      "step": 803
    },
    {
      "epoch": 0.9840881272949816,
      "grad_norm": 0.8872786972909009,
      "learning_rate": 4.947087622606775e-06,
      "loss": 0.9852,
      "step": 804
    },
    {
      "epoch": 0.98531211750306,
      "grad_norm": 0.7688120320974523,
      "learning_rate": 4.946956385647896e-06,
      "loss": 1.2714,
      "step": 805
    },
    {
      "epoch": 0.9865361077111383,
      "grad_norm": 0.7701508714382308,
      "learning_rate": 4.946824987884095e-06,
      "loss": 0.6108,
      "step": 806
    },
    {
      "epoch": 0.9877600979192166,
      "grad_norm": 0.5248408647757767,
      "learning_rate": 4.946693429324005e-06,
      "loss": 0.5792,
      "step": 807
    },
    {
      "epoch": 0.988984088127295,
      "grad_norm": 0.9737646168351266,
      "learning_rate": 4.9465617099762725e-06,
      "loss": 0.5862,
      "step": 808
    },
    {
      "epoch": 0.9902080783353733,
      "grad_norm": 0.7606404708012034,
      "learning_rate": 4.946429829849554e-06,
      "loss": 0.6566,
      "step": 809
    },
    {
      "epoch": 0.9914320685434517,
      "grad_norm": 0.5894997176752026,
      "learning_rate": 4.946297788952515e-06,
      "loss": 0.4674,
      "step": 810
    },
    {
      "epoch": 0.99265605875153,
      "grad_norm": 0.7956348703033926,
      "learning_rate": 4.946165587293835e-06,
      "loss": 0.6732,
      "step": 811
    },
    {
      "epoch": 0.9938800489596084,
      "grad_norm": 0.9052988526417667,
      "learning_rate": 4.946033224882199e-06,
      "loss": 1.6674,
      "step": 812
    },
    {
      "epoch": 0.9951040391676866,
      "grad_norm": 0.8035133481225467,
      "learning_rate": 4.945900701726307e-06,
      "loss": 0.5189,
      "step": 813
    },
    {
      "epoch": 0.996328029375765,
      "grad_norm": 0.8606738785392671,
      "learning_rate": 4.945768017834868e-06,
      "loss": 0.6192,
      "step": 814
    },
    {
      "epoch": 0.9975520195838433,
      "grad_norm": 0.6238537090922615,
      "learning_rate": 4.945635173216602e-06,
      "loss": 0.6999,
      "step": 815
    },
    {
      "epoch": 0.9987760097919217,
      "grad_norm": 0.6203579342173314,
      "learning_rate": 4.945502167880236e-06,
      "loss": 0.7971,
      "step": 816
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5693544736753935,
      "learning_rate": 4.9453690018345144e-06,
      "loss": 0.6203,
      "step": 817
    },
    {
      "epoch": 1.0012239902080784,
      "grad_norm": 0.8661940272854449,
      "learning_rate": 4.945235675088187e-06,
      "loss": 0.6973,
      "step": 818
    },
    {
      "epoch": 1.0024479804161566,
      "grad_norm": 0.707151983898238,
      "learning_rate": 4.945102187650015e-06,
      "loss": 0.836,
      "step": 819
    },
    {
      "epoch": 1.003671970624235,
      "grad_norm": 0.6446105970479922,
      "learning_rate": 4.94496853952877e-06,
      "loss": 0.9218,
      "step": 820
    },
    {
      "epoch": 1.0048959608323134,
      "grad_norm": 0.5477801538704142,
      "learning_rate": 4.944834730733237e-06,
      "loss": 0.4513,
      "step": 821
    },
    {
      "epoch": 1.0061199510403918,
      "grad_norm": 0.6368417317347421,
      "learning_rate": 4.9447007612722076e-06,
      "loss": 0.9051,
      "step": 822
    },
    {
      "epoch": 1.00734394124847,
      "grad_norm": 0.7613256766593812,
      "learning_rate": 4.944566631154487e-06,
      "loss": 0.6626,
      "step": 823
    },
    {
      "epoch": 1.0085679314565483,
      "grad_norm": 0.9616628961324019,
      "learning_rate": 4.944432340388889e-06,
      "loss": 0.7408,
      "step": 824
    },
    {
      "epoch": 1.0097919216646267,
      "grad_norm": 0.8207080720176745,
      "learning_rate": 4.944297888984239e-06,
      "loss": 0.5055,
      "step": 825
    },
    {
      "epoch": 1.0110159118727051,
      "grad_norm": 0.6978541379200914,
      "learning_rate": 4.944163276949373e-06,
      "loss": 1.6658,
      "step": 826
    },
    {
      "epoch": 1.0122399020807833,
      "grad_norm": 0.40682640339588866,
      "learning_rate": 4.944028504293136e-06,
      "loss": 0.4815,
      "step": 827
    },
    {
      "epoch": 1.0134638922888617,
      "grad_norm": 0.5904825770432894,
      "learning_rate": 4.943893571024386e-06,
      "loss": 0.5393,
      "step": 828
    },
    {
      "epoch": 1.01468788249694,
      "grad_norm": 0.8637420173350128,
      "learning_rate": 4.94375847715199e-06,
      "loss": 0.7088,
      "step": 829
    },
    {
      "epoch": 1.0159118727050183,
      "grad_norm": 0.7083996965236148,
      "learning_rate": 4.9436232226848254e-06,
      "loss": 0.8026,
      "step": 830
    },
    {
      "epoch": 1.0171358629130967,
      "grad_norm": 1.1715644372132843,
      "learning_rate": 4.94348780763178e-06,
      "loss": 1.1236,
      "step": 831
    },
    {
      "epoch": 1.018359853121175,
      "grad_norm": 0.4170623593069568,
      "learning_rate": 4.943352232001755e-06,
      "loss": 0.5134,
      "step": 832
    },
    {
      "epoch": 1.0195838433292534,
      "grad_norm": 1.1046852518147066,
      "learning_rate": 4.943216495803659e-06,
      "loss": 1.5398,
      "step": 833
    },
    {
      "epoch": 1.0208078335373316,
      "grad_norm": 0.6988946062657729,
      "learning_rate": 4.943080599046411e-06,
      "loss": 0.5888,
      "step": 834
    },
    {
      "epoch": 1.02203182374541,
      "grad_norm": 0.6141841076150183,
      "learning_rate": 4.9429445417389425e-06,
      "loss": 0.4347,
      "step": 835
    },
    {
      "epoch": 1.0232558139534884,
      "grad_norm": 0.7861586283093154,
      "learning_rate": 4.9428083238901945e-06,
      "loss": 0.795,
      "step": 836
    },
    {
      "epoch": 1.0244798041615668,
      "grad_norm": 0.4737187818914004,
      "learning_rate": 4.942671945509119e-06,
      "loss": 0.6727,
      "step": 837
    },
    {
      "epoch": 1.025703794369645,
      "grad_norm": 0.8658438555873381,
      "learning_rate": 4.942535406604678e-06,
      "loss": 1.1304,
      "step": 838
    },
    {
      "epoch": 1.0269277845777234,
      "grad_norm": 1.2823582828519138,
      "learning_rate": 4.942398707185844e-06,
      "loss": 0.8813,
      "step": 839
    },
    {
      "epoch": 1.0281517747858018,
      "grad_norm": 0.8294051139830289,
      "learning_rate": 4.9422618472616005e-06,
      "loss": 0.5225,
      "step": 840
    },
    {
      "epoch": 1.02937576499388,
      "grad_norm": 1.1458218559403601,
      "learning_rate": 4.9421248268409425e-06,
      "loss": 0.67,
      "step": 841
    },
    {
      "epoch": 1.0305997552019583,
      "grad_norm": 0.9507148784282786,
      "learning_rate": 4.941987645932873e-06,
      "loss": 0.728,
      "step": 842
    },
    {
      "epoch": 1.0318237454100367,
      "grad_norm": 1.0953952073990505,
      "learning_rate": 4.9418503045464085e-06,
      "loss": 0.681,
      "step": 843
    },
    {
      "epoch": 1.0330477356181151,
      "grad_norm": 0.9183838020477175,
      "learning_rate": 4.941712802690573e-06,
      "loss": 0.9376,
      "step": 844
    },
    {
      "epoch": 1.0342717258261933,
      "grad_norm": 1.131657359103838,
      "learning_rate": 4.941575140374404e-06,
      "loss": 1.0094,
      "step": 845
    },
    {
      "epoch": 1.0354957160342717,
      "grad_norm": 0.4836443208439507,
      "learning_rate": 4.941437317606948e-06,
      "loss": 0.697,
      "step": 846
    },
    {
      "epoch": 1.03671970624235,
      "grad_norm": 0.5312915295698105,
      "learning_rate": 4.94129933439726e-06,
      "loss": 0.6277,
      "step": 847
    },
    {
      "epoch": 1.0379436964504285,
      "grad_norm": 0.9088395020591802,
      "learning_rate": 4.941161190754411e-06,
      "loss": 2.0003,
      "step": 848
    },
    {
      "epoch": 1.0391676866585067,
      "grad_norm": 0.7397669026028801,
      "learning_rate": 4.941022886687477e-06,
      "loss": 0.9658,
      "step": 849
    },
    {
      "epoch": 1.040391676866585,
      "grad_norm": 0.9475645784237857,
      "learning_rate": 4.940884422205548e-06,
      "loss": 1.502,
      "step": 850
    },
    {
      "epoch": 1.0416156670746635,
      "grad_norm": 1.2027277459322063,
      "learning_rate": 4.940745797317723e-06,
      "loss": 0.8304,
      "step": 851
    },
    {
      "epoch": 1.0428396572827416,
      "grad_norm": 1.4366594569596243,
      "learning_rate": 4.9406070120331116e-06,
      "loss": 0.636,
      "step": 852
    },
    {
      "epoch": 1.04406364749082,
      "grad_norm": 0.4464224155188248,
      "learning_rate": 4.940468066360835e-06,
      "loss": 0.4806,
      "step": 853
    },
    {
      "epoch": 1.0452876376988984,
      "grad_norm": 1.3012907038125006,
      "learning_rate": 4.940328960310023e-06,
      "loss": 0.7299,
      "step": 854
    },
    {
      "epoch": 1.0465116279069768,
      "grad_norm": 0.7355836555123412,
      "learning_rate": 4.940189693889819e-06,
      "loss": 0.6743,
      "step": 855
    },
    {
      "epoch": 1.047735618115055,
      "grad_norm": 0.6992330112851228,
      "learning_rate": 4.940050267109373e-06,
      "loss": 0.7331,
      "step": 856
    },
    {
      "epoch": 1.0489596083231334,
      "grad_norm": 0.6033412559576553,
      "learning_rate": 4.939910679977848e-06,
      "loss": 0.6466,
      "step": 857
    },
    {
      "epoch": 1.0501835985312118,
      "grad_norm": 0.7488411337841201,
      "learning_rate": 4.939770932504419e-06,
      "loss": 0.6794,
      "step": 858
    },
    {
      "epoch": 1.0514075887392902,
      "grad_norm": 0.5800918156185451,
      "learning_rate": 4.9396310246982685e-06,
      "loss": 0.6641,
      "step": 859
    },
    {
      "epoch": 1.0526315789473684,
      "grad_norm": 0.7551573359193208,
      "learning_rate": 4.939490956568589e-06,
      "loss": 1.195,
      "step": 860
    },
    {
      "epoch": 1.0538555691554468,
      "grad_norm": 0.9180580663687439,
      "learning_rate": 4.939350728124588e-06,
      "loss": 0.547,
      "step": 861
    },
    {
      "epoch": 1.0550795593635252,
      "grad_norm": 0.812652605834241,
      "learning_rate": 4.93921033937548e-06,
      "loss": 0.6669,
      "step": 862
    },
    {
      "epoch": 1.0563035495716033,
      "grad_norm": 0.6000139939882762,
      "learning_rate": 4.93906979033049e-06,
      "loss": 0.6664,
      "step": 863
    },
    {
      "epoch": 1.0575275397796817,
      "grad_norm": 0.9157930175254086,
      "learning_rate": 4.938929080998856e-06,
      "loss": 0.6129,
      "step": 864
    },
    {
      "epoch": 1.0587515299877601,
      "grad_norm": 0.794185060925577,
      "learning_rate": 4.938788211389822e-06,
      "loss": 1.1158,
      "step": 865
    },
    {
      "epoch": 1.0599755201958385,
      "grad_norm": 0.8064743589840888,
      "learning_rate": 4.938647181512648e-06,
      "loss": 0.6668,
      "step": 866
    },
    {
      "epoch": 1.0611995104039167,
      "grad_norm": 0.9588730797191946,
      "learning_rate": 4.938505991376601e-06,
      "loss": 1.1179,
      "step": 867
    },
    {
      "epoch": 1.062423500611995,
      "grad_norm": 0.6374079362232818,
      "learning_rate": 4.9383646409909604e-06,
      "loss": 0.6922,
      "step": 868
    },
    {
      "epoch": 1.0636474908200735,
      "grad_norm": 1.0498529267851326,
      "learning_rate": 4.938223130365014e-06,
      "loss": 1.0353,
      "step": 869
    },
    {
      "epoch": 1.0648714810281519,
      "grad_norm": 0.8120040043969616,
      "learning_rate": 4.938081459508061e-06,
      "loss": 1.3931,
      "step": 870
    },
    {
      "epoch": 1.06609547123623,
      "grad_norm": 0.9150882229283052,
      "learning_rate": 4.9379396284294135e-06,
      "loss": 0.9144,
      "step": 871
    },
    {
      "epoch": 1.0673194614443084,
      "grad_norm": 0.7500586163669113,
      "learning_rate": 4.937797637138391e-06,
      "loss": 0.4016,
      "step": 872
    },
    {
      "epoch": 1.0685434516523868,
      "grad_norm": 0.8078203460896909,
      "learning_rate": 4.937655485644324e-06,
      "loss": 0.4653,
      "step": 873
    },
    {
      "epoch": 1.069767441860465,
      "grad_norm": 0.8067060929128912,
      "learning_rate": 4.937513173956554e-06,
      "loss": 0.4519,
      "step": 874
    },
    {
      "epoch": 1.0709914320685434,
      "grad_norm": 0.6180085262754264,
      "learning_rate": 4.937370702084435e-06,
      "loss": 0.8411,
      "step": 875
    },
    {
      "epoch": 1.0722154222766218,
      "grad_norm": 1.1061653049333586,
      "learning_rate": 4.937228070037329e-06,
      "loss": 0.9889,
      "step": 876
    },
    {
      "epoch": 1.0734394124847002,
      "grad_norm": 1.1043132113659924,
      "learning_rate": 4.937085277824608e-06,
      "loss": 1.071,
      "step": 877
    },
    {
      "epoch": 1.0746634026927784,
      "grad_norm": 0.6688194414997779,
      "learning_rate": 4.936942325455657e-06,
      "loss": 1.1218,
      "step": 878
    },
    {
      "epoch": 1.0758873929008568,
      "grad_norm": 1.3979019482131052,
      "learning_rate": 4.93679921293987e-06,
      "loss": 0.6765,
      "step": 879
    },
    {
      "epoch": 1.0771113831089352,
      "grad_norm": 0.8891508978142024,
      "learning_rate": 4.936655940286652e-06,
      "loss": 0.7434,
      "step": 880
    },
    {
      "epoch": 1.0783353733170133,
      "grad_norm": 0.6170598773984476,
      "learning_rate": 4.936512507505418e-06,
      "loss": 0.7343,
      "step": 881
    },
    {
      "epoch": 1.0795593635250917,
      "grad_norm": 0.6601043549706049,
      "learning_rate": 4.936368914605593e-06,
      "loss": 0.6579,
      "step": 882
    },
    {
      "epoch": 1.0807833537331701,
      "grad_norm": 0.797528036624534,
      "learning_rate": 4.9362251615966165e-06,
      "loss": 0.5999,
      "step": 883
    },
    {
      "epoch": 1.0820073439412485,
      "grad_norm": 0.7302283080473125,
      "learning_rate": 4.936081248487932e-06,
      "loss": 0.4804,
      "step": 884
    },
    {
      "epoch": 1.0832313341493267,
      "grad_norm": 0.8000642381307068,
      "learning_rate": 4.935937175288999e-06,
      "loss": 0.6468,
      "step": 885
    },
    {
      "epoch": 1.084455324357405,
      "grad_norm": 0.7482638787732815,
      "learning_rate": 4.935792942009285e-06,
      "loss": 0.6066,
      "step": 886
    },
    {
      "epoch": 1.0856793145654835,
      "grad_norm": 0.7839314095512564,
      "learning_rate": 4.9356485486582676e-06,
      "loss": 0.5721,
      "step": 887
    },
    {
      "epoch": 1.086903304773562,
      "grad_norm": 0.5233299131994322,
      "learning_rate": 4.935503995245435e-06,
      "loss": 0.4908,
      "step": 888
    },
    {
      "epoch": 1.08812729498164,
      "grad_norm": 0.6099284488229789,
      "learning_rate": 4.93535928178029e-06,
      "loss": 0.4673,
      "step": 889
    },
    {
      "epoch": 1.0893512851897185,
      "grad_norm": 0.48434984226035793,
      "learning_rate": 4.935214408272341e-06,
      "loss": 0.6289,
      "step": 890
    },
    {
      "epoch": 1.0905752753977969,
      "grad_norm": 0.7076133083344578,
      "learning_rate": 4.935069374731108e-06,
      "loss": 0.4804,
      "step": 891
    },
    {
      "epoch": 1.091799265605875,
      "grad_norm": 0.63015760369857,
      "learning_rate": 4.934924181166121e-06,
      "loss": 0.692,
      "step": 892
    },
    {
      "epoch": 1.0930232558139534,
      "grad_norm": 0.722134084925986,
      "learning_rate": 4.9347788275869245e-06,
      "loss": 1.3578,
      "step": 893
    },
    {
      "epoch": 1.0942472460220318,
      "grad_norm": 0.8147020330197138,
      "learning_rate": 4.9346333140030685e-06,
      "loss": 0.8277,
      "step": 894
    },
    {
      "epoch": 1.0954712362301102,
      "grad_norm": 0.6722225050176123,
      "learning_rate": 4.934487640424116e-06,
      "loss": 0.5787,
      "step": 895
    },
    {
      "epoch": 1.0966952264381884,
      "grad_norm": 1.1436531291375829,
      "learning_rate": 4.934341806859641e-06,
      "loss": 1.1474,
      "step": 896
    },
    {
      "epoch": 1.0979192166462668,
      "grad_norm": 0.4830185150974974,
      "learning_rate": 4.934195813319226e-06,
      "loss": 0.5948,
      "step": 897
    },
    {
      "epoch": 1.0991432068543452,
      "grad_norm": 1.0204265174128586,
      "learning_rate": 4.934049659812466e-06,
      "loss": 0.6106,
      "step": 898
    },
    {
      "epoch": 1.1003671970624236,
      "grad_norm": 0.717775102907032,
      "learning_rate": 4.933903346348965e-06,
      "loss": 1.1727,
      "step": 899
    },
    {
      "epoch": 1.1015911872705018,
      "grad_norm": 1.183887331450676,
      "learning_rate": 4.933756872938338e-06,
      "loss": 0.7355,
      "step": 900
    },
    {
      "epoch": 1.1028151774785802,
      "grad_norm": 0.6266520398342849,
      "learning_rate": 4.933610239590212e-06,
      "loss": 0.5325,
      "step": 901
    },
    {
      "epoch": 1.1040391676866586,
      "grad_norm": 0.7839937195979166,
      "learning_rate": 4.933463446314221e-06,
      "loss": 0.8373,
      "step": 902
    },
    {
      "epoch": 1.1052631578947367,
      "grad_norm": 1.2403158864800392,
      "learning_rate": 4.933316493120015e-06,
      "loss": 0.5419,
      "step": 903
    },
    {
      "epoch": 1.1064871481028151,
      "grad_norm": 1.0968796147664983,
      "learning_rate": 4.933169380017249e-06,
      "loss": 1.1853,
      "step": 904
    },
    {
      "epoch": 1.1077111383108935,
      "grad_norm": 0.9760553254367574,
      "learning_rate": 4.9330221070155895e-06,
      "loss": 1.2385,
      "step": 905
    },
    {
      "epoch": 1.108935128518972,
      "grad_norm": 0.5933915925866321,
      "learning_rate": 4.932874674124717e-06,
      "loss": 0.9547,
      "step": 906
    },
    {
      "epoch": 1.11015911872705,
      "grad_norm": 0.5260163594124148,
      "learning_rate": 4.9327270813543195e-06,
      "loss": 0.7735,
      "step": 907
    },
    {
      "epoch": 1.1113831089351285,
      "grad_norm": 0.7198693072522813,
      "learning_rate": 4.932579328714097e-06,
      "loss": 0.6696,
      "step": 908
    },
    {
      "epoch": 1.1126070991432069,
      "grad_norm": 0.8423682346092232,
      "learning_rate": 4.932431416213757e-06,
      "loss": 0.5328,
      "step": 909
    },
    {
      "epoch": 1.1138310893512853,
      "grad_norm": 0.7805678363954234,
      "learning_rate": 4.932283343863023e-06,
      "loss": 1.0152,
      "step": 910
    },
    {
      "epoch": 1.1150550795593634,
      "grad_norm": 0.8329413704405583,
      "learning_rate": 4.932135111671622e-06,
      "loss": 1.1024,
      "step": 911
    },
    {
      "epoch": 1.1162790697674418,
      "grad_norm": 0.5943061743713192,
      "learning_rate": 4.931986719649298e-06,
      "loss": 0.4314,
      "step": 912
    },
    {
      "epoch": 1.1175030599755202,
      "grad_norm": 0.794108142697148,
      "learning_rate": 4.931838167805804e-06,
      "loss": 1.7536,
      "step": 913
    },
    {
      "epoch": 1.1187270501835984,
      "grad_norm": 0.746695015450399,
      "learning_rate": 4.9316894561508985e-06,
      "loss": 1.2597,
      "step": 914
    },
    {
      "epoch": 1.1199510403916768,
      "grad_norm": 0.6997059029735759,
      "learning_rate": 4.931540584694356e-06,
      "loss": 0.4579,
      "step": 915
    },
    {
      "epoch": 1.1211750305997552,
      "grad_norm": 0.6928252230203336,
      "learning_rate": 4.931391553445961e-06,
      "loss": 0.5626,
      "step": 916
    },
    {
      "epoch": 1.1223990208078336,
      "grad_norm": 1.0488506948189316,
      "learning_rate": 4.931242362415506e-06,
      "loss": 0.7599,
      "step": 917
    },
    {
      "epoch": 1.1236230110159118,
      "grad_norm": 0.46909806642517743,
      "learning_rate": 4.931093011612795e-06,
      "loss": 0.4386,
      "step": 918
    },
    {
      "epoch": 1.1248470012239902,
      "grad_norm": 1.0453864809975137,
      "learning_rate": 4.930943501047644e-06,
      "loss": 0.7751,
      "step": 919
    },
    {
      "epoch": 1.1260709914320686,
      "grad_norm": 0.6272981133245432,
      "learning_rate": 4.930793830729877e-06,
      "loss": 0.9235,
      "step": 920
    },
    {
      "epoch": 1.127294981640147,
      "grad_norm": 0.7740683320030309,
      "learning_rate": 4.93064400066933e-06,
      "loss": 0.6552,
      "step": 921
    },
    {
      "epoch": 1.1285189718482251,
      "grad_norm": 1.5976515114142773,
      "learning_rate": 4.9304940108758496e-06,
      "loss": 0.542,
      "step": 922
    },
    {
      "epoch": 1.1297429620563035,
      "grad_norm": 1.3665461061409951,
      "learning_rate": 4.9303438613592925e-06,
      "loss": 0.4761,
      "step": 923
    },
    {
      "epoch": 1.130966952264382,
      "grad_norm": 1.032797170774786,
      "learning_rate": 4.930193552129527e-06,
      "loss": 0.631,
      "step": 924
    },
    {
      "epoch": 1.13219094247246,
      "grad_norm": 0.5868325194223467,
      "learning_rate": 4.9300430831964286e-06,
      "loss": 0.7676,
      "step": 925
    },
    {
      "epoch": 1.1334149326805385,
      "grad_norm": 1.4125796974973803,
      "learning_rate": 4.929892454569888e-06,
      "loss": 0.8429,
      "step": 926
    },
    {
      "epoch": 1.134638922888617,
      "grad_norm": 1.0415909203562288,
      "learning_rate": 4.9297416662598025e-06,
      "loss": 0.6605,
      "step": 927
    },
    {
      "epoch": 1.1358629130966953,
      "grad_norm": 0.6958759036489013,
      "learning_rate": 4.929590718276082e-06,
      "loss": 0.4051,
      "step": 928
    },
    {
      "epoch": 1.1370869033047735,
      "grad_norm": 0.8563740613252238,
      "learning_rate": 4.929439610628645e-06,
      "loss": 1.0091,
      "step": 929
    },
    {
      "epoch": 1.1383108935128519,
      "grad_norm": 0.5959915597772225,
      "learning_rate": 4.929288343327423e-06,
      "loss": 0.6246,
      "step": 930
    },
    {
      "epoch": 1.1395348837209303,
      "grad_norm": 0.4289902223630486,
      "learning_rate": 4.929136916382356e-06,
      "loss": 0.5524,
      "step": 931
    },
    {
      "epoch": 1.1407588739290087,
      "grad_norm": 0.6468380010275062,
      "learning_rate": 4.928985329803397e-06,
      "loss": 0.7002,
      "step": 932
    },
    {
      "epoch": 1.1419828641370868,
      "grad_norm": 0.5693465792602754,
      "learning_rate": 4.928833583600505e-06,
      "loss": 0.5987,
      "step": 933
    },
    {
      "epoch": 1.1432068543451652,
      "grad_norm": 0.7824804311255348,
      "learning_rate": 4.928681677783654e-06,
      "loss": 0.6367,
      "step": 934
    },
    {
      "epoch": 1.1444308445532436,
      "grad_norm": 0.8154337392216127,
      "learning_rate": 4.928529612362827e-06,
      "loss": 0.6176,
      "step": 935
    },
    {
      "epoch": 1.1456548347613218,
      "grad_norm": 0.5610969221997649,
      "learning_rate": 4.928377387348016e-06,
      "loss": 0.5611,
      "step": 936
    },
    {
      "epoch": 1.1468788249694002,
      "grad_norm": 0.7099537279375294,
      "learning_rate": 4.928225002749224e-06,
      "loss": 0.8004,
      "step": 937
    },
    {
      "epoch": 1.1481028151774786,
      "grad_norm": 0.7894720183270739,
      "learning_rate": 4.928072458576467e-06,
      "loss": 0.565,
      "step": 938
    },
    {
      "epoch": 1.149326805385557,
      "grad_norm": 0.5316884827058088,
      "learning_rate": 4.92791975483977e-06,
      "loss": 0.5073,
      "step": 939
    },
    {
      "epoch": 1.1505507955936352,
      "grad_norm": 0.8910257373498645,
      "learning_rate": 4.927766891549165e-06,
      "loss": 0.4202,
      "step": 940
    },
    {
      "epoch": 1.1517747858017136,
      "grad_norm": 0.7859968254384013,
      "learning_rate": 4.9276138687147e-06,
      "loss": 0.4868,
      "step": 941
    },
    {
      "epoch": 1.152998776009792,
      "grad_norm": 0.9733159649971784,
      "learning_rate": 4.927460686346431e-06,
      "loss": 1.1687,
      "step": 942
    },
    {
      "epoch": 1.1542227662178703,
      "grad_norm": 0.5069621260234324,
      "learning_rate": 4.927307344454424e-06,
      "loss": 0.7368,
      "step": 943
    },
    {
      "epoch": 1.1554467564259485,
      "grad_norm": 0.9543918791709688,
      "learning_rate": 4.9271538430487565e-06,
      "loss": 1.5965,
      "step": 944
    },
    {
      "epoch": 1.156670746634027,
      "grad_norm": 0.6038739711408615,
      "learning_rate": 4.927000182139516e-06,
      "loss": 0.578,
      "step": 945
    },
    {
      "epoch": 1.1578947368421053,
      "grad_norm": 0.7634198474197418,
      "learning_rate": 4.9268463617368e-06,
      "loss": 0.7874,
      "step": 946
    },
    {
      "epoch": 1.1591187270501835,
      "grad_norm": 0.6256737297145649,
      "learning_rate": 4.926692381850718e-06,
      "loss": 1.1065,
      "step": 947
    },
    {
      "epoch": 1.1603427172582619,
      "grad_norm": 0.8339957897002079,
      "learning_rate": 4.926538242491388e-06,
      "loss": 1.2655,
      "step": 948
    },
    {
      "epoch": 1.1615667074663403,
      "grad_norm": 0.6127153345448116,
      "learning_rate": 4.92638394366894e-06,
      "loss": 0.6249,
      "step": 949
    },
    {
      "epoch": 1.1627906976744187,
      "grad_norm": 1.2542859836633777,
      "learning_rate": 4.926229485393513e-06,
      "loss": 0.5003,
      "step": 950
    },
    {
      "epoch": 1.1640146878824968,
      "grad_norm": 0.5927348994236576,
      "learning_rate": 4.9260748676752595e-06,
      "loss": 0.7237,
      "step": 951
    },
    {
      "epoch": 1.1652386780905752,
      "grad_norm": 1.2866449929619959,
      "learning_rate": 4.925920090524338e-06,
      "loss": 0.7623,
      "step": 952
    },
    {
      "epoch": 1.1664626682986536,
      "grad_norm": 0.6551881947629175,
      "learning_rate": 4.925765153950922e-06,
      "loss": 0.7113,
      "step": 953
    },
    {
      "epoch": 1.167686658506732,
      "grad_norm": 1.0664162593744582,
      "learning_rate": 4.925610057965193e-06,
      "loss": 0.5734,
      "step": 954
    },
    {
      "epoch": 1.1689106487148102,
      "grad_norm": 0.8377440063233088,
      "learning_rate": 4.925454802577341e-06,
      "loss": 0.6356,
      "step": 955
    },
    {
      "epoch": 1.1701346389228886,
      "grad_norm": 0.958904077070117,
      "learning_rate": 4.925299387797571e-06,
      "loss": 1.5659,
      "step": 956
    },
    {
      "epoch": 1.171358629130967,
      "grad_norm": 0.720541162087234,
      "learning_rate": 4.9251438136360956e-06,
      "loss": 0.627,
      "step": 957
    },
    {
      "epoch": 1.1725826193390452,
      "grad_norm": 0.8027211485843715,
      "learning_rate": 4.924988080103139e-06,
      "loss": 0.5878,
      "step": 958
    },
    {
      "epoch": 1.1738066095471236,
      "grad_norm": 0.680947294238266,
      "learning_rate": 4.924832187208936e-06,
      "loss": 0.8529,
      "step": 959
    },
    {
      "epoch": 1.175030599755202,
      "grad_norm": 0.7789477347455837,
      "learning_rate": 4.924676134963729e-06,
      "loss": 0.6991,
      "step": 960
    },
    {
      "epoch": 1.1762545899632804,
      "grad_norm": 0.5647351597644509,
      "learning_rate": 4.924519923377775e-06,
      "loss": 0.6869,
      "step": 961
    },
    {
      "epoch": 1.1774785801713585,
      "grad_norm": 0.8041850804465166,
      "learning_rate": 4.9243635524613405e-06,
      "loss": 0.5852,
      "step": 962
    },
    {
      "epoch": 1.178702570379437,
      "grad_norm": 0.7761495342909841,
      "learning_rate": 4.9242070222247e-06,
      "loss": 0.6551,
      "step": 963
    },
    {
      "epoch": 1.1799265605875153,
      "grad_norm": 0.5208367555983227,
      "learning_rate": 4.9240503326781395e-06,
      "loss": 0.7174,
      "step": 964
    },
    {
      "epoch": 1.1811505507955937,
      "grad_norm": 0.897098027578854,
      "learning_rate": 4.923893483831957e-06,
      "loss": 0.7096,
      "step": 965
    },
    {
      "epoch": 1.182374541003672,
      "grad_norm": 0.9420163277029596,
      "learning_rate": 4.923736475696462e-06,
      "loss": 0.6526,
      "step": 966
    },
    {
      "epoch": 1.1835985312117503,
      "grad_norm": 1.138976690639025,
      "learning_rate": 4.923579308281969e-06,
      "loss": 1.0771,
      "step": 967
    },
    {
      "epoch": 1.1848225214198287,
      "grad_norm": 0.8356549551399949,
      "learning_rate": 4.923421981598808e-06,
      "loss": 0.804,
      "step": 968
    },
    {
      "epoch": 1.1860465116279069,
      "grad_norm": 0.7107893489698167,
      "learning_rate": 4.923264495657319e-06,
      "loss": 0.7593,
      "step": 969
    },
    {
      "epoch": 1.1872705018359853,
      "grad_norm": 0.9764157104302571,
      "learning_rate": 4.923106850467851e-06,
      "loss": 0.5815,
      "step": 970
    },
    {
      "epoch": 1.1884944920440637,
      "grad_norm": 0.7543816850733417,
      "learning_rate": 4.922949046040761e-06,
      "loss": 1.0227,
      "step": 971
    },
    {
      "epoch": 1.189718482252142,
      "grad_norm": 0.5591102336542225,
      "learning_rate": 4.9227910823864225e-06,
      "loss": 0.8137,
      "step": 972
    },
    {
      "epoch": 1.1909424724602202,
      "grad_norm": 0.7500602509810964,
      "learning_rate": 4.922632959515215e-06,
      "loss": 1.0798,
      "step": 973
    },
    {
      "epoch": 1.1921664626682986,
      "grad_norm": 0.9627577952203146,
      "learning_rate": 4.922474677437532e-06,
      "loss": 0.5036,
      "step": 974
    },
    {
      "epoch": 1.193390452876377,
      "grad_norm": 0.973304754817475,
      "learning_rate": 4.922316236163771e-06,
      "loss": 0.7579,
      "step": 975
    },
    {
      "epoch": 1.1946144430844554,
      "grad_norm": 0.8126311162827432,
      "learning_rate": 4.9221576357043475e-06,
      "loss": 1.2265,
      "step": 976
    },
    {
      "epoch": 1.1958384332925336,
      "grad_norm": 1.2418598819945843,
      "learning_rate": 4.921998876069684e-06,
      "loss": 0.5119,
      "step": 977
    },
    {
      "epoch": 1.197062423500612,
      "grad_norm": 0.7086188588681058,
      "learning_rate": 4.921839957270211e-06,
      "loss": 0.6529,
      "step": 978
    },
    {
      "epoch": 1.1982864137086904,
      "grad_norm": 0.723585265704103,
      "learning_rate": 4.921680879316375e-06,
      "loss": 0.9056,
      "step": 979
    },
    {
      "epoch": 1.1995104039167686,
      "grad_norm": 0.7950226442863032,
      "learning_rate": 4.921521642218628e-06,
      "loss": 1.3962,
      "step": 980
    },
    {
      "epoch": 1.200734394124847,
      "grad_norm": 0.8648744807066634,
      "learning_rate": 4.921362245987436e-06,
      "loss": 0.8437,
      "step": 981
    },
    {
      "epoch": 1.2019583843329253,
      "grad_norm": 0.6128650184941334,
      "learning_rate": 4.921202690633272e-06,
      "loss": 1.0406,
      "step": 982
    },
    {
      "epoch": 1.2031823745410037,
      "grad_norm": 1.1443531703227012,
      "learning_rate": 4.921042976166623e-06,
      "loss": 0.9109,
      "step": 983
    },
    {
      "epoch": 1.204406364749082,
      "grad_norm": 0.8344225431890311,
      "learning_rate": 4.9208831025979854e-06,
      "loss": 0.6978,
      "step": 984
    },
    {
      "epoch": 1.2056303549571603,
      "grad_norm": 1.1195495602100733,
      "learning_rate": 4.9207230699378636e-06,
      "loss": 0.7325,
      "step": 985
    },
    {
      "epoch": 1.2068543451652387,
      "grad_norm": 0.39588294193338325,
      "learning_rate": 4.920562878196776e-06,
      "loss": 0.3968,
      "step": 986
    },
    {
      "epoch": 1.208078335373317,
      "grad_norm": 0.9760070512472677,
      "learning_rate": 4.920402527385248e-06,
      "loss": 0.5608,
      "step": 987
    },
    {
      "epoch": 1.2093023255813953,
      "grad_norm": 1.2293153634384493,
      "learning_rate": 4.92024201751382e-06,
      "loss": 0.6443,
      "step": 988
    },
    {
      "epoch": 1.2105263157894737,
      "grad_norm": 0.6030495268527476,
      "learning_rate": 4.920081348593038e-06,
      "loss": 0.55,
      "step": 989
    },
    {
      "epoch": 1.211750305997552,
      "grad_norm": 0.7126068612496806,
      "learning_rate": 4.9199205206334595e-06,
      "loss": 1.3686,
      "step": 990
    },
    {
      "epoch": 1.2129742962056302,
      "grad_norm": 0.7740471075086655,
      "learning_rate": 4.919759533645656e-06,
      "loss": 0.8094,
      "step": 991
    },
    {
      "epoch": 1.2141982864137086,
      "grad_norm": 0.515580783456274,
      "learning_rate": 4.919598387640207e-06,
      "loss": 0.6673,
      "step": 992
    },
    {
      "epoch": 1.215422276621787,
      "grad_norm": 0.780976307398921,
      "learning_rate": 4.919437082627701e-06,
      "loss": 0.641,
      "step": 993
    },
    {
      "epoch": 1.2166462668298654,
      "grad_norm": 0.9544816299190854,
      "learning_rate": 4.919275618618737e-06,
      "loss": 1.4979,
      "step": 994
    },
    {
      "epoch": 1.2178702570379436,
      "grad_norm": 0.5106496870118997,
      "learning_rate": 4.919113995623929e-06,
      "loss": 0.6591,
      "step": 995
    },
    {
      "epoch": 1.219094247246022,
      "grad_norm": 0.6762304323937639,
      "learning_rate": 4.918952213653897e-06,
      "loss": 0.6253,
      "step": 996
    },
    {
      "epoch": 1.2203182374541004,
      "grad_norm": 0.6838575604461612,
      "learning_rate": 4.918790272719273e-06,
      "loss": 1.1172,
      "step": 997
    },
    {
      "epoch": 1.2215422276621788,
      "grad_norm": 0.8246161308497401,
      "learning_rate": 4.918628172830698e-06,
      "loss": 0.7533,
      "step": 998
    },
    {
      "epoch": 1.222766217870257,
      "grad_norm": 1.0863318043310828,
      "learning_rate": 4.918465913998825e-06,
      "loss": 0.4049,
      "step": 999
    },
    {
      "epoch": 1.2239902080783354,
      "grad_norm": 0.6096340900185448,
      "learning_rate": 4.918303496234318e-06,
      "loss": 0.8322,
      "step": 1000
    },
    {
      "epoch": 1.2252141982864138,
      "grad_norm": 0.8645840083019114,
      "learning_rate": 4.91814091954785e-06,
      "loss": 0.4845,
      "step": 1001
    },
    {
      "epoch": 1.226438188494492,
      "grad_norm": 0.8102115911830429,
      "learning_rate": 4.917978183950105e-06,
      "loss": 0.6439,
      "step": 1002
    },
    {
      "epoch": 1.2276621787025703,
      "grad_norm": 0.7512809015329637,
      "learning_rate": 4.917815289451777e-06,
      "loss": 1.3333,
      "step": 1003
    },
    {
      "epoch": 1.2288861689106487,
      "grad_norm": 0.6431051421878352,
      "learning_rate": 4.917652236063572e-06,
      "loss": 0.7275,
      "step": 1004
    },
    {
      "epoch": 1.2301101591187271,
      "grad_norm": 0.6448254471673616,
      "learning_rate": 4.917489023796202e-06,
      "loss": 0.8449,
      "step": 1005
    },
    {
      "epoch": 1.2313341493268053,
      "grad_norm": 1.330971742119266,
      "learning_rate": 4.9173256526603975e-06,
      "loss": 0.7213,
      "step": 1006
    },
    {
      "epoch": 1.2325581395348837,
      "grad_norm": 0.8432932452798065,
      "learning_rate": 4.917162122666891e-06,
      "loss": 0.6551,
      "step": 1007
    },
    {
      "epoch": 1.233782129742962,
      "grad_norm": 0.9338181929205969,
      "learning_rate": 4.91699843382643e-06,
      "loss": 0.594,
      "step": 1008
    },
    {
      "epoch": 1.2350061199510405,
      "grad_norm": 0.47003880427296724,
      "learning_rate": 4.916834586149772e-06,
      "loss": 0.5829,
      "step": 1009
    },
    {
      "epoch": 1.2362301101591187,
      "grad_norm": 1.1175818271109208,
      "learning_rate": 4.9166705796476846e-06,
      "loss": 0.8135,
      "step": 1010
    },
    {
      "epoch": 1.237454100367197,
      "grad_norm": 0.5461684548674339,
      "learning_rate": 4.916506414330945e-06,
      "loss": 0.8966,
      "step": 1011
    },
    {
      "epoch": 1.2386780905752754,
      "grad_norm": 0.8189207716880389,
      "learning_rate": 4.916342090210342e-06,
      "loss": 1.0068,
      "step": 1012
    },
    {
      "epoch": 1.2399020807833536,
      "grad_norm": 1.019437584202964,
      "learning_rate": 4.916177607296675e-06,
      "loss": 0.5515,
      "step": 1013
    },
    {
      "epoch": 1.241126070991432,
      "grad_norm": 0.929667499792301,
      "learning_rate": 4.916012965600752e-06,
      "loss": 0.5994,
      "step": 1014
    },
    {
      "epoch": 1.2423500611995104,
      "grad_norm": 0.5304088553117589,
      "learning_rate": 4.915848165133392e-06,
      "loss": 0.6411,
      "step": 1015
    },
    {
      "epoch": 1.2435740514075888,
      "grad_norm": 0.5282803546863084,
      "learning_rate": 4.915683205905427e-06,
      "loss": 0.7026,
      "step": 1016
    },
    {
      "epoch": 1.244798041615667,
      "grad_norm": 0.7887128091764675,
      "learning_rate": 4.9155180879276974e-06,
      "loss": 0.83,
      "step": 1017
    },
    {
      "epoch": 1.2460220318237454,
      "grad_norm": 0.8038974396610025,
      "learning_rate": 4.915352811211053e-06,
      "loss": 0.9488,
      "step": 1018
    },
    {
      "epoch": 1.2472460220318238,
      "grad_norm": 1.2872357738168838,
      "learning_rate": 4.915187375766355e-06,
      "loss": 0.5598,
      "step": 1019
    },
    {
      "epoch": 1.2484700122399022,
      "grad_norm": 0.7219431286340979,
      "learning_rate": 4.915021781604476e-06,
      "loss": 0.2874,
      "step": 1020
    },
    {
      "epoch": 1.2496940024479803,
      "grad_norm": 0.6710068681026446,
      "learning_rate": 4.914856028736298e-06,
      "loss": 0.5839,
      "step": 1021
    },
    {
      "epoch": 1.2509179926560587,
      "grad_norm": 1.2432570458466436,
      "learning_rate": 4.9146901171727144e-06,
      "loss": 0.5486,
      "step": 1022
    },
    {
      "epoch": 1.2521419828641371,
      "grad_norm": 0.8151537942527669,
      "learning_rate": 4.914524046924627e-06,
      "loss": 0.711,
      "step": 1023
    },
    {
      "epoch": 1.2533659730722153,
      "grad_norm": 0.9504461739681449,
      "learning_rate": 4.91435781800295e-06,
      "loss": 1.2027,
      "step": 1024
    },
    {
      "epoch": 1.2545899632802937,
      "grad_norm": 0.90578730800974,
      "learning_rate": 4.914191430418608e-06,
      "loss": 0.6249,
      "step": 1025
    },
    {
      "epoch": 1.255813953488372,
      "grad_norm": 1.4054020932479068,
      "learning_rate": 4.914024884182535e-06,
      "loss": 0.8146,
      "step": 1026
    },
    {
      "epoch": 1.2570379436964505,
      "grad_norm": 0.6786586298943523,
      "learning_rate": 4.9138581793056735e-06,
      "loss": 0.7505,
      "step": 1027
    },
    {
      "epoch": 1.258261933904529,
      "grad_norm": 0.761767346758803,
      "learning_rate": 4.9136913157989825e-06,
      "loss": 1.2721,
      "step": 1028
    },
    {
      "epoch": 1.259485924112607,
      "grad_norm": 1.1651046257041635,
      "learning_rate": 4.9135242936734255e-06,
      "loss": 0.9971,
      "step": 1029
    },
    {
      "epoch": 1.2607099143206855,
      "grad_norm": 1.1112997733405057,
      "learning_rate": 4.913357112939978e-06,
      "loss": 0.5412,
      "step": 1030
    },
    {
      "epoch": 1.2619339045287639,
      "grad_norm": 1.077518229420999,
      "learning_rate": 4.913189773609628e-06,
      "loss": 0.6806,
      "step": 1031
    },
    {
      "epoch": 1.263157894736842,
      "grad_norm": 0.7799422346648465,
      "learning_rate": 4.913022275693372e-06,
      "loss": 1.6532,
      "step": 1032
    },
    {
      "epoch": 1.2643818849449204,
      "grad_norm": 1.2686396516334226,
      "learning_rate": 4.912854619202218e-06,
      "loss": 0.5767,
      "step": 1033
    },
    {
      "epoch": 1.2656058751529988,
      "grad_norm": 0.9762570273001033,
      "learning_rate": 4.912686804147181e-06,
      "loss": 0.4087,
      "step": 1034
    },
    {
      "epoch": 1.266829865361077,
      "grad_norm": 0.6009351904915886,
      "learning_rate": 4.912518830539292e-06,
      "loss": 0.7852,
      "step": 1035
    },
    {
      "epoch": 1.2680538555691554,
      "grad_norm": 0.5706649725043963,
      "learning_rate": 4.91235069838959e-06,
      "loss": 0.6202,
      "step": 1036
    },
    {
      "epoch": 1.2692778457772338,
      "grad_norm": 0.48467864123474236,
      "learning_rate": 4.91218240770912e-06,
      "loss": 0.5806,
      "step": 1037
    },
    {
      "epoch": 1.2705018359853122,
      "grad_norm": 0.6223687908401416,
      "learning_rate": 4.912013958508947e-06,
      "loss": 0.6051,
      "step": 1038
    },
    {
      "epoch": 1.2717258261933906,
      "grad_norm": 1.5454483233117453,
      "learning_rate": 4.9118453508001375e-06,
      "loss": 0.5579,
      "step": 1039
    },
    {
      "epoch": 1.2729498164014688,
      "grad_norm": 0.5855177011950922,
      "learning_rate": 4.9116765845937715e-06,
      "loss": 0.575,
      "step": 1040
    },
    {
      "epoch": 1.2741738066095472,
      "grad_norm": 0.5714269294840185,
      "learning_rate": 4.91150765990094e-06,
      "loss": 0.6914,
      "step": 1041
    },
    {
      "epoch": 1.2753977968176256,
      "grad_norm": 0.8462659370753556,
      "learning_rate": 4.911338576732746e-06,
      "loss": 0.8168,
      "step": 1042
    },
    {
      "epoch": 1.2766217870257037,
      "grad_norm": 1.1846517633620257,
      "learning_rate": 4.9111693351003e-06,
      "loss": 0.4928,
      "step": 1043
    },
    {
      "epoch": 1.2778457772337821,
      "grad_norm": 0.6396779225132486,
      "learning_rate": 4.910999935014722e-06,
      "loss": 0.6902,
      "step": 1044
    },
    {
      "epoch": 1.2790697674418605,
      "grad_norm": 1.1272876104119254,
      "learning_rate": 4.910830376487147e-06,
      "loss": 0.6584,
      "step": 1045
    },
    {
      "epoch": 1.2802937576499387,
      "grad_norm": 0.94587504450619,
      "learning_rate": 4.910660659528716e-06,
      "loss": 1.4931,
      "step": 1046
    },
    {
      "epoch": 1.281517747858017,
      "grad_norm": 0.8434396433757076,
      "learning_rate": 4.910490784150583e-06,
      "loss": 0.7663,
      "step": 1047
    },
    {
      "epoch": 1.2827417380660955,
      "grad_norm": 0.7249595657518458,
      "learning_rate": 4.910320750363912e-06,
      "loss": 0.6905,
      "step": 1048
    },
    {
      "epoch": 1.2839657282741739,
      "grad_norm": 0.39607086271841424,
      "learning_rate": 4.910150558179877e-06,
      "loss": 0.2548,
      "step": 1049
    },
    {
      "epoch": 1.2851897184822523,
      "grad_norm": 0.6255905752065372,
      "learning_rate": 4.909980207609661e-06,
      "loss": 0.6511,
      "step": 1050
    },
    {
      "epoch": 1.2864137086903304,
      "grad_norm": 1.2164878690095329,
      "learning_rate": 4.9098096986644595e-06,
      "loss": 1.0885,
      "step": 1051
    },
    {
      "epoch": 1.2876376988984088,
      "grad_norm": 0.8338084387779549,
      "learning_rate": 4.909639031355478e-06,
      "loss": 1.0243,
      "step": 1052
    },
    {
      "epoch": 1.2888616891064872,
      "grad_norm": 0.9307022124869818,
      "learning_rate": 4.909468205693932e-06,
      "loss": 1.4632,
      "step": 1053
    },
    {
      "epoch": 1.2900856793145654,
      "grad_norm": 0.5907288912057423,
      "learning_rate": 4.909297221691047e-06,
      "loss": 0.6851,
      "step": 1054
    },
    {
      "epoch": 1.2913096695226438,
      "grad_norm": 0.7859393991764065,
      "learning_rate": 4.909126079358061e-06,
      "loss": 1.659,
      "step": 1055
    },
    {
      "epoch": 1.2925336597307222,
      "grad_norm": 0.6946272145980266,
      "learning_rate": 4.90895477870622e-06,
      "loss": 0.8927,
      "step": 1056
    },
    {
      "epoch": 1.2937576499388004,
      "grad_norm": 0.8647521894953364,
      "learning_rate": 4.90878331974678e-06,
      "loss": 0.5832,
      "step": 1057
    },
    {
      "epoch": 1.2949816401468788,
      "grad_norm": 1.3439474574306063,
      "learning_rate": 4.90861170249101e-06,
      "loss": 0.9269,
      "step": 1058
    },
    {
      "epoch": 1.2962056303549572,
      "grad_norm": 0.7759666294897798,
      "learning_rate": 4.908439926950188e-06,
      "loss": 0.501,
      "step": 1059
    },
    {
      "epoch": 1.2974296205630356,
      "grad_norm": 0.7457974282048464,
      "learning_rate": 4.908267993135602e-06,
      "loss": 0.6254,
      "step": 1060
    },
    {
      "epoch": 1.298653610771114,
      "grad_norm": 1.2092145958070974,
      "learning_rate": 4.90809590105855e-06,
      "loss": 0.5816,
      "step": 1061
    },
    {
      "epoch": 1.2998776009791921,
      "grad_norm": 0.5944068061053872,
      "learning_rate": 4.907923650730344e-06,
      "loss": 0.7543,
      "step": 1062
    },
    {
      "epoch": 1.3011015911872705,
      "grad_norm": 1.157101291382137,
      "learning_rate": 4.907751242162302e-06,
      "loss": 1.3014,
      "step": 1063
    },
    {
      "epoch": 1.302325581395349,
      "grad_norm": 1.1164781148840996,
      "learning_rate": 4.907578675365753e-06,
      "loss": 0.7546,
      "step": 1064
    },
    {
      "epoch": 1.303549571603427,
      "grad_norm": 0.5310689856584393,
      "learning_rate": 4.907405950352039e-06,
      "loss": 0.568,
      "step": 1065
    },
    {
      "epoch": 1.3047735618115055,
      "grad_norm": 1.1875873221422717,
      "learning_rate": 4.90723306713251e-06,
      "loss": 0.3372,
      "step": 1066
    },
    {
      "epoch": 1.305997552019584,
      "grad_norm": 1.0307019916100078,
      "learning_rate": 4.907060025718528e-06,
      "loss": 0.7595,
      "step": 1067
    },
    {
      "epoch": 1.307221542227662,
      "grad_norm": 0.6578066026318019,
      "learning_rate": 4.906886826121464e-06,
      "loss": 0.6299,
      "step": 1068
    },
    {
      "epoch": 1.3084455324357405,
      "grad_norm": 0.6371688410083587,
      "learning_rate": 4.9067134683527e-06,
      "loss": 0.6252,
      "step": 1069
    },
    {
      "epoch": 1.3096695226438189,
      "grad_norm": 0.7943568486224012,
      "learning_rate": 4.906539952423629e-06,
      "loss": 0.5642,
      "step": 1070
    },
    {
      "epoch": 1.3108935128518973,
      "grad_norm": 1.085381880539723,
      "learning_rate": 4.906366278345653e-06,
      "loss": 0.6709,
      "step": 1071
    },
    {
      "epoch": 1.3121175030599757,
      "grad_norm": 0.9280479231043428,
      "learning_rate": 4.906192446130187e-06,
      "loss": 1.1794,
      "step": 1072
    },
    {
      "epoch": 1.3133414932680538,
      "grad_norm": 0.7631110346505048,
      "learning_rate": 4.906018455788652e-06,
      "loss": 0.606,
      "step": 1073
    },
    {
      "epoch": 1.3145654834761322,
      "grad_norm": 0.7274024351807775,
      "learning_rate": 4.905844307332484e-06,
      "loss": 0.5015,
      "step": 1074
    },
    {
      "epoch": 1.3157894736842106,
      "grad_norm": 0.7971073514761351,
      "learning_rate": 4.905670000773126e-06,
      "loss": 0.5541,
      "step": 1075
    },
    {
      "epoch": 1.3170134638922888,
      "grad_norm": 0.5524365760657188,
      "learning_rate": 4.905495536122034e-06,
      "loss": 0.6665,
      "step": 1076
    },
    {
      "epoch": 1.3182374541003672,
      "grad_norm": 0.9903957545500455,
      "learning_rate": 4.905320913390673e-06,
      "loss": 0.6868,
      "step": 1077
    },
    {
      "epoch": 1.3194614443084456,
      "grad_norm": 1.4867549530108757,
      "learning_rate": 4.905146132590518e-06,
      "loss": 0.5044,
      "step": 1078
    },
    {
      "epoch": 1.3206854345165238,
      "grad_norm": 0.8079198359312023,
      "learning_rate": 4.904971193733055e-06,
      "loss": 1.1417,
      "step": 1079
    },
    {
      "epoch": 1.3219094247246022,
      "grad_norm": 0.7705632401976016,
      "learning_rate": 4.904796096829781e-06,
      "loss": 0.9595,
      "step": 1080
    },
    {
      "epoch": 1.3231334149326806,
      "grad_norm": 0.7735109978261849,
      "learning_rate": 4.904620841892201e-06,
      "loss": 0.6258,
      "step": 1081
    },
    {
      "epoch": 1.324357405140759,
      "grad_norm": 0.8609514439504001,
      "learning_rate": 4.9044454289318345e-06,
      "loss": 1.4805,
      "step": 1082
    },
    {
      "epoch": 1.3255813953488373,
      "grad_norm": 0.8790860607024982,
      "learning_rate": 4.904269857960208e-06,
      "loss": 0.6894,
      "step": 1083
    },
    {
      "epoch": 1.3268053855569155,
      "grad_norm": 1.1010200781840789,
      "learning_rate": 4.9040941289888585e-06,
      "loss": 0.5819,
      "step": 1084
    },
    {
      "epoch": 1.328029375764994,
      "grad_norm": 1.01638728777305,
      "learning_rate": 4.903918242029335e-06,
      "loss": 0.8175,
      "step": 1085
    },
    {
      "epoch": 1.3292533659730723,
      "grad_norm": 0.5946861261723128,
      "learning_rate": 4.903742197093197e-06,
      "loss": 0.6346,
      "step": 1086
    },
    {
      "epoch": 1.3304773561811505,
      "grad_norm": 0.9277566765201711,
      "learning_rate": 4.903565994192012e-06,
      "loss": 0.5345,
      "step": 1087
    },
    {
      "epoch": 1.3317013463892289,
      "grad_norm": 0.8782016445114257,
      "learning_rate": 4.90338963333736e-06,
      "loss": 1.2902,
      "step": 1088
    },
    {
      "epoch": 1.3329253365973073,
      "grad_norm": 0.6864683930465169,
      "learning_rate": 4.903213114540831e-06,
      "loss": 0.8311,
      "step": 1089
    },
    {
      "epoch": 1.3341493268053854,
      "grad_norm": 0.8963701147015143,
      "learning_rate": 4.903036437814025e-06,
      "loss": 1.285,
      "step": 1090
    },
    {
      "epoch": 1.3353733170134638,
      "grad_norm": 1.2402983899185305,
      "learning_rate": 4.902859603168552e-06,
      "loss": 1.0214,
      "step": 1091
    },
    {
      "epoch": 1.3365973072215422,
      "grad_norm": 0.9044587068239511,
      "learning_rate": 4.902682610616034e-06,
      "loss": 0.5171,
      "step": 1092
    },
    {
      "epoch": 1.3378212974296206,
      "grad_norm": 0.6492804779132865,
      "learning_rate": 4.902505460168101e-06,
      "loss": 0.7111,
      "step": 1093
    },
    {
      "epoch": 1.339045287637699,
      "grad_norm": 0.931608547087862,
      "learning_rate": 4.9023281518363965e-06,
      "loss": 1.2952,
      "step": 1094
    },
    {
      "epoch": 1.3402692778457772,
      "grad_norm": 1.4355971234240854,
      "learning_rate": 4.902150685632571e-06,
      "loss": 1.1625,
      "step": 1095
    },
    {
      "epoch": 1.3414932680538556,
      "grad_norm": 0.6497840413474866,
      "learning_rate": 4.901973061568287e-06,
      "loss": 0.447,
      "step": 1096
    },
    {
      "epoch": 1.342717258261934,
      "grad_norm": 0.8431309503813132,
      "learning_rate": 4.901795279655218e-06,
      "loss": 1.7044,
      "step": 1097
    },
    {
      "epoch": 1.3439412484700122,
      "grad_norm": 0.5682078592186636,
      "learning_rate": 4.901617339905047e-06,
      "loss": 0.5916,
      "step": 1098
    },
    {
      "epoch": 1.3451652386780906,
      "grad_norm": 0.8486957061312206,
      "learning_rate": 4.901439242329467e-06,
      "loss": 1.6226,
      "step": 1099
    },
    {
      "epoch": 1.346389228886169,
      "grad_norm": 0.9450196687163396,
      "learning_rate": 4.9012609869401826e-06,
      "loss": 0.6011,
      "step": 1100
    },
    {
      "epoch": 1.3476132190942471,
      "grad_norm": 0.792601072258462,
      "learning_rate": 4.901082573748907e-06,
      "loss": 0.7509,
      "step": 1101
    },
    {
      "epoch": 1.3488372093023255,
      "grad_norm": 0.7975393702893371,
      "learning_rate": 4.900904002767367e-06,
      "loss": 0.5591,
      "step": 1102
    },
    {
      "epoch": 1.350061199510404,
      "grad_norm": 0.622536101146563,
      "learning_rate": 4.900725274007296e-06,
      "loss": 0.6398,
      "step": 1103
    },
    {
      "epoch": 1.3512851897184823,
      "grad_norm": 1.027752781787025,
      "learning_rate": 4.900546387480439e-06,
      "loss": 2.3774,
      "step": 1104
    },
    {
      "epoch": 1.3525091799265607,
      "grad_norm": 0.9324896453336858,
      "learning_rate": 4.9003673431985525e-06,
      "loss": 0.5183,
      "step": 1105
    },
    {
      "epoch": 1.353733170134639,
      "grad_norm": 1.146571080251268,
      "learning_rate": 4.900188141173403e-06,
      "loss": 0.3474,
      "step": 1106
    },
    {
      "epoch": 1.3549571603427173,
      "grad_norm": 1.3718909876753118,
      "learning_rate": 4.9000087814167665e-06,
      "loss": 0.4625,
      "step": 1107
    },
    {
      "epoch": 1.3561811505507957,
      "grad_norm": 0.868832683744956,
      "learning_rate": 4.899829263940429e-06,
      "loss": 0.6029,
      "step": 1108
    },
    {
      "epoch": 1.3574051407588739,
      "grad_norm": 0.6763350045508673,
      "learning_rate": 4.89964958875619e-06,
      "loss": 0.7569,
      "step": 1109
    },
    {
      "epoch": 1.3586291309669523,
      "grad_norm": 1.278962063921407,
      "learning_rate": 4.899469755875855e-06,
      "loss": 0.6489,
      "step": 1110
    },
    {
      "epoch": 1.3598531211750307,
      "grad_norm": 1.0905033742375256,
      "learning_rate": 4.899289765311243e-06,
      "loss": 0.4515,
      "step": 1111
    },
    {
      "epoch": 1.3610771113831088,
      "grad_norm": 1.4747607544809413,
      "learning_rate": 4.899109617074181e-06,
      "loss": 1.1834,
      "step": 1112
    },
    {
      "epoch": 1.3623011015911872,
      "grad_norm": 0.9530163198519706,
      "learning_rate": 4.89892931117651e-06,
      "loss": 1.0856,
      "step": 1113
    },
    {
      "epoch": 1.3635250917992656,
      "grad_norm": 0.6995087334805631,
      "learning_rate": 4.8987488476300765e-06,
      "loss": 0.4748,
      "step": 1114
    },
    {
      "epoch": 1.364749082007344,
      "grad_norm": 0.8860788685367288,
      "learning_rate": 4.898568226446742e-06,
      "loss": 1.6588,
      "step": 1115
    },
    {
      "epoch": 1.3659730722154224,
      "grad_norm": 0.7267940445380268,
      "learning_rate": 4.898387447638374e-06,
      "loss": 0.5623,
      "step": 1116
    },
    {
      "epoch": 1.3671970624235006,
      "grad_norm": 0.7041345906784646,
      "learning_rate": 4.898206511216855e-06,
      "loss": 0.8569,
      "step": 1117
    },
    {
      "epoch": 1.368421052631579,
      "grad_norm": 0.681304298029581,
      "learning_rate": 4.898025417194075e-06,
      "loss": 0.7717,
      "step": 1118
    },
    {
      "epoch": 1.3696450428396574,
      "grad_norm": 0.6131686214160815,
      "learning_rate": 4.897844165581934e-06,
      "loss": 0.441,
      "step": 1119
    },
    {
      "epoch": 1.3708690330477356,
      "grad_norm": 1.2188223275005334,
      "learning_rate": 4.897662756392343e-06,
      "loss": 1.0677,
      "step": 1120
    },
    {
      "epoch": 1.372093023255814,
      "grad_norm": 1.0282830482489482,
      "learning_rate": 4.897481189637224e-06,
      "loss": 0.552,
      "step": 1121
    },
    {
      "epoch": 1.3733170134638923,
      "grad_norm": 0.7282006820265837,
      "learning_rate": 4.897299465328509e-06,
      "loss": 0.5896,
      "step": 1122
    },
    {
      "epoch": 1.3745410036719705,
      "grad_norm": 0.8610131147096707,
      "learning_rate": 4.89711758347814e-06,
      "loss": 0.5657,
      "step": 1123
    },
    {
      "epoch": 1.375764993880049,
      "grad_norm": 0.7278065809704583,
      "learning_rate": 4.896935544098069e-06,
      "loss": 0.5613,
      "step": 1124
    },
    {
      "epoch": 1.3769889840881273,
      "grad_norm": 1.0723322353464118,
      "learning_rate": 4.896753347200261e-06,
      "loss": 0.6708,
      "step": 1125
    },
    {
      "epoch": 1.3782129742962057,
      "grad_norm": 1.0095410744038804,
      "learning_rate": 4.8965709927966875e-06,
      "loss": 0.7943,
      "step": 1126
    },
    {
      "epoch": 1.379436964504284,
      "grad_norm": 0.9330199231654827,
      "learning_rate": 4.896388480899333e-06,
      "loss": 1.3037,
      "step": 1127
    },
    {
      "epoch": 1.3806609547123623,
      "grad_norm": 1.242765047001813,
      "learning_rate": 4.896205811520191e-06,
      "loss": 0.5315,
      "step": 1128
    },
    {
      "epoch": 1.3818849449204407,
      "grad_norm": 0.7328236421804898,
      "learning_rate": 4.896022984671266e-06,
      "loss": 0.5529,
      "step": 1129
    },
    {
      "epoch": 1.383108935128519,
      "grad_norm": 1.4924446837060636,
      "learning_rate": 4.895840000364572e-06,
      "loss": 0.6041,
      "step": 1130
    },
    {
      "epoch": 1.3843329253365972,
      "grad_norm": 0.5750379305200451,
      "learning_rate": 4.895656858612136e-06,
      "loss": 0.6575,
      "step": 1131
    },
    {
      "epoch": 1.3855569155446756,
      "grad_norm": 1.0475845842601774,
      "learning_rate": 4.895473559425991e-06,
      "loss": 0.4975,
      "step": 1132
    },
    {
      "epoch": 1.386780905752754,
      "grad_norm": 0.5311708201710875,
      "learning_rate": 4.895290102818184e-06,
      "loss": 0.5973,
      "step": 1133
    },
    {
      "epoch": 1.3880048959608322,
      "grad_norm": 0.47767838262645773,
      "learning_rate": 4.895106488800772e-06,
      "loss": 0.3936,
      "step": 1134
    },
    {
      "epoch": 1.3892288861689106,
      "grad_norm": 0.7361757544350555,
      "learning_rate": 4.89492271738582e-06,
      "loss": 1.0734,
      "step": 1135
    },
    {
      "epoch": 1.390452876376989,
      "grad_norm": 0.8097075253269312,
      "learning_rate": 4.894738788585404e-06,
      "loss": 1.5948,
      "step": 1136
    },
    {
      "epoch": 1.3916768665850674,
      "grad_norm": 0.5547630807214372,
      "learning_rate": 4.894554702411615e-06,
      "loss": 0.6451,
      "step": 1137
    },
    {
      "epoch": 1.3929008567931458,
      "grad_norm": 0.785935352703157,
      "learning_rate": 4.894370458876546e-06,
      "loss": 0.5716,
      "step": 1138
    },
    {
      "epoch": 1.394124847001224,
      "grad_norm": 1.6645337053339961,
      "learning_rate": 4.894186057992307e-06,
      "loss": 0.8786,
      "step": 1139
    },
    {
      "epoch": 1.3953488372093024,
      "grad_norm": 1.0299728094290033,
      "learning_rate": 4.894001499771015e-06,
      "loss": 0.5646,
      "step": 1140
    },
    {
      "epoch": 1.3965728274173808,
      "grad_norm": 0.6884106902635764,
      "learning_rate": 4.8938167842248e-06,
      "loss": 1.1515,
      "step": 1141
    },
    {
      "epoch": 1.397796817625459,
      "grad_norm": 0.5440666764789626,
      "learning_rate": 4.893631911365801e-06,
      "loss": 0.6304,
      "step": 1142
    },
    {
      "epoch": 1.3990208078335373,
      "grad_norm": 1.0067017257150366,
      "learning_rate": 4.893446881206164e-06,
      "loss": 0.625,
      "step": 1143
    },
    {
      "epoch": 1.4002447980416157,
      "grad_norm": 0.7455584968586297,
      "learning_rate": 4.893261693758052e-06,
      "loss": 0.6521,
      "step": 1144
    },
    {
      "epoch": 1.401468788249694,
      "grad_norm": 1.090075140968715,
      "learning_rate": 4.8930763490336334e-06,
      "loss": 0.4371,
      "step": 1145
    },
    {
      "epoch": 1.4026927784577723,
      "grad_norm": 0.5981417185342028,
      "learning_rate": 4.892890847045089e-06,
      "loss": 0.5328,
      "step": 1146
    },
    {
      "epoch": 1.4039167686658507,
      "grad_norm": 0.5182221339513697,
      "learning_rate": 4.892705187804608e-06,
      "loss": 0.6609,
      "step": 1147
    },
    {
      "epoch": 1.405140758873929,
      "grad_norm": 0.8568359870937143,
      "learning_rate": 4.892519371324393e-06,
      "loss": 0.6739,
      "step": 1148
    },
    {
      "epoch": 1.4063647490820075,
      "grad_norm": 1.0313713154038737,
      "learning_rate": 4.892333397616653e-06,
      "loss": 0.7129,
      "step": 1149
    },
    {
      "epoch": 1.4075887392900857,
      "grad_norm": 1.430588490396873,
      "learning_rate": 4.892147266693611e-06,
      "loss": 0.4512,
      "step": 1150
    },
    {
      "epoch": 1.408812729498164,
      "grad_norm": 0.6693619914780443,
      "learning_rate": 4.891960978567499e-06,
      "loss": 0.7809,
      "step": 1151
    },
    {
      "epoch": 1.4100367197062424,
      "grad_norm": 0.6572112029500279,
      "learning_rate": 4.891774533250559e-06,
      "loss": 0.8061,
      "step": 1152
    },
    {
      "epoch": 1.4112607099143206,
      "grad_norm": 0.6249133135188174,
      "learning_rate": 4.8915879307550426e-06,
      "loss": 0.8543,
      "step": 1153
    },
    {
      "epoch": 1.412484700122399,
      "grad_norm": 1.3646176291362138,
      "learning_rate": 4.891401171093214e-06,
      "loss": 0.4781,
      "step": 1154
    },
    {
      "epoch": 1.4137086903304774,
      "grad_norm": 0.7817140234578278,
      "learning_rate": 4.891214254277345e-06,
      "loss": 0.5528,
      "step": 1155
    },
    {
      "epoch": 1.4149326805385556,
      "grad_norm": 0.8346509314599219,
      "learning_rate": 4.8910271803197195e-06,
      "loss": 0.7754,
      "step": 1156
    },
    {
      "epoch": 1.416156670746634,
      "grad_norm": 0.8610127899815279,
      "learning_rate": 4.890839949232632e-06,
      "loss": 1.5457,
      "step": 1157
    },
    {
      "epoch": 1.4173806609547124,
      "grad_norm": 1.4054426726173574,
      "learning_rate": 4.890652561028386e-06,
      "loss": 0.6113,
      "step": 1158
    },
    {
      "epoch": 1.4186046511627908,
      "grad_norm": 0.8273339290234721,
      "learning_rate": 4.890465015719297e-06,
      "loss": 0.3913,
      "step": 1159
    },
    {
      "epoch": 1.4198286413708692,
      "grad_norm": 0.8687774648520845,
      "learning_rate": 4.890277313317687e-06,
      "loss": 0.7198,
      "step": 1160
    },
    {
      "epoch": 1.4210526315789473,
      "grad_norm": 0.8946357741561024,
      "learning_rate": 4.890089453835894e-06,
      "loss": 1.1938,
      "step": 1161
    },
    {
      "epoch": 1.4222766217870257,
      "grad_norm": 0.8192884566968174,
      "learning_rate": 4.889901437286263e-06,
      "loss": 0.6155,
      "step": 1162
    },
    {
      "epoch": 1.4235006119951041,
      "grad_norm": 1.4366569292063802,
      "learning_rate": 4.889713263681149e-06,
      "loss": 0.5785,
      "step": 1163
    },
    {
      "epoch": 1.4247246022031823,
      "grad_norm": 0.5206450031276049,
      "learning_rate": 4.8895249330329175e-06,
      "loss": 0.5796,
      "step": 1164
    },
    {
      "epoch": 1.4259485924112607,
      "grad_norm": 0.8038562597100775,
      "learning_rate": 4.889336445353946e-06,
      "loss": 0.5033,
      "step": 1165
    },
    {
      "epoch": 1.427172582619339,
      "grad_norm": 1.0745047680459017,
      "learning_rate": 4.889147800656621e-06,
      "loss": 1.0874,
      "step": 1166
    },
    {
      "epoch": 1.4283965728274173,
      "grad_norm": 0.6879359687175218,
      "learning_rate": 4.88895899895334e-06,
      "loss": 0.5625,
      "step": 1167
    },
    {
      "epoch": 1.4296205630354957,
      "grad_norm": 0.528064917488075,
      "learning_rate": 4.888770040256508e-06,
      "loss": 0.599,
      "step": 1168
    },
    {
      "epoch": 1.430844553243574,
      "grad_norm": 0.9789527907994929,
      "learning_rate": 4.8885809245785454e-06,
      "loss": 1.1339,
      "step": 1169
    },
    {
      "epoch": 1.4320685434516525,
      "grad_norm": 0.6678071626329536,
      "learning_rate": 4.888391651931879e-06,
      "loss": 0.5274,
      "step": 1170
    },
    {
      "epoch": 1.4332925336597309,
      "grad_norm": 1.4293396018544062,
      "learning_rate": 4.888202222328948e-06,
      "loss": 0.523,
      "step": 1171
    },
    {
      "epoch": 1.434516523867809,
      "grad_norm": 0.8669687575262846,
      "learning_rate": 4.888012635782199e-06,
      "loss": 0.9041,
      "step": 1172
    },
    {
      "epoch": 1.4357405140758874,
      "grad_norm": 0.8151743014181194,
      "learning_rate": 4.887822892304094e-06,
      "loss": 0.521,
      "step": 1173
    },
    {
      "epoch": 1.4369645042839658,
      "grad_norm": 1.1762451119100759,
      "learning_rate": 4.887632991907099e-06,
      "loss": 0.6112,
      "step": 1174
    },
    {
      "epoch": 1.438188494492044,
      "grad_norm": 0.828996560860172,
      "learning_rate": 4.887442934603696e-06,
      "loss": 0.6509,
      "step": 1175
    },
    {
      "epoch": 1.4394124847001224,
      "grad_norm": 0.8886545296001926,
      "learning_rate": 4.887252720406373e-06,
      "loss": 0.5955,
      "step": 1176
    },
    {
      "epoch": 1.4406364749082008,
      "grad_norm": 0.5845138679853458,
      "learning_rate": 4.887062349327631e-06,
      "loss": 0.7097,
      "step": 1177
    },
    {
      "epoch": 1.441860465116279,
      "grad_norm": 0.6676544280319279,
      "learning_rate": 4.886871821379981e-06,
      "loss": 0.6868,
      "step": 1178
    },
    {
      "epoch": 1.4430844553243574,
      "grad_norm": 1.2585786290411753,
      "learning_rate": 4.886681136575943e-06,
      "loss": 0.7097,
      "step": 1179
    },
    {
      "epoch": 1.4443084455324358,
      "grad_norm": 1.0857247107635473,
      "learning_rate": 4.886490294928049e-06,
      "loss": 0.7021,
      "step": 1180
    },
    {
      "epoch": 1.4455324357405142,
      "grad_norm": 1.1375788518200016,
      "learning_rate": 4.8862992964488395e-06,
      "loss": 0.5486,
      "step": 1181
    },
    {
      "epoch": 1.4467564259485923,
      "grad_norm": 0.5713654211162366,
      "learning_rate": 4.886108141150866e-06,
      "loss": 0.5461,
      "step": 1182
    },
    {
      "epoch": 1.4479804161566707,
      "grad_norm": 1.2211191859404338,
      "learning_rate": 4.885916829046692e-06,
      "loss": 0.6173,
      "step": 1183
    },
    {
      "epoch": 1.4492044063647491,
      "grad_norm": 0.7682521127007785,
      "learning_rate": 4.885725360148888e-06,
      "loss": 0.5721,
      "step": 1184
    },
    {
      "epoch": 1.4504283965728275,
      "grad_norm": 0.8476181582721496,
      "learning_rate": 4.885533734470039e-06,
      "loss": 1.0153,
      "step": 1185
    },
    {
      "epoch": 1.4516523867809057,
      "grad_norm": 1.2083400167422327,
      "learning_rate": 4.885341952022735e-06,
      "loss": 0.8671,
      "step": 1186
    },
    {
      "epoch": 1.452876376988984,
      "grad_norm": 0.765224846936888,
      "learning_rate": 4.885150012819582e-06,
      "loss": 0.5466,
      "step": 1187
    },
    {
      "epoch": 1.4541003671970625,
      "grad_norm": 0.7271323706539619,
      "learning_rate": 4.884957916873191e-06,
      "loss": 0.6565,
      "step": 1188
    },
    {
      "epoch": 1.4553243574051407,
      "grad_norm": 1.2275074254847234,
      "learning_rate": 4.884765664196187e-06,
      "loss": 0.4773,
      "step": 1189
    },
    {
      "epoch": 1.456548347613219,
      "grad_norm": 1.1887938606969513,
      "learning_rate": 4.884573254801205e-06,
      "loss": 0.6802,
      "step": 1190
    },
    {
      "epoch": 1.4577723378212974,
      "grad_norm": 0.7637351692957838,
      "learning_rate": 4.884380688700888e-06,
      "loss": 0.7452,
      "step": 1191
    },
    {
      "epoch": 1.4589963280293758,
      "grad_norm": 0.8663856818816139,
      "learning_rate": 4.8841879659078915e-06,
      "loss": 0.7114,
      "step": 1192
    },
    {
      "epoch": 1.460220318237454,
      "grad_norm": 0.8545206546836961,
      "learning_rate": 4.883995086434881e-06,
      "loss": 0.6757,
      "step": 1193
    },
    {
      "epoch": 1.4614443084455324,
      "grad_norm": 1.099454532209251,
      "learning_rate": 4.8838020502945305e-06,
      "loss": 1.1227,
      "step": 1194
    },
    {
      "epoch": 1.4626682986536108,
      "grad_norm": 0.9471838652399067,
      "learning_rate": 4.883608857499526e-06,
      "loss": 0.7475,
      "step": 1195
    },
    {
      "epoch": 1.4638922888616892,
      "grad_norm": 0.5817100965156184,
      "learning_rate": 4.883415508062565e-06,
      "loss": 0.5643,
      "step": 1196
    },
    {
      "epoch": 1.4651162790697674,
      "grad_norm": 0.5242312017787263,
      "learning_rate": 4.883222001996352e-06,
      "loss": 0.6021,
      "step": 1197
    },
    {
      "epoch": 1.4663402692778458,
      "grad_norm": 0.6112608879238469,
      "learning_rate": 4.883028339313603e-06,
      "loss": 0.5515,
      "step": 1198
    },
    {
      "epoch": 1.4675642594859242,
      "grad_norm": 0.746363745826184,
      "learning_rate": 4.8828345200270465e-06,
      "loss": 0.6575,
      "step": 1199
    },
    {
      "epoch": 1.4687882496940023,
      "grad_norm": 0.5441400220762983,
      "learning_rate": 4.8826405441494185e-06,
      "loss": 0.6424,
      "step": 1200
    },
    {
      "epoch": 1.4700122399020807,
      "grad_norm": 0.7418069408837273,
      "learning_rate": 4.8824464116934675e-06,
      "loss": 0.5822,
      "step": 1201
    },
    {
      "epoch": 1.4712362301101591,
      "grad_norm": 1.0771289447214616,
      "learning_rate": 4.882252122671949e-06,
      "loss": 0.6028,
      "step": 1202
    },
    {
      "epoch": 1.4724602203182375,
      "grad_norm": 0.9410223309511013,
      "learning_rate": 4.882057677097633e-06,
      "loss": 0.5943,
      "step": 1203
    },
    {
      "epoch": 1.4736842105263157,
      "grad_norm": 0.5799904346053545,
      "learning_rate": 4.881863074983298e-06,
      "loss": 0.6269,
      "step": 1204
    },
    {
      "epoch": 1.474908200734394,
      "grad_norm": 0.9798184438144371,
      "learning_rate": 4.881668316341731e-06,
      "loss": 0.9567,
      "step": 1205
    },
    {
      "epoch": 1.4761321909424725,
      "grad_norm": 1.0055329204246881,
      "learning_rate": 4.881473401185732e-06,
      "loss": 0.6094,
      "step": 1206
    },
    {
      "epoch": 1.477356181150551,
      "grad_norm": 0.9763515548928199,
      "learning_rate": 4.881278329528108e-06,
      "loss": 0.6275,
      "step": 1207
    },
    {
      "epoch": 1.478580171358629,
      "grad_norm": 1.174518265429184,
      "learning_rate": 4.881083101381681e-06,
      "loss": 0.4906,
      "step": 1208
    },
    {
      "epoch": 1.4798041615667075,
      "grad_norm": 1.1596627219040332,
      "learning_rate": 4.8808877167592795e-06,
      "loss": 0.5106,
      "step": 1209
    },
    {
      "epoch": 1.4810281517747859,
      "grad_norm": 0.851382671798185,
      "learning_rate": 4.880692175673743e-06,
      "loss": 1.221,
      "step": 1210
    },
    {
      "epoch": 1.482252141982864,
      "grad_norm": 0.3815843895462902,
      "learning_rate": 4.880496478137923e-06,
      "loss": 0.2825,
      "step": 1211
    },
    {
      "epoch": 1.4834761321909424,
      "grad_norm": 0.8071900253370294,
      "learning_rate": 4.880300624164679e-06,
      "loss": 0.7536,
      "step": 1212
    },
    {
      "epoch": 1.4847001223990208,
      "grad_norm": 0.9276840768926988,
      "learning_rate": 4.880104613766881e-06,
      "loss": 0.4521,
      "step": 1213
    },
    {
      "epoch": 1.4859241126070992,
      "grad_norm": 1.4711334313118787,
      "learning_rate": 4.879908446957413e-06,
      "loss": 1.046,
      "step": 1214
    },
    {
      "epoch": 1.4871481028151774,
      "grad_norm": 0.8187000320715278,
      "learning_rate": 4.879712123749164e-06,
      "loss": 0.5378,
      "step": 1215
    },
    {
      "epoch": 1.4883720930232558,
      "grad_norm": 1.6789672103265632,
      "learning_rate": 4.879515644155036e-06,
      "loss": 0.5873,
      "step": 1216
    },
    {
      "epoch": 1.4895960832313342,
      "grad_norm": 0.7475266800258944,
      "learning_rate": 4.879319008187942e-06,
      "loss": 1.388,
      "step": 1217
    },
    {
      "epoch": 1.4908200734394126,
      "grad_norm": 0.8530002499718742,
      "learning_rate": 4.879122215860802e-06,
      "loss": 0.5782,
      "step": 1218
    },
    {
      "epoch": 1.4920440636474908,
      "grad_norm": 0.9760552501347369,
      "learning_rate": 4.878925267186549e-06,
      "loss": 0.3886,
      "step": 1219
    },
    {
      "epoch": 1.4932680538555692,
      "grad_norm": 0.7262251767219555,
      "learning_rate": 4.878728162178128e-06,
      "loss": 0.5767,
      "step": 1220
    },
    {
      "epoch": 1.4944920440636476,
      "grad_norm": 0.8496532393192555,
      "learning_rate": 4.8785309008484905e-06,
      "loss": 1.4828,
      "step": 1221
    },
    {
      "epoch": 1.4957160342717257,
      "grad_norm": 0.818644098190518,
      "learning_rate": 4.878333483210599e-06,
      "loss": 0.5851,
      "step": 1222
    },
    {
      "epoch": 1.4969400244798041,
      "grad_norm": 0.7737479488375891,
      "learning_rate": 4.878135909277428e-06,
      "loss": 0.6536,
      "step": 1223
    },
    {
      "epoch": 1.4981640146878825,
      "grad_norm": 1.1221044816124206,
      "learning_rate": 4.877938179061962e-06,
      "loss": 0.7076,
      "step": 1224
    },
    {
      "epoch": 1.499388004895961,
      "grad_norm": 1.3181398891150558,
      "learning_rate": 4.877740292577194e-06,
      "loss": 0.4751,
      "step": 1225
    },
    {
      "epoch": 1.5006119951040393,
      "grad_norm": 1.386905200834611,
      "learning_rate": 4.877542249836127e-06,
      "loss": 0.9189,
      "step": 1226
    },
    {
      "epoch": 1.5018359853121175,
      "grad_norm": 1.3514418295926585,
      "learning_rate": 4.877344050851779e-06,
      "loss": 1.283,
      "step": 1227
    },
    {
      "epoch": 1.5030599755201959,
      "grad_norm": 0.6288279384701069,
      "learning_rate": 4.877145695637173e-06,
      "loss": 0.5059,
      "step": 1228
    },
    {
      "epoch": 1.5042839657282743,
      "grad_norm": 0.7333657087818135,
      "learning_rate": 4.8769471842053436e-06,
      "loss": 0.946,
      "step": 1229
    },
    {
      "epoch": 1.5055079559363524,
      "grad_norm": 0.5685984509039165,
      "learning_rate": 4.876748516569337e-06,
      "loss": 0.6067,
      "step": 1230
    },
    {
      "epoch": 1.5067319461444308,
      "grad_norm": 0.7669290361892817,
      "learning_rate": 4.876549692742209e-06,
      "loss": 0.5138,
      "step": 1231
    },
    {
      "epoch": 1.5079559363525092,
      "grad_norm": 1.0123100175840016,
      "learning_rate": 4.876350712737026e-06,
      "loss": 0.5887,
      "step": 1232
    },
    {
      "epoch": 1.5091799265605874,
      "grad_norm": 0.7024717228794907,
      "learning_rate": 4.876151576566863e-06,
      "loss": 0.6919,
      "step": 1233
    },
    {
      "epoch": 1.5104039167686658,
      "grad_norm": 0.9258911198930364,
      "learning_rate": 4.875952284244808e-06,
      "loss": 1.2447,
      "step": 1234
    },
    {
      "epoch": 1.5116279069767442,
      "grad_norm": 1.0980471908498237,
      "learning_rate": 4.875752835783957e-06,
      "loss": 1.4557,
      "step": 1235
    },
    {
      "epoch": 1.5128518971848224,
      "grad_norm": 0.6739579246101375,
      "learning_rate": 4.875553231197416e-06,
      "loss": 0.5888,
      "step": 1236
    },
    {
      "epoch": 1.514075887392901,
      "grad_norm": 1.5045137449631847,
      "learning_rate": 4.875353470498303e-06,
      "loss": 0.6895,
      "step": 1237
    },
    {
      "epoch": 1.5152998776009792,
      "grad_norm": 1.0304562114394713,
      "learning_rate": 4.8751535536997465e-06,
      "loss": 0.6715,
      "step": 1238
    },
    {
      "epoch": 1.5165238678090576,
      "grad_norm": 1.1798508958139775,
      "learning_rate": 4.8749534808148835e-06,
      "loss": 0.5468,
      "step": 1239
    },
    {
      "epoch": 1.517747858017136,
      "grad_norm": 1.30341616347032,
      "learning_rate": 4.874753251856862e-06,
      "loss": 0.5787,
      "step": 1240
    },
    {
      "epoch": 1.5189718482252141,
      "grad_norm": 1.0958475763129394,
      "learning_rate": 4.87455286683884e-06,
      "loss": 0.4981,
      "step": 1241
    },
    {
      "epoch": 1.5201958384332925,
      "grad_norm": 0.6226745407956138,
      "learning_rate": 4.8743523257739866e-06,
      "loss": 0.5477,
      "step": 1242
    },
    {
      "epoch": 1.521419828641371,
      "grad_norm": 1.1042885954704351,
      "learning_rate": 4.874151628675481e-06,
      "loss": 0.5586,
      "step": 1243
    },
    {
      "epoch": 1.522643818849449,
      "grad_norm": 0.6993281146757927,
      "learning_rate": 4.873950775556511e-06,
      "loss": 0.5808,
      "step": 1244
    },
    {
      "epoch": 1.5238678090575275,
      "grad_norm": 1.811794119264975,
      "learning_rate": 4.873749766430277e-06,
      "loss": 0.5287,
      "step": 1245
    },
    {
      "epoch": 1.525091799265606,
      "grad_norm": 0.519143181601484,
      "learning_rate": 4.873548601309988e-06,
      "loss": 0.5732,
      "step": 1246
    },
    {
      "epoch": 1.526315789473684,
      "grad_norm": 0.5053337785484447,
      "learning_rate": 4.873347280208866e-06,
      "loss": 0.4506,
      "step": 1247
    },
    {
      "epoch": 1.5275397796817627,
      "grad_norm": 1.1466487211396454,
      "learning_rate": 4.873145803140137e-06,
      "loss": 0.9317,
      "step": 1248
    },
    {
      "epoch": 1.5287637698898409,
      "grad_norm": 1.3975054358871863,
      "learning_rate": 4.872944170117044e-06,
      "loss": 0.6666,
      "step": 1249
    },
    {
      "epoch": 1.5299877600979193,
      "grad_norm": 0.9055931967858868,
      "learning_rate": 4.872742381152837e-06,
      "loss": 1.4555,
      "step": 1250
    },
    {
      "epoch": 1.5312117503059977,
      "grad_norm": 1.2105732945510717,
      "learning_rate": 4.872540436260776e-06,
      "loss": 0.7988,
      "step": 1251
    },
    {
      "epoch": 1.5324357405140758,
      "grad_norm": 0.6746884816320633,
      "learning_rate": 4.872338335454134e-06,
      "loss": 0.6277,
      "step": 1252
    },
    {
      "epoch": 1.5336597307221542,
      "grad_norm": 1.1093992332542735,
      "learning_rate": 4.872136078746191e-06,
      "loss": 1.2359,
      "step": 1253
    },
    {
      "epoch": 1.5348837209302326,
      "grad_norm": 0.8547103999645286,
      "learning_rate": 4.871933666150239e-06,
      "loss": 1.305,
      "step": 1254
    },
    {
      "epoch": 1.5361077111383108,
      "grad_norm": 0.7343524561546985,
      "learning_rate": 4.87173109767958e-06,
      "loss": 1.1699,
      "step": 1255
    },
    {
      "epoch": 1.5373317013463892,
      "grad_norm": 0.8962267139467814,
      "learning_rate": 4.871528373347525e-06,
      "loss": 1.2598,
      "step": 1256
    },
    {
      "epoch": 1.5385556915544676,
      "grad_norm": 1.3428711439563348,
      "learning_rate": 4.871325493167398e-06,
      "loss": 0.6084,
      "step": 1257
    },
    {
      "epoch": 1.5397796817625458,
      "grad_norm": 0.6398627843784108,
      "learning_rate": 4.87112245715253e-06,
      "loss": 0.6951,
      "step": 1258
    },
    {
      "epoch": 1.5410036719706244,
      "grad_norm": 0.7893622728968009,
      "learning_rate": 4.870919265316264e-06,
      "loss": 0.8219,
      "step": 1259
    },
    {
      "epoch": 1.5422276621787026,
      "grad_norm": 0.7824666346718955,
      "learning_rate": 4.870715917671953e-06,
      "loss": 1.1929,
      "step": 1260
    },
    {
      "epoch": 1.543451652386781,
      "grad_norm": 0.9235517577337293,
      "learning_rate": 4.870512414232961e-06,
      "loss": 0.6401,
      "step": 1261
    },
    {
      "epoch": 1.5446756425948593,
      "grad_norm": 0.8740233896553848,
      "learning_rate": 4.870308755012661e-06,
      "loss": 1.0384,
      "step": 1262
    },
    {
      "epoch": 1.5458996328029375,
      "grad_norm": 1.1131375093209825,
      "learning_rate": 4.870104940024437e-06,
      "loss": 1.2314,
      "step": 1263
    },
    {
      "epoch": 1.547123623011016,
      "grad_norm": 0.9155993202320094,
      "learning_rate": 4.869900969281683e-06,
      "loss": 0.5068,
      "step": 1264
    },
    {
      "epoch": 1.5483476132190943,
      "grad_norm": 0.7877223716528418,
      "learning_rate": 4.869696842797802e-06,
      "loss": 0.8321,
      "step": 1265
    },
    {
      "epoch": 1.5495716034271725,
      "grad_norm": 1.0729437717819634,
      "learning_rate": 4.86949256058621e-06,
      "loss": 0.9288,
      "step": 1266
    },
    {
      "epoch": 1.5507955936352509,
      "grad_norm": 0.8140952636405555,
      "learning_rate": 4.869288122660331e-06,
      "loss": 0.5267,
      "step": 1267
    },
    {
      "epoch": 1.5520195838433293,
      "grad_norm": 0.6779073107237319,
      "learning_rate": 4.869083529033599e-06,
      "loss": 0.8976,
      "step": 1268
    },
    {
      "epoch": 1.5532435740514074,
      "grad_norm": 1.0657664771955375,
      "learning_rate": 4.868878779719462e-06,
      "loss": 1.1118,
      "step": 1269
    },
    {
      "epoch": 1.554467564259486,
      "grad_norm": 0.9703439580190897,
      "learning_rate": 4.868673874731372e-06,
      "loss": 1.26,
      "step": 1270
    },
    {
      "epoch": 1.5556915544675642,
      "grad_norm": 0.7240747517427476,
      "learning_rate": 4.868468814082796e-06,
      "loss": 0.399,
      "step": 1271
    },
    {
      "epoch": 1.5569155446756426,
      "grad_norm": 0.8279338496867802,
      "learning_rate": 4.868263597787211e-06,
      "loss": 1.0394,
      "step": 1272
    },
    {
      "epoch": 1.558139534883721,
      "grad_norm": 0.697447343309313,
      "learning_rate": 4.8680582258581016e-06,
      "loss": 1.0164,
      "step": 1273
    },
    {
      "epoch": 1.5593635250917992,
      "grad_norm": 0.6791974737894272,
      "learning_rate": 4.867852698308964e-06,
      "loss": 0.9366,
      "step": 1274
    },
    {
      "epoch": 1.5605875152998776,
      "grad_norm": 1.1429456597771954,
      "learning_rate": 4.867647015153306e-06,
      "loss": 0.6507,
      "step": 1275
    },
    {
      "epoch": 1.561811505507956,
      "grad_norm": 0.8627963461149404,
      "learning_rate": 4.8674411764046425e-06,
      "loss": 0.701,
      "step": 1276
    },
    {
      "epoch": 1.5630354957160342,
      "grad_norm": 0.9307743538200832,
      "learning_rate": 4.867235182076502e-06,
      "loss": 0.8259,
      "step": 1277
    },
    {
      "epoch": 1.5642594859241126,
      "grad_norm": 0.49952956018201805,
      "learning_rate": 4.867029032182421e-06,
      "loss": 0.5092,
      "step": 1278
    },
    {
      "epoch": 1.565483476132191,
      "grad_norm": 0.9086549916176242,
      "learning_rate": 4.866822726735947e-06,
      "loss": 0.5722,
      "step": 1279
    },
    {
      "epoch": 1.5667074663402691,
      "grad_norm": 0.8128093759307843,
      "learning_rate": 4.866616265750637e-06,
      "loss": 1.3979,
      "step": 1280
    },
    {
      "epoch": 1.5679314565483478,
      "grad_norm": 1.2887211920427077,
      "learning_rate": 4.8664096492400604e-06,
      "loss": 0.6293,
      "step": 1281
    },
    {
      "epoch": 1.569155446756426,
      "grad_norm": 0.7472808379577176,
      "learning_rate": 4.8662028772177945e-06,
      "loss": 0.7293,
      "step": 1282
    },
    {
      "epoch": 1.5703794369645043,
      "grad_norm": 1.2209972349362828,
      "learning_rate": 4.8659959496974276e-06,
      "loss": 0.6568,
      "step": 1283
    },
    {
      "epoch": 1.5716034271725827,
      "grad_norm": 0.8050873513630448,
      "learning_rate": 4.865788866692557e-06,
      "loss": 0.5319,
      "step": 1284
    },
    {
      "epoch": 1.572827417380661,
      "grad_norm": 1.080940081244901,
      "learning_rate": 4.865581628216793e-06,
      "loss": 0.5959,
      "step": 1285
    },
    {
      "epoch": 1.5740514075887393,
      "grad_norm": 1.2188637016420278,
      "learning_rate": 4.865374234283755e-06,
      "loss": 1.058,
      "step": 1286
    },
    {
      "epoch": 1.5752753977968177,
      "grad_norm": 0.5802900643246232,
      "learning_rate": 4.86516668490707e-06,
      "loss": 0.4426,
      "step": 1287
    },
    {
      "epoch": 1.5764993880048959,
      "grad_norm": 0.67331000299375,
      "learning_rate": 4.864958980100379e-06,
      "loss": 0.4751,
      "step": 1288
    },
    {
      "epoch": 1.5777233782129743,
      "grad_norm": 0.620216303123961,
      "learning_rate": 4.864751119877332e-06,
      "loss": 0.6445,
      "step": 1289
    },
    {
      "epoch": 1.5789473684210527,
      "grad_norm": 0.5603524188150197,
      "learning_rate": 4.864543104251587e-06,
      "loss": 0.5166,
      "step": 1290
    },
    {
      "epoch": 1.5801713586291308,
      "grad_norm": 0.9628117918710384,
      "learning_rate": 4.864334933236816e-06,
      "loss": 0.669,
      "step": 1291
    },
    {
      "epoch": 1.5813953488372094,
      "grad_norm": 0.7747258513436537,
      "learning_rate": 4.864126606846697e-06,
      "loss": 0.6797,
      "step": 1292
    },
    {
      "epoch": 1.5826193390452876,
      "grad_norm": 1.0510986552259636,
      "learning_rate": 4.8639181250949225e-06,
      "loss": 1.0228,
      "step": 1293
    },
    {
      "epoch": 1.583843329253366,
      "grad_norm": 1.1323773196729516,
      "learning_rate": 4.863709487995192e-06,
      "loss": 1.399,
      "step": 1294
    },
    {
      "epoch": 1.5850673194614444,
      "grad_norm": 1.054771893563801,
      "learning_rate": 4.8635006955612175e-06,
      "loss": 0.6225,
      "step": 1295
    },
    {
      "epoch": 1.5862913096695226,
      "grad_norm": 1.289721539358316,
      "learning_rate": 4.863291747806718e-06,
      "loss": 0.5786,
      "step": 1296
    },
    {
      "epoch": 1.587515299877601,
      "grad_norm": 1.066661357290134,
      "learning_rate": 4.863082644745427e-06,
      "loss": 0.5561,
      "step": 1297
    },
    {
      "epoch": 1.5887392900856794,
      "grad_norm": 0.8861876581718572,
      "learning_rate": 4.862873386391085e-06,
      "loss": 0.5817,
      "step": 1298
    },
    {
      "epoch": 1.5899632802937576,
      "grad_norm": 1.2134201505155071,
      "learning_rate": 4.862663972757443e-06,
      "loss": 0.5431,
      "step": 1299
    },
    {
      "epoch": 1.591187270501836,
      "grad_norm": 0.6824973037816972,
      "learning_rate": 4.862454403858264e-06,
      "loss": 0.5988,
      "step": 1300
    },
    {
      "epoch": 1.5924112607099143,
      "grad_norm": 0.9118784821892911,
      "learning_rate": 4.86224467970732e-06,
      "loss": 0.9899,
      "step": 1301
    },
    {
      "epoch": 1.5936352509179925,
      "grad_norm": 0.8463844548075125,
      "learning_rate": 4.862034800318392e-06,
      "loss": 1.147,
      "step": 1302
    },
    {
      "epoch": 1.5948592411260711,
      "grad_norm": 0.8838515084760562,
      "learning_rate": 4.861824765705275e-06,
      "loss": 0.5707,
      "step": 1303
    },
    {
      "epoch": 1.5960832313341493,
      "grad_norm": 0.5586838901577673,
      "learning_rate": 4.861614575881769e-06,
      "loss": 0.509,
      "step": 1304
    },
    {
      "epoch": 1.5973072215422277,
      "grad_norm": 0.7278865696787127,
      "learning_rate": 4.861404230861688e-06,
      "loss": 0.5505,
      "step": 1305
    },
    {
      "epoch": 1.598531211750306,
      "grad_norm": 1.209919730231102,
      "learning_rate": 4.861193730658856e-06,
      "loss": 0.4917,
      "step": 1306
    },
    {
      "epoch": 1.5997552019583843,
      "grad_norm": 1.0902870395851068,
      "learning_rate": 4.8609830752871055e-06,
      "loss": 0.527,
      "step": 1307
    },
    {
      "epoch": 1.6009791921664627,
      "grad_norm": 0.5400794698971968,
      "learning_rate": 4.860772264760279e-06,
      "loss": 0.5704,
      "step": 1308
    },
    {
      "epoch": 1.602203182374541,
      "grad_norm": 0.9534456903405749,
      "learning_rate": 4.8605612990922326e-06,
      "loss": 1.5033,
      "step": 1309
    },
    {
      "epoch": 1.6034271725826192,
      "grad_norm": 1.6361443891689702,
      "learning_rate": 4.860350178296828e-06,
      "loss": 0.482,
      "step": 1310
    },
    {
      "epoch": 1.6046511627906976,
      "grad_norm": 1.2937913155461236,
      "learning_rate": 4.8601389023879395e-06,
      "loss": 0.4973,
      "step": 1311
    },
    {
      "epoch": 1.605875152998776,
      "grad_norm": 1.225940152786754,
      "learning_rate": 4.859927471379452e-06,
      "loss": 0.5878,
      "step": 1312
    },
    {
      "epoch": 1.6070991432068542,
      "grad_norm": 0.8729862588247709,
      "learning_rate": 4.85971588528526e-06,
      "loss": 0.6194,
      "step": 1313
    },
    {
      "epoch": 1.6083231334149328,
      "grad_norm": 0.9912373185567547,
      "learning_rate": 4.859504144119268e-06,
      "loss": 0.4919,
      "step": 1314
    },
    {
      "epoch": 1.609547123623011,
      "grad_norm": 0.49511696212569967,
      "learning_rate": 4.8592922478953915e-06,
      "loss": 0.4917,
      "step": 1315
    },
    {
      "epoch": 1.6107711138310894,
      "grad_norm": 0.5970535155490498,
      "learning_rate": 4.859080196627554e-06,
      "loss": 0.5607,
      "step": 1316
    },
    {
      "epoch": 1.6119951040391678,
      "grad_norm": 1.0702180327365347,
      "learning_rate": 4.858867990329692e-06,
      "loss": 0.7249,
      "step": 1317
    },
    {
      "epoch": 1.613219094247246,
      "grad_norm": 1.2623727071558004,
      "learning_rate": 4.85865562901575e-06,
      "loss": 0.6018,
      "step": 1318
    },
    {
      "epoch": 1.6144430844553244,
      "grad_norm": 0.8862716295358708,
      "learning_rate": 4.858443112699685e-06,
      "loss": 0.7971,
      "step": 1319
    },
    {
      "epoch": 1.6156670746634028,
      "grad_norm": 0.6785540271818395,
      "learning_rate": 4.858230441395462e-06,
      "loss": 0.4315,
      "step": 1320
    },
    {
      "epoch": 1.616891064871481,
      "grad_norm": 1.0348099374879327,
      "learning_rate": 4.8580176151170565e-06,
      "loss": 1.5543,
      "step": 1321
    },
    {
      "epoch": 1.6181150550795593,
      "grad_norm": 0.859196625423065,
      "learning_rate": 4.857804633878455e-06,
      "loss": 0.9792,
      "step": 1322
    },
    {
      "epoch": 1.6193390452876377,
      "grad_norm": 0.7252994247653778,
      "learning_rate": 4.857591497693654e-06,
      "loss": 0.5629,
      "step": 1323
    },
    {
      "epoch": 1.620563035495716,
      "grad_norm": 1.2377112659141427,
      "learning_rate": 4.857378206576659e-06,
      "loss": 0.6221,
      "step": 1324
    },
    {
      "epoch": 1.6217870257037945,
      "grad_norm": 1.2621367558751118,
      "learning_rate": 4.857164760541489e-06,
      "loss": 0.7073,
      "step": 1325
    },
    {
      "epoch": 1.6230110159118727,
      "grad_norm": 1.081962728710649,
      "learning_rate": 4.856951159602168e-06,
      "loss": 1.0375,
      "step": 1326
    },
    {
      "epoch": 1.624235006119951,
      "grad_norm": 1.3287630685401857,
      "learning_rate": 4.856737403772735e-06,
      "loss": 0.6238,
      "step": 1327
    },
    {
      "epoch": 1.6254589963280295,
      "grad_norm": 0.5749454756818798,
      "learning_rate": 4.856523493067238e-06,
      "loss": 0.5984,
      "step": 1328
    },
    {
      "epoch": 1.6266829865361077,
      "grad_norm": 0.914600103989306,
      "learning_rate": 4.8563094274997325e-06,
      "loss": 0.5569,
      "step": 1329
    },
    {
      "epoch": 1.627906976744186,
      "grad_norm": 0.9279012652409752,
      "learning_rate": 4.856095207084287e-06,
      "loss": 0.8186,
      "step": 1330
    },
    {
      "epoch": 1.6291309669522644,
      "grad_norm": 0.7431589824602952,
      "learning_rate": 4.855880831834978e-06,
      "loss": 0.5676,
      "step": 1331
    },
    {
      "epoch": 1.6303549571603426,
      "grad_norm": 1.209577241666578,
      "learning_rate": 4.855666301765895e-06,
      "loss": 0.5479,
      "step": 1332
    },
    {
      "epoch": 1.631578947368421,
      "grad_norm": 1.5426648423845288,
      "learning_rate": 4.8554516168911364e-06,
      "loss": 1.0857,
      "step": 1333
    },
    {
      "epoch": 1.6328029375764994,
      "grad_norm": 1.1627382551047727,
      "learning_rate": 4.85523677722481e-06,
      "loss": 0.6013,
      "step": 1334
    },
    {
      "epoch": 1.6340269277845776,
      "grad_norm": 0.9986436889437172,
      "learning_rate": 4.855021782781033e-06,
      "loss": 0.6065,
      "step": 1335
    },
    {
      "epoch": 1.6352509179926562,
      "grad_norm": 0.7767789280488503,
      "learning_rate": 4.8548066335739355e-06,
      "loss": 0.9064,
      "step": 1336
    },
    {
      "epoch": 1.6364749082007344,
      "grad_norm": 0.8492255677426758,
      "learning_rate": 4.854591329617655e-06,
      "loss": 1.3151,
      "step": 1337
    },
    {
      "epoch": 1.6376988984088128,
      "grad_norm": 0.989116711650738,
      "learning_rate": 4.854375870926342e-06,
      "loss": 1.6304,
      "step": 1338
    },
    {
      "epoch": 1.6389228886168912,
      "grad_norm": 0.9441605227066312,
      "learning_rate": 4.854160257514155e-06,
      "loss": 0.4936,
      "step": 1339
    },
    {
      "epoch": 1.6401468788249693,
      "grad_norm": 0.7577852722313807,
      "learning_rate": 4.853944489395264e-06,
      "loss": 1.1823,
      "step": 1340
    },
    {
      "epoch": 1.6413708690330477,
      "grad_norm": 1.0699013530541237,
      "learning_rate": 4.853728566583847e-06,
      "loss": 1.3895,
      "step": 1341
    },
    {
      "epoch": 1.6425948592411261,
      "grad_norm": 1.0177459759367598,
      "learning_rate": 4.8535124890940944e-06,
      "loss": 0.6413,
      "step": 1342
    },
    {
      "epoch": 1.6438188494492043,
      "grad_norm": 0.7484579873313579,
      "learning_rate": 4.853296256940206e-06,
      "loss": 0.7953,
      "step": 1343
    },
    {
      "epoch": 1.6450428396572827,
      "grad_norm": 0.7065984335070344,
      "learning_rate": 4.853079870136392e-06,
      "loss": 0.7967,
      "step": 1344
    },
    {
      "epoch": 1.646266829865361,
      "grad_norm": 0.6754733251334394,
      "learning_rate": 4.852863328696873e-06,
      "loss": 0.4691,
      "step": 1345
    },
    {
      "epoch": 1.6474908200734393,
      "grad_norm": 0.7535483018816074,
      "learning_rate": 4.852646632635878e-06,
      "loss": 0.6419,
      "step": 1346
    },
    {
      "epoch": 1.648714810281518,
      "grad_norm": 0.9335048575988064,
      "learning_rate": 4.8524297819676494e-06,
      "loss": 0.463,
      "step": 1347
    },
    {
      "epoch": 1.649938800489596,
      "grad_norm": 0.4788823849364188,
      "learning_rate": 4.8522127767064344e-06,
      "loss": 0.4536,
      "step": 1348
    },
    {
      "epoch": 1.6511627906976745,
      "grad_norm": 0.6990374625548642,
      "learning_rate": 4.851995616866497e-06,
      "loss": 0.508,
      "step": 1349
    },
    {
      "epoch": 1.6523867809057529,
      "grad_norm": 0.9700323317300212,
      "learning_rate": 4.851778302462108e-06,
      "loss": 0.5853,
      "step": 1350
    },
    {
      "epoch": 1.653610771113831,
      "grad_norm": 0.7357349048416832,
      "learning_rate": 4.851560833507546e-06,
      "loss": 0.6819,
      "step": 1351
    },
    {
      "epoch": 1.6548347613219094,
      "grad_norm": 0.6174928789709433,
      "learning_rate": 4.8513432100171044e-06,
      "loss": 0.7344,
      "step": 1352
    },
    {
      "epoch": 1.6560587515299878,
      "grad_norm": 0.41941156105343674,
      "learning_rate": 4.851125432005085e-06,
      "loss": 0.3967,
      "step": 1353
    },
    {
      "epoch": 1.657282741738066,
      "grad_norm": 1.10855360030447,
      "learning_rate": 4.850907499485798e-06,
      "loss": 1.1499,
      "step": 1354
    },
    {
      "epoch": 1.6585067319461444,
      "grad_norm": 0.8030494616017138,
      "learning_rate": 4.850689412473565e-06,
      "loss": 0.6579,
      "step": 1355
    },
    {
      "epoch": 1.6597307221542228,
      "grad_norm": 0.6652803920591223,
      "learning_rate": 4.850471170982718e-06,
      "loss": 0.6075,
      "step": 1356
    },
    {
      "epoch": 1.660954712362301,
      "grad_norm": 0.8485450368081902,
      "learning_rate": 4.8502527750276e-06,
      "loss": 0.5829,
      "step": 1357
    },
    {
      "epoch": 1.6621787025703796,
      "grad_norm": 0.8486727341664393,
      "learning_rate": 4.850034224622563e-06,
      "loss": 0.5789,
      "step": 1358
    },
    {
      "epoch": 1.6634026927784578,
      "grad_norm": 1.2916547002119936,
      "learning_rate": 4.849815519781969e-06,
      "loss": 0.7959,
      "step": 1359
    },
    {
      "epoch": 1.6646266829865362,
      "grad_norm": 1.3670052703137243,
      "learning_rate": 4.8495966605201895e-06,
      "loss": 0.4788,
      "step": 1360
    },
    {
      "epoch": 1.6658506731946146,
      "grad_norm": 0.8459446527470325,
      "learning_rate": 4.849377646851609e-06,
      "loss": 0.9401,
      "step": 1361
    },
    {
      "epoch": 1.6670746634026927,
      "grad_norm": 0.8850094395833262,
      "learning_rate": 4.849158478790618e-06,
      "loss": 0.9929,
      "step": 1362
    },
    {
      "epoch": 1.6682986536107711,
      "grad_norm": 0.82629077446571,
      "learning_rate": 4.848939156351622e-06,
      "loss": 0.5203,
      "step": 1363
    },
    {
      "epoch": 1.6695226438188495,
      "grad_norm": 0.984167554488913,
      "learning_rate": 4.848719679549032e-06,
      "loss": 0.7205,
      "step": 1364
    },
    {
      "epoch": 1.6707466340269277,
      "grad_norm": 1.2906276270549286,
      "learning_rate": 4.8485000483972724e-06,
      "loss": 0.4931,
      "step": 1365
    },
    {
      "epoch": 1.671970624235006,
      "grad_norm": 0.8568283075229575,
      "learning_rate": 4.848280262910776e-06,
      "loss": 0.5456,
      "step": 1366
    },
    {
      "epoch": 1.6731946144430845,
      "grad_norm": 1.0378493466396788,
      "learning_rate": 4.8480603231039855e-06,
      "loss": 1.0458,
      "step": 1367
    },
    {
      "epoch": 1.6744186046511627,
      "grad_norm": 1.5545427442070008,
      "learning_rate": 4.8478402289913566e-06,
      "loss": 0.5829,
      "step": 1368
    },
    {
      "epoch": 1.6756425948592413,
      "grad_norm": 1.2146483118697817,
      "learning_rate": 4.847619980587351e-06,
      "loss": 0.5282,
      "step": 1369
    },
    {
      "epoch": 1.6768665850673194,
      "grad_norm": 1.3057177265577697,
      "learning_rate": 4.847399577906445e-06,
      "loss": 0.5385,
      "step": 1370
    },
    {
      "epoch": 1.6780905752753978,
      "grad_norm": 1.3010363198457944,
      "learning_rate": 4.84717902096312e-06,
      "loss": 1.0415,
      "step": 1371
    },
    {
      "epoch": 1.6793145654834762,
      "grad_norm": 1.097423872542746,
      "learning_rate": 4.846958309771872e-06,
      "loss": 0.5982,
      "step": 1372
    },
    {
      "epoch": 1.6805385556915544,
      "grad_norm": 0.9190234955374704,
      "learning_rate": 4.846737444347204e-06,
      "loss": 0.8408,
      "step": 1373
    },
    {
      "epoch": 1.6817625458996328,
      "grad_norm": 0.9539080579963868,
      "learning_rate": 4.8465164247036315e-06,
      "loss": 0.6833,
      "step": 1374
    },
    {
      "epoch": 1.6829865361077112,
      "grad_norm": 1.0189941618103304,
      "learning_rate": 4.8462952508556795e-06,
      "loss": 0.5425,
      "step": 1375
    },
    {
      "epoch": 1.6842105263157894,
      "grad_norm": 0.9546078906645562,
      "learning_rate": 4.846073922817881e-06,
      "loss": 0.7074,
      "step": 1376
    },
    {
      "epoch": 1.6854345165238678,
      "grad_norm": 1.318250133445118,
      "learning_rate": 4.8458524406047814e-06,
      "loss": 0.59,
      "step": 1377
    },
    {
      "epoch": 1.6866585067319462,
      "grad_norm": 0.8396172369658257,
      "learning_rate": 4.845630804230937e-06,
      "loss": 0.3801,
      "step": 1378
    },
    {
      "epoch": 1.6878824969400243,
      "grad_norm": 0.5053025695981047,
      "learning_rate": 4.845409013710912e-06,
      "loss": 0.4293,
      "step": 1379
    },
    {
      "epoch": 1.689106487148103,
      "grad_norm": 1.107202519002441,
      "learning_rate": 4.8451870690592815e-06,
      "loss": 0.5315,
      "step": 1380
    },
    {
      "epoch": 1.6903304773561811,
      "grad_norm": 0.8321144207482137,
      "learning_rate": 4.8449649702906305e-06,
      "loss": 1.1851,
      "step": 1381
    },
    {
      "epoch": 1.6915544675642595,
      "grad_norm": 0.5432489801135765,
      "learning_rate": 4.844742717419555e-06,
      "loss": 0.4244,
      "step": 1382
    },
    {
      "epoch": 1.692778457772338,
      "grad_norm": 0.7461345531536079,
      "learning_rate": 4.8445203104606616e-06,
      "loss": 0.4797,
      "step": 1383
    },
    {
      "epoch": 1.694002447980416,
      "grad_norm": 1.0772964135869134,
      "learning_rate": 4.8442977494285645e-06,
      "loss": 1.4066,
      "step": 1384
    },
    {
      "epoch": 1.6952264381884945,
      "grad_norm": 0.8885310766162761,
      "learning_rate": 4.844075034337891e-06,
      "loss": 1.6176,
      "step": 1385
    },
    {
      "epoch": 1.696450428396573,
      "grad_norm": 0.8881540751736681,
      "learning_rate": 4.843852165203276e-06,
      "loss": 1.2524,
      "step": 1386
    },
    {
      "epoch": 1.697674418604651,
      "grad_norm": 1.1335553658619664,
      "learning_rate": 4.843629142039366e-06,
      "loss": 0.6461,
      "step": 1387
    },
    {
      "epoch": 1.6988984088127295,
      "grad_norm": 1.1289339218566845,
      "learning_rate": 4.843405964860818e-06,
      "loss": 0.5618,
      "step": 1388
    },
    {
      "epoch": 1.7001223990208079,
      "grad_norm": 0.8939417067555648,
      "learning_rate": 4.8431826336822965e-06,
      "loss": 1.3365,
      "step": 1389
    },
    {
      "epoch": 1.701346389228886,
      "grad_norm": 0.9214500333377122,
      "learning_rate": 4.8429591485184795e-06,
      "loss": 0.6453,
      "step": 1390
    },
    {
      "epoch": 1.7025703794369647,
      "grad_norm": 0.8788926841480005,
      "learning_rate": 4.842735509384053e-06,
      "loss": 0.7287,
      "step": 1391
    },
    {
      "epoch": 1.7037943696450428,
      "grad_norm": 0.9457172859574038,
      "learning_rate": 4.842511716293715e-06,
      "loss": 0.6348,
      "step": 1392
    },
    {
      "epoch": 1.7050183598531212,
      "grad_norm": 1.1734931553643246,
      "learning_rate": 4.8422877692621705e-06,
      "loss": 0.53,
      "step": 1393
    },
    {
      "epoch": 1.7062423500611996,
      "grad_norm": 1.3285801745847947,
      "learning_rate": 4.842063668304138e-06,
      "loss": 0.501,
      "step": 1394
    },
    {
      "epoch": 1.7074663402692778,
      "grad_norm": 0.7199472703807621,
      "learning_rate": 4.841839413434343e-06,
      "loss": 0.6997,
      "step": 1395
    },
    {
      "epoch": 1.7086903304773562,
      "grad_norm": 1.83276289893966,
      "learning_rate": 4.841615004667524e-06,
      "loss": 0.5097,
      "step": 1396
    },
    {
      "epoch": 1.7099143206854346,
      "grad_norm": 0.7052925368181556,
      "learning_rate": 4.841390442018428e-06,
      "loss": 0.8261,
      "step": 1397
    },
    {
      "epoch": 1.7111383108935128,
      "grad_norm": 0.7598471345316804,
      "learning_rate": 4.841165725501812e-06,
      "loss": 0.7082,
      "step": 1398
    },
    {
      "epoch": 1.7123623011015912,
      "grad_norm": 0.7840692605783667,
      "learning_rate": 4.840940855132445e-06,
      "loss": 0.5775,
      "step": 1399
    },
    {
      "epoch": 1.7135862913096696,
      "grad_norm": 0.57584935461918,
      "learning_rate": 4.840715830925102e-06,
      "loss": 0.5684,
      "step": 1400
    },
    {
      "epoch": 1.7148102815177477,
      "grad_norm": 0.6617358250441628,
      "learning_rate": 4.8404906528945736e-06,
      "loss": 0.4991,
      "step": 1401
    },
    {
      "epoch": 1.7160342717258263,
      "grad_norm": 1.1445618470947627,
      "learning_rate": 4.840265321055656e-06,
      "loss": 0.5051,
      "step": 1402
    },
    {
      "epoch": 1.7172582619339045,
      "grad_norm": 1.538271508159586,
      "learning_rate": 4.840039835423157e-06,
      "loss": 0.654,
      "step": 1403
    },
    {
      "epoch": 1.718482252141983,
      "grad_norm": 0.7134922011799449,
      "learning_rate": 4.839814196011896e-06,
      "loss": 1.2648,
      "step": 1404
    },
    {
      "epoch": 1.7197062423500613,
      "grad_norm": 0.969705780432247,
      "learning_rate": 4.8395884028366994e-06,
      "loss": 0.539,
      "step": 1405
    },
    {
      "epoch": 1.7209302325581395,
      "grad_norm": 0.7386756837251552,
      "learning_rate": 4.8393624559124074e-06,
      "loss": 0.5594,
      "step": 1406
    },
    {
      "epoch": 1.7221542227662179,
      "grad_norm": 0.6629138978784509,
      "learning_rate": 4.8391363552538674e-06,
      "loss": 0.5322,
      "step": 1407
    },
    {
      "epoch": 1.7233782129742963,
      "grad_norm": 1.372697965929999,
      "learning_rate": 4.838910100875937e-06,
      "loss": 0.6041,
      "step": 1408
    },
    {
      "epoch": 1.7246022031823744,
      "grad_norm": 1.1665020399550567,
      "learning_rate": 4.838683692793487e-06,
      "loss": 0.5951,
      "step": 1409
    },
    {
      "epoch": 1.7258261933904528,
      "grad_norm": 1.00397553516941,
      "learning_rate": 4.838457131021394e-06,
      "loss": 0.7124,
      "step": 1410
    },
    {
      "epoch": 1.7270501835985312,
      "grad_norm": 1.0035554133006386,
      "learning_rate": 4.838230415574549e-06,
      "loss": 0.6264,
      "step": 1411
    },
    {
      "epoch": 1.7282741738066094,
      "grad_norm": 0.8629602372173336,
      "learning_rate": 4.8380035464678485e-06,
      "loss": 0.6809,
      "step": 1412
    },
    {
      "epoch": 1.729498164014688,
      "grad_norm": 1.1364577666201288,
      "learning_rate": 4.837776523716203e-06,
      "loss": 0.6161,
      "step": 1413
    },
    {
      "epoch": 1.7307221542227662,
      "grad_norm": 1.1951282054629713,
      "learning_rate": 4.837549347334532e-06,
      "loss": 0.6011,
      "step": 1414
    },
    {
      "epoch": 1.7319461444308446,
      "grad_norm": 0.7444852547295959,
      "learning_rate": 4.837322017337762e-06,
      "loss": 0.6868,
      "step": 1415
    },
    {
      "epoch": 1.733170134638923,
      "grad_norm": 1.2031672218003941,
      "learning_rate": 4.8370945337408356e-06,
      "loss": 1.1329,
      "step": 1416
    },
    {
      "epoch": 1.7343941248470012,
      "grad_norm": 0.5380183744441042,
      "learning_rate": 4.836866896558701e-06,
      "loss": 0.3613,
      "step": 1417
    },
    {
      "epoch": 1.7356181150550796,
      "grad_norm": 1.1858491090616687,
      "learning_rate": 4.836639105806316e-06,
      "loss": 0.7785,
      "step": 1418
    },
    {
      "epoch": 1.736842105263158,
      "grad_norm": 0.9596609725048824,
      "learning_rate": 4.836411161498653e-06,
      "loss": 0.5446,
      "step": 1419
    },
    {
      "epoch": 1.7380660954712361,
      "grad_norm": 1.812871766843271,
      "learning_rate": 4.836183063650689e-06,
      "loss": 0.4096,
      "step": 1420
    },
    {
      "epoch": 1.7392900856793145,
      "grad_norm": 1.3994264445112043,
      "learning_rate": 4.835954812277416e-06,
      "loss": 0.6902,
      "step": 1421
    },
    {
      "epoch": 1.740514075887393,
      "grad_norm": 1.6970485240903936,
      "learning_rate": 4.835726407393833e-06,
      "loss": 0.4778,
      "step": 1422
    },
    {
      "epoch": 1.741738066095471,
      "grad_norm": 1.3219431743593213,
      "learning_rate": 4.835497849014948e-06,
      "loss": 0.5155,
      "step": 1423
    },
    {
      "epoch": 1.7429620563035497,
      "grad_norm": 0.6777142142697299,
      "learning_rate": 4.835269137155785e-06,
      "loss": 0.5073,
      "step": 1424
    },
    {
      "epoch": 1.744186046511628,
      "grad_norm": 0.9467826782412418,
      "learning_rate": 4.835040271831371e-06,
      "loss": 1.704,
      "step": 1425
    },
    {
      "epoch": 1.7454100367197063,
      "grad_norm": 0.8935283934972048,
      "learning_rate": 4.834811253056746e-06,
      "loss": 1.4113,
      "step": 1426
    },
    {
      "epoch": 1.7466340269277847,
      "grad_norm": 0.8144381789045311,
      "learning_rate": 4.834582080846962e-06,
      "loss": 0.5345,
      "step": 1427
    },
    {
      "epoch": 1.7478580171358629,
      "grad_norm": 0.5978291395144247,
      "learning_rate": 4.834352755217079e-06,
      "loss": 0.5682,
      "step": 1428
    },
    {
      "epoch": 1.7490820073439413,
      "grad_norm": 1.0298927491557819,
      "learning_rate": 4.834123276182167e-06,
      "loss": 0.4752,
      "step": 1429
    },
    {
      "epoch": 1.7503059975520197,
      "grad_norm": 1.0757422131080239,
      "learning_rate": 4.833893643757307e-06,
      "loss": 1.0708,
      "step": 1430
    },
    {
      "epoch": 1.7515299877600978,
      "grad_norm": 1.140193493363014,
      "learning_rate": 4.833663857957589e-06,
      "loss": 0.6584,
      "step": 1431
    },
    {
      "epoch": 1.7527539779681762,
      "grad_norm": 1.0498651202795317,
      "learning_rate": 4.833433918798114e-06,
      "loss": 0.8691,
      "step": 1432
    },
    {
      "epoch": 1.7539779681762546,
      "grad_norm": 0.29051884009607687,
      "learning_rate": 4.8332038262939914e-06,
      "loss": 0.1424,
      "step": 1433
    },
    {
      "epoch": 1.7552019583843328,
      "grad_norm": 1.003747979684293,
      "learning_rate": 4.832973580460345e-06,
      "loss": 1.6384,
      "step": 1434
    },
    {
      "epoch": 1.7564259485924114,
      "grad_norm": 1.0275523068284698,
      "learning_rate": 4.832743181312303e-06,
      "loss": 0.8051,
      "step": 1435
    },
    {
      "epoch": 1.7576499388004896,
      "grad_norm": 0.8802168865799148,
      "learning_rate": 4.832512628865007e-06,
      "loss": 0.5273,
      "step": 1436
    },
    {
      "epoch": 1.758873929008568,
      "grad_norm": 1.1090781196950992,
      "learning_rate": 4.832281923133609e-06,
      "loss": 0.8023,
      "step": 1437
    },
    {
      "epoch": 1.7600979192166464,
      "grad_norm": 0.7877038327645125,
      "learning_rate": 4.83205106413327e-06,
      "loss": 0.6419,
      "step": 1438
    },
    {
      "epoch": 1.7613219094247246,
      "grad_norm": 0.9633117173854182,
      "learning_rate": 4.83182005187916e-06,
      "loss": 0.7026,
      "step": 1439
    },
    {
      "epoch": 1.762545899632803,
      "grad_norm": 0.9315482359424593,
      "learning_rate": 4.8315888863864615e-06,
      "loss": 0.4907,
      "step": 1440
    },
    {
      "epoch": 1.7637698898408813,
      "grad_norm": 1.266559758771324,
      "learning_rate": 4.831357567670365e-06,
      "loss": 0.3877,
      "step": 1441
    },
    {
      "epoch": 1.7649938800489595,
      "grad_norm": 1.0711176323070952,
      "learning_rate": 4.831126095746072e-06,
      "loss": 0.5213,
      "step": 1442
    },
    {
      "epoch": 1.766217870257038,
      "grad_norm": 0.560452411556726,
      "learning_rate": 4.830894470628794e-06,
      "loss": 0.5893,
      "step": 1443
    },
    {
      "epoch": 1.7674418604651163,
      "grad_norm": 0.9624596295314435,
      "learning_rate": 4.830662692333754e-06,
      "loss": 0.6874,
      "step": 1444
    },
    {
      "epoch": 1.7686658506731945,
      "grad_norm": 1.1601756221193327,
      "learning_rate": 4.830430760876181e-06,
      "loss": 0.7026,
      "step": 1445
    },
    {
      "epoch": 1.769889840881273,
      "grad_norm": 0.7653526766814143,
      "learning_rate": 4.830198676271319e-06,
      "loss": 1.3045,
      "step": 1446
    },
    {
      "epoch": 1.7711138310893513,
      "grad_norm": 0.836608977438489,
      "learning_rate": 4.8299664385344184e-06,
      "loss": 1.0041,
      "step": 1447
    },
    {
      "epoch": 1.7723378212974297,
      "grad_norm": 0.7163530598483613,
      "learning_rate": 4.829734047680742e-06,
      "loss": 0.7731,
      "step": 1448
    },
    {
      "epoch": 1.773561811505508,
      "grad_norm": 0.7535189244824811,
      "learning_rate": 4.82950150372556e-06,
      "loss": 0.7461,
      "step": 1449
    },
    {
      "epoch": 1.7747858017135862,
      "grad_norm": 0.7676280899418488,
      "learning_rate": 4.829268806684156e-06,
      "loss": 0.8682,
      "step": 1450
    },
    {
      "epoch": 1.7760097919216646,
      "grad_norm": 0.5607757510867017,
      "learning_rate": 4.829035956571821e-06,
      "loss": 0.7716,
      "step": 1451
    },
    {
      "epoch": 1.777233782129743,
      "grad_norm": 0.5288249183296992,
      "learning_rate": 4.828802953403858e-06,
      "loss": 0.6514,
      "step": 1452
    },
    {
      "epoch": 1.7784577723378212,
      "grad_norm": 0.8543669651055841,
      "learning_rate": 4.8285697971955785e-06,
      "loss": 0.7092,
      "step": 1453
    },
    {
      "epoch": 1.7796817625458996,
      "grad_norm": 1.236002560873317,
      "learning_rate": 4.828336487962305e-06,
      "loss": 1.2197,
      "step": 1454
    },
    {
      "epoch": 1.780905752753978,
      "grad_norm": 0.7611961135179246,
      "learning_rate": 4.828103025719368e-06,
      "loss": 0.4352,
      "step": 1455
    },
    {
      "epoch": 1.7821297429620562,
      "grad_norm": 0.585270506211237,
      "learning_rate": 4.827869410482113e-06,
      "loss": 0.6359,
      "step": 1456
    },
    {
      "epoch": 1.7833537331701348,
      "grad_norm": 0.5691500029885712,
      "learning_rate": 4.82763564226589e-06,
      "loss": 0.6929,
      "step": 1457
    },
    {
      "epoch": 1.784577723378213,
      "grad_norm": 1.2692134179564167,
      "learning_rate": 4.8274017210860614e-06,
      "loss": 0.5831,
      "step": 1458
    },
    {
      "epoch": 1.7858017135862914,
      "grad_norm": 1.1968943413430078,
      "learning_rate": 4.827167646958e-06,
      "loss": 1.1208,
      "step": 1459
    },
    {
      "epoch": 1.7870257037943698,
      "grad_norm": 0.9007292695455923,
      "learning_rate": 4.826933419897089e-06,
      "loss": 0.5554,
      "step": 1460
    },
    {
      "epoch": 1.788249694002448,
      "grad_norm": 1.315736156666934,
      "learning_rate": 4.826699039918721e-06,
      "loss": 0.5642,
      "step": 1461
    },
    {
      "epoch": 1.7894736842105263,
      "grad_norm": 0.6890804498886349,
      "learning_rate": 4.8264645070382964e-06,
      "loss": 0.6224,
      "step": 1462
    },
    {
      "epoch": 1.7906976744186047,
      "grad_norm": 0.893253330229824,
      "learning_rate": 4.82622982127123e-06,
      "loss": 0.7738,
      "step": 1463
    },
    {
      "epoch": 1.791921664626683,
      "grad_norm": 0.7496760166368227,
      "learning_rate": 4.825994982632944e-06,
      "loss": 0.6149,
      "step": 1464
    },
    {
      "epoch": 1.7931456548347613,
      "grad_norm": 1.1939569135824306,
      "learning_rate": 4.825759991138872e-06,
      "loss": 1.2779,
      "step": 1465
    },
    {
      "epoch": 1.7943696450428397,
      "grad_norm": 1.2956294888148754,
      "learning_rate": 4.825524846804455e-06,
      "loss": 1.1265,
      "step": 1466
    },
    {
      "epoch": 1.7955936352509179,
      "grad_norm": 0.7180266129586983,
      "learning_rate": 4.8252895496451455e-06,
      "loss": 0.7777,
      "step": 1467
    },
    {
      "epoch": 1.7968176254589965,
      "grad_norm": 0.7136187038557338,
      "learning_rate": 4.8250540996764086e-06,
      "loss": 0.6305,
      "step": 1468
    },
    {
      "epoch": 1.7980416156670747,
      "grad_norm": 1.0062599563324495,
      "learning_rate": 4.824818496913716e-06,
      "loss": 1.904,
      "step": 1469
    },
    {
      "epoch": 1.799265605875153,
      "grad_norm": 0.6918545327681261,
      "learning_rate": 4.824582741372551e-06,
      "loss": 0.5781,
      "step": 1470
    },
    {
      "epoch": 1.8004895960832314,
      "grad_norm": 1.0430502695215231,
      "learning_rate": 4.824346833068405e-06,
      "loss": 0.6479,
      "step": 1471
    },
    {
      "epoch": 1.8017135862913096,
      "grad_norm": 0.7457556451687253,
      "learning_rate": 4.824110772016784e-06,
      "loss": 0.4822,
      "step": 1472
    },
    {
      "epoch": 1.802937576499388,
      "grad_norm": 1.052174678771876,
      "learning_rate": 4.823874558233198e-06,
      "loss": 0.6481,
      "step": 1473
    },
    {
      "epoch": 1.8041615667074664,
      "grad_norm": 0.9083250478784275,
      "learning_rate": 4.823638191733172e-06,
      "loss": 0.5417,
      "step": 1474
    },
    {
      "epoch": 1.8053855569155446,
      "grad_norm": 1.1439026159882684,
      "learning_rate": 4.823401672532239e-06,
      "loss": 0.858,
      "step": 1475
    },
    {
      "epoch": 1.806609547123623,
      "grad_norm": 1.256538574235264,
      "learning_rate": 4.8231650006459405e-06,
      "loss": 0.5352,
      "step": 1476
    },
    {
      "epoch": 1.8078335373317014,
      "grad_norm": 0.7919996035872724,
      "learning_rate": 4.822928176089832e-06,
      "loss": 0.6042,
      "step": 1477
    },
    {
      "epoch": 1.8090575275397796,
      "grad_norm": 1.109389395009973,
      "learning_rate": 4.822691198879474e-06,
      "loss": 0.4746,
      "step": 1478
    },
    {
      "epoch": 1.8102815177478582,
      "grad_norm": 1.304222745961119,
      "learning_rate": 4.822454069030443e-06,
      "loss": 0.4661,
      "step": 1479
    },
    {
      "epoch": 1.8115055079559363,
      "grad_norm": 1.5182068433934577,
      "learning_rate": 4.822216786558322e-06,
      "loss": 0.5989,
      "step": 1480
    },
    {
      "epoch": 1.8127294981640147,
      "grad_norm": 0.515145698306268,
      "learning_rate": 4.8219793514787006e-06,
      "loss": 0.4915,
      "step": 1481
    },
    {
      "epoch": 1.8139534883720931,
      "grad_norm": 1.435381972900211,
      "learning_rate": 4.821741763807186e-06,
      "loss": 0.4254,
      "step": 1482
    },
    {
      "epoch": 1.8151774785801713,
      "grad_norm": 1.109797096943933,
      "learning_rate": 4.82150402355939e-06,
      "loss": 0.7823,
      "step": 1483
    },
    {
      "epoch": 1.8164014687882497,
      "grad_norm": 1.0654999109953989,
      "learning_rate": 4.821266130750936e-06,
      "loss": 0.7398,
      "step": 1484
    },
    {
      "epoch": 1.817625458996328,
      "grad_norm": 0.7074367822925478,
      "learning_rate": 4.821028085397458e-06,
      "loss": 0.3936,
      "step": 1485
    },
    {
      "epoch": 1.8188494492044063,
      "grad_norm": 0.854122269515197,
      "learning_rate": 4.820789887514599e-06,
      "loss": 0.7649,
      "step": 1486
    },
    {
      "epoch": 1.8200734394124847,
      "grad_norm": 1.130758457983194,
      "learning_rate": 4.820551537118012e-06,
      "loss": 1.4075,
      "step": 1487
    },
    {
      "epoch": 1.821297429620563,
      "grad_norm": 1.1303211458507376,
      "learning_rate": 4.820313034223362e-06,
      "loss": 0.5834,
      "step": 1488
    },
    {
      "epoch": 1.8225214198286412,
      "grad_norm": 0.8877457788280062,
      "learning_rate": 4.82007437884632e-06,
      "loss": 1.0032,
      "step": 1489
    },
    {
      "epoch": 1.8237454100367199,
      "grad_norm": 0.9739734926993258,
      "learning_rate": 4.819835571002573e-06,
      "loss": 0.4644,
      "step": 1490
    },
    {
      "epoch": 1.824969400244798,
      "grad_norm": 1.623434946798665,
      "learning_rate": 4.819596610707812e-06,
      "loss": 1.0311,
      "step": 1491
    },
    {
      "epoch": 1.8261933904528764,
      "grad_norm": 0.8748455082565944,
      "learning_rate": 4.819357497977741e-06,
      "loss": 0.7445,
      "step": 1492
    },
    {
      "epoch": 1.8274173806609548,
      "grad_norm": 0.6040657590077547,
      "learning_rate": 4.819118232828075e-06,
      "loss": 0.5022,
      "step": 1493
    },
    {
      "epoch": 1.828641370869033,
      "grad_norm": 1.0165450930958981,
      "learning_rate": 4.818878815274536e-06,
      "loss": 1.0947,
      "step": 1494
    },
    {
      "epoch": 1.8298653610771114,
      "grad_norm": 1.0663356289941008,
      "learning_rate": 4.8186392453328576e-06,
      "loss": 1.0436,
      "step": 1495
    },
    {
      "epoch": 1.8310893512851898,
      "grad_norm": 1.107881590568128,
      "learning_rate": 4.818399523018785e-06,
      "loss": 0.6394,
      "step": 1496
    },
    {
      "epoch": 1.832313341493268,
      "grad_norm": 0.7738151914790905,
      "learning_rate": 4.8181596483480704e-06,
      "loss": 1.2337,
      "step": 1497
    },
    {
      "epoch": 1.8335373317013464,
      "grad_norm": 0.894704045032589,
      "learning_rate": 4.817919621336479e-06,
      "loss": 0.7724,
      "step": 1498
    },
    {
      "epoch": 1.8347613219094248,
      "grad_norm": 1.0613202505320993,
      "learning_rate": 4.817679441999782e-06,
      "loss": 0.5273,
      "step": 1499
    },
    {
      "epoch": 1.835985312117503,
      "grad_norm": 0.7053683671503921,
      "learning_rate": 4.8174391103537655e-06,
      "loss": 0.6449,
      "step": 1500
    },
    {
      "epoch": 1.8372093023255816,
      "grad_norm": 1.0700410780135006,
      "learning_rate": 4.817198626414222e-06,
      "loss": 0.7081,
      "step": 1501
    },
    {
      "epoch": 1.8384332925336597,
      "grad_norm": 0.8864625877201399,
      "learning_rate": 4.8169579901969556e-06,
      "loss": 1.1952,
      "step": 1502
    },
    {
      "epoch": 1.8396572827417381,
      "grad_norm": 0.9498111240616791,
      "learning_rate": 4.816717201717779e-06,
      "loss": 0.4213,
      "step": 1503
    },
    {
      "epoch": 1.8408812729498165,
      "grad_norm": 0.7801217545876808,
      "learning_rate": 4.816476260992518e-06,
      "loss": 0.3375,
      "step": 1504
    },
    {
      "epoch": 1.8421052631578947,
      "grad_norm": 0.7235709244331405,
      "learning_rate": 4.8162351680370046e-06,
      "loss": 0.6006,
      "step": 1505
    },
    {
      "epoch": 1.843329253365973,
      "grad_norm": 1.0925782755243532,
      "learning_rate": 4.815993922867084e-06,
      "loss": 0.9456,
      "step": 1506
    },
    {
      "epoch": 1.8445532435740515,
      "grad_norm": 0.8754926016871207,
      "learning_rate": 4.815752525498608e-06,
      "loss": 1.3869,
      "step": 1507
    },
    {
      "epoch": 1.8457772337821297,
      "grad_norm": 0.8328080498002697,
      "learning_rate": 4.8155109759474415e-06,
      "loss": 0.5503,
      "step": 1508
    },
    {
      "epoch": 1.847001223990208,
      "grad_norm": 1.2455584044321122,
      "learning_rate": 4.815269274229459e-06,
      "loss": 0.4463,
      "step": 1509
    },
    {
      "epoch": 1.8482252141982864,
      "grad_norm": 0.9275182298774276,
      "learning_rate": 4.8150274203605425e-06,
      "loss": 0.4408,
      "step": 1510
    },
    {
      "epoch": 1.8494492044063646,
      "grad_norm": 1.0840100737439364,
      "learning_rate": 4.814785414356587e-06,
      "loss": 0.4433,
      "step": 1511
    },
    {
      "epoch": 1.8506731946144432,
      "grad_norm": 1.0552100637568758,
      "learning_rate": 4.814543256233496e-06,
      "loss": 0.6814,
      "step": 1512
    },
    {
      "epoch": 1.8518971848225214,
      "grad_norm": 1.1488548126990605,
      "learning_rate": 4.8143009460071825e-06,
      "loss": 1.219,
      "step": 1513
    },
    {
      "epoch": 1.8531211750305998,
      "grad_norm": 0.9831063796409565,
      "learning_rate": 4.814058483693571e-06,
      "loss": 1.6901,
      "step": 1514
    },
    {
      "epoch": 1.8543451652386782,
      "grad_norm": 1.020444989883295,
      "learning_rate": 4.8138158693085955e-06,
      "loss": 0.514,
      "step": 1515
    },
    {
      "epoch": 1.8555691554467564,
      "grad_norm": 0.8803074916524232,
      "learning_rate": 4.8135731028681995e-06,
      "loss": 0.6051,
      "step": 1516
    },
    {
      "epoch": 1.8567931456548348,
      "grad_norm": 1.0944912254157153,
      "learning_rate": 4.813330184388336e-06,
      "loss": 0.9437,
      "step": 1517
    },
    {
      "epoch": 1.8580171358629132,
      "grad_norm": 0.8017896115993902,
      "learning_rate": 4.813087113884969e-06,
      "loss": 0.3534,
      "step": 1518
    },
    {
      "epoch": 1.8592411260709913,
      "grad_norm": 1.3391569596956512,
      "learning_rate": 4.812843891374073e-06,
      "loss": 0.9792,
      "step": 1519
    },
    {
      "epoch": 1.8604651162790697,
      "grad_norm": 0.9575500128466213,
      "learning_rate": 4.812600516871631e-06,
      "loss": 0.6995,
      "step": 1520
    },
    {
      "epoch": 1.8616891064871481,
      "grad_norm": 0.6607800837320198,
      "learning_rate": 4.8123569903936365e-06,
      "loss": 0.6451,
      "step": 1521
    },
    {
      "epoch": 1.8629130966952263,
      "grad_norm": 0.869912539710866,
      "learning_rate": 4.812113311956092e-06,
      "loss": 1.0551,
      "step": 1522
    },
    {
      "epoch": 1.864137086903305,
      "grad_norm": 0.5838790234659531,
      "learning_rate": 4.811869481575015e-06,
      "loss": 0.522,
      "step": 1523
    },
    {
      "epoch": 1.865361077111383,
      "grad_norm": 1.350590569104206,
      "learning_rate": 4.811625499266426e-06,
      "loss": 0.6209,
      "step": 1524
    },
    {
      "epoch": 1.8665850673194615,
      "grad_norm": 1.6371713950579037,
      "learning_rate": 4.811381365046358e-06,
      "loss": 0.5374,
      "step": 1525
    },
    {
      "epoch": 1.86780905752754,
      "grad_norm": 1.098951158727059,
      "learning_rate": 4.8111370789308575e-06,
      "loss": 0.5957,
      "step": 1526
    },
    {
      "epoch": 1.869033047735618,
      "grad_norm": 0.7839450633508284,
      "learning_rate": 4.810892640935976e-06,
      "loss": 1.0284,
      "step": 1527
    },
    {
      "epoch": 1.8702570379436965,
      "grad_norm": 1.6708137430281147,
      "learning_rate": 4.810648051077777e-06,
      "loss": 0.9328,
      "step": 1528
    },
    {
      "epoch": 1.8714810281517749,
      "grad_norm": 1.250364662763702,
      "learning_rate": 4.810403309372334e-06,
      "loss": 1.1808,
      "step": 1529
    },
    {
      "epoch": 1.872705018359853,
      "grad_norm": 1.0168547795883442,
      "learning_rate": 4.810158415835733e-06,
      "loss": 0.6585,
      "step": 1530
    },
    {
      "epoch": 1.8739290085679314,
      "grad_norm": 0.6385948502943634,
      "learning_rate": 4.809913370484064e-06,
      "loss": 0.7035,
      "step": 1531
    },
    {
      "epoch": 1.8751529987760098,
      "grad_norm": 0.7029505573606275,
      "learning_rate": 4.809668173333433e-06,
      "loss": 0.3702,
      "step": 1532
    },
    {
      "epoch": 1.876376988984088,
      "grad_norm": 0.7027859663405749,
      "learning_rate": 4.809422824399953e-06,
      "loss": 1.1989,
      "step": 1533
    },
    {
      "epoch": 1.8776009791921666,
      "grad_norm": 0.9574615135006147,
      "learning_rate": 4.809177323699745e-06,
      "loss": 1.21,
      "step": 1534
    },
    {
      "epoch": 1.8788249694002448,
      "grad_norm": 0.735155950074949,
      "learning_rate": 4.808931671248946e-06,
      "loss": 1.4036,
      "step": 1535
    },
    {
      "epoch": 1.880048959608323,
      "grad_norm": 1.0676006437755676,
      "learning_rate": 4.808685867063697e-06,
      "loss": 1.0464,
      "step": 1536
    },
    {
      "epoch": 1.8812729498164016,
      "grad_norm": 0.842146943442167,
      "learning_rate": 4.808439911160153e-06,
      "loss": 0.5602,
      "step": 1537
    },
    {
      "epoch": 1.8824969400244798,
      "grad_norm": 1.3556074807724596,
      "learning_rate": 4.808193803554475e-06,
      "loss": 0.9214,
      "step": 1538
    },
    {
      "epoch": 1.8837209302325582,
      "grad_norm": 1.5543090043681407,
      "learning_rate": 4.807947544262838e-06,
      "loss": 1.0266,
      "step": 1539
    },
    {
      "epoch": 1.8849449204406366,
      "grad_norm": 1.260015606551513,
      "learning_rate": 4.8077011333014255e-06,
      "loss": 0.6517,
      "step": 1540
    },
    {
      "epoch": 1.8861689106487147,
      "grad_norm": 1.362238409990797,
      "learning_rate": 4.80745457068643e-06,
      "loss": 1.0566,
      "step": 1541
    },
    {
      "epoch": 1.8873929008567931,
      "grad_norm": 0.9673539197161027,
      "learning_rate": 4.8072078564340544e-06,
      "loss": 0.8399,
      "step": 1542
    },
    {
      "epoch": 1.8886168910648715,
      "grad_norm": 0.6942370848274635,
      "learning_rate": 4.806960990560513e-06,
      "loss": 0.5736,
      "step": 1543
    },
    {
      "epoch": 1.8898408812729497,
      "grad_norm": 1.120104529714829,
      "learning_rate": 4.806713973082027e-06,
      "loss": 0.7519,
      "step": 1544
    },
    {
      "epoch": 1.8910648714810283,
      "grad_norm": 0.9220660548016286,
      "learning_rate": 4.8064668040148315e-06,
      "loss": 0.5115,
      "step": 1545
    },
    {
      "epoch": 1.8922888616891065,
      "grad_norm": 0.849610342061057,
      "learning_rate": 4.806219483375168e-06,
      "loss": 1.0824,
      "step": 1546
    },
    {
      "epoch": 1.8935128518971847,
      "grad_norm": 0.7855186118456127,
      "learning_rate": 4.805972011179291e-06,
      "loss": 0.8608,
      "step": 1547
    },
    {
      "epoch": 1.8947368421052633,
      "grad_norm": 0.6359533911782205,
      "learning_rate": 4.8057243874434625e-06,
      "loss": 0.6822,
      "step": 1548
    },
    {
      "epoch": 1.8959608323133414,
      "grad_norm": 1.2259476116977583,
      "learning_rate": 4.805476612183955e-06,
      "loss": 1.1406,
      "step": 1549
    },
    {
      "epoch": 1.8971848225214198,
      "grad_norm": 0.8488143973244074,
      "learning_rate": 4.805228685417052e-06,
      "loss": 0.9291,
      "step": 1550
    },
    {
      "epoch": 1.8984088127294982,
      "grad_norm": 1.0973264929642157,
      "learning_rate": 4.804980607159047e-06,
      "loss": 0.4934,
      "step": 1551
    },
    {
      "epoch": 1.8996328029375764,
      "grad_norm": 0.5853289191125773,
      "learning_rate": 4.804732377426241e-06,
      "loss": 0.5552,
      "step": 1552
    },
    {
      "epoch": 1.9008567931456548,
      "grad_norm": 1.1341593121674785,
      "learning_rate": 4.804483996234949e-06,
      "loss": 1.2437,
      "step": 1553
    },
    {
      "epoch": 1.9020807833537332,
      "grad_norm": 1.0490810724715849,
      "learning_rate": 4.804235463601491e-06,
      "loss": 0.4737,
      "step": 1554
    },
    {
      "epoch": 1.9033047735618114,
      "grad_norm": 1.3341499778955537,
      "learning_rate": 4.8039867795422015e-06,
      "loss": 0.622,
      "step": 1555
    },
    {
      "epoch": 1.90452876376989,
      "grad_norm": 0.5906945024274528,
      "learning_rate": 4.803737944073423e-06,
      "loss": 0.6012,
      "step": 1556
    },
    {
      "epoch": 1.9057527539779682,
      "grad_norm": 1.0220042730977423,
      "learning_rate": 4.803488957211508e-06,
      "loss": 0.7194,
      "step": 1557
    },
    {
      "epoch": 1.9069767441860463,
      "grad_norm": 0.8406068750683137,
      "learning_rate": 4.803239818972818e-06,
      "loss": 0.6668,
      "step": 1558
    },
    {
      "epoch": 1.908200734394125,
      "grad_norm": 0.8329151244038531,
      "learning_rate": 4.8029905293737265e-06,
      "loss": 1.2737,
      "step": 1559
    },
    {
      "epoch": 1.9094247246022031,
      "grad_norm": 0.7697497840113672,
      "learning_rate": 4.802741088430616e-06,
      "loss": 0.6155,
      "step": 1560
    },
    {
      "epoch": 1.9106487148102815,
      "grad_norm": 0.9161043715658023,
      "learning_rate": 4.802491496159878e-06,
      "loss": 0.6098,
      "step": 1561
    },
    {
      "epoch": 1.91187270501836,
      "grad_norm": 1.548788154384532,
      "learning_rate": 4.8022417525779155e-06,
      "loss": 0.9378,
      "step": 1562
    },
    {
      "epoch": 1.913096695226438,
      "grad_norm": 0.9954150013599182,
      "learning_rate": 4.80199185770114e-06,
      "loss": 0.6086,
      "step": 1563
    },
    {
      "epoch": 1.9143206854345165,
      "grad_norm": 1.237058750275263,
      "learning_rate": 4.801741811545975e-06,
      "loss": 0.6909,
      "step": 1564
    },
    {
      "epoch": 1.915544675642595,
      "grad_norm": 0.9480509960415106,
      "learning_rate": 4.801491614128851e-06,
      "loss": 0.6204,
      "step": 1565
    },
    {
      "epoch": 1.916768665850673,
      "grad_norm": 1.2017033704707512,
      "learning_rate": 4.801241265466211e-06,
      "loss": 0.5929,
      "step": 1566
    },
    {
      "epoch": 1.9179926560587517,
      "grad_norm": 1.093453887715438,
      "learning_rate": 4.800990765574507e-06,
      "loss": 0.8205,
      "step": 1567
    },
    {
      "epoch": 1.9192166462668299,
      "grad_norm": 0.5727770073428716,
      "learning_rate": 4.800740114470199e-06,
      "loss": 0.4438,
      "step": 1568
    },
    {
      "epoch": 1.920440636474908,
      "grad_norm": 1.1251755045828478,
      "learning_rate": 4.8004893121697626e-06,
      "loss": 0.4136,
      "step": 1569
    },
    {
      "epoch": 1.9216646266829867,
      "grad_norm": 0.7555333676743492,
      "learning_rate": 4.800238358689676e-06,
      "loss": 0.6418,
      "step": 1570
    },
    {
      "epoch": 1.9228886168910648,
      "grad_norm": 1.2527800001417086,
      "learning_rate": 4.799987254046433e-06,
      "loss": 0.4479,
      "step": 1571
    },
    {
      "epoch": 1.9241126070991432,
      "grad_norm": 1.0645839689689491,
      "learning_rate": 4.799735998256535e-06,
      "loss": 1.0947,
      "step": 1572
    },
    {
      "epoch": 1.9253365973072216,
      "grad_norm": 0.5916297141513027,
      "learning_rate": 4.799484591336494e-06,
      "loss": 0.6436,
      "step": 1573
    },
    {
      "epoch": 1.9265605875152998,
      "grad_norm": 0.6514642287801761,
      "learning_rate": 4.79923303330283e-06,
      "loss": 0.4481,
      "step": 1574
    },
    {
      "epoch": 1.9277845777233782,
      "grad_norm": 0.896296188970787,
      "learning_rate": 4.798981324172075e-06,
      "loss": 0.8211,
      "step": 1575
    },
    {
      "epoch": 1.9290085679314566,
      "grad_norm": 0.7983919089512396,
      "learning_rate": 4.798729463960771e-06,
      "loss": 0.5918,
      "step": 1576
    },
    {
      "epoch": 1.9302325581395348,
      "grad_norm": 1.1981620844381733,
      "learning_rate": 4.79847745268547e-06,
      "loss": 1.2933,
      "step": 1577
    },
    {
      "epoch": 1.9314565483476134,
      "grad_norm": 0.9958421783718234,
      "learning_rate": 4.798225290362731e-06,
      "loss": 0.8313,
      "step": 1578
    },
    {
      "epoch": 1.9326805385556916,
      "grad_norm": 0.795154614038978,
      "learning_rate": 4.797972977009127e-06,
      "loss": 0.723,
      "step": 1579
    },
    {
      "epoch": 1.9339045287637697,
      "grad_norm": 0.8489395754953905,
      "learning_rate": 4.797720512641238e-06,
      "loss": 0.6306,
      "step": 1580
    },
    {
      "epoch": 1.9351285189718483,
      "grad_norm": 0.8163265507577604,
      "learning_rate": 4.797467897275656e-06,
      "loss": 0.5754,
      "step": 1581
    },
    {
      "epoch": 1.9363525091799265,
      "grad_norm": 1.251260743323998,
      "learning_rate": 4.797215130928982e-06,
      "loss": 0.5843,
      "step": 1582
    },
    {
      "epoch": 1.937576499388005,
      "grad_norm": 0.7792404795394278,
      "learning_rate": 4.796962213617826e-06,
      "loss": 0.6807,
      "step": 1583
    },
    {
      "epoch": 1.9388004895960833,
      "grad_norm": 0.9382435937172323,
      "learning_rate": 4.796709145358809e-06,
      "loss": 0.5393,
      "step": 1584
    },
    {
      "epoch": 1.9400244798041615,
      "grad_norm": 0.9484565418253064,
      "learning_rate": 4.796455926168561e-06,
      "loss": 0.5472,
      "step": 1585
    },
    {
      "epoch": 1.9412484700122399,
      "grad_norm": 0.788175646007282,
      "learning_rate": 4.796202556063726e-06,
      "loss": 0.5138,
      "step": 1586
    },
    {
      "epoch": 1.9424724602203183,
      "grad_norm": 0.5167376654052457,
      "learning_rate": 4.79594903506095e-06,
      "loss": 0.4768,
      "step": 1587
    },
    {
      "epoch": 1.9436964504283964,
      "grad_norm": 1.3124482978543974,
      "learning_rate": 4.795695363176897e-06,
      "loss": 0.7975,
      "step": 1588
    },
    {
      "epoch": 1.944920440636475,
      "grad_norm": 1.499050422895394,
      "learning_rate": 4.795441540428235e-06,
      "loss": 1.3848,
      "step": 1589
    },
    {
      "epoch": 1.9461444308445532,
      "grad_norm": 1.5736573751410727,
      "learning_rate": 4.795187566831645e-06,
      "loss": 0.4929,
      "step": 1590
    },
    {
      "epoch": 1.9473684210526314,
      "grad_norm": 0.9282396318941775,
      "learning_rate": 4.794933442403818e-06,
      "loss": 0.5113,
      "step": 1591
    },
    {
      "epoch": 1.94859241126071,
      "grad_norm": 0.5668914444564315,
      "learning_rate": 4.794679167161453e-06,
      "loss": 0.6605,
      "step": 1592
    },
    {
      "epoch": 1.9498164014687882,
      "grad_norm": 0.9413994777064197,
      "learning_rate": 4.794424741121261e-06,
      "loss": 0.7124,
      "step": 1593
    },
    {
      "epoch": 1.9510403916768666,
      "grad_norm": 1.4751485691641153,
      "learning_rate": 4.794170164299962e-06,
      "loss": 0.568,
      "step": 1594
    },
    {
      "epoch": 1.952264381884945,
      "grad_norm": 0.6632059960631201,
      "learning_rate": 4.793915436714285e-06,
      "loss": 0.968,
      "step": 1595
    },
    {
      "epoch": 1.9534883720930232,
      "grad_norm": 0.9436766785972045,
      "learning_rate": 4.793660558380969e-06,
      "loss": 0.6869,
      "step": 1596
    },
    {
      "epoch": 1.9547123623011016,
      "grad_norm": 0.7545899323255287,
      "learning_rate": 4.7934055293167655e-06,
      "loss": 0.6668,
      "step": 1597
    },
    {
      "epoch": 1.95593635250918,
      "grad_norm": 0.796718361455369,
      "learning_rate": 4.793150349538434e-06,
      "loss": 0.922,
      "step": 1598
    },
    {
      "epoch": 1.9571603427172581,
      "grad_norm": 0.8838904072449117,
      "learning_rate": 4.792895019062742e-06,
      "loss": 0.9931,
      "step": 1599
    },
    {
      "epoch": 1.9583843329253368,
      "grad_norm": 1.3610229943698242,
      "learning_rate": 4.792639537906472e-06,
      "loss": 0.9976,
      "step": 1600
    },
    {
      "epoch": 1.959608323133415,
      "grad_norm": 0.873640372990564,
      "learning_rate": 4.79238390608641e-06,
      "loss": 1.0471,
      "step": 1601
    },
    {
      "epoch": 1.960832313341493,
      "grad_norm": 1.106150886064866,
      "learning_rate": 4.792128123619357e-06,
      "loss": 0.8431,
      "step": 1602
    },
    {
      "epoch": 1.9620563035495717,
      "grad_norm": 1.1566886780408674,
      "learning_rate": 4.791872190522122e-06,
      "loss": 0.6454,
      "step": 1603
    },
    {
      "epoch": 1.96328029375765,
      "grad_norm": 0.952404860746821,
      "learning_rate": 4.791616106811523e-06,
      "loss": 0.6135,
      "step": 1604
    },
    {
      "epoch": 1.9645042839657283,
      "grad_norm": 0.6047281057345922,
      "learning_rate": 4.79135987250439e-06,
      "loss": 0.6614,
      "step": 1605
    },
    {
      "epoch": 1.9657282741738067,
      "grad_norm": 1.072351043382434,
      "learning_rate": 4.791103487617562e-06,
      "loss": 0.6474,
      "step": 1606
    },
    {
      "epoch": 1.9669522643818849,
      "grad_norm": 1.4877300163071183,
      "learning_rate": 4.790846952167886e-06,
      "loss": 0.5281,
      "step": 1607
    },
    {
      "epoch": 1.9681762545899633,
      "grad_norm": 0.8475132727790834,
      "learning_rate": 4.790590266172222e-06,
      "loss": 0.6702,
      "step": 1608
    },
    {
      "epoch": 1.9694002447980417,
      "grad_norm": 0.6904125037000479,
      "learning_rate": 4.790333429647438e-06,
      "loss": 0.5328,
      "step": 1609
    },
    {
      "epoch": 1.9706242350061198,
      "grad_norm": 0.5738387721390603,
      "learning_rate": 4.790076442610412e-06,
      "loss": 0.7358,
      "step": 1610
    },
    {
      "epoch": 1.9718482252141984,
      "grad_norm": 0.9940065467747108,
      "learning_rate": 4.789819305078033e-06,
      "loss": 1.1867,
      "step": 1611
    },
    {
      "epoch": 1.9730722154222766,
      "grad_norm": 0.7741968807200302,
      "learning_rate": 4.789562017067199e-06,
      "loss": 0.6888,
      "step": 1612
    },
    {
      "epoch": 1.9742962056303548,
      "grad_norm": 1.0157883415025875,
      "learning_rate": 4.7893045785948165e-06,
      "loss": 0.6599,
      "step": 1613
    },
    {
      "epoch": 1.9755201958384334,
      "grad_norm": 1.365293167549576,
      "learning_rate": 4.789046989677805e-06,
      "loss": 0.6171,
      "step": 1614
    },
    {
      "epoch": 1.9767441860465116,
      "grad_norm": 1.0348273602315656,
      "learning_rate": 4.7887892503330936e-06,
      "loss": 0.9952,
      "step": 1615
    },
    {
      "epoch": 1.97796817625459,
      "grad_norm": 0.5772740264910727,
      "learning_rate": 4.7885313605776166e-06,
      "loss": 0.4676,
      "step": 1616
    },
    {
      "epoch": 1.9791921664626684,
      "grad_norm": 1.1462268746002024,
      "learning_rate": 4.7882733204283236e-06,
      "loss": 1.0766,
      "step": 1617
    },
    {
      "epoch": 1.9804161566707466,
      "grad_norm": 1.3045196276006517,
      "learning_rate": 4.788015129902173e-06,
      "loss": 0.8134,
      "step": 1618
    },
    {
      "epoch": 1.981640146878825,
      "grad_norm": 0.5746352940970983,
      "learning_rate": 4.787756789016129e-06,
      "loss": 0.5776,
      "step": 1619
    },
    {
      "epoch": 1.9828641370869033,
      "grad_norm": 1.2871927748316516,
      "learning_rate": 4.787498297787172e-06,
      "loss": 1.3033,
      "step": 1620
    },
    {
      "epoch": 1.9840881272949815,
      "grad_norm": 0.5250215967251644,
      "learning_rate": 4.7872396562322865e-06,
      "loss": 0.6282,
      "step": 1621
    },
    {
      "epoch": 1.9853121175030601,
      "grad_norm": 0.8117094005468507,
      "learning_rate": 4.786980864368471e-06,
      "loss": 0.4296,
      "step": 1622
    },
    {
      "epoch": 1.9865361077111383,
      "grad_norm": 0.7246604017053855,
      "learning_rate": 4.786721922212733e-06,
      "loss": 0.5986,
      "step": 1623
    },
    {
      "epoch": 1.9877600979192165,
      "grad_norm": 0.8103264306157301,
      "learning_rate": 4.786462829782088e-06,
      "loss": 0.6901,
      "step": 1624
    },
    {
      "epoch": 1.988984088127295,
      "grad_norm": 1.0144325822234537,
      "learning_rate": 4.786203587093562e-06,
      "loss": 1.4275,
      "step": 1625
    },
    {
      "epoch": 1.9902080783353733,
      "grad_norm": 1.1449075940255176,
      "learning_rate": 4.785944194164193e-06,
      "loss": 1.1992,
      "step": 1626
    },
    {
      "epoch": 1.9914320685434517,
      "grad_norm": 0.9127127368532034,
      "learning_rate": 4.785684651011027e-06,
      "loss": 1.5022,
      "step": 1627
    },
    {
      "epoch": 1.99265605875153,
      "grad_norm": 0.6696578934519518,
      "learning_rate": 4.785424957651119e-06,
      "loss": 0.7072,
      "step": 1628
    },
    {
      "epoch": 1.9938800489596082,
      "grad_norm": 1.3549650773850697,
      "learning_rate": 4.785165114101536e-06,
      "loss": 0.8635,
      "step": 1629
    },
    {
      "epoch": 1.9951040391676866,
      "grad_norm": 1.1412482767167145,
      "learning_rate": 4.784905120379355e-06,
      "loss": 0.8452,
      "step": 1630
    },
    {
      "epoch": 1.996328029375765,
      "grad_norm": 1.1973599530072898,
      "learning_rate": 4.78464497650166e-06,
      "loss": 0.5399,
      "step": 1631
    },
    {
      "epoch": 1.9975520195838432,
      "grad_norm": 0.7731379418226847,
      "learning_rate": 4.784384682485548e-06,
      "loss": 0.6433,
      "step": 1632
    },
    {
      "epoch": 1.9987760097919218,
      "grad_norm": 0.7525596065093401,
      "learning_rate": 4.7841242383481235e-06,
      "loss": 0.632,
      "step": 1633
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.030655347898543,
      "learning_rate": 4.783863644106502e-06,
      "loss": 1.2715,
      "step": 1634
    },
    {
      "epoch": 2.001223990208078,
      "grad_norm": 1.3490444287734904,
      "learning_rate": 4.783602899777811e-06,
      "loss": 0.595,
      "step": 1635
    },
    {
      "epoch": 2.002447980416157,
      "grad_norm": 0.7923855008059066,
      "learning_rate": 4.783342005379182e-06,
      "loss": 0.6415,
      "step": 1636
    },
    {
      "epoch": 2.003671970624235,
      "grad_norm": 0.840601862691673,
      "learning_rate": 4.783080960927763e-06,
      "loss": 0.5996,
      "step": 1637
    },
    {
      "epoch": 2.004895960832313,
      "grad_norm": 1.391136439821326,
      "learning_rate": 4.782819766440707e-06,
      "loss": 0.5675,
      "step": 1638
    },
    {
      "epoch": 2.0061199510403918,
      "grad_norm": 0.7652098606946581,
      "learning_rate": 4.78255842193518e-06,
      "loss": 0.9135,
      "step": 1639
    },
    {
      "epoch": 2.00734394124847,
      "grad_norm": 1.013382164786699,
      "learning_rate": 4.782296927428356e-06,
      "loss": 0.9813,
      "step": 1640
    },
    {
      "epoch": 2.0085679314565485,
      "grad_norm": 0.9079413204020284,
      "learning_rate": 4.78203528293742e-06,
      "loss": 0.853,
      "step": 1641
    },
    {
      "epoch": 2.0097919216646267,
      "grad_norm": 0.9123261391193673,
      "learning_rate": 4.7817734884795656e-06,
      "loss": 0.555,
      "step": 1642
    },
    {
      "epoch": 2.011015911872705,
      "grad_norm": 1.3492668717845504,
      "learning_rate": 4.781511544071996e-06,
      "loss": 0.6342,
      "step": 1643
    },
    {
      "epoch": 2.0122399020807835,
      "grad_norm": 1.4964263381362626,
      "learning_rate": 4.781249449731928e-06,
      "loss": 0.508,
      "step": 1644
    },
    {
      "epoch": 2.0134638922888617,
      "grad_norm": 0.5764661455109725,
      "learning_rate": 4.780987205476584e-06,
      "loss": 0.438,
      "step": 1645
    },
    {
      "epoch": 2.01468788249694,
      "grad_norm": 0.5169239546489464,
      "learning_rate": 4.7807248113231966e-06,
      "loss": 0.4727,
      "step": 1646
    },
    {
      "epoch": 2.0159118727050185,
      "grad_norm": 0.7646726079291667,
      "learning_rate": 4.780462267289011e-06,
      "loss": 0.8894,
      "step": 1647
    },
    {
      "epoch": 2.0171358629130967,
      "grad_norm": 1.2525118759256464,
      "learning_rate": 4.78019957339128e-06,
      "loss": 0.5727,
      "step": 1648
    },
    {
      "epoch": 2.018359853121175,
      "grad_norm": 1.689618777322081,
      "learning_rate": 4.779936729647267e-06,
      "loss": 1.0123,
      "step": 1649
    },
    {
      "epoch": 2.0195838433292534,
      "grad_norm": 1.0700814104922753,
      "learning_rate": 4.779673736074245e-06,
      "loss": 0.7087,
      "step": 1650
    },
    {
      "epoch": 2.0208078335373316,
      "grad_norm": 1.1335791592403512,
      "learning_rate": 4.7794105926894965e-06,
      "loss": 0.7027,
      "step": 1651
    },
    {
      "epoch": 2.0220318237454102,
      "grad_norm": 0.9448495370841044,
      "learning_rate": 4.779147299510316e-06,
      "loss": 0.4908,
      "step": 1652
    },
    {
      "epoch": 2.0232558139534884,
      "grad_norm": 0.8304644268133966,
      "learning_rate": 4.7788838565540044e-06,
      "loss": 1.6364,
      "step": 1653
    },
    {
      "epoch": 2.0244798041615666,
      "grad_norm": 1.2690986488399714,
      "learning_rate": 4.778620263837875e-06,
      "loss": 0.3258,
      "step": 1654
    },
    {
      "epoch": 2.025703794369645,
      "grad_norm": 0.7254185937017236,
      "learning_rate": 4.778356521379248e-06,
      "loss": 0.6692,
      "step": 1655
    },
    {
      "epoch": 2.0269277845777234,
      "grad_norm": 0.8991220592918276,
      "learning_rate": 4.778092629195459e-06,
      "loss": 0.6414,
      "step": 1656
    },
    {
      "epoch": 2.0281517747858016,
      "grad_norm": 1.0098488404769435,
      "learning_rate": 4.777828587303849e-06,
      "loss": 0.6886,
      "step": 1657
    },
    {
      "epoch": 2.02937576499388,
      "grad_norm": 0.8998552000188978,
      "learning_rate": 4.777564395721768e-06,
      "loss": 1.29,
      "step": 1658
    },
    {
      "epoch": 2.0305997552019583,
      "grad_norm": 0.7900915476144522,
      "learning_rate": 4.77730005446658e-06,
      "loss": 1.0966,
      "step": 1659
    },
    {
      "epoch": 2.0318237454100365,
      "grad_norm": 1.274376228013902,
      "learning_rate": 4.777035563555655e-06,
      "loss": 0.5093,
      "step": 1660
    },
    {
      "epoch": 2.033047735618115,
      "grad_norm": 0.8947602390354029,
      "learning_rate": 4.776770923006376e-06,
      "loss": 1.2073,
      "step": 1661
    },
    {
      "epoch": 2.0342717258261933,
      "grad_norm": 0.6396272574357137,
      "learning_rate": 4.776506132836131e-06,
      "loss": 0.6155,
      "step": 1662
    },
    {
      "epoch": 2.035495716034272,
      "grad_norm": 0.5786500263985648,
      "learning_rate": 4.7762411930623255e-06,
      "loss": 0.463,
      "step": 1663
    },
    {
      "epoch": 2.03671970624235,
      "grad_norm": 1.1313863117217278,
      "learning_rate": 4.775976103702367e-06,
      "loss": 1.213,
      "step": 1664
    },
    {
      "epoch": 2.0379436964504283,
      "grad_norm": 0.5733360418805645,
      "learning_rate": 4.775710864773677e-06,
      "loss": 0.6554,
      "step": 1665
    },
    {
      "epoch": 2.039167686658507,
      "grad_norm": 0.7735539350818456,
      "learning_rate": 4.775445476293688e-06,
      "loss": 0.6618,
      "step": 1666
    },
    {
      "epoch": 2.040391676866585,
      "grad_norm": 0.9418886170050217,
      "learning_rate": 4.775179938279836e-06,
      "loss": 0.6685,
      "step": 1667
    },
    {
      "epoch": 2.0416156670746632,
      "grad_norm": 0.6861068909865007,
      "learning_rate": 4.774914250749575e-06,
      "loss": 0.4329,
      "step": 1668
    },
    {
      "epoch": 2.042839657282742,
      "grad_norm": 1.5389060247313142,
      "learning_rate": 4.774648413720364e-06,
      "loss": 1.013,
      "step": 1669
    },
    {
      "epoch": 2.04406364749082,
      "grad_norm": 1.5102522542072925,
      "learning_rate": 4.774382427209672e-06,
      "loss": 0.429,
      "step": 1670
    },
    {
      "epoch": 2.045287637698898,
      "grad_norm": 1.025940607447305,
      "learning_rate": 4.77411629123498e-06,
      "loss": 1.2591,
      "step": 1671
    },
    {
      "epoch": 2.046511627906977,
      "grad_norm": 0.711185166481529,
      "learning_rate": 4.773850005813776e-06,
      "loss": 0.6222,
      "step": 1672
    },
    {
      "epoch": 2.047735618115055,
      "grad_norm": 0.8524740497303297,
      "learning_rate": 4.77358357096356e-06,
      "loss": 0.7012,
      "step": 1673
    },
    {
      "epoch": 2.0489596083231336,
      "grad_norm": 0.9624710180554567,
      "learning_rate": 4.7733169867018414e-06,
      "loss": 1.6047,
      "step": 1674
    },
    {
      "epoch": 2.050183598531212,
      "grad_norm": 1.2788530532638447,
      "learning_rate": 4.773050253046139e-06,
      "loss": 0.6269,
      "step": 1675
    },
    {
      "epoch": 2.05140758873929,
      "grad_norm": 1.0217457862202248,
      "learning_rate": 4.772783370013981e-06,
      "loss": 1.8886,
      "step": 1676
    },
    {
      "epoch": 2.0526315789473686,
      "grad_norm": 1.4765882368011554,
      "learning_rate": 4.772516337622907e-06,
      "loss": 0.6272,
      "step": 1677
    },
    {
      "epoch": 2.0538555691554468,
      "grad_norm": 0.5717420280638605,
      "learning_rate": 4.772249155890464e-06,
      "loss": 0.5839,
      "step": 1678
    },
    {
      "epoch": 2.055079559363525,
      "grad_norm": 0.8855425505193519,
      "learning_rate": 4.771981824834211e-06,
      "loss": 0.6591,
      "step": 1679
    },
    {
      "epoch": 2.0563035495716036,
      "grad_norm": 1.3007485590081038,
      "learning_rate": 4.771714344471716e-06,
      "loss": 1.2758,
      "step": 1680
    },
    {
      "epoch": 2.0575275397796817,
      "grad_norm": 0.8867001622432548,
      "learning_rate": 4.771446714820558e-06,
      "loss": 0.4784,
      "step": 1681
    },
    {
      "epoch": 2.05875152998776,
      "grad_norm": 0.8220415597183386,
      "learning_rate": 4.771178935898322e-06,
      "loss": 0.5838,
      "step": 1682
    },
    {
      "epoch": 2.0599755201958385,
      "grad_norm": 1.1879024691303384,
      "learning_rate": 4.770911007722607e-06,
      "loss": 0.6057,
      "step": 1683
    },
    {
      "epoch": 2.0611995104039167,
      "grad_norm": 0.8032418789970761,
      "learning_rate": 4.7706429303110205e-06,
      "loss": 0.6386,
      "step": 1684
    },
    {
      "epoch": 2.0624235006119953,
      "grad_norm": 1.3515686114454204,
      "learning_rate": 4.770374703681178e-06,
      "loss": 0.9821,
      "step": 1685
    },
    {
      "epoch": 2.0636474908200735,
      "grad_norm": 1.643732420666673,
      "learning_rate": 4.770106327850709e-06,
      "loss": 0.4771,
      "step": 1686
    },
    {
      "epoch": 2.0648714810281517,
      "grad_norm": 0.8870433627561697,
      "learning_rate": 4.769837802837248e-06,
      "loss": 0.5062,
      "step": 1687
    },
    {
      "epoch": 2.0660954712362303,
      "grad_norm": 0.8277681308651833,
      "learning_rate": 4.769569128658443e-06,
      "loss": 0.875,
      "step": 1688
    },
    {
      "epoch": 2.0673194614443084,
      "grad_norm": 0.9305066892829389,
      "learning_rate": 4.769300305331949e-06,
      "loss": 0.6364,
      "step": 1689
    },
    {
      "epoch": 2.0685434516523866,
      "grad_norm": 1.3562157517338775,
      "learning_rate": 4.7690313328754316e-06,
      "loss": 0.5846,
      "step": 1690
    },
    {
      "epoch": 2.0697674418604652,
      "grad_norm": 1.1746921621008894,
      "learning_rate": 4.768762211306568e-06,
      "loss": 1.1463,
      "step": 1691
    },
    {
      "epoch": 2.0709914320685434,
      "grad_norm": 0.6298041034646288,
      "learning_rate": 4.768492940643043e-06,
      "loss": 0.6953,
      "step": 1692
    },
    {
      "epoch": 2.0722154222766216,
      "grad_norm": 0.681923862205774,
      "learning_rate": 4.768223520902554e-06,
      "loss": 0.7025,
      "step": 1693
    },
    {
      "epoch": 2.0734394124847,
      "grad_norm": 0.5175011734226519,
      "learning_rate": 4.767953952102804e-06,
      "loss": 0.4999,
      "step": 1694
    },
    {
      "epoch": 2.0746634026927784,
      "grad_norm": 0.9013754959204043,
      "learning_rate": 4.767684234261509e-06,
      "loss": 0.3814,
      "step": 1695
    },
    {
      "epoch": 2.075887392900857,
      "grad_norm": 0.8899917581311273,
      "learning_rate": 4.767414367396393e-06,
      "loss": 1.1774,
      "step": 1696
    },
    {
      "epoch": 2.077111383108935,
      "grad_norm": 0.776151371660974,
      "learning_rate": 4.767144351525192e-06,
      "loss": 0.6757,
      "step": 1697
    },
    {
      "epoch": 2.0783353733170133,
      "grad_norm": 0.7296713265867719,
      "learning_rate": 4.7668741866656495e-06,
      "loss": 0.6192,
      "step": 1698
    },
    {
      "epoch": 2.079559363525092,
      "grad_norm": 0.6413799606073156,
      "learning_rate": 4.76660387283552e-06,
      "loss": 0.5596,
      "step": 1699
    },
    {
      "epoch": 2.08078335373317,
      "grad_norm": 1.2070980034873404,
      "learning_rate": 4.766333410052568e-06,
      "loss": 0.6761,
      "step": 1700
    },
    {
      "epoch": 2.0820073439412483,
      "grad_norm": 0.8205271201463079,
      "learning_rate": 4.766062798334566e-06,
      "loss": 0.7142,
      "step": 1701
    },
    {
      "epoch": 2.083231334149327,
      "grad_norm": 0.849099388361979,
      "learning_rate": 4.765792037699299e-06,
      "loss": 0.5057,
      "step": 1702
    },
    {
      "epoch": 2.084455324357405,
      "grad_norm": 1.4807478302959003,
      "learning_rate": 4.765521128164561e-06,
      "loss": 1.26,
      "step": 1703
    },
    {
      "epoch": 2.0856793145654833,
      "grad_norm": 0.7557006985387202,
      "learning_rate": 4.765250069748153e-06,
      "loss": 0.6342,
      "step": 1704
    },
    {
      "epoch": 2.086903304773562,
      "grad_norm": 0.9478472773592295,
      "learning_rate": 4.764978862467889e-06,
      "loss": 1.5716,
      "step": 1705
    },
    {
      "epoch": 2.08812729498164,
      "grad_norm": 0.8707528682370116,
      "learning_rate": 4.764707506341592e-06,
      "loss": 1.052,
      "step": 1706
    },
    {
      "epoch": 2.0893512851897187,
      "grad_norm": 0.8855340741221631,
      "learning_rate": 4.764436001387094e-06,
      "loss": 0.7315,
      "step": 1707
    },
    {
      "epoch": 2.090575275397797,
      "grad_norm": 0.9717173172079296,
      "learning_rate": 4.764164347622237e-06,
      "loss": 0.7863,
      "step": 1708
    },
    {
      "epoch": 2.091799265605875,
      "grad_norm": 0.9952359706463395,
      "learning_rate": 4.763892545064874e-06,
      "loss": 0.8258,
      "step": 1709
    },
    {
      "epoch": 2.0930232558139537,
      "grad_norm": 0.6909781689289004,
      "learning_rate": 4.763620593732867e-06,
      "loss": 0.8363,
      "step": 1710
    },
    {
      "epoch": 2.094247246022032,
      "grad_norm": 1.226308147176626,
      "learning_rate": 4.763348493644087e-06,
      "loss": 0.5291,
      "step": 1711
    },
    {
      "epoch": 2.09547123623011,
      "grad_norm": 0.8356326441068579,
      "learning_rate": 4.763076244816414e-06,
      "loss": 0.2731,
      "step": 1712
    },
    {
      "epoch": 2.0966952264381886,
      "grad_norm": 0.8618034249405533,
      "learning_rate": 4.762803847267743e-06,
      "loss": 1.2402,
      "step": 1713
    },
    {
      "epoch": 2.097919216646267,
      "grad_norm": 1.6486283167088436,
      "learning_rate": 4.762531301015972e-06,
      "loss": 0.5007,
      "step": 1714
    },
    {
      "epoch": 2.099143206854345,
      "grad_norm": 0.9386617708445711,
      "learning_rate": 4.762258606079012e-06,
      "loss": 1.1188,
      "step": 1715
    },
    {
      "epoch": 2.1003671970624236,
      "grad_norm": 0.978306173900155,
      "learning_rate": 4.761985762474784e-06,
      "loss": 0.6813,
      "step": 1716
    },
    {
      "epoch": 2.1015911872705018,
      "grad_norm": 0.7897540880502508,
      "learning_rate": 4.761712770221219e-06,
      "loss": 0.4696,
      "step": 1717
    },
    {
      "epoch": 2.1028151774785804,
      "grad_norm": 1.2469524536855359,
      "learning_rate": 4.7614396293362555e-06,
      "loss": 0.6841,
      "step": 1718
    },
    {
      "epoch": 2.1040391676866586,
      "grad_norm": 0.7982074074636537,
      "learning_rate": 4.761166339837844e-06,
      "loss": 1.2923,
      "step": 1719
    },
    {
      "epoch": 2.1052631578947367,
      "grad_norm": 1.16803009625026,
      "learning_rate": 4.760892901743944e-06,
      "loss": 0.6524,
      "step": 1720
    },
    {
      "epoch": 2.1064871481028153,
      "grad_norm": 1.5263695940141289,
      "learning_rate": 4.760619315072525e-06,
      "loss": 0.4485,
      "step": 1721
    },
    {
      "epoch": 2.1077111383108935,
      "grad_norm": 0.6229222575548571,
      "learning_rate": 4.7603455798415664e-06,
      "loss": 0.6148,
      "step": 1722
    },
    {
      "epoch": 2.1089351285189717,
      "grad_norm": 1.1816401904189082,
      "learning_rate": 4.760071696069057e-06,
      "loss": 0.5042,
      "step": 1723
    },
    {
      "epoch": 2.1101591187270503,
      "grad_norm": 1.3656047303957681,
      "learning_rate": 4.759797663772995e-06,
      "loss": 0.9532,
      "step": 1724
    },
    {
      "epoch": 2.1113831089351285,
      "grad_norm": 0.9026007881199491,
      "learning_rate": 4.759523482971389e-06,
      "loss": 0.6274,
      "step": 1725
    },
    {
      "epoch": 2.1126070991432067,
      "grad_norm": 0.6925339490668826,
      "learning_rate": 4.759249153682256e-06,
      "loss": 0.9589,
      "step": 1726
    },
    {
      "epoch": 2.1138310893512853,
      "grad_norm": 1.092159800471345,
      "learning_rate": 4.758974675923625e-06,
      "loss": 1.0809,
      "step": 1727
    },
    {
      "epoch": 2.1150550795593634,
      "grad_norm": 0.9609230031064179,
      "learning_rate": 4.758700049713535e-06,
      "loss": 1.6841,
      "step": 1728
    },
    {
      "epoch": 2.116279069767442,
      "grad_norm": 1.3613335228678929,
      "learning_rate": 4.758425275070032e-06,
      "loss": 0.5831,
      "step": 1729
    },
    {
      "epoch": 2.1175030599755202,
      "grad_norm": 1.4067945891132145,
      "learning_rate": 4.758150352011172e-06,
      "loss": 0.4991,
      "step": 1730
    },
    {
      "epoch": 2.1187270501835984,
      "grad_norm": 0.5826754910591225,
      "learning_rate": 4.757875280555024e-06,
      "loss": 0.5654,
      "step": 1731
    },
    {
      "epoch": 2.119951040391677,
      "grad_norm": 0.6713054679309671,
      "learning_rate": 4.757600060719665e-06,
      "loss": 0.5381,
      "step": 1732
    },
    {
      "epoch": 2.121175030599755,
      "grad_norm": 0.6603022635318611,
      "learning_rate": 4.757324692523179e-06,
      "loss": 0.5967,
      "step": 1733
    },
    {
      "epoch": 2.1223990208078334,
      "grad_norm": 1.739955682879349,
      "learning_rate": 4.757049175983663e-06,
      "loss": 0.6657,
      "step": 1734
    },
    {
      "epoch": 2.123623011015912,
      "grad_norm": 1.4928719853892538,
      "learning_rate": 4.756773511119225e-06,
      "loss": 0.5505,
      "step": 1735
    },
    {
      "epoch": 2.12484700122399,
      "grad_norm": 1.0014239746136204,
      "learning_rate": 4.756497697947978e-06,
      "loss": 0.5667,
      "step": 1736
    },
    {
      "epoch": 2.1260709914320683,
      "grad_norm": 1.4587697882417965,
      "learning_rate": 4.756221736488048e-06,
      "loss": 0.3107,
      "step": 1737
    },
    {
      "epoch": 2.127294981640147,
      "grad_norm": 0.9744809671776056,
      "learning_rate": 4.755945626757572e-06,
      "loss": 0.4052,
      "step": 1738
    },
    {
      "epoch": 2.128518971848225,
      "grad_norm": 1.0826318594087034,
      "learning_rate": 4.755669368774693e-06,
      "loss": 0.3694,
      "step": 1739
    },
    {
      "epoch": 2.1297429620563038,
      "grad_norm": 1.0485519413956328,
      "learning_rate": 4.755392962557565e-06,
      "loss": 1.2753,
      "step": 1740
    },
    {
      "epoch": 2.130966952264382,
      "grad_norm": 0.5297184378343576,
      "learning_rate": 4.755116408124355e-06,
      "loss": 0.4809,
      "step": 1741
    },
    {
      "epoch": 2.13219094247246,
      "grad_norm": 1.5161854409880913,
      "learning_rate": 4.754839705493236e-06,
      "loss": 0.4472,
      "step": 1742
    },
    {
      "epoch": 2.1334149326805387,
      "grad_norm": 0.9053565036262434,
      "learning_rate": 4.75456285468239e-06,
      "loss": 0.3276,
      "step": 1743
    },
    {
      "epoch": 2.134638922888617,
      "grad_norm": 1.0484293334816646,
      "learning_rate": 4.754285855710014e-06,
      "loss": 1.4149,
      "step": 1744
    },
    {
      "epoch": 2.135862913096695,
      "grad_norm": 0.7804640642118915,
      "learning_rate": 4.754008708594308e-06,
      "loss": 0.6731,
      "step": 1745
    },
    {
      "epoch": 2.1370869033047737,
      "grad_norm": 1.2699250970577678,
      "learning_rate": 4.753731413353488e-06,
      "loss": 0.6442,
      "step": 1746
    },
    {
      "epoch": 2.138310893512852,
      "grad_norm": 0.7752691507762891,
      "learning_rate": 4.753453970005776e-06,
      "loss": 0.9262,
      "step": 1747
    },
    {
      "epoch": 2.13953488372093,
      "grad_norm": 0.9047863492717602,
      "learning_rate": 4.753176378569403e-06,
      "loss": 0.724,
      "step": 1748
    },
    {
      "epoch": 2.1407588739290087,
      "grad_norm": 0.815756485854861,
      "learning_rate": 4.752898639062612e-06,
      "loss": 0.5425,
      "step": 1749
    },
    {
      "epoch": 2.141982864137087,
      "grad_norm": 0.7466341475371909,
      "learning_rate": 4.752620751503656e-06,
      "loss": 0.46,
      "step": 1750
    },
    {
      "epoch": 2.1432068543451654,
      "grad_norm": 0.8330271685404932,
      "learning_rate": 4.752342715910796e-06,
      "loss": 0.344,
      "step": 1751
    },
    {
      "epoch": 2.1444308445532436,
      "grad_norm": 1.295484219311035,
      "learning_rate": 4.752064532302303e-06,
      "loss": 0.5263,
      "step": 1752
    },
    {
      "epoch": 2.145654834761322,
      "grad_norm": 0.7708465461630388,
      "learning_rate": 4.751786200696459e-06,
      "loss": 0.6004,
      "step": 1753
    },
    {
      "epoch": 2.1468788249694004,
      "grad_norm": 0.5625380049418784,
      "learning_rate": 4.751507721111555e-06,
      "loss": 0.5671,
      "step": 1754
    },
    {
      "epoch": 2.1481028151774786,
      "grad_norm": 0.7821827935292218,
      "learning_rate": 4.751229093565891e-06,
      "loss": 0.887,
      "step": 1755
    },
    {
      "epoch": 2.1493268053855568,
      "grad_norm": 0.6416714853485103,
      "learning_rate": 4.750950318077778e-06,
      "loss": 0.5931,
      "step": 1756
    },
    {
      "epoch": 2.1505507955936354,
      "grad_norm": 0.8894509783296423,
      "learning_rate": 4.750671394665536e-06,
      "loss": 0.6835,
      "step": 1757
    },
    {
      "epoch": 2.1517747858017136,
      "grad_norm": 0.9435729025265643,
      "learning_rate": 4.750392323347494e-06,
      "loss": 1.4662,
      "step": 1758
    },
    {
      "epoch": 2.1529987760097917,
      "grad_norm": 1.174664139125529,
      "learning_rate": 4.750113104141992e-06,
      "loss": 0.4315,
      "step": 1759
    },
    {
      "epoch": 2.1542227662178703,
      "grad_norm": 1.199311624186637,
      "learning_rate": 4.74983373706738e-06,
      "loss": 0.473,
      "step": 1760
    },
    {
      "epoch": 2.1554467564259485,
      "grad_norm": 1.5508949010600261,
      "learning_rate": 4.7495542221420165e-06,
      "loss": 1.3696,
      "step": 1761
    },
    {
      "epoch": 2.1566707466340267,
      "grad_norm": 1.4530554672289604,
      "learning_rate": 4.749274559384269e-06,
      "loss": 0.9923,
      "step": 1762
    },
    {
      "epoch": 2.1578947368421053,
      "grad_norm": 1.2894182484962242,
      "learning_rate": 4.748994748812518e-06,
      "loss": 1.2662,
      "step": 1763
    },
    {
      "epoch": 2.1591187270501835,
      "grad_norm": 1.7444268844262065,
      "learning_rate": 4.74871479044515e-06,
      "loss": 0.453,
      "step": 1764
    },
    {
      "epoch": 2.160342717258262,
      "grad_norm": 1.22194839718637,
      "learning_rate": 4.748434684300564e-06,
      "loss": 0.5124,
      "step": 1765
    },
    {
      "epoch": 2.1615667074663403,
      "grad_norm": 0.8780388823093234,
      "learning_rate": 4.748154430397166e-06,
      "loss": 0.5523,
      "step": 1766
    },
    {
      "epoch": 2.1627906976744184,
      "grad_norm": 1.2301432478643375,
      "learning_rate": 4.747874028753375e-06,
      "loss": 0.6919,
      "step": 1767
    },
    {
      "epoch": 2.164014687882497,
      "grad_norm": 1.3219349160942884,
      "learning_rate": 4.7475934793876165e-06,
      "loss": 0.9689,
      "step": 1768
    },
    {
      "epoch": 2.1652386780905752,
      "grad_norm": 1.2618463601765646,
      "learning_rate": 4.747312782318329e-06,
      "loss": 0.6402,
      "step": 1769
    },
    {
      "epoch": 2.1664626682986534,
      "grad_norm": 1.0022874580219454,
      "learning_rate": 4.747031937563956e-06,
      "loss": 0.4487,
      "step": 1770
    },
    {
      "epoch": 2.167686658506732,
      "grad_norm": 0.9257859139413773,
      "learning_rate": 4.746750945142956e-06,
      "loss": 0.366,
      "step": 1771
    },
    {
      "epoch": 2.16891064871481,
      "grad_norm": 0.6704583104659286,
      "learning_rate": 4.746469805073794e-06,
      "loss": 0.6701,
      "step": 1772
    },
    {
      "epoch": 2.170134638922889,
      "grad_norm": 1.1285864852383078,
      "learning_rate": 4.746188517374946e-06,
      "loss": 0.6736,
      "step": 1773
    },
    {
      "epoch": 2.171358629130967,
      "grad_norm": 0.8850309489675169,
      "learning_rate": 4.745907082064897e-06,
      "loss": 1.2971,
      "step": 1774
    },
    {
      "epoch": 2.172582619339045,
      "grad_norm": 1.1156511712503216,
      "learning_rate": 4.74562549916214e-06,
      "loss": 0.9061,
      "step": 1775
    },
    {
      "epoch": 2.173806609547124,
      "grad_norm": 0.8329642642581615,
      "learning_rate": 4.745343768685182e-06,
      "loss": 0.547,
      "step": 1776
    },
    {
      "epoch": 2.175030599755202,
      "grad_norm": 1.5172666336688354,
      "learning_rate": 4.745061890652537e-06,
      "loss": 0.62,
      "step": 1777
    },
    {
      "epoch": 2.17625458996328,
      "grad_norm": 0.9027662793665578,
      "learning_rate": 4.744779865082728e-06,
      "loss": 0.7572,
      "step": 1778
    },
    {
      "epoch": 2.1774785801713588,
      "grad_norm": 0.9873706669697835,
      "learning_rate": 4.744497691994289e-06,
      "loss": 0.7273,
      "step": 1779
    },
    {
      "epoch": 2.178702570379437,
      "grad_norm": 1.303181055798484,
      "learning_rate": 4.744215371405764e-06,
      "loss": 0.5969,
      "step": 1780
    },
    {
      "epoch": 2.179926560587515,
      "grad_norm": 1.3134831564590304,
      "learning_rate": 4.7439329033357055e-06,
      "loss": 1.0466,
      "step": 1781
    },
    {
      "epoch": 2.1811505507955937,
      "grad_norm": 0.8101837763511736,
      "learning_rate": 4.743650287802676e-06,
      "loss": 0.8579,
      "step": 1782
    },
    {
      "epoch": 2.182374541003672,
      "grad_norm": 1.0879644823716874,
      "learning_rate": 4.743367524825249e-06,
      "loss": 1.0159,
      "step": 1783
    },
    {
      "epoch": 2.18359853121175,
      "grad_norm": 1.1428176183079704,
      "learning_rate": 4.743084614422004e-06,
      "loss": 0.4788,
      "step": 1784
    },
    {
      "epoch": 2.1848225214198287,
      "grad_norm": 1.577010398002274,
      "learning_rate": 4.7428015566115365e-06,
      "loss": 0.722,
      "step": 1785
    },
    {
      "epoch": 2.186046511627907,
      "grad_norm": 0.5479610683692664,
      "learning_rate": 4.742518351412446e-06,
      "loss": 0.4881,
      "step": 1786
    },
    {
      "epoch": 2.1872705018359855,
      "grad_norm": 1.0831106431688409,
      "learning_rate": 4.7422349988433435e-06,
      "loss": 1.0277,
      "step": 1787
    },
    {
      "epoch": 2.1884944920440637,
      "grad_norm": 0.8137059457381444,
      "learning_rate": 4.7419514989228506e-06,
      "loss": 0.6348,
      "step": 1788
    },
    {
      "epoch": 2.189718482252142,
      "grad_norm": 1.109160614594669,
      "learning_rate": 4.7416678516695976e-06,
      "loss": 1.0812,
      "step": 1789
    },
    {
      "epoch": 2.1909424724602204,
      "grad_norm": 1.5617374900405647,
      "learning_rate": 4.741384057102224e-06,
      "loss": 0.6711,
      "step": 1790
    },
    {
      "epoch": 2.1921664626682986,
      "grad_norm": 0.9958576810548486,
      "learning_rate": 4.741100115239382e-06,
      "loss": 1.1966,
      "step": 1791
    },
    {
      "epoch": 2.193390452876377,
      "grad_norm": 1.3368785616852292,
      "learning_rate": 4.740816026099728e-06,
      "loss": 0.5177,
      "step": 1792
    },
    {
      "epoch": 2.1946144430844554,
      "grad_norm": 1.2459621065853588,
      "learning_rate": 4.740531789701934e-06,
      "loss": 0.5726,
      "step": 1793
    },
    {
      "epoch": 2.1958384332925336,
      "grad_norm": 1.1450589630973276,
      "learning_rate": 4.740247406064678e-06,
      "loss": 0.651,
      "step": 1794
    },
    {
      "epoch": 2.197062423500612,
      "grad_norm": 1.0751465594882477,
      "learning_rate": 4.7399628752066486e-06,
      "loss": 0.4384,
      "step": 1795
    },
    {
      "epoch": 2.1982864137086904,
      "grad_norm": 0.8376073995816036,
      "learning_rate": 4.739678197146544e-06,
      "loss": 0.5452,
      "step": 1796
    },
    {
      "epoch": 2.1995104039167686,
      "grad_norm": 1.0701691025973727,
      "learning_rate": 4.7393933719030715e-06,
      "loss": 0.3894,
      "step": 1797
    },
    {
      "epoch": 2.200734394124847,
      "grad_norm": 0.9215621066118259,
      "learning_rate": 4.739108399494951e-06,
      "loss": 0.8002,
      "step": 1798
    },
    {
      "epoch": 2.2019583843329253,
      "grad_norm": 0.6412729534694632,
      "learning_rate": 4.738823279940907e-06,
      "loss": 0.6726,
      "step": 1799
    },
    {
      "epoch": 2.2031823745410035,
      "grad_norm": 0.7004628880447266,
      "learning_rate": 4.7385380132596785e-06,
      "loss": 0.551,
      "step": 1800
    },
    {
      "epoch": 2.204406364749082,
      "grad_norm": 0.9262003534126843,
      "learning_rate": 4.7382525994700115e-06,
      "loss": 0.8,
      "step": 1801
    },
    {
      "epoch": 2.2056303549571603,
      "grad_norm": 1.0710788151352677,
      "learning_rate": 4.737967038590662e-06,
      "loss": 0.4904,
      "step": 1802
    },
    {
      "epoch": 2.2068543451652385,
      "grad_norm": 0.7062019626708327,
      "learning_rate": 4.737681330640397e-06,
      "loss": 0.6984,
      "step": 1803
    },
    {
      "epoch": 2.208078335373317,
      "grad_norm": 1.0565224955452448,
      "learning_rate": 4.737395475637991e-06,
      "loss": 0.592,
      "step": 1804
    },
    {
      "epoch": 2.2093023255813953,
      "grad_norm": 0.8346772761359592,
      "learning_rate": 4.73710947360223e-06,
      "loss": 0.8892,
      "step": 1805
    },
    {
      "epoch": 2.2105263157894735,
      "grad_norm": 0.7791482841512478,
      "learning_rate": 4.736823324551909e-06,
      "loss": 0.9354,
      "step": 1806
    },
    {
      "epoch": 2.211750305997552,
      "grad_norm": 0.728324620056421,
      "learning_rate": 4.736537028505832e-06,
      "loss": 0.4012,
      "step": 1807
    },
    {
      "epoch": 2.2129742962056302,
      "grad_norm": 0.9839303655234137,
      "learning_rate": 4.736250585482815e-06,
      "loss": 0.555,
      "step": 1808
    },
    {
      "epoch": 2.214198286413709,
      "grad_norm": 0.9180249157922745,
      "learning_rate": 4.735963995501681e-06,
      "loss": 0.4999,
      "step": 1809
    },
    {
      "epoch": 2.215422276621787,
      "grad_norm": 1.6989421209791185,
      "learning_rate": 4.7356772585812625e-06,
      "loss": 0.6278,
      "step": 1810
    },
    {
      "epoch": 2.216646266829865,
      "grad_norm": 0.9378942900986411,
      "learning_rate": 4.735390374740404e-06,
      "loss": 1.3353,
      "step": 1811
    },
    {
      "epoch": 2.217870257037944,
      "grad_norm": 1.0856616670887254,
      "learning_rate": 4.735103343997958e-06,
      "loss": 1.4561,
      "step": 1812
    },
    {
      "epoch": 2.219094247246022,
      "grad_norm": 0.8746120336894221,
      "learning_rate": 4.734816166372788e-06,
      "loss": 0.5586,
      "step": 1813
    },
    {
      "epoch": 2.2203182374541,
      "grad_norm": 1.065969943422191,
      "learning_rate": 4.734528841883765e-06,
      "loss": 1.2255,
      "step": 1814
    },
    {
      "epoch": 2.221542227662179,
      "grad_norm": 1.0522994747609946,
      "learning_rate": 4.734241370549771e-06,
      "loss": 0.425,
      "step": 1815
    },
    {
      "epoch": 2.222766217870257,
      "grad_norm": 1.3282550160336022,
      "learning_rate": 4.7339537523896985e-06,
      "loss": 0.5816,
      "step": 1816
    },
    {
      "epoch": 2.2239902080783356,
      "grad_norm": 1.2656975598846336,
      "learning_rate": 4.733665987422449e-06,
      "loss": 0.6368,
      "step": 1817
    },
    {
      "epoch": 2.2252141982864138,
      "grad_norm": 1.0060139398567456,
      "learning_rate": 4.733378075666931e-06,
      "loss": 0.7547,
      "step": 1818
    },
    {
      "epoch": 2.226438188494492,
      "grad_norm": 0.7553384049883551,
      "learning_rate": 4.733090017142068e-06,
      "loss": 0.4207,
      "step": 1819
    },
    {
      "epoch": 2.2276621787025706,
      "grad_norm": 1.3278405854798798,
      "learning_rate": 4.732801811866789e-06,
      "loss": 0.4105,
      "step": 1820
    },
    {
      "epoch": 2.2288861689106487,
      "grad_norm": 1.2089961065140573,
      "learning_rate": 4.732513459860032e-06,
      "loss": 1.0333,
      "step": 1821
    },
    {
      "epoch": 2.230110159118727,
      "grad_norm": 0.8469317936601658,
      "learning_rate": 4.732224961140748e-06,
      "loss": 0.7319,
      "step": 1822
    },
    {
      "epoch": 2.2313341493268055,
      "grad_norm": 2.211180444328651,
      "learning_rate": 4.731936315727896e-06,
      "loss": 0.4932,
      "step": 1823
    },
    {
      "epoch": 2.2325581395348837,
      "grad_norm": 0.7674983221769965,
      "learning_rate": 4.731647523640446e-06,
      "loss": 0.4976,
      "step": 1824
    },
    {
      "epoch": 2.233782129742962,
      "grad_norm": 0.7409032210537124,
      "learning_rate": 4.7313585848973745e-06,
      "loss": 0.6004,
      "step": 1825
    },
    {
      "epoch": 2.2350061199510405,
      "grad_norm": 0.9638518210593149,
      "learning_rate": 4.73106949951767e-06,
      "loss": 0.5089,
      "step": 1826
    },
    {
      "epoch": 2.2362301101591187,
      "grad_norm": 0.9509339026916247,
      "learning_rate": 4.73078026752033e-06,
      "loss": 0.5697,
      "step": 1827
    },
    {
      "epoch": 2.237454100367197,
      "grad_norm": 0.9738475333198687,
      "learning_rate": 4.730490888924361e-06,
      "loss": 1.0584,
      "step": 1828
    },
    {
      "epoch": 2.2386780905752754,
      "grad_norm": 1.4582233294507194,
      "learning_rate": 4.730201363748782e-06,
      "loss": 1.084,
      "step": 1829
    },
    {
      "epoch": 2.2399020807833536,
      "grad_norm": 1.7042668918675679,
      "learning_rate": 4.729911692012617e-06,
      "loss": 0.5606,
      "step": 1830
    },
    {
      "epoch": 2.2411260709914322,
      "grad_norm": 0.7581074151542866,
      "learning_rate": 4.729621873734906e-06,
      "loss": 0.5963,
      "step": 1831
    },
    {
      "epoch": 2.2423500611995104,
      "grad_norm": 1.3360313786455533,
      "learning_rate": 4.72933190893469e-06,
      "loss": 0.4877,
      "step": 1832
    },
    {
      "epoch": 2.2435740514075886,
      "grad_norm": 0.7184313254117661,
      "learning_rate": 4.729041797631027e-06,
      "loss": 0.7677,
      "step": 1833
    },
    {
      "epoch": 2.244798041615667,
      "grad_norm": 1.1645894847080243,
      "learning_rate": 4.7287515398429824e-06,
      "loss": 0.5799,
      "step": 1834
    },
    {
      "epoch": 2.2460220318237454,
      "grad_norm": 0.8479011547862644,
      "learning_rate": 4.72846113558963e-06,
      "loss": 0.6823,
      "step": 1835
    },
    {
      "epoch": 2.2472460220318236,
      "grad_norm": 0.8898998548033978,
      "learning_rate": 4.728170584890054e-06,
      "loss": 0.6253,
      "step": 1836
    },
    {
      "epoch": 2.248470012239902,
      "grad_norm": 0.4561413170871365,
      "learning_rate": 4.727879887763348e-06,
      "loss": 0.2452,
      "step": 1837
    },
    {
      "epoch": 2.2496940024479803,
      "grad_norm": 1.3255084834600868,
      "learning_rate": 4.7275890442286175e-06,
      "loss": 0.6177,
      "step": 1838
    },
    {
      "epoch": 2.250917992656059,
      "grad_norm": 1.1361557952232664,
      "learning_rate": 4.727298054304974e-06,
      "loss": 0.6927,
      "step": 1839
    },
    {
      "epoch": 2.252141982864137,
      "grad_norm": 0.6445646635251346,
      "learning_rate": 4.72700691801154e-06,
      "loss": 0.6932,
      "step": 1840
    },
    {
      "epoch": 2.2533659730722153,
      "grad_norm": 1.1869827340444223,
      "learning_rate": 4.726715635367448e-06,
      "loss": 0.5297,
      "step": 1841
    },
    {
      "epoch": 2.254589963280294,
      "grad_norm": 1.1262635701193886,
      "learning_rate": 4.726424206391842e-06,
      "loss": 0.6857,
      "step": 1842
    },
    {
      "epoch": 2.255813953488372,
      "grad_norm": 1.1251204405610844,
      "learning_rate": 4.726132631103871e-06,
      "loss": 0.6287,
      "step": 1843
    },
    {
      "epoch": 2.2570379436964503,
      "grad_norm": 1.2808548883599133,
      "learning_rate": 4.725840909522698e-06,
      "loss": 0.505,
      "step": 1844
    },
    {
      "epoch": 2.258261933904529,
      "grad_norm": 1.2573748060361178,
      "learning_rate": 4.725549041667492e-06,
      "loss": 0.4732,
      "step": 1845
    },
    {
      "epoch": 2.259485924112607,
      "grad_norm": 0.625782663898054,
      "learning_rate": 4.725257027557436e-06,
      "loss": 0.5549,
      "step": 1846
    },
    {
      "epoch": 2.2607099143206852,
      "grad_norm": 0.8962879415661794,
      "learning_rate": 4.724964867211719e-06,
      "loss": 0.5514,
      "step": 1847
    },
    {
      "epoch": 2.261933904528764,
      "grad_norm": 1.1857627338796528,
      "learning_rate": 4.724672560649539e-06,
      "loss": 1.032,
      "step": 1848
    },
    {
      "epoch": 2.263157894736842,
      "grad_norm": 0.9898677015952242,
      "learning_rate": 4.724380107890109e-06,
      "loss": 0.65,
      "step": 1849
    },
    {
      "epoch": 2.26438188494492,
      "grad_norm": 1.2647551099992795,
      "learning_rate": 4.724087508952644e-06,
      "loss": 0.7948,
      "step": 1850
    },
    {
      "epoch": 2.265605875152999,
      "grad_norm": 1.4254645426893446,
      "learning_rate": 4.723794763856374e-06,
      "loss": 1.1283,
      "step": 1851
    },
    {
      "epoch": 2.266829865361077,
      "grad_norm": 0.9517438484247206,
      "learning_rate": 4.723501872620539e-06,
      "loss": 0.4815,
      "step": 1852
    },
    {
      "epoch": 2.2680538555691556,
      "grad_norm": 0.8302880719195052,
      "learning_rate": 4.7232088352643835e-06,
      "loss": 0.8382,
      "step": 1853
    },
    {
      "epoch": 2.269277845777234,
      "grad_norm": 1.1880042075613113,
      "learning_rate": 4.722915651807168e-06,
      "loss": 0.8823,
      "step": 1854
    },
    {
      "epoch": 2.270501835985312,
      "grad_norm": 1.959710865197276,
      "learning_rate": 4.722622322268157e-06,
      "loss": 0.8529,
      "step": 1855
    },
    {
      "epoch": 2.2717258261933906,
      "grad_norm": 1.3383385208293768,
      "learning_rate": 4.722328846666627e-06,
      "loss": 0.7134,
      "step": 1856
    },
    {
      "epoch": 2.2729498164014688,
      "grad_norm": 0.7894993022537544,
      "learning_rate": 4.722035225021866e-06,
      "loss": 0.8994,
      "step": 1857
    },
    {
      "epoch": 2.274173806609547,
      "grad_norm": 1.4010546573473848,
      "learning_rate": 4.721741457353169e-06,
      "loss": 0.874,
      "step": 1858
    },
    {
      "epoch": 2.2753977968176256,
      "grad_norm": 0.846692059695791,
      "learning_rate": 4.72144754367984e-06,
      "loss": 0.551,
      "step": 1859
    },
    {
      "epoch": 2.2766217870257037,
      "grad_norm": 1.138376636031738,
      "learning_rate": 4.721153484021196e-06,
      "loss": 0.5056,
      "step": 1860
    },
    {
      "epoch": 2.2778457772337823,
      "grad_norm": 1.3194817659277083,
      "learning_rate": 4.72085927839656e-06,
      "loss": 1.2871,
      "step": 1861
    },
    {
      "epoch": 2.2790697674418605,
      "grad_norm": 0.8756462663047224,
      "learning_rate": 4.720564926825267e-06,
      "loss": 0.9803,
      "step": 1862
    },
    {
      "epoch": 2.2802937576499387,
      "grad_norm": 1.3930737784578227,
      "learning_rate": 4.7202704293266605e-06,
      "loss": 0.7733,
      "step": 1863
    },
    {
      "epoch": 2.2815177478580173,
      "grad_norm": 1.2695267347626382,
      "learning_rate": 4.719975785920093e-06,
      "loss": 0.5501,
      "step": 1864
    },
    {
      "epoch": 2.2827417380660955,
      "grad_norm": 1.53738085378353,
      "learning_rate": 4.719680996624928e-06,
      "loss": 1.1468,
      "step": 1865
    },
    {
      "epoch": 2.2839657282741737,
      "grad_norm": 1.299574984351455,
      "learning_rate": 4.719386061460539e-06,
      "loss": 0.4306,
      "step": 1866
    },
    {
      "epoch": 2.2851897184822523,
      "grad_norm": 1.0925311283029786,
      "learning_rate": 4.719090980446306e-06,
      "loss": 0.5851,
      "step": 1867
    },
    {
      "epoch": 2.2864137086903304,
      "grad_norm": 1.063449925527497,
      "learning_rate": 4.718795753601622e-06,
      "loss": 1.435,
      "step": 1868
    },
    {
      "epoch": 2.2876376988984086,
      "grad_norm": 1.4497983828572243,
      "learning_rate": 4.718500380945887e-06,
      "loss": 0.5128,
      "step": 1869
    },
    {
      "epoch": 2.2888616891064872,
      "grad_norm": 1.2062776153243528,
      "learning_rate": 4.718204862498514e-06,
      "loss": 0.6857,
      "step": 1870
    },
    {
      "epoch": 2.2900856793145654,
      "grad_norm": 0.9012660885233963,
      "learning_rate": 4.71790919827892e-06,
      "loss": 1.6296,
      "step": 1871
    },
    {
      "epoch": 2.2913096695226436,
      "grad_norm": 0.9715708676625618,
      "learning_rate": 4.717613388306539e-06,
      "loss": 1.1728,
      "step": 1872
    },
    {
      "epoch": 2.292533659730722,
      "grad_norm": 1.0263480235301417,
      "learning_rate": 4.7173174326008075e-06,
      "loss": 0.6868,
      "step": 1873
    },
    {
      "epoch": 2.2937576499388004,
      "grad_norm": 0.6194327692280582,
      "learning_rate": 4.7170213311811756e-06,
      "loss": 0.5867,
      "step": 1874
    },
    {
      "epoch": 2.294981640146879,
      "grad_norm": 0.891155756844536,
      "learning_rate": 4.716725084067102e-06,
      "loss": 0.5361,
      "step": 1875
    },
    {
      "epoch": 2.296205630354957,
      "grad_norm": 1.1769091125333857,
      "learning_rate": 4.716428691278056e-06,
      "loss": 1.1276,
      "step": 1876
    },
    {
      "epoch": 2.2974296205630353,
      "grad_norm": 0.8029776505209408,
      "learning_rate": 4.716132152833513e-06,
      "loss": 1.1354,
      "step": 1877
    },
    {
      "epoch": 2.298653610771114,
      "grad_norm": 1.196895501195475,
      "learning_rate": 4.7158354687529625e-06,
      "loss": 1.4678,
      "step": 1878
    },
    {
      "epoch": 2.299877600979192,
      "grad_norm": 1.3165146202300035,
      "learning_rate": 4.715538639055901e-06,
      "loss": 0.7601,
      "step": 1879
    },
    {
      "epoch": 2.3011015911872703,
      "grad_norm": 1.1039384845770621,
      "learning_rate": 4.7152416637618335e-06,
      "loss": 0.5752,
      "step": 1880
    },
    {
      "epoch": 2.302325581395349,
      "grad_norm": 0.7269891027360053,
      "learning_rate": 4.7149445428902794e-06,
      "loss": 0.7948,
      "step": 1881
    },
    {
      "epoch": 2.303549571603427,
      "grad_norm": 0.8998994953010416,
      "learning_rate": 4.714647276460762e-06,
      "loss": 0.5645,
      "step": 1882
    },
    {
      "epoch": 2.3047735618115057,
      "grad_norm": 1.2568270616569326,
      "learning_rate": 4.714349864492816e-06,
      "loss": 0.997,
      "step": 1883
    },
    {
      "epoch": 2.305997552019584,
      "grad_norm": 0.9724051828838117,
      "learning_rate": 4.714052307005987e-06,
      "loss": 0.922,
      "step": 1884
    },
    {
      "epoch": 2.307221542227662,
      "grad_norm": 0.7563659597943564,
      "learning_rate": 4.713754604019831e-06,
      "loss": 0.4572,
      "step": 1885
    },
    {
      "epoch": 2.3084455324357407,
      "grad_norm": 1.1748432216588744,
      "learning_rate": 4.71345675555391e-06,
      "loss": 0.5626,
      "step": 1886
    },
    {
      "epoch": 2.309669522643819,
      "grad_norm": 1.6688177462208922,
      "learning_rate": 4.713158761627799e-06,
      "loss": 1.0607,
      "step": 1887
    },
    {
      "epoch": 2.310893512851897,
      "grad_norm": 0.9631109577110486,
      "learning_rate": 4.712860622261078e-06,
      "loss": 0.5539,
      "step": 1888
    },
    {
      "epoch": 2.3121175030599757,
      "grad_norm": 1.0912333202113949,
      "learning_rate": 4.712562337473343e-06,
      "loss": 0.4901,
      "step": 1889
    },
    {
      "epoch": 2.313341493268054,
      "grad_norm": 0.9151982095186041,
      "learning_rate": 4.712263907284195e-06,
      "loss": 0.6658,
      "step": 1890
    },
    {
      "epoch": 2.314565483476132,
      "grad_norm": 0.7630881977432663,
      "learning_rate": 4.7119653317132454e-06,
      "loss": 0.4936,
      "step": 1891
    },
    {
      "epoch": 2.3157894736842106,
      "grad_norm": 0.5097942777190032,
      "learning_rate": 4.711666610780115e-06,
      "loss": 0.4609,
      "step": 1892
    },
    {
      "epoch": 2.317013463892289,
      "grad_norm": 0.8108320508497355,
      "learning_rate": 4.711367744504436e-06,
      "loss": 0.6758,
      "step": 1893
    },
    {
      "epoch": 2.318237454100367,
      "grad_norm": 0.7435477627947105,
      "learning_rate": 4.711068732905848e-06,
      "loss": 0.5171,
      "step": 1894
    },
    {
      "epoch": 2.3194614443084456,
      "grad_norm": 0.8959268130708758,
      "learning_rate": 4.710769576004001e-06,
      "loss": 0.4453,
      "step": 1895
    },
    {
      "epoch": 2.3206854345165238,
      "grad_norm": 0.7471499052525399,
      "learning_rate": 4.710470273818555e-06,
      "loss": 1.2508,
      "step": 1896
    },
    {
      "epoch": 2.3219094247246024,
      "grad_norm": 1.1633638260446977,
      "learning_rate": 4.710170826369178e-06,
      "loss": 1.0517,
      "step": 1897
    },
    {
      "epoch": 2.3231334149326806,
      "grad_norm": 0.608115559424718,
      "learning_rate": 4.709871233675549e-06,
      "loss": 0.5565,
      "step": 1898
    },
    {
      "epoch": 2.3243574051407587,
      "grad_norm": 0.8005951613983394,
      "learning_rate": 4.709571495757357e-06,
      "loss": 0.7041,
      "step": 1899
    },
    {
      "epoch": 2.3255813953488373,
      "grad_norm": 0.9610517655945263,
      "learning_rate": 4.709271612634299e-06,
      "loss": 0.9996,
      "step": 1900
    },
    {
      "epoch": 2.3268053855569155,
      "grad_norm": 0.8326697093095426,
      "learning_rate": 4.708971584326081e-06,
      "loss": 1.0881,
      "step": 1901
    },
    {
      "epoch": 2.3280293757649937,
      "grad_norm": 0.7542930072225869,
      "learning_rate": 4.7086714108524215e-06,
      "loss": 0.5679,
      "step": 1902
    },
    {
      "epoch": 2.3292533659730723,
      "grad_norm": 0.8384071664102889,
      "learning_rate": 4.708371092233046e-06,
      "loss": 0.7764,
      "step": 1903
    },
    {
      "epoch": 2.3304773561811505,
      "grad_norm": 1.1740951557909265,
      "learning_rate": 4.708070628487691e-06,
      "loss": 0.7425,
      "step": 1904
    },
    {
      "epoch": 2.331701346389229,
      "grad_norm": 0.639505116805248,
      "learning_rate": 4.707770019636101e-06,
      "loss": 0.6238,
      "step": 1905
    },
    {
      "epoch": 2.3329253365973073,
      "grad_norm": 1.7116632004338355,
      "learning_rate": 4.707469265698031e-06,
      "loss": 0.7828,
      "step": 1906
    },
    {
      "epoch": 2.3341493268053854,
      "grad_norm": 1.555643459821072,
      "learning_rate": 4.707168366693247e-06,
      "loss": 0.8961,
      "step": 1907
    },
    {
      "epoch": 2.335373317013464,
      "grad_norm": 0.8023299135409482,
      "learning_rate": 4.70686732264152e-06,
      "loss": 1.0553,
      "step": 1908
    },
    {
      "epoch": 2.3365973072215422,
      "grad_norm": 1.0094992980005395,
      "learning_rate": 4.706566133562636e-06,
      "loss": 0.733,
      "step": 1909
    },
    {
      "epoch": 2.3378212974296204,
      "grad_norm": 1.1427769200077593,
      "learning_rate": 4.706264799476388e-06,
      "loss": 1.2053,
      "step": 1910
    },
    {
      "epoch": 2.339045287637699,
      "grad_norm": 0.9930890466331854,
      "learning_rate": 4.705963320402577e-06,
      "loss": 0.5195,
      "step": 1911
    },
    {
      "epoch": 2.340269277845777,
      "grad_norm": 0.8653098939490641,
      "learning_rate": 4.705661696361016e-06,
      "loss": 0.777,
      "step": 1912
    },
    {
      "epoch": 2.3414932680538554,
      "grad_norm": 0.6873472466625226,
      "learning_rate": 4.705359927371526e-06,
      "loss": 0.572,
      "step": 1913
    },
    {
      "epoch": 2.342717258261934,
      "grad_norm": 1.3607697033237565,
      "learning_rate": 4.7050580134539395e-06,
      "loss": 0.6323,
      "step": 1914
    },
    {
      "epoch": 2.343941248470012,
      "grad_norm": 1.2451884434956817,
      "learning_rate": 4.7047559546280965e-06,
      "loss": 0.7687,
      "step": 1915
    },
    {
      "epoch": 2.3451652386780903,
      "grad_norm": 0.9453088368988347,
      "learning_rate": 4.704453750913846e-06,
      "loss": 0.5769,
      "step": 1916
    },
    {
      "epoch": 2.346389228886169,
      "grad_norm": 0.79296426484685,
      "learning_rate": 4.704151402331049e-06,
      "loss": 0.5068,
      "step": 1917
    },
    {
      "epoch": 2.347613219094247,
      "grad_norm": 0.7502610101843505,
      "learning_rate": 4.703848908899574e-06,
      "loss": 0.6193,
      "step": 1918
    },
    {
      "epoch": 2.3488372093023258,
      "grad_norm": 1.664090265732174,
      "learning_rate": 4.703546270639301e-06,
      "loss": 0.9136,
      "step": 1919
    },
    {
      "epoch": 2.350061199510404,
      "grad_norm": 1.481465404552778,
      "learning_rate": 4.7032434875701176e-06,
      "loss": 0.5476,
      "step": 1920
    },
    {
      "epoch": 2.351285189718482,
      "grad_norm": 1.0184048897291256,
      "learning_rate": 4.702940559711921e-06,
      "loss": 0.5005,
      "step": 1921
    },
    {
      "epoch": 2.3525091799265607,
      "grad_norm": 1.0554609294124078,
      "learning_rate": 4.702637487084618e-06,
      "loss": 0.9804,
      "step": 1922
    },
    {
      "epoch": 2.353733170134639,
      "grad_norm": 1.4399412085210204,
      "learning_rate": 4.7023342697081276e-06,
      "loss": 0.6211,
      "step": 1923
    },
    {
      "epoch": 2.354957160342717,
      "grad_norm": 0.7170198421669076,
      "learning_rate": 4.702030907602375e-06,
      "loss": 0.4592,
      "step": 1924
    },
    {
      "epoch": 2.3561811505507957,
      "grad_norm": 0.9968701763978323,
      "learning_rate": 4.701727400787295e-06,
      "loss": 0.5919,
      "step": 1925
    },
    {
      "epoch": 2.357405140758874,
      "grad_norm": 0.9385233850263351,
      "learning_rate": 4.701423749282834e-06,
      "loss": 0.7543,
      "step": 1926
    },
    {
      "epoch": 2.3586291309669525,
      "grad_norm": 0.8955818205284788,
      "learning_rate": 4.701119953108947e-06,
      "loss": 0.6333,
      "step": 1927
    },
    {
      "epoch": 2.3598531211750307,
      "grad_norm": 0.7063996029116164,
      "learning_rate": 4.700816012285598e-06,
      "loss": 0.8812,
      "step": 1928
    },
    {
      "epoch": 2.361077111383109,
      "grad_norm": 1.570893375814074,
      "learning_rate": 4.700511926832762e-06,
      "loss": 0.5483,
      "step": 1929
    },
    {
      "epoch": 2.3623011015911874,
      "grad_norm": 1.0496391730387182,
      "learning_rate": 4.700207696770419e-06,
      "loss": 1.5592,
      "step": 1930
    },
    {
      "epoch": 2.3635250917992656,
      "grad_norm": 1.1805859725287076,
      "learning_rate": 4.699903322118566e-06,
      "loss": 0.9241,
      "step": 1931
    },
    {
      "epoch": 2.364749082007344,
      "grad_norm": 1.1521708308442629,
      "learning_rate": 4.699598802897203e-06,
      "loss": 0.7245,
      "step": 1932
    },
    {
      "epoch": 2.3659730722154224,
      "grad_norm": 0.9394921217044425,
      "learning_rate": 4.699294139126343e-06,
      "loss": 0.6159,
      "step": 1933
    },
    {
      "epoch": 2.3671970624235006,
      "grad_norm": 1.1761606310691894,
      "learning_rate": 4.698989330826005e-06,
      "loss": 0.3799,
      "step": 1934
    },
    {
      "epoch": 2.3684210526315788,
      "grad_norm": 1.4233267404878251,
      "learning_rate": 4.698684378016223e-06,
      "loss": 0.9572,
      "step": 1935
    },
    {
      "epoch": 2.3696450428396574,
      "grad_norm": 0.6451691659753587,
      "learning_rate": 4.698379280717036e-06,
      "loss": 0.6448,
      "step": 1936
    },
    {
      "epoch": 2.3708690330477356,
      "grad_norm": 1.439779100423735,
      "learning_rate": 4.698074038948494e-06,
      "loss": 0.7716,
      "step": 1937
    },
    {
      "epoch": 2.3720930232558137,
      "grad_norm": 1.3128047291081741,
      "learning_rate": 4.697768652730656e-06,
      "loss": 0.3677,
      "step": 1938
    },
    {
      "epoch": 2.3733170134638923,
      "grad_norm": 0.8545334307708811,
      "learning_rate": 4.6974631220835905e-06,
      "loss": 0.4209,
      "step": 1939
    },
    {
      "epoch": 2.3745410036719705,
      "grad_norm": 0.789638408852278,
      "learning_rate": 4.697157447027378e-06,
      "loss": 1.3923,
      "step": 1940
    },
    {
      "epoch": 2.375764993880049,
      "grad_norm": 1.7246724080929752,
      "learning_rate": 4.696851627582104e-06,
      "loss": 0.8318,
      "step": 1941
    },
    {
      "epoch": 2.3769889840881273,
      "grad_norm": 1.456222956380552,
      "learning_rate": 4.696545663767867e-06,
      "loss": 1.0376,
      "step": 1942
    },
    {
      "epoch": 2.3782129742962055,
      "grad_norm": 0.6127381225023172,
      "learning_rate": 4.696239555604773e-06,
      "loss": 0.4118,
      "step": 1943
    },
    {
      "epoch": 2.379436964504284,
      "grad_norm": 1.0696777199392271,
      "learning_rate": 4.695933303112939e-06,
      "loss": 1.0755,
      "step": 1944
    },
    {
      "epoch": 2.3806609547123623,
      "grad_norm": 1.0209308036731126,
      "learning_rate": 4.6956269063124895e-06,
      "loss": 0.6175,
      "step": 1945
    },
    {
      "epoch": 2.3818849449204405,
      "grad_norm": 0.9339617373988369,
      "learning_rate": 4.6953203652235615e-06,
      "loss": 0.9893,
      "step": 1946
    },
    {
      "epoch": 2.383108935128519,
      "grad_norm": 1.214925742772279,
      "learning_rate": 4.6950136798663e-06,
      "loss": 0.5943,
      "step": 1947
    },
    {
      "epoch": 2.3843329253365972,
      "grad_norm": 0.8132353543630525,
      "learning_rate": 4.694706850260857e-06,
      "loss": 0.7598,
      "step": 1948
    },
    {
      "epoch": 2.385556915544676,
      "grad_norm": 1.3373094365979548,
      "learning_rate": 4.694399876427397e-06,
      "loss": 0.627,
      "step": 1949
    },
    {
      "epoch": 2.386780905752754,
      "grad_norm": 1.4439702640169259,
      "learning_rate": 4.694092758386095e-06,
      "loss": 0.5952,
      "step": 1950
    },
    {
      "epoch": 2.388004895960832,
      "grad_norm": 0.9176361644503913,
      "learning_rate": 4.693785496157132e-06,
      "loss": 1.3798,
      "step": 1951
    },
    {
      "epoch": 2.389228886168911,
      "grad_norm": 0.6360087069726685,
      "learning_rate": 4.6934780897607005e-06,
      "loss": 0.5893,
      "step": 1952
    },
    {
      "epoch": 2.390452876376989,
      "grad_norm": 1.1873039082776884,
      "learning_rate": 4.6931705392170016e-06,
      "loss": 0.5028,
      "step": 1953
    },
    {
      "epoch": 2.391676866585067,
      "grad_norm": 0.8976245880576597,
      "learning_rate": 4.692862844546246e-06,
      "loss": 0.6801,
      "step": 1954
    },
    {
      "epoch": 2.392900856793146,
      "grad_norm": 1.0239295740757415,
      "learning_rate": 4.692555005768656e-06,
      "loss": 1.6771,
      "step": 1955
    },
    {
      "epoch": 2.394124847001224,
      "grad_norm": 0.8201907697460017,
      "learning_rate": 4.692247022904461e-06,
      "loss": 0.75,
      "step": 1956
    },
    {
      "epoch": 2.395348837209302,
      "grad_norm": 1.1062417697379563,
      "learning_rate": 4.691938895973901e-06,
      "loss": 1.5318,
      "step": 1957
    },
    {
      "epoch": 2.3965728274173808,
      "grad_norm": 1.4255719949443622,
      "learning_rate": 4.691630624997222e-06,
      "loss": 0.5687,
      "step": 1958
    },
    {
      "epoch": 2.397796817625459,
      "grad_norm": 0.723361749038174,
      "learning_rate": 4.691322209994687e-06,
      "loss": 0.5326,
      "step": 1959
    },
    {
      "epoch": 2.399020807833537,
      "grad_norm": 0.9656153034686662,
      "learning_rate": 4.69101365098656e-06,
      "loss": 1.392,
      "step": 1960
    },
    {
      "epoch": 2.4002447980416157,
      "grad_norm": 0.632142485940512,
      "learning_rate": 4.69070494799312e-06,
      "loss": 0.6782,
      "step": 1961
    },
    {
      "epoch": 2.401468788249694,
      "grad_norm": 1.4857979129260754,
      "learning_rate": 4.690396101034654e-06,
      "loss": 0.4955,
      "step": 1962
    },
    {
      "epoch": 2.4026927784577725,
      "grad_norm": 1.2001526909005433,
      "learning_rate": 4.690087110131458e-06,
      "loss": 0.5794,
      "step": 1963
    },
    {
      "epoch": 2.4039167686658507,
      "grad_norm": 0.5613649425225692,
      "learning_rate": 4.689777975303837e-06,
      "loss": 0.3814,
      "step": 1964
    },
    {
      "epoch": 2.405140758873929,
      "grad_norm": 0.6756331894660321,
      "learning_rate": 4.689468696572108e-06,
      "loss": 0.6698,
      "step": 1965
    },
    {
      "epoch": 2.4063647490820075,
      "grad_norm": 1.0462189008787655,
      "learning_rate": 4.689159273956593e-06,
      "loss": 0.4957,
      "step": 1966
    },
    {
      "epoch": 2.4075887392900857,
      "grad_norm": 0.9070338221332254,
      "learning_rate": 4.688849707477629e-06,
      "loss": 0.6409,
      "step": 1967
    },
    {
      "epoch": 2.408812729498164,
      "grad_norm": 0.9967960041788031,
      "learning_rate": 4.688539997155558e-06,
      "loss": 0.686,
      "step": 1968
    },
    {
      "epoch": 2.4100367197062424,
      "grad_norm": 1.2869814660815038,
      "learning_rate": 4.688230143010733e-06,
      "loss": 0.8089,
      "step": 1969
    },
    {
      "epoch": 2.4112607099143206,
      "grad_norm": 0.6453206958110141,
      "learning_rate": 4.687920145063517e-06,
      "loss": 0.5461,
      "step": 1970
    },
    {
      "epoch": 2.4124847001223992,
      "grad_norm": 1.403224757460005,
      "learning_rate": 4.6876100033342805e-06,
      "loss": 1.0907,
      "step": 1971
    },
    {
      "epoch": 2.4137086903304774,
      "grad_norm": 0.6486978902986142,
      "learning_rate": 4.687299717843407e-06,
      "loss": 0.5121,
      "step": 1972
    },
    {
      "epoch": 2.4149326805385556,
      "grad_norm": 0.7575360742798415,
      "learning_rate": 4.686989288611287e-06,
      "loss": 0.5722,
      "step": 1973
    },
    {
      "epoch": 2.416156670746634,
      "grad_norm": 1.2238088568054706,
      "learning_rate": 4.686678715658319e-06,
      "loss": 1.2408,
      "step": 1974
    },
    {
      "epoch": 2.4173806609547124,
      "grad_norm": 1.2586215422424256,
      "learning_rate": 4.686367999004914e-06,
      "loss": 1.0632,
      "step": 1975
    },
    {
      "epoch": 2.4186046511627906,
      "grad_norm": 0.6021748916105605,
      "learning_rate": 4.6860571386714905e-06,
      "loss": 0.5588,
      "step": 1976
    },
    {
      "epoch": 2.419828641370869,
      "grad_norm": 0.6611145951086846,
      "learning_rate": 4.685746134678478e-06,
      "loss": 0.5291,
      "step": 1977
    },
    {
      "epoch": 2.4210526315789473,
      "grad_norm": 0.8229597056603816,
      "learning_rate": 4.685434987046314e-06,
      "loss": 0.7084,
      "step": 1978
    },
    {
      "epoch": 2.4222766217870255,
      "grad_norm": 0.7819356406782624,
      "learning_rate": 4.685123695795446e-06,
      "loss": 1.1465,
      "step": 1979
    },
    {
      "epoch": 2.423500611995104,
      "grad_norm": 1.6516904878339431,
      "learning_rate": 4.684812260946331e-06,
      "loss": 1.1279,
      "step": 1980
    },
    {
      "epoch": 2.4247246022031823,
      "grad_norm": 0.781169797088301,
      "learning_rate": 4.684500682519435e-06,
      "loss": 0.5803,
      "step": 1981
    },
    {
      "epoch": 2.4259485924112605,
      "grad_norm": 0.8255248329447696,
      "learning_rate": 4.684188960535234e-06,
      "loss": 0.5327,
      "step": 1982
    },
    {
      "epoch": 2.427172582619339,
      "grad_norm": 0.9664449380806159,
      "learning_rate": 4.683877095014215e-06,
      "loss": 0.6499,
      "step": 1983
    },
    {
      "epoch": 2.4283965728274173,
      "grad_norm": 0.9511864570644972,
      "learning_rate": 4.6835650859768685e-06,
      "loss": 0.5007,
      "step": 1984
    },
    {
      "epoch": 2.429620563035496,
      "grad_norm": 0.8630653387578063,
      "learning_rate": 4.683252933443701e-06,
      "loss": 0.6229,
      "step": 1985
    },
    {
      "epoch": 2.430844553243574,
      "grad_norm": 1.087426350518753,
      "learning_rate": 4.682940637435228e-06,
      "loss": 0.6085,
      "step": 1986
    },
    {
      "epoch": 2.4320685434516522,
      "grad_norm": 0.976540682236335,
      "learning_rate": 4.682628197971969e-06,
      "loss": 1.2311,
      "step": 1987
    },
    {
      "epoch": 2.433292533659731,
      "grad_norm": 0.8362601926282501,
      "learning_rate": 4.682315615074459e-06,
      "loss": 0.5455,
      "step": 1988
    },
    {
      "epoch": 2.434516523867809,
      "grad_norm": 1.2600123674086403,
      "learning_rate": 4.682002888763237e-06,
      "loss": 0.537,
      "step": 1989
    },
    {
      "epoch": 2.435740514075887,
      "grad_norm": 1.241308544882274,
      "learning_rate": 4.681690019058857e-06,
      "loss": 0.5755,
      "step": 1990
    },
    {
      "epoch": 2.436964504283966,
      "grad_norm": 0.9879516652685221,
      "learning_rate": 4.681377005981878e-06,
      "loss": 1.0684,
      "step": 1991
    },
    {
      "epoch": 2.438188494492044,
      "grad_norm": 1.2137035156314764,
      "learning_rate": 4.681063849552871e-06,
      "loss": 0.7647,
      "step": 1992
    },
    {
      "epoch": 2.4394124847001226,
      "grad_norm": 1.2958304187030811,
      "learning_rate": 4.680750549792415e-06,
      "loss": 0.5229,
      "step": 1993
    },
    {
      "epoch": 2.440636474908201,
      "grad_norm": 0.8033045091645395,
      "learning_rate": 4.680437106721098e-06,
      "loss": 0.5968,
      "step": 1994
    },
    {
      "epoch": 2.441860465116279,
      "grad_norm": 0.9231230495744421,
      "learning_rate": 4.68012352035952e-06,
      "loss": 0.6566,
      "step": 1995
    },
    {
      "epoch": 2.4430844553243576,
      "grad_norm": 0.8793167928133118,
      "learning_rate": 4.679809790728287e-06,
      "loss": 1.3367,
      "step": 1996
    },
    {
      "epoch": 2.4443084455324358,
      "grad_norm": 0.953736568385416,
      "learning_rate": 4.679495917848018e-06,
      "loss": 0.8771,
      "step": 1997
    },
    {
      "epoch": 2.445532435740514,
      "grad_norm": 1.6317287910534855,
      "learning_rate": 4.679181901739339e-06,
      "loss": 0.4566,
      "step": 1998
    },
    {
      "epoch": 2.4467564259485926,
      "grad_norm": 0.8093634260542532,
      "learning_rate": 4.678867742422885e-06,
      "loss": 0.6829,
      "step": 1999
    },
    {
      "epoch": 2.4479804161566707,
      "grad_norm": 1.5105886740632646,
      "learning_rate": 4.6785534399193015e-06,
      "loss": 0.4565,
      "step": 2000
    },
    {
      "epoch": 2.449204406364749,
      "grad_norm": 1.4643503858923508,
      "learning_rate": 4.678238994249245e-06,
      "loss": 0.5366,
      "step": 2001
    },
    {
      "epoch": 2.4504283965728275,
      "grad_norm": 0.9391872800730385,
      "learning_rate": 4.677924405433378e-06,
      "loss": 0.6659,
      "step": 2002
    },
    {
      "epoch": 2.4516523867809057,
      "grad_norm": 0.689648363482712,
      "learning_rate": 4.677609673492375e-06,
      "loss": 0.6167,
      "step": 2003
    },
    {
      "epoch": 2.452876376988984,
      "grad_norm": 1.0244745242005668,
      "learning_rate": 4.677294798446918e-06,
      "loss": 0.6969,
      "step": 2004
    },
    {
      "epoch": 2.4541003671970625,
      "grad_norm": 1.391714878445633,
      "learning_rate": 4.676979780317701e-06,
      "loss": 0.7948,
      "step": 2005
    },
    {
      "epoch": 2.4553243574051407,
      "grad_norm": 0.7441361607029013,
      "learning_rate": 4.676664619125424e-06,
      "loss": 0.7332,
      "step": 2006
    },
    {
      "epoch": 2.4565483476132193,
      "grad_norm": 0.9360468581899858,
      "learning_rate": 4.6763493148907995e-06,
      "loss": 1.0396,
      "step": 2007
    },
    {
      "epoch": 2.4577723378212974,
      "grad_norm": 0.8179059763595751,
      "learning_rate": 4.676033867634548e-06,
      "loss": 0.9257,
      "step": 2008
    },
    {
      "epoch": 2.4589963280293756,
      "grad_norm": 0.9103874259138527,
      "learning_rate": 4.675718277377399e-06,
      "loss": 0.3834,
      "step": 2009
    },
    {
      "epoch": 2.4602203182374542,
      "grad_norm": 0.8265545041437229,
      "learning_rate": 4.675402544140093e-06,
      "loss": 0.7303,
      "step": 2010
    },
    {
      "epoch": 2.4614443084455324,
      "grad_norm": 0.9088141147919799,
      "learning_rate": 4.675086667943377e-06,
      "loss": 0.7313,
      "step": 2011
    },
    {
      "epoch": 2.4626682986536106,
      "grad_norm": 1.1528736987929922,
      "learning_rate": 4.674770648808011e-06,
      "loss": 0.5642,
      "step": 2012
    },
    {
      "epoch": 2.463892288861689,
      "grad_norm": 0.7593355888294113,
      "learning_rate": 4.674454486754762e-06,
      "loss": 0.4517,
      "step": 2013
    },
    {
      "epoch": 2.4651162790697674,
      "grad_norm": 0.7312213833910401,
      "learning_rate": 4.674138181804406e-06,
      "loss": 0.4202,
      "step": 2014
    },
    {
      "epoch": 2.466340269277846,
      "grad_norm": 0.7117683056305699,
      "learning_rate": 4.673821733977731e-06,
      "loss": 0.6088,
      "step": 2015
    },
    {
      "epoch": 2.467564259485924,
      "grad_norm": 0.7600211202310814,
      "learning_rate": 4.6735051432955314e-06,
      "loss": 0.5758,
      "step": 2016
    },
    {
      "epoch": 2.4687882496940023,
      "grad_norm": 1.0044817658988496,
      "learning_rate": 4.673188409778614e-06,
      "loss": 0.6097,
      "step": 2017
    },
    {
      "epoch": 2.470012239902081,
      "grad_norm": 1.6430019153827742,
      "learning_rate": 4.672871533447791e-06,
      "loss": 0.4027,
      "step": 2018
    },
    {
      "epoch": 2.471236230110159,
      "grad_norm": 1.2899556293953571,
      "learning_rate": 4.672554514323889e-06,
      "loss": 0.5659,
      "step": 2019
    },
    {
      "epoch": 2.4724602203182373,
      "grad_norm": 1.3482809537612783,
      "learning_rate": 4.67223735242774e-06,
      "loss": 0.4588,
      "step": 2020
    },
    {
      "epoch": 2.473684210526316,
      "grad_norm": 1.0982380787179564,
      "learning_rate": 4.6719200477801865e-06,
      "loss": 0.4683,
      "step": 2021
    },
    {
      "epoch": 2.474908200734394,
      "grad_norm": 1.0897631691509069,
      "learning_rate": 4.67160260040208e-06,
      "loss": 0.4694,
      "step": 2022
    },
    {
      "epoch": 2.4761321909424723,
      "grad_norm": 1.024320284884599,
      "learning_rate": 4.671285010314284e-06,
      "loss": 1.1598,
      "step": 2023
    },
    {
      "epoch": 2.477356181150551,
      "grad_norm": 1.1545270238147962,
      "learning_rate": 4.670967277537667e-06,
      "loss": 0.6288,
      "step": 2024
    },
    {
      "epoch": 2.478580171358629,
      "grad_norm": 1.0798534668809237,
      "learning_rate": 4.670649402093112e-06,
      "loss": 0.61,
      "step": 2025
    },
    {
      "epoch": 2.4798041615667072,
      "grad_norm": 1.0764644794708718,
      "learning_rate": 4.6703313840015055e-06,
      "loss": 1.1621,
      "step": 2026
    },
    {
      "epoch": 2.481028151774786,
      "grad_norm": 0.7943273138271846,
      "learning_rate": 4.670013223283748e-06,
      "loss": 0.7624,
      "step": 2027
    },
    {
      "epoch": 2.482252141982864,
      "grad_norm": 0.8512146712941421,
      "learning_rate": 4.669694919960749e-06,
      "loss": 0.4063,
      "step": 2028
    },
    {
      "epoch": 2.4834761321909427,
      "grad_norm": 1.056282043894564,
      "learning_rate": 4.669376474053424e-06,
      "loss": 0.6716,
      "step": 2029
    },
    {
      "epoch": 2.484700122399021,
      "grad_norm": 1.1315707906117647,
      "learning_rate": 4.6690578855827006e-06,
      "loss": 0.6443,
      "step": 2030
    },
    {
      "epoch": 2.485924112607099,
      "grad_norm": 1.3745625296415231,
      "learning_rate": 4.668739154569517e-06,
      "loss": 0.6216,
      "step": 2031
    },
    {
      "epoch": 2.4871481028151776,
      "grad_norm": 0.905764903604571,
      "learning_rate": 4.668420281034817e-06,
      "loss": 0.4719,
      "step": 2032
    },
    {
      "epoch": 2.488372093023256,
      "grad_norm": 1.0881152181161204,
      "learning_rate": 4.668101264999557e-06,
      "loss": 0.9763,
      "step": 2033
    },
    {
      "epoch": 2.489596083231334,
      "grad_norm": 1.0819726592251493,
      "learning_rate": 4.667782106484701e-06,
      "loss": 1.4903,
      "step": 2034
    },
    {
      "epoch": 2.4908200734394126,
      "grad_norm": 0.9690189951718694,
      "learning_rate": 4.667462805511223e-06,
      "loss": 0.5254,
      "step": 2035
    },
    {
      "epoch": 2.4920440636474908,
      "grad_norm": 1.014980905433294,
      "learning_rate": 4.667143362100106e-06,
      "loss": 1.2322,
      "step": 2036
    },
    {
      "epoch": 2.4932680538555694,
      "grad_norm": 0.8157045803424169,
      "learning_rate": 4.666823776272344e-06,
      "loss": 0.6549,
      "step": 2037
    },
    {
      "epoch": 2.4944920440636476,
      "grad_norm": 0.6659207752470652,
      "learning_rate": 4.666504048048937e-06,
      "loss": 0.6322,
      "step": 2038
    },
    {
      "epoch": 2.4957160342717257,
      "grad_norm": 1.7911628920858291,
      "learning_rate": 4.666184177450897e-06,
      "loss": 0.7994,
      "step": 2039
    },
    {
      "epoch": 2.4969400244798043,
      "grad_norm": 0.9831500745888958,
      "learning_rate": 4.665864164499245e-06,
      "loss": 0.5262,
      "step": 2040
    },
    {
      "epoch": 2.4981640146878825,
      "grad_norm": 1.1964928512229351,
      "learning_rate": 4.665544009215012e-06,
      "loss": 0.4512,
      "step": 2041
    },
    {
      "epoch": 2.4993880048959607,
      "grad_norm": 1.3722437336116062,
      "learning_rate": 4.665223711619236e-06,
      "loss": 1.4931,
      "step": 2042
    },
    {
      "epoch": 2.5006119951040393,
      "grad_norm": 0.7502108691050444,
      "learning_rate": 4.664903271732966e-06,
      "loss": 0.6247,
      "step": 2043
    },
    {
      "epoch": 2.5018359853121175,
      "grad_norm": 0.9127226801361672,
      "learning_rate": 4.664582689577261e-06,
      "loss": 0.9907,
      "step": 2044
    },
    {
      "epoch": 2.5030599755201957,
      "grad_norm": 0.9087741259363168,
      "learning_rate": 4.664261965173188e-06,
      "loss": 0.9293,
      "step": 2045
    },
    {
      "epoch": 2.5042839657282743,
      "grad_norm": 0.8928036186180087,
      "learning_rate": 4.663941098541823e-06,
      "loss": 1.365,
      "step": 2046
    },
    {
      "epoch": 2.5055079559363524,
      "grad_norm": 1.6154146938554772,
      "learning_rate": 4.663620089704253e-06,
      "loss": 0.9113,
      "step": 2047
    },
    {
      "epoch": 2.5067319461444306,
      "grad_norm": 1.5325606043101374,
      "learning_rate": 4.6632989386815744e-06,
      "loss": 0.5367,
      "step": 2048
    },
    {
      "epoch": 2.5079559363525092,
      "grad_norm": 0.7396270937517353,
      "learning_rate": 4.662977645494889e-06,
      "loss": 0.5482,
      "step": 2049
    },
    {
      "epoch": 2.5091799265605874,
      "grad_norm": 1.405465285131659,
      "learning_rate": 4.662656210165315e-06,
      "loss": 0.6416,
      "step": 2050
    },
    {
      "epoch": 2.5104039167686656,
      "grad_norm": 0.80710487969568,
      "learning_rate": 4.662334632713973e-06,
      "loss": 0.5832,
      "step": 2051
    },
    {
      "epoch": 2.511627906976744,
      "grad_norm": 1.5059245057990465,
      "learning_rate": 4.662012913161998e-06,
      "loss": 0.512,
      "step": 2052
    },
    {
      "epoch": 2.5128518971848224,
      "grad_norm": 0.857311595280918,
      "learning_rate": 4.661691051530529e-06,
      "loss": 1.0003,
      "step": 2053
    },
    {
      "epoch": 2.514075887392901,
      "grad_norm": 1.1764782213359095,
      "learning_rate": 4.661369047840722e-06,
      "loss": 0.69,
      "step": 2054
    },
    {
      "epoch": 2.515299877600979,
      "grad_norm": 1.053797392608861,
      "learning_rate": 4.661046902113734e-06,
      "loss": 1.4593,
      "step": 2055
    },
    {
      "epoch": 2.516523867809058,
      "grad_norm": 1.3665686703365592,
      "learning_rate": 4.660724614370737e-06,
      "loss": 0.6689,
      "step": 2056
    },
    {
      "epoch": 2.517747858017136,
      "grad_norm": 1.204794022962197,
      "learning_rate": 4.66040218463291e-06,
      "loss": 0.5165,
      "step": 2057
    },
    {
      "epoch": 2.518971848225214,
      "grad_norm": 2.0784461150173845,
      "learning_rate": 4.660079612921443e-06,
      "loss": 0.3873,
      "step": 2058
    },
    {
      "epoch": 2.5201958384332928,
      "grad_norm": 1.3119630268259508,
      "learning_rate": 4.659756899257532e-06,
      "loss": 0.9024,
      "step": 2059
    },
    {
      "epoch": 2.521419828641371,
      "grad_norm": 0.9963898559279143,
      "learning_rate": 4.659434043662388e-06,
      "loss": 0.7723,
      "step": 2060
    },
    {
      "epoch": 2.522643818849449,
      "grad_norm": 0.9156178165680909,
      "learning_rate": 4.659111046157225e-06,
      "loss": 0.6725,
      "step": 2061
    },
    {
      "epoch": 2.5238678090575277,
      "grad_norm": 0.9334890440525432,
      "learning_rate": 4.65878790676327e-06,
      "loss": 0.5992,
      "step": 2062
    },
    {
      "epoch": 2.525091799265606,
      "grad_norm": 0.9955109799015857,
      "learning_rate": 4.658464625501759e-06,
      "loss": 0.5329,
      "step": 2063
    },
    {
      "epoch": 2.526315789473684,
      "grad_norm": 0.656079145237901,
      "learning_rate": 4.658141202393935e-06,
      "loss": 0.726,
      "step": 2064
    },
    {
      "epoch": 2.5275397796817627,
      "grad_norm": 1.2085286500971073,
      "learning_rate": 4.6578176374610545e-06,
      "loss": 0.7309,
      "step": 2065
    },
    {
      "epoch": 2.528763769889841,
      "grad_norm": 1.0343651850060889,
      "learning_rate": 4.65749393072438e-06,
      "loss": 0.8806,
      "step": 2066
    },
    {
      "epoch": 2.529987760097919,
      "grad_norm": 1.0430910166940934,
      "learning_rate": 4.657170082205185e-06,
      "loss": 0.618,
      "step": 2067
    },
    {
      "epoch": 2.5312117503059977,
      "grad_norm": 0.9736605867913639,
      "learning_rate": 4.6568460919247495e-06,
      "loss": 0.9745,
      "step": 2068
    },
    {
      "epoch": 2.532435740514076,
      "grad_norm": 1.5203476957695676,
      "learning_rate": 4.656521959904368e-06,
      "loss": 0.4698,
      "step": 2069
    },
    {
      "epoch": 2.533659730722154,
      "grad_norm": 0.9442290007645968,
      "learning_rate": 4.656197686165339e-06,
      "loss": 0.7653,
      "step": 2070
    },
    {
      "epoch": 2.5348837209302326,
      "grad_norm": 1.1070459465137836,
      "learning_rate": 4.655873270728973e-06,
      "loss": 1.6179,
      "step": 2071
    },
    {
      "epoch": 2.536107711138311,
      "grad_norm": 1.0707753501436956,
      "learning_rate": 4.655548713616591e-06,
      "loss": 0.5692,
      "step": 2072
    },
    {
      "epoch": 2.537331701346389,
      "grad_norm": 0.981975084742688,
      "learning_rate": 4.655224014849519e-06,
      "loss": 0.5867,
      "step": 2073
    },
    {
      "epoch": 2.5385556915544676,
      "grad_norm": 1.1371537352711047,
      "learning_rate": 4.654899174449097e-06,
      "loss": 0.5847,
      "step": 2074
    },
    {
      "epoch": 2.5397796817625458,
      "grad_norm": 1.3683627396272005,
      "learning_rate": 4.654574192436672e-06,
      "loss": 1.0362,
      "step": 2075
    },
    {
      "epoch": 2.5410036719706244,
      "grad_norm": 0.930662798267102,
      "learning_rate": 4.6542490688336e-06,
      "loss": 0.5217,
      "step": 2076
    },
    {
      "epoch": 2.5422276621787026,
      "grad_norm": 1.5138177920209508,
      "learning_rate": 4.653923803661246e-06,
      "loss": 0.472,
      "step": 2077
    },
    {
      "epoch": 2.543451652386781,
      "grad_norm": 1.7615472395071672,
      "learning_rate": 4.653598396940988e-06,
      "loss": 0.5755,
      "step": 2078
    },
    {
      "epoch": 2.5446756425948593,
      "grad_norm": 1.5124888994398478,
      "learning_rate": 4.653272848694208e-06,
      "loss": 0.3637,
      "step": 2079
    },
    {
      "epoch": 2.5458996328029375,
      "grad_norm": 1.3073437015323743,
      "learning_rate": 4.652947158942301e-06,
      "loss": 1.0822,
      "step": 2080
    },
    {
      "epoch": 2.547123623011016,
      "grad_norm": 1.2861262227901857,
      "learning_rate": 4.65262132770667e-06,
      "loss": 1.4235,
      "step": 2081
    },
    {
      "epoch": 2.5483476132190943,
      "grad_norm": 0.613898787695729,
      "learning_rate": 4.652295355008728e-06,
      "loss": 0.6397,
      "step": 2082
    },
    {
      "epoch": 2.5495716034271725,
      "grad_norm": 1.4213442493853043,
      "learning_rate": 4.651969240869895e-06,
      "loss": 0.4613,
      "step": 2083
    },
    {
      "epoch": 2.550795593635251,
      "grad_norm": 0.7241423234863972,
      "learning_rate": 4.651642985311604e-06,
      "loss": 0.6286,
      "step": 2084
    },
    {
      "epoch": 2.5520195838433293,
      "grad_norm": 1.3002659757692996,
      "learning_rate": 4.651316588355294e-06,
      "loss": 0.596,
      "step": 2085
    },
    {
      "epoch": 2.5532435740514074,
      "grad_norm": 0.8958377942237526,
      "learning_rate": 4.650990050022415e-06,
      "loss": 0.5603,
      "step": 2086
    },
    {
      "epoch": 2.554467564259486,
      "grad_norm": 0.8046171671990715,
      "learning_rate": 4.650663370334426e-06,
      "loss": 0.4431,
      "step": 2087
    },
    {
      "epoch": 2.5556915544675642,
      "grad_norm": 1.4507892131212132,
      "learning_rate": 4.6503365493127945e-06,
      "loss": 0.6924,
      "step": 2088
    },
    {
      "epoch": 2.5569155446756424,
      "grad_norm": 2.065000829353021,
      "learning_rate": 4.650009586978999e-06,
      "loss": 0.4829,
      "step": 2089
    },
    {
      "epoch": 2.558139534883721,
      "grad_norm": 1.2757677493302193,
      "learning_rate": 4.6496824833545254e-06,
      "loss": 0.7579,
      "step": 2090
    },
    {
      "epoch": 2.559363525091799,
      "grad_norm": 1.212731790994818,
      "learning_rate": 4.649355238460871e-06,
      "loss": 1.027,
      "step": 2091
    },
    {
      "epoch": 2.5605875152998774,
      "grad_norm": 1.0888089467420163,
      "learning_rate": 4.6490278523195395e-06,
      "loss": 0.9339,
      "step": 2092
    },
    {
      "epoch": 2.561811505507956,
      "grad_norm": 0.5234815759610121,
      "learning_rate": 4.6487003249520455e-06,
      "loss": 0.4572,
      "step": 2093
    },
    {
      "epoch": 2.563035495716034,
      "grad_norm": 0.6491394404938936,
      "learning_rate": 4.648372656379915e-06,
      "loss": 0.6298,
      "step": 2094
    },
    {
      "epoch": 2.5642594859241123,
      "grad_norm": 1.1445492213594157,
      "learning_rate": 4.64804484662468e-06,
      "loss": 0.6563,
      "step": 2095
    },
    {
      "epoch": 2.565483476132191,
      "grad_norm": 1.3857417468957276,
      "learning_rate": 4.647716895707881e-06,
      "loss": 0.5703,
      "step": 2096
    },
    {
      "epoch": 2.566707466340269,
      "grad_norm": 1.3535704607565835,
      "learning_rate": 4.647388803651073e-06,
      "loss": 0.5064,
      "step": 2097
    },
    {
      "epoch": 2.5679314565483478,
      "grad_norm": 1.1238696519048372,
      "learning_rate": 4.6470605704758155e-06,
      "loss": 0.5099,
      "step": 2098
    },
    {
      "epoch": 2.569155446756426,
      "grad_norm": 1.5234737344430358,
      "learning_rate": 4.646732196203679e-06,
      "loss": 0.5968,
      "step": 2099
    },
    {
      "epoch": 2.5703794369645045,
      "grad_norm": 0.5834258593778913,
      "learning_rate": 4.646403680856241e-06,
      "loss": 0.5622,
      "step": 2100
    },
    {
      "epoch": 2.5716034271725827,
      "grad_norm": 1.4431406327834029,
      "learning_rate": 4.646075024455093e-06,
      "loss": 0.5165,
      "step": 2101
    },
    {
      "epoch": 2.572827417380661,
      "grad_norm": 1.3054570412267728,
      "learning_rate": 4.645746227021833e-06,
      "loss": 0.6803,
      "step": 2102
    },
    {
      "epoch": 2.5740514075887395,
      "grad_norm": 1.038043509922551,
      "learning_rate": 4.645417288578066e-06,
      "loss": 0.4522,
      "step": 2103
    },
    {
      "epoch": 2.5752753977968177,
      "grad_norm": 1.2321864109612957,
      "learning_rate": 4.645088209145411e-06,
      "loss": 0.829,
      "step": 2104
    },
    {
      "epoch": 2.576499388004896,
      "grad_norm": 0.5519622620062806,
      "learning_rate": 4.644758988745493e-06,
      "loss": 0.5296,
      "step": 2105
    },
    {
      "epoch": 2.5777233782129745,
      "grad_norm": 1.3820656366175297,
      "learning_rate": 4.644429627399947e-06,
      "loss": 1.1597,
      "step": 2106
    },
    {
      "epoch": 2.5789473684210527,
      "grad_norm": 0.8421848227348346,
      "learning_rate": 4.644100125130418e-06,
      "loss": 0.5617,
      "step": 2107
    },
    {
      "epoch": 2.580171358629131,
      "grad_norm": 0.973204490085165,
      "learning_rate": 4.6437704819585595e-06,
      "loss": 0.5479,
      "step": 2108
    },
    {
      "epoch": 2.5813953488372094,
      "grad_norm": 1.138924017652126,
      "learning_rate": 4.643440697906033e-06,
      "loss": 0.4387,
      "step": 2109
    },
    {
      "epoch": 2.5826193390452876,
      "grad_norm": 0.9412391192922551,
      "learning_rate": 4.6431107729945126e-06,
      "loss": 0.5049,
      "step": 2110
    },
    {
      "epoch": 2.583843329253366,
      "grad_norm": 0.9261500306384691,
      "learning_rate": 4.642780707245679e-06,
      "loss": 0.6582,
      "step": 2111
    },
    {
      "epoch": 2.5850673194614444,
      "grad_norm": 0.9744912971493168,
      "learning_rate": 4.6424505006812225e-06,
      "loss": 0.9399,
      "step": 2112
    },
    {
      "epoch": 2.5862913096695226,
      "grad_norm": 1.7347745171051403,
      "learning_rate": 4.642120153322845e-06,
      "loss": 0.6806,
      "step": 2113
    },
    {
      "epoch": 2.5875152998776008,
      "grad_norm": 1.428678385152879,
      "learning_rate": 4.641789665192252e-06,
      "loss": 0.498,
      "step": 2114
    },
    {
      "epoch": 2.5887392900856794,
      "grad_norm": 0.8689637453607042,
      "learning_rate": 4.641459036311166e-06,
      "loss": 1.3624,
      "step": 2115
    },
    {
      "epoch": 2.5899632802937576,
      "grad_norm": 0.8764753770634701,
      "learning_rate": 4.641128266701311e-06,
      "loss": 0.7269,
      "step": 2116
    },
    {
      "epoch": 2.5911872705018357,
      "grad_norm": 0.9266386198802584,
      "learning_rate": 4.6407973563844274e-06,
      "loss": 0.5307,
      "step": 2117
    },
    {
      "epoch": 2.5924112607099143,
      "grad_norm": 1.264306659919823,
      "learning_rate": 4.640466305382259e-06,
      "loss": 0.5019,
      "step": 2118
    },
    {
      "epoch": 2.5936352509179925,
      "grad_norm": 0.8249426463975164,
      "learning_rate": 4.640135113716562e-06,
      "loss": 0.732,
      "step": 2119
    },
    {
      "epoch": 2.594859241126071,
      "grad_norm": 1.2102716489196568,
      "learning_rate": 4.639803781409102e-06,
      "loss": 0.554,
      "step": 2120
    },
    {
      "epoch": 2.5960832313341493,
      "grad_norm": 1.8983204772016684,
      "learning_rate": 4.639472308481651e-06,
      "loss": 0.4465,
      "step": 2121
    },
    {
      "epoch": 2.597307221542228,
      "grad_norm": 1.053892085320467,
      "learning_rate": 4.639140694955993e-06,
      "loss": 1.2047,
      "step": 2122
    },
    {
      "epoch": 2.598531211750306,
      "grad_norm": 1.3322411663293574,
      "learning_rate": 4.638808940853923e-06,
      "loss": 1.3746,
      "step": 2123
    },
    {
      "epoch": 2.5997552019583843,
      "grad_norm": 1.113072607181846,
      "learning_rate": 4.638477046197238e-06,
      "loss": 0.556,
      "step": 2124
    },
    {
      "epoch": 2.600979192166463,
      "grad_norm": 1.0452558193492478,
      "learning_rate": 4.638145011007752e-06,
      "loss": 1.5915,
      "step": 2125
    },
    {
      "epoch": 2.602203182374541,
      "grad_norm": 1.7101631569556102,
      "learning_rate": 4.637812835307284e-06,
      "loss": 0.8983,
      "step": 2126
    },
    {
      "epoch": 2.6034271725826192,
      "grad_norm": 1.0478733685906347,
      "learning_rate": 4.637480519117664e-06,
      "loss": 1.7102,
      "step": 2127
    },
    {
      "epoch": 2.604651162790698,
      "grad_norm": 0.9827442203121775,
      "learning_rate": 4.63714806246073e-06,
      "loss": 1.128,
      "step": 2128
    },
    {
      "epoch": 2.605875152998776,
      "grad_norm": 1.1787403419208504,
      "learning_rate": 4.63681546535833e-06,
      "loss": 0.6064,
      "step": 2129
    },
    {
      "epoch": 2.607099143206854,
      "grad_norm": 1.5854246342053935,
      "learning_rate": 4.636482727832322e-06,
      "loss": 0.4502,
      "step": 2130
    },
    {
      "epoch": 2.608323133414933,
      "grad_norm": 0.811103833794266,
      "learning_rate": 4.636149849904571e-06,
      "loss": 0.4228,
      "step": 2131
    },
    {
      "epoch": 2.609547123623011,
      "grad_norm": 1.3060986833270367,
      "learning_rate": 4.6358168315969536e-06,
      "loss": 0.7204,
      "step": 2132
    },
    {
      "epoch": 2.610771113831089,
      "grad_norm": 1.0307046918668448,
      "learning_rate": 4.6354836729313525e-06,
      "loss": 0.9887,
      "step": 2133
    },
    {
      "epoch": 2.611995104039168,
      "grad_norm": 0.9994328272538338,
      "learning_rate": 4.635150373929663e-06,
      "loss": 0.6914,
      "step": 2134
    },
    {
      "epoch": 2.613219094247246,
      "grad_norm": 1.3154941480864162,
      "learning_rate": 4.63481693461379e-06,
      "loss": 1.1409,
      "step": 2135
    },
    {
      "epoch": 2.614443084455324,
      "grad_norm": 1.60115088296653,
      "learning_rate": 4.634483355005642e-06,
      "loss": 0.4494,
      "step": 2136
    },
    {
      "epoch": 2.6156670746634028,
      "grad_norm": 1.076399643997741,
      "learning_rate": 4.634149635127144e-06,
      "loss": 0.5958,
      "step": 2137
    },
    {
      "epoch": 2.616891064871481,
      "grad_norm": 1.0069916996313146,
      "learning_rate": 4.633815775000226e-06,
      "loss": 0.5668,
      "step": 2138
    },
    {
      "epoch": 2.618115055079559,
      "grad_norm": 1.0801143560442514,
      "learning_rate": 4.633481774646827e-06,
      "loss": 0.6644,
      "step": 2139
    },
    {
      "epoch": 2.6193390452876377,
      "grad_norm": 1.1115074894600387,
      "learning_rate": 4.633147634088897e-06,
      "loss": 1.6084,
      "step": 2140
    },
    {
      "epoch": 2.620563035495716,
      "grad_norm": 0.915699651577624,
      "learning_rate": 4.632813353348395e-06,
      "loss": 0.688,
      "step": 2141
    },
    {
      "epoch": 2.6217870257037945,
      "grad_norm": 0.6651109822995419,
      "learning_rate": 4.632478932447288e-06,
      "loss": 0.4301,
      "step": 2142
    },
    {
      "epoch": 2.6230110159118727,
      "grad_norm": 0.5451588635291792,
      "learning_rate": 4.632144371407553e-06,
      "loss": 0.4424,
      "step": 2143
    },
    {
      "epoch": 2.6242350061199513,
      "grad_norm": 0.8060212855822413,
      "learning_rate": 4.631809670251176e-06,
      "loss": 0.5628,
      "step": 2144
    },
    {
      "epoch": 2.6254589963280295,
      "grad_norm": 0.8267720826299793,
      "learning_rate": 4.631474829000152e-06,
      "loss": 0.6211,
      "step": 2145
    },
    {
      "epoch": 2.6266829865361077,
      "grad_norm": 1.2719840398177333,
      "learning_rate": 4.631139847676487e-06,
      "loss": 0.4524,
      "step": 2146
    },
    {
      "epoch": 2.6279069767441863,
      "grad_norm": 1.147529101825175,
      "learning_rate": 4.630804726302194e-06,
      "loss": 0.6644,
      "step": 2147
    },
    {
      "epoch": 2.6291309669522644,
      "grad_norm": 0.9910113537444569,
      "learning_rate": 4.630469464899294e-06,
      "loss": 0.6042,
      "step": 2148
    },
    {
      "epoch": 2.6303549571603426,
      "grad_norm": 0.8194425000132584,
      "learning_rate": 4.630134063489822e-06,
      "loss": 1.1833,
      "step": 2149
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 1.7692079267470275,
      "learning_rate": 4.629798522095818e-06,
      "loss": 0.4135,
      "step": 2150
    },
    {
      "epoch": 2.6328029375764994,
      "grad_norm": 0.7210987684959724,
      "learning_rate": 4.629462840739333e-06,
      "loss": 0.719,
      "step": 2151
    },
    {
      "epoch": 2.6340269277845776,
      "grad_norm": 1.3266655232925069,
      "learning_rate": 4.629127019442426e-06,
      "loss": 0.6338,
      "step": 2152
    },
    {
      "epoch": 2.635250917992656,
      "grad_norm": 1.4842063916601462,
      "learning_rate": 4.628791058227167e-06,
      "loss": 0.6736,
      "step": 2153
    },
    {
      "epoch": 2.6364749082007344,
      "grad_norm": 1.5476080308147022,
      "learning_rate": 4.6284549571156325e-06,
      "loss": 0.5478,
      "step": 2154
    },
    {
      "epoch": 2.6376988984088126,
      "grad_norm": 0.9730951880799199,
      "learning_rate": 4.6281187161299115e-06,
      "loss": 0.7473,
      "step": 2155
    },
    {
      "epoch": 2.638922888616891,
      "grad_norm": 1.0073362179897392,
      "learning_rate": 4.627782335292099e-06,
      "loss": 1.3489,
      "step": 2156
    },
    {
      "epoch": 2.6401468788249693,
      "grad_norm": 1.0157339041814044,
      "learning_rate": 4.6274458146243026e-06,
      "loss": 0.8173,
      "step": 2157
    },
    {
      "epoch": 2.6413708690330475,
      "grad_norm": 1.0605597767282686,
      "learning_rate": 4.627109154148636e-06,
      "loss": 0.5332,
      "step": 2158
    },
    {
      "epoch": 2.642594859241126,
      "grad_norm": 1.084474440279894,
      "learning_rate": 4.626772353887223e-06,
      "loss": 1.2364,
      "step": 2159
    },
    {
      "epoch": 2.6438188494492043,
      "grad_norm": 1.0577835527442867,
      "learning_rate": 4.626435413862198e-06,
      "loss": 0.5233,
      "step": 2160
    },
    {
      "epoch": 2.6450428396572825,
      "grad_norm": 0.8876364224576155,
      "learning_rate": 4.626098334095703e-06,
      "loss": 0.8672,
      "step": 2161
    },
    {
      "epoch": 2.646266829865361,
      "grad_norm": 0.8010456062850537,
      "learning_rate": 4.625761114609888e-06,
      "loss": 0.6291,
      "step": 2162
    },
    {
      "epoch": 2.6474908200734393,
      "grad_norm": 1.1315509599497875,
      "learning_rate": 4.625423755426915e-06,
      "loss": 0.6149,
      "step": 2163
    },
    {
      "epoch": 2.648714810281518,
      "grad_norm": 1.021956061330222,
      "learning_rate": 4.6250862565689555e-06,
      "loss": 1.3148,
      "step": 2164
    },
    {
      "epoch": 2.649938800489596,
      "grad_norm": 1.3257506690009997,
      "learning_rate": 4.624748618058187e-06,
      "loss": 1.1076,
      "step": 2165
    },
    {
      "epoch": 2.6511627906976747,
      "grad_norm": 1.1823709950591617,
      "learning_rate": 4.624410839916798e-06,
      "loss": 0.8471,
      "step": 2166
    },
    {
      "epoch": 2.652386780905753,
      "grad_norm": 0.6507203521302883,
      "learning_rate": 4.624072922166987e-06,
      "loss": 0.5786,
      "step": 2167
    },
    {
      "epoch": 2.653610771113831,
      "grad_norm": 0.966142696279852,
      "learning_rate": 4.6237348648309585e-06,
      "loss": 0.5926,
      "step": 2168
    },
    {
      "epoch": 2.6548347613219097,
      "grad_norm": 0.6724944900159797,
      "learning_rate": 4.62339666793093e-06,
      "loss": 0.649,
      "step": 2169
    },
    {
      "epoch": 2.656058751529988,
      "grad_norm": 1.168565142666766,
      "learning_rate": 4.623058331489127e-06,
      "loss": 1.6315,
      "step": 2170
    },
    {
      "epoch": 2.657282741738066,
      "grad_norm": 1.2719328927201698,
      "learning_rate": 4.622719855527784e-06,
      "loss": 0.616,
      "step": 2171
    },
    {
      "epoch": 2.6585067319461446,
      "grad_norm": 1.344854066804907,
      "learning_rate": 4.622381240069142e-06,
      "loss": 1.0501,
      "step": 2172
    },
    {
      "epoch": 2.659730722154223,
      "grad_norm": 1.2125485941656362,
      "learning_rate": 4.6220424851354565e-06,
      "loss": 0.611,
      "step": 2173
    },
    {
      "epoch": 2.660954712362301,
      "grad_norm": 1.4218069751659992,
      "learning_rate": 4.621703590748987e-06,
      "loss": 0.7035,
      "step": 2174
    },
    {
      "epoch": 2.6621787025703796,
      "grad_norm": 1.0594397384972658,
      "learning_rate": 4.621364556932005e-06,
      "loss": 0.5976,
      "step": 2175
    },
    {
      "epoch": 2.6634026927784578,
      "grad_norm": 0.7992949863966209,
      "learning_rate": 4.621025383706791e-06,
      "loss": 0.3578,
      "step": 2176
    },
    {
      "epoch": 2.664626682986536,
      "grad_norm": 0.7823159089600693,
      "learning_rate": 4.620686071095634e-06,
      "loss": 0.7954,
      "step": 2177
    },
    {
      "epoch": 2.6658506731946146,
      "grad_norm": 1.3886335651871597,
      "learning_rate": 4.620346619120831e-06,
      "loss": 0.5775,
      "step": 2178
    },
    {
      "epoch": 2.6670746634026927,
      "grad_norm": 1.175061970440519,
      "learning_rate": 4.620007027804692e-06,
      "loss": 0.7875,
      "step": 2179
    },
    {
      "epoch": 2.668298653610771,
      "grad_norm": 1.1149459901827092,
      "learning_rate": 4.619667297169532e-06,
      "loss": 0.5361,
      "step": 2180
    },
    {
      "epoch": 2.6695226438188495,
      "grad_norm": 1.2546370272138538,
      "learning_rate": 4.619327427237678e-06,
      "loss": 1.3836,
      "step": 2181
    },
    {
      "epoch": 2.6707466340269277,
      "grad_norm": 1.2139980903232455,
      "learning_rate": 4.618987418031464e-06,
      "loss": 1.1049,
      "step": 2182
    },
    {
      "epoch": 2.671970624235006,
      "grad_norm": 0.8111423860226766,
      "learning_rate": 4.618647269573234e-06,
      "loss": 0.787,
      "step": 2183
    },
    {
      "epoch": 2.6731946144430845,
      "grad_norm": 1.0052890159337822,
      "learning_rate": 4.618306981885343e-06,
      "loss": 0.8031,
      "step": 2184
    },
    {
      "epoch": 2.6744186046511627,
      "grad_norm": 0.718479367574715,
      "learning_rate": 4.617966554990151e-06,
      "loss": 0.6532,
      "step": 2185
    },
    {
      "epoch": 2.6756425948592413,
      "grad_norm": 0.9307897768065131,
      "learning_rate": 4.617625988910031e-06,
      "loss": 1.02,
      "step": 2186
    },
    {
      "epoch": 2.6768665850673194,
      "grad_norm": 0.7953223888724118,
      "learning_rate": 4.6172852836673645e-06,
      "loss": 0.6864,
      "step": 2187
    },
    {
      "epoch": 2.678090575275398,
      "grad_norm": 1.116667893658052,
      "learning_rate": 4.61694443928454e-06,
      "loss": 0.4732,
      "step": 2188
    },
    {
      "epoch": 2.6793145654834762,
      "grad_norm": 1.3605382580754222,
      "learning_rate": 4.6166034557839564e-06,
      "loss": 0.5321,
      "step": 2189
    },
    {
      "epoch": 2.6805385556915544,
      "grad_norm": 0.9116926354367513,
      "learning_rate": 4.616262333188023e-06,
      "loss": 0.5599,
      "step": 2190
    },
    {
      "epoch": 2.681762545899633,
      "grad_norm": 1.2314200615173392,
      "learning_rate": 4.615921071519156e-06,
      "loss": 0.7003,
      "step": 2191
    },
    {
      "epoch": 2.682986536107711,
      "grad_norm": 0.9877864767465424,
      "learning_rate": 4.615579670799783e-06,
      "loss": 0.5615,
      "step": 2192
    },
    {
      "epoch": 2.6842105263157894,
      "grad_norm": 0.9090137941070238,
      "learning_rate": 4.615238131052339e-06,
      "loss": 0.5243,
      "step": 2193
    },
    {
      "epoch": 2.685434516523868,
      "grad_norm": 1.1611064069556565,
      "learning_rate": 4.614896452299269e-06,
      "loss": 1.0496,
      "step": 2194
    },
    {
      "epoch": 2.686658506731946,
      "grad_norm": 1.7218852099072395,
      "learning_rate": 4.614554634563026e-06,
      "loss": 0.8886,
      "step": 2195
    },
    {
      "epoch": 2.6878824969400243,
      "grad_norm": 0.9583011073411407,
      "learning_rate": 4.614212677866073e-06,
      "loss": 0.5189,
      "step": 2196
    },
    {
      "epoch": 2.689106487148103,
      "grad_norm": 1.179327323885605,
      "learning_rate": 4.6138705822308845e-06,
      "loss": 1.1716,
      "step": 2197
    },
    {
      "epoch": 2.690330477356181,
      "grad_norm": 0.8044303267553937,
      "learning_rate": 4.613528347679938e-06,
      "loss": 1.0175,
      "step": 2198
    },
    {
      "epoch": 2.6915544675642593,
      "grad_norm": 1.3198960272056917,
      "learning_rate": 4.613185974235728e-06,
      "loss": 1.1914,
      "step": 2199
    },
    {
      "epoch": 2.692778457772338,
      "grad_norm": 1.150438155135514,
      "learning_rate": 4.61284346192075e-06,
      "loss": 0.6387,
      "step": 2200
    },
    {
      "epoch": 2.694002447980416,
      "grad_norm": 0.8815506167757904,
      "learning_rate": 4.6125008107575165e-06,
      "loss": 0.6185,
      "step": 2201
    },
    {
      "epoch": 2.6952264381884943,
      "grad_norm": 1.8015984886502048,
      "learning_rate": 4.612158020768542e-06,
      "loss": 0.5164,
      "step": 2202
    },
    {
      "epoch": 2.696450428396573,
      "grad_norm": 0.7105515922917889,
      "learning_rate": 4.611815091976355e-06,
      "loss": 0.4845,
      "step": 2203
    },
    {
      "epoch": 2.697674418604651,
      "grad_norm": 1.263268621234841,
      "learning_rate": 4.611472024403491e-06,
      "loss": 0.5095,
      "step": 2204
    },
    {
      "epoch": 2.6988984088127292,
      "grad_norm": 1.1299113743075342,
      "learning_rate": 4.611128818072496e-06,
      "loss": 0.5199,
      "step": 2205
    },
    {
      "epoch": 2.700122399020808,
      "grad_norm": 0.998282353987232,
      "learning_rate": 4.610785473005923e-06,
      "loss": 0.49,
      "step": 2206
    },
    {
      "epoch": 2.701346389228886,
      "grad_norm": 1.661492000975268,
      "learning_rate": 4.610441989226336e-06,
      "loss": 0.4674,
      "step": 2207
    },
    {
      "epoch": 2.7025703794369647,
      "grad_norm": 0.8247924981345542,
      "learning_rate": 4.610098366756308e-06,
      "loss": 0.7629,
      "step": 2208
    },
    {
      "epoch": 2.703794369645043,
      "grad_norm": 1.086138293508475,
      "learning_rate": 4.609754605618419e-06,
      "loss": 1.2668,
      "step": 2209
    },
    {
      "epoch": 2.7050183598531214,
      "grad_norm": 1.2785328721665017,
      "learning_rate": 4.609410705835261e-06,
      "loss": 0.5645,
      "step": 2210
    },
    {
      "epoch": 2.7062423500611996,
      "grad_norm": 0.9603712212398866,
      "learning_rate": 4.609066667429435e-06,
      "loss": 0.6446,
      "step": 2211
    },
    {
      "epoch": 2.707466340269278,
      "grad_norm": 0.8926265725118112,
      "learning_rate": 4.6087224904235475e-06,
      "loss": 0.7601,
      "step": 2212
    },
    {
      "epoch": 2.7086903304773564,
      "grad_norm": 0.857920306137564,
      "learning_rate": 4.6083781748402165e-06,
      "loss": 0.7179,
      "step": 2213
    },
    {
      "epoch": 2.7099143206854346,
      "grad_norm": 0.8287416172133364,
      "learning_rate": 4.608033720702072e-06,
      "loss": 0.3788,
      "step": 2214
    },
    {
      "epoch": 2.7111383108935128,
      "grad_norm": 2.0643575744888283,
      "learning_rate": 4.607689128031747e-06,
      "loss": 0.5858,
      "step": 2215
    },
    {
      "epoch": 2.7123623011015914,
      "grad_norm": 1.1590016676509982,
      "learning_rate": 4.6073443968518894e-06,
      "loss": 0.8128,
      "step": 2216
    },
    {
      "epoch": 2.7135862913096696,
      "grad_norm": 1.117295246761789,
      "learning_rate": 4.606999527185152e-06,
      "loss": 1.6738,
      "step": 2217
    },
    {
      "epoch": 2.7148102815177477,
      "grad_norm": 1.4590982023502928,
      "learning_rate": 4.606654519054199e-06,
      "loss": 0.425,
      "step": 2218
    },
    {
      "epoch": 2.7160342717258263,
      "grad_norm": 0.7570633227803636,
      "learning_rate": 4.606309372481703e-06,
      "loss": 0.5234,
      "step": 2219
    },
    {
      "epoch": 2.7172582619339045,
      "grad_norm": 0.6862282775088852,
      "learning_rate": 4.605964087490346e-06,
      "loss": 0.4874,
      "step": 2220
    },
    {
      "epoch": 2.7184822521419827,
      "grad_norm": 0.8190324031988453,
      "learning_rate": 4.605618664102818e-06,
      "loss": 0.4168,
      "step": 2221
    },
    {
      "epoch": 2.7197062423500613,
      "grad_norm": 0.8403264398358936,
      "learning_rate": 4.60527310234182e-06,
      "loss": 0.615,
      "step": 2222
    },
    {
      "epoch": 2.7209302325581395,
      "grad_norm": 1.413856088711858,
      "learning_rate": 4.604927402230061e-06,
      "loss": 0.8246,
      "step": 2223
    },
    {
      "epoch": 2.7221542227662177,
      "grad_norm": 1.0743724007112436,
      "learning_rate": 4.604581563790258e-06,
      "loss": 0.7938,
      "step": 2224
    },
    {
      "epoch": 2.7233782129742963,
      "grad_norm": 1.3102138301835118,
      "learning_rate": 4.60423558704514e-06,
      "loss": 0.457,
      "step": 2225
    },
    {
      "epoch": 2.7246022031823744,
      "grad_norm": 1.0664313571969872,
      "learning_rate": 4.603889472017441e-06,
      "loss": 2.3426,
      "step": 2226
    },
    {
      "epoch": 2.7258261933904526,
      "grad_norm": 1.3224402311265275,
      "learning_rate": 4.603543218729908e-06,
      "loss": 0.4438,
      "step": 2227
    },
    {
      "epoch": 2.7270501835985312,
      "grad_norm": 0.9699778630985946,
      "learning_rate": 4.603196827205295e-06,
      "loss": 1.0323,
      "step": 2228
    },
    {
      "epoch": 2.7282741738066094,
      "grad_norm": 0.7747721059239158,
      "learning_rate": 4.602850297466366e-06,
      "loss": 1.0796,
      "step": 2229
    },
    {
      "epoch": 2.729498164014688,
      "grad_norm": 1.2622997414562305,
      "learning_rate": 4.602503629535893e-06,
      "loss": 0.8345,
      "step": 2230
    },
    {
      "epoch": 2.730722154222766,
      "grad_norm": 1.0969158387174398,
      "learning_rate": 4.6021568234366585e-06,
      "loss": 0.6197,
      "step": 2231
    },
    {
      "epoch": 2.731946144430845,
      "grad_norm": 1.040415349158388,
      "learning_rate": 4.601809879191452e-06,
      "loss": 0.4532,
      "step": 2232
    },
    {
      "epoch": 2.733170134638923,
      "grad_norm": 0.7887428701695565,
      "learning_rate": 4.6014627968230755e-06,
      "loss": 0.8087,
      "step": 2233
    },
    {
      "epoch": 2.734394124847001,
      "grad_norm": 0.9648800372527141,
      "learning_rate": 4.601115576354336e-06,
      "loss": 0.5799,
      "step": 2234
    },
    {
      "epoch": 2.73561811505508,
      "grad_norm": 1.1590714379858513,
      "learning_rate": 4.600768217808053e-06,
      "loss": 0.5621,
      "step": 2235
    },
    {
      "epoch": 2.736842105263158,
      "grad_norm": 1.0741543526118316,
      "learning_rate": 4.600420721207053e-06,
      "loss": 0.557,
      "step": 2236
    },
    {
      "epoch": 2.738066095471236,
      "grad_norm": 0.9037531903754719,
      "learning_rate": 4.600073086574171e-06,
      "loss": 0.4592,
      "step": 2237
    },
    {
      "epoch": 2.7392900856793148,
      "grad_norm": 1.0856531550094826,
      "learning_rate": 4.599725313932254e-06,
      "loss": 1.0754,
      "step": 2238
    },
    {
      "epoch": 2.740514075887393,
      "grad_norm": 0.6668802893545525,
      "learning_rate": 4.599377403304156e-06,
      "loss": 0.4936,
      "step": 2239
    },
    {
      "epoch": 2.741738066095471,
      "grad_norm": 1.148784380288063,
      "learning_rate": 4.59902935471274e-06,
      "loss": 0.6099,
      "step": 2240
    },
    {
      "epoch": 2.7429620563035497,
      "grad_norm": 0.8239190422818863,
      "learning_rate": 4.598681168180879e-06,
      "loss": 0.6652,
      "step": 2241
    },
    {
      "epoch": 2.744186046511628,
      "grad_norm": 1.3137469417793242,
      "learning_rate": 4.598332843731453e-06,
      "loss": 0.5213,
      "step": 2242
    },
    {
      "epoch": 2.745410036719706,
      "grad_norm": 0.9468270621698734,
      "learning_rate": 4.597984381387354e-06,
      "loss": 0.9203,
      "step": 2243
    },
    {
      "epoch": 2.7466340269277847,
      "grad_norm": 1.0880887778647201,
      "learning_rate": 4.597635781171482e-06,
      "loss": 1.5164,
      "step": 2244
    },
    {
      "epoch": 2.747858017135863,
      "grad_norm": 0.5814260077638449,
      "learning_rate": 4.5972870431067445e-06,
      "loss": 0.4183,
      "step": 2245
    },
    {
      "epoch": 2.749082007343941,
      "grad_norm": 1.310491874131183,
      "learning_rate": 4.5969381672160595e-06,
      "loss": 1.2752,
      "step": 2246
    },
    {
      "epoch": 2.7503059975520197,
      "grad_norm": 1.6686413193395975,
      "learning_rate": 4.596589153522355e-06,
      "loss": 0.5235,
      "step": 2247
    },
    {
      "epoch": 2.751529987760098,
      "grad_norm": 0.7873642646520717,
      "learning_rate": 4.5962400020485665e-06,
      "loss": 0.8017,
      "step": 2248
    },
    {
      "epoch": 2.752753977968176,
      "grad_norm": 1.2956362462731228,
      "learning_rate": 4.595890712817638e-06,
      "loss": 0.8232,
      "step": 2249
    },
    {
      "epoch": 2.7539779681762546,
      "grad_norm": 0.822849908110494,
      "learning_rate": 4.5955412858525234e-06,
      "loss": 0.8067,
      "step": 2250
    },
    {
      "epoch": 2.755201958384333,
      "grad_norm": 1.2634016887887665,
      "learning_rate": 4.595191721176187e-06,
      "loss": 0.3891,
      "step": 2251
    },
    {
      "epoch": 2.7564259485924114,
      "grad_norm": 0.6899623472126725,
      "learning_rate": 4.5948420188116e-06,
      "loss": 0.4871,
      "step": 2252
    },
    {
      "epoch": 2.7576499388004896,
      "grad_norm": 0.658962596982903,
      "learning_rate": 4.594492178781744e-06,
      "loss": 0.6795,
      "step": 2253
    },
    {
      "epoch": 2.758873929008568,
      "grad_norm": 1.0216352817911467,
      "learning_rate": 4.5941422011096085e-06,
      "loss": 0.5991,
      "step": 2254
    },
    {
      "epoch": 2.7600979192166464,
      "grad_norm": 1.035605098942707,
      "learning_rate": 4.593792085818194e-06,
      "loss": 1.365,
      "step": 2255
    },
    {
      "epoch": 2.7613219094247246,
      "grad_norm": 0.804696328358046,
      "learning_rate": 4.593441832930507e-06,
      "loss": 0.7315,
      "step": 2256
    },
    {
      "epoch": 2.762545899632803,
      "grad_norm": 1.0066518364921733,
      "learning_rate": 4.593091442469567e-06,
      "loss": 0.6171,
      "step": 2257
    },
    {
      "epoch": 2.7637698898408813,
      "grad_norm": 1.0754940594927935,
      "learning_rate": 4.592740914458399e-06,
      "loss": 0.6011,
      "step": 2258
    },
    {
      "epoch": 2.7649938800489595,
      "grad_norm": 1.3505625346939523,
      "learning_rate": 4.5923902489200375e-06,
      "loss": 1.0946,
      "step": 2259
    },
    {
      "epoch": 2.766217870257038,
      "grad_norm": 0.7686285404360617,
      "learning_rate": 4.59203944587753e-06,
      "loss": 0.5548,
      "step": 2260
    },
    {
      "epoch": 2.7674418604651163,
      "grad_norm": 1.9457788722992384,
      "learning_rate": 4.5916885053539265e-06,
      "loss": 0.4643,
      "step": 2261
    },
    {
      "epoch": 2.7686658506731945,
      "grad_norm": 0.6454646889900293,
      "learning_rate": 4.591337427372291e-06,
      "loss": 0.4094,
      "step": 2262
    },
    {
      "epoch": 2.769889840881273,
      "grad_norm": 0.9204165251327752,
      "learning_rate": 4.590986211955696e-06,
      "loss": 0.5843,
      "step": 2263
    },
    {
      "epoch": 2.7711138310893513,
      "grad_norm": 1.6162005424265886,
      "learning_rate": 4.590634859127221e-06,
      "loss": 0.601,
      "step": 2264
    },
    {
      "epoch": 2.7723378212974294,
      "grad_norm": 0.8096770319473827,
      "learning_rate": 4.590283368909955e-06,
      "loss": 0.5159,
      "step": 2265
    },
    {
      "epoch": 2.773561811505508,
      "grad_norm": 1.1541709038458718,
      "learning_rate": 4.5899317413269985e-06,
      "loss": 1.428,
      "step": 2266
    },
    {
      "epoch": 2.7747858017135862,
      "grad_norm": 1.7885055755938009,
      "learning_rate": 4.589579976401457e-06,
      "loss": 0.5631,
      "step": 2267
    },
    {
      "epoch": 2.7760097919216644,
      "grad_norm": 1.032566638707591,
      "learning_rate": 4.589228074156449e-06,
      "loss": 0.5885,
      "step": 2268
    },
    {
      "epoch": 2.777233782129743,
      "grad_norm": 0.7628429913513141,
      "learning_rate": 4.5888760346150996e-06,
      "loss": 0.6279,
      "step": 2269
    },
    {
      "epoch": 2.778457772337821,
      "grad_norm": 0.7455540974133628,
      "learning_rate": 4.588523857800543e-06,
      "loss": 0.6725,
      "step": 2270
    },
    {
      "epoch": 2.7796817625458994,
      "grad_norm": 0.8248441580188699,
      "learning_rate": 4.588171543735923e-06,
      "loss": 0.5995,
      "step": 2271
    },
    {
      "epoch": 2.780905752753978,
      "grad_norm": 0.7433673929799363,
      "learning_rate": 4.587819092444393e-06,
      "loss": 0.5333,
      "step": 2272
    },
    {
      "epoch": 2.782129742962056,
      "grad_norm": 1.519701228297172,
      "learning_rate": 4.587466503949115e-06,
      "loss": 0.768,
      "step": 2273
    },
    {
      "epoch": 2.783353733170135,
      "grad_norm": 1.1656685465472054,
      "learning_rate": 4.587113778273259e-06,
      "loss": 1.1489,
      "step": 2274
    },
    {
      "epoch": 2.784577723378213,
      "grad_norm": 1.3151148098914216,
      "learning_rate": 4.586760915440006e-06,
      "loss": 1.1794,
      "step": 2275
    },
    {
      "epoch": 2.7858017135862916,
      "grad_norm": 1.309896871608821,
      "learning_rate": 4.5864079154725434e-06,
      "loss": 1.2013,
      "step": 2276
    },
    {
      "epoch": 2.7870257037943698,
      "grad_norm": 0.9688787160185232,
      "learning_rate": 4.5860547783940695e-06,
      "loss": 0.9835,
      "step": 2277
    },
    {
      "epoch": 2.788249694002448,
      "grad_norm": 1.5116010840969039,
      "learning_rate": 4.585701504227792e-06,
      "loss": 1.1028,
      "step": 2278
    },
    {
      "epoch": 2.7894736842105265,
      "grad_norm": 1.2119840777978987,
      "learning_rate": 4.585348092996925e-06,
      "loss": 1.3838,
      "step": 2279
    },
    {
      "epoch": 2.7906976744186047,
      "grad_norm": 1.2702887399958727,
      "learning_rate": 4.584994544724695e-06,
      "loss": 0.6521,
      "step": 2280
    },
    {
      "epoch": 2.791921664626683,
      "grad_norm": 1.284024645936518,
      "learning_rate": 4.584640859434336e-06,
      "loss": 1.223,
      "step": 2281
    },
    {
      "epoch": 2.7931456548347615,
      "grad_norm": 0.6033871075379328,
      "learning_rate": 4.58428703714909e-06,
      "loss": 0.6341,
      "step": 2282
    },
    {
      "epoch": 2.7943696450428397,
      "grad_norm": 1.0820106944975485,
      "learning_rate": 4.5839330778922085e-06,
      "loss": 0.5292,
      "step": 2283
    },
    {
      "epoch": 2.795593635250918,
      "grad_norm": 1.119611054153639,
      "learning_rate": 4.583578981686953e-06,
      "loss": 0.6569,
      "step": 2284
    },
    {
      "epoch": 2.7968176254589965,
      "grad_norm": 1.0904514881992535,
      "learning_rate": 4.583224748556595e-06,
      "loss": 0.536,
      "step": 2285
    },
    {
      "epoch": 2.7980416156670747,
      "grad_norm": 1.1186132328737495,
      "learning_rate": 4.58287037852441e-06,
      "loss": 0.6915,
      "step": 2286
    },
    {
      "epoch": 2.799265605875153,
      "grad_norm": 1.227743888880496,
      "learning_rate": 4.582515871613688e-06,
      "loss": 1.3577,
      "step": 2287
    },
    {
      "epoch": 2.8004895960832314,
      "grad_norm": 0.7959143564589592,
      "learning_rate": 4.582161227847727e-06,
      "loss": 1.0136,
      "step": 2288
    },
    {
      "epoch": 2.8017135862913096,
      "grad_norm": 0.981771929690956,
      "learning_rate": 4.58180644724983e-06,
      "loss": 0.9127,
      "step": 2289
    },
    {
      "epoch": 2.802937576499388,
      "grad_norm": 1.5852002396349985,
      "learning_rate": 4.581451529843315e-06,
      "loss": 1.0343,
      "step": 2290
    },
    {
      "epoch": 2.8041615667074664,
      "grad_norm": 0.9295811380355233,
      "learning_rate": 4.581096475651503e-06,
      "loss": 0.5537,
      "step": 2291
    },
    {
      "epoch": 2.8053855569155446,
      "grad_norm": 0.7689469187901813,
      "learning_rate": 4.5807412846977284e-06,
      "loss": 0.4361,
      "step": 2292
    },
    {
      "epoch": 2.8066095471236228,
      "grad_norm": 1.4073082078724117,
      "learning_rate": 4.580385957005332e-06,
      "loss": 0.5486,
      "step": 2293
    },
    {
      "epoch": 2.8078335373317014,
      "grad_norm": 1.3424227123587704,
      "learning_rate": 4.580030492597666e-06,
      "loss": 0.7112,
      "step": 2294
    },
    {
      "epoch": 2.8090575275397796,
      "grad_norm": 0.6976926937454897,
      "learning_rate": 4.579674891498089e-06,
      "loss": 0.7067,
      "step": 2295
    },
    {
      "epoch": 2.810281517747858,
      "grad_norm": 1.2006421543190517,
      "learning_rate": 4.579319153729971e-06,
      "loss": 1.9528,
      "step": 2296
    },
    {
      "epoch": 2.8115055079559363,
      "grad_norm": 0.8914891056007683,
      "learning_rate": 4.578963279316689e-06,
      "loss": 1.1309,
      "step": 2297
    },
    {
      "epoch": 2.812729498164015,
      "grad_norm": 1.0439420133212902,
      "learning_rate": 4.578607268281629e-06,
      "loss": 0.7695,
      "step": 2298
    },
    {
      "epoch": 2.813953488372093,
      "grad_norm": 1.284679320225621,
      "learning_rate": 4.578251120648188e-06,
      "loss": 0.9219,
      "step": 2299
    },
    {
      "epoch": 2.8151774785801713,
      "grad_norm": 1.818793314465752,
      "learning_rate": 4.577894836439771e-06,
      "loss": 0.4947,
      "step": 2300
    },
    {
      "epoch": 2.81640146878825,
      "grad_norm": 1.0572098724423684,
      "learning_rate": 4.5775384156797904e-06,
      "loss": 0.5801,
      "step": 2301
    },
    {
      "epoch": 2.817625458996328,
      "grad_norm": 1.9335240813462298,
      "learning_rate": 4.57718185839167e-06,
      "loss": 0.5433,
      "step": 2302
    },
    {
      "epoch": 2.8188494492044063,
      "grad_norm": 1.0827248861118537,
      "learning_rate": 4.57682516459884e-06,
      "loss": 1.2535,
      "step": 2303
    },
    {
      "epoch": 2.820073439412485,
      "grad_norm": 1.2316971626640572,
      "learning_rate": 4.576468334324742e-06,
      "loss": 0.625,
      "step": 2304
    },
    {
      "epoch": 2.821297429620563,
      "grad_norm": 0.8461271391270831,
      "learning_rate": 4.576111367592825e-06,
      "loss": 0.5924,
      "step": 2305
    },
    {
      "epoch": 2.8225214198286412,
      "grad_norm": 1.1401648420785262,
      "learning_rate": 4.575754264426548e-06,
      "loss": 0.7052,
      "step": 2306
    },
    {
      "epoch": 2.82374541003672,
      "grad_norm": 0.782319223172386,
      "learning_rate": 4.57539702484938e-06,
      "loss": 0.4864,
      "step": 2307
    },
    {
      "epoch": 2.824969400244798,
      "grad_norm": 0.9288651647549095,
      "learning_rate": 4.575039648884795e-06,
      "loss": 1.2571,
      "step": 2308
    },
    {
      "epoch": 2.826193390452876,
      "grad_norm": 0.6459261678319496,
      "learning_rate": 4.574682136556278e-06,
      "loss": 0.3472,
      "step": 2309
    },
    {
      "epoch": 2.827417380660955,
      "grad_norm": 0.8977162777791443,
      "learning_rate": 4.574324487887326e-06,
      "loss": 0.5938,
      "step": 2310
    },
    {
      "epoch": 2.828641370869033,
      "grad_norm": 1.0766223020863774,
      "learning_rate": 4.5739667029014416e-06,
      "loss": 0.7092,
      "step": 2311
    },
    {
      "epoch": 2.829865361077111,
      "grad_norm": 2.120350913128021,
      "learning_rate": 4.573608781622136e-06,
      "loss": 0.5432,
      "step": 2312
    },
    {
      "epoch": 2.83108935128519,
      "grad_norm": 1.0656715219005652,
      "learning_rate": 4.57325072407293e-06,
      "loss": 0.5306,
      "step": 2313
    },
    {
      "epoch": 2.832313341493268,
      "grad_norm": 1.710331354721117,
      "learning_rate": 4.5728925302773555e-06,
      "loss": 0.5961,
      "step": 2314
    },
    {
      "epoch": 2.833537331701346,
      "grad_norm": 1.2285609946303768,
      "learning_rate": 4.572534200258951e-06,
      "loss": 0.4577,
      "step": 2315
    },
    {
      "epoch": 2.8347613219094248,
      "grad_norm": 1.45828512526589,
      "learning_rate": 4.572175734041265e-06,
      "loss": 0.5597,
      "step": 2316
    },
    {
      "epoch": 2.835985312117503,
      "grad_norm": 0.8745618706108752,
      "learning_rate": 4.5718171316478536e-06,
      "loss": 0.6259,
      "step": 2317
    },
    {
      "epoch": 2.8372093023255816,
      "grad_norm": 1.2185259850896366,
      "learning_rate": 4.571458393102284e-06,
      "loss": 0.8016,
      "step": 2318
    },
    {
      "epoch": 2.8384332925336597,
      "grad_norm": 1.0738050366547107,
      "learning_rate": 4.571099518428131e-06,
      "loss": 0.5175,
      "step": 2319
    },
    {
      "epoch": 2.8396572827417383,
      "grad_norm": 1.032307356219277,
      "learning_rate": 4.570740507648977e-06,
      "loss": 1.2303,
      "step": 2320
    },
    {
      "epoch": 2.8408812729498165,
      "grad_norm": 1.0457654557525675,
      "learning_rate": 4.570381360788416e-06,
      "loss": 0.547,
      "step": 2321
    },
    {
      "epoch": 2.8421052631578947,
      "grad_norm": 1.0563061778528389,
      "learning_rate": 4.570022077870051e-06,
      "loss": 0.4668,
      "step": 2322
    },
    {
      "epoch": 2.8433292533659733,
      "grad_norm": 0.9950091682488957,
      "learning_rate": 4.56966265891749e-06,
      "loss": 1.4581,
      "step": 2323
    },
    {
      "epoch": 2.8445532435740515,
      "grad_norm": 1.242646184661988,
      "learning_rate": 4.569303103954355e-06,
      "loss": 1.089,
      "step": 2324
    },
    {
      "epoch": 2.8457772337821297,
      "grad_norm": 1.1771126463862922,
      "learning_rate": 4.568943413004274e-06,
      "loss": 0.4421,
      "step": 2325
    },
    {
      "epoch": 2.8470012239902083,
      "grad_norm": 1.2617355097657599,
      "learning_rate": 4.568583586090884e-06,
      "loss": 1.1798,
      "step": 2326
    },
    {
      "epoch": 2.8482252141982864,
      "grad_norm": 1.206577388495166,
      "learning_rate": 4.568223623237832e-06,
      "loss": 0.4521,
      "step": 2327
    },
    {
      "epoch": 2.8494492044063646,
      "grad_norm": 0.7399314375368015,
      "learning_rate": 4.5678635244687745e-06,
      "loss": 0.5042,
      "step": 2328
    },
    {
      "epoch": 2.8506731946144432,
      "grad_norm": 0.8142641778886158,
      "learning_rate": 4.567503289807373e-06,
      "loss": 0.545,
      "step": 2329
    },
    {
      "epoch": 2.8518971848225214,
      "grad_norm": 1.0524887624241896,
      "learning_rate": 4.567142919277303e-06,
      "loss": 0.4204,
      "step": 2330
    },
    {
      "epoch": 2.8531211750305996,
      "grad_norm": 0.9067088699247055,
      "learning_rate": 4.566782412902246e-06,
      "loss": 0.6439,
      "step": 2331
    },
    {
      "epoch": 2.854345165238678,
      "grad_norm": 0.6372786725649205,
      "learning_rate": 4.5664217707058935e-06,
      "loss": 0.6351,
      "step": 2332
    },
    {
      "epoch": 2.8555691554467564,
      "grad_norm": 0.8330396436852358,
      "learning_rate": 4.566060992711946e-06,
      "loss": 0.8174,
      "step": 2333
    },
    {
      "epoch": 2.8567931456548346,
      "grad_norm": 1.066306959265128,
      "learning_rate": 4.565700078944111e-06,
      "loss": 0.967,
      "step": 2334
    },
    {
      "epoch": 2.858017135862913,
      "grad_norm": 1.3147404227688513,
      "learning_rate": 4.565339029426108e-06,
      "loss": 0.6488,
      "step": 2335
    },
    {
      "epoch": 2.8592411260709913,
      "grad_norm": 0.8669759813068143,
      "learning_rate": 4.564977844181662e-06,
      "loss": 0.6648,
      "step": 2336
    },
    {
      "epoch": 2.8604651162790695,
      "grad_norm": 1.575808157742632,
      "learning_rate": 4.564616523234511e-06,
      "loss": 0.4335,
      "step": 2337
    },
    {
      "epoch": 2.861689106487148,
      "grad_norm": 0.7252501385526152,
      "learning_rate": 4.564255066608398e-06,
      "loss": 0.5861,
      "step": 2338
    },
    {
      "epoch": 2.8629130966952263,
      "grad_norm": 1.1591517750813822,
      "learning_rate": 4.563893474327077e-06,
      "loss": 0.4805,
      "step": 2339
    },
    {
      "epoch": 2.864137086903305,
      "grad_norm": 0.6418717544926285,
      "learning_rate": 4.563531746414311e-06,
      "loss": 0.7536,
      "step": 2340
    },
    {
      "epoch": 2.865361077111383,
      "grad_norm": 0.8198777016790953,
      "learning_rate": 4.563169882893872e-06,
      "loss": 0.5976,
      "step": 2341
    },
    {
      "epoch": 2.8665850673194617,
      "grad_norm": 1.4521369915398115,
      "learning_rate": 4.5628078837895385e-06,
      "loss": 0.5725,
      "step": 2342
    },
    {
      "epoch": 2.86780905752754,
      "grad_norm": 1.1000540769783658,
      "learning_rate": 4.5624457491251e-06,
      "loss": 0.6789,
      "step": 2343
    },
    {
      "epoch": 2.869033047735618,
      "grad_norm": 1.1173505982683034,
      "learning_rate": 4.562083478924357e-06,
      "loss": 0.5505,
      "step": 2344
    },
    {
      "epoch": 2.8702570379436967,
      "grad_norm": 1.0904083827904085,
      "learning_rate": 4.561721073211114e-06,
      "loss": 0.5911,
      "step": 2345
    },
    {
      "epoch": 2.871481028151775,
      "grad_norm": 1.2463425666514405,
      "learning_rate": 4.561358532009188e-06,
      "loss": 0.5286,
      "step": 2346
    },
    {
      "epoch": 2.872705018359853,
      "grad_norm": 1.5983601229943538,
      "learning_rate": 4.560995855342404e-06,
      "loss": 0.4641,
      "step": 2347
    },
    {
      "epoch": 2.8739290085679317,
      "grad_norm": 1.063584641560283,
      "learning_rate": 4.560633043234595e-06,
      "loss": 0.3687,
      "step": 2348
    },
    {
      "epoch": 2.87515299877601,
      "grad_norm": 1.127785796403911,
      "learning_rate": 4.560270095709603e-06,
      "loss": 0.6608,
      "step": 2349
    },
    {
      "epoch": 2.876376988984088,
      "grad_norm": 0.6763194253536473,
      "learning_rate": 4.559907012791283e-06,
      "loss": 0.6137,
      "step": 2350
    },
    {
      "epoch": 2.8776009791921666,
      "grad_norm": 0.7638915188348195,
      "learning_rate": 4.559543794503492e-06,
      "loss": 0.74,
      "step": 2351
    },
    {
      "epoch": 2.878824969400245,
      "grad_norm": 0.95191459958052,
      "learning_rate": 4.5591804408701e-06,
      "loss": 0.5305,
      "step": 2352
    },
    {
      "epoch": 2.880048959608323,
      "grad_norm": 1.3619140292302907,
      "learning_rate": 4.5588169519149875e-06,
      "loss": 0.5195,
      "step": 2353
    },
    {
      "epoch": 2.8812729498164016,
      "grad_norm": 0.7175503780433844,
      "learning_rate": 4.558453327662039e-06,
      "loss": 0.6324,
      "step": 2354
    },
    {
      "epoch": 2.8824969400244798,
      "grad_norm": 0.7727625588792004,
      "learning_rate": 4.558089568135151e-06,
      "loss": 0.7007,
      "step": 2355
    },
    {
      "epoch": 2.883720930232558,
      "grad_norm": 1.5578224079401806,
      "learning_rate": 4.5577256733582296e-06,
      "loss": 0.5038,
      "step": 2356
    },
    {
      "epoch": 2.8849449204406366,
      "grad_norm": 0.92841940576592,
      "learning_rate": 4.557361643355187e-06,
      "loss": 0.6211,
      "step": 2357
    },
    {
      "epoch": 2.8861689106487147,
      "grad_norm": 1.1360295222483738,
      "learning_rate": 4.556997478149948e-06,
      "loss": 1.254,
      "step": 2358
    },
    {
      "epoch": 2.887392900856793,
      "grad_norm": 0.847869620891708,
      "learning_rate": 4.556633177766442e-06,
      "loss": 0.7773,
      "step": 2359
    },
    {
      "epoch": 2.8886168910648715,
      "grad_norm": 0.6852002056695864,
      "learning_rate": 4.556268742228612e-06,
      "loss": 0.6383,
      "step": 2360
    },
    {
      "epoch": 2.8898408812729497,
      "grad_norm": 0.5339475913659488,
      "learning_rate": 4.5559041715604045e-06,
      "loss": 0.3754,
      "step": 2361
    },
    {
      "epoch": 2.8910648714810283,
      "grad_norm": 0.9599298115980407,
      "learning_rate": 4.55553946578578e-06,
      "loss": 0.7596,
      "step": 2362
    },
    {
      "epoch": 2.8922888616891065,
      "grad_norm": 1.6468359131972943,
      "learning_rate": 4.555174624928704e-06,
      "loss": 1.0203,
      "step": 2363
    },
    {
      "epoch": 2.8935128518971847,
      "grad_norm": 1.4858262907575457,
      "learning_rate": 4.554809649013154e-06,
      "loss": 0.5134,
      "step": 2364
    },
    {
      "epoch": 2.8947368421052633,
      "grad_norm": 0.6596207313888333,
      "learning_rate": 4.554444538063113e-06,
      "loss": 0.5809,
      "step": 2365
    },
    {
      "epoch": 2.8959608323133414,
      "grad_norm": 1.1567952590568915,
      "learning_rate": 4.5540792921025765e-06,
      "loss": 0.6176,
      "step": 2366
    },
    {
      "epoch": 2.89718482252142,
      "grad_norm": 1.0533160458229922,
      "learning_rate": 4.553713911155547e-06,
      "loss": 1.1694,
      "step": 2367
    },
    {
      "epoch": 2.8984088127294982,
      "grad_norm": 0.6257114251881628,
      "learning_rate": 4.553348395246035e-06,
      "loss": 0.4777,
      "step": 2368
    },
    {
      "epoch": 2.8996328029375764,
      "grad_norm": 1.7274436100513606,
      "learning_rate": 4.552982744398061e-06,
      "loss": 0.5079,
      "step": 2369
    },
    {
      "epoch": 2.900856793145655,
      "grad_norm": 1.0437321355575,
      "learning_rate": 4.5526169586356535e-06,
      "loss": 0.8015,
      "step": 2370
    },
    {
      "epoch": 2.902080783353733,
      "grad_norm": 0.973767746575319,
      "learning_rate": 4.552251037982854e-06,
      "loss": 0.5242,
      "step": 2371
    },
    {
      "epoch": 2.9033047735618114,
      "grad_norm": 1.1091322131351278,
      "learning_rate": 4.551884982463705e-06,
      "loss": 0.7936,
      "step": 2372
    },
    {
      "epoch": 2.90452876376989,
      "grad_norm": 1.096043265919907,
      "learning_rate": 4.551518792102266e-06,
      "loss": 1.5329,
      "step": 2373
    },
    {
      "epoch": 2.905752753977968,
      "grad_norm": 1.1852800498459155,
      "learning_rate": 4.551152466922598e-06,
      "loss": 0.8082,
      "step": 2374
    },
    {
      "epoch": 2.9069767441860463,
      "grad_norm": 1.2646509913115278,
      "learning_rate": 4.550786006948778e-06,
      "loss": 0.6118,
      "step": 2375
    },
    {
      "epoch": 2.908200734394125,
      "grad_norm": 0.8868003466719138,
      "learning_rate": 4.5504194122048865e-06,
      "loss": 0.5518,
      "step": 2376
    },
    {
      "epoch": 2.909424724602203,
      "grad_norm": 1.1667983221973468,
      "learning_rate": 4.550052682715015e-06,
      "loss": 0.4869,
      "step": 2377
    },
    {
      "epoch": 2.9106487148102813,
      "grad_norm": 0.9050304363250656,
      "learning_rate": 4.5496858185032645e-06,
      "loss": 0.6564,
      "step": 2378
    },
    {
      "epoch": 2.91187270501836,
      "grad_norm": 0.9858405482451021,
      "learning_rate": 4.5493188195937425e-06,
      "loss": 0.5589,
      "step": 2379
    },
    {
      "epoch": 2.913096695226438,
      "grad_norm": 0.347388076962206,
      "learning_rate": 4.548951686010568e-06,
      "loss": 0.1392,
      "step": 2380
    },
    {
      "epoch": 2.9143206854345163,
      "grad_norm": 0.647667656072901,
      "learning_rate": 4.548584417777867e-06,
      "loss": 0.5714,
      "step": 2381
    },
    {
      "epoch": 2.915544675642595,
      "grad_norm": 0.9567819190547915,
      "learning_rate": 4.548217014919776e-06,
      "loss": 0.8294,
      "step": 2382
    },
    {
      "epoch": 2.916768665850673,
      "grad_norm": 1.5405976942438115,
      "learning_rate": 4.5478494774604375e-06,
      "loss": 0.5506,
      "step": 2383
    },
    {
      "epoch": 2.9179926560587517,
      "grad_norm": 1.203331396455515,
      "learning_rate": 4.547481805424007e-06,
      "loss": 0.5577,
      "step": 2384
    },
    {
      "epoch": 2.91921664626683,
      "grad_norm": 0.9631931362738453,
      "learning_rate": 4.5471139988346445e-06,
      "loss": 1.6193,
      "step": 2385
    },
    {
      "epoch": 2.920440636474908,
      "grad_norm": 1.14367129385976,
      "learning_rate": 4.546746057716522e-06,
      "loss": 1.0145,
      "step": 2386
    },
    {
      "epoch": 2.9216646266829867,
      "grad_norm": 2.0506114370874857,
      "learning_rate": 4.546377982093819e-06,
      "loss": 0.6053,
      "step": 2387
    },
    {
      "epoch": 2.922888616891065,
      "grad_norm": 1.1707923347272948,
      "learning_rate": 4.546009771990724e-06,
      "loss": 1.0609,
      "step": 2388
    },
    {
      "epoch": 2.9241126070991434,
      "grad_norm": 1.3842248201092033,
      "learning_rate": 4.545641427431434e-06,
      "loss": 0.4095,
      "step": 2389
    },
    {
      "epoch": 2.9253365973072216,
      "grad_norm": 2.271198187921925,
      "learning_rate": 4.545272948440157e-06,
      "loss": 0.4835,
      "step": 2390
    },
    {
      "epoch": 2.9265605875153,
      "grad_norm": 0.8365925933227671,
      "learning_rate": 4.544904335041106e-06,
      "loss": 0.6076,
      "step": 2391
    },
    {
      "epoch": 2.9277845777233784,
      "grad_norm": 1.0592938201503375,
      "learning_rate": 4.544535587258505e-06,
      "loss": 0.5135,
      "step": 2392
    },
    {
      "epoch": 2.9290085679314566,
      "grad_norm": 0.6737639960939096,
      "learning_rate": 4.5441667051165884e-06,
      "loss": 0.5423,
      "step": 2393
    },
    {
      "epoch": 2.9302325581395348,
      "grad_norm": 1.1510359449089556,
      "learning_rate": 4.543797688639596e-06,
      "loss": 0.9885,
      "step": 2394
    },
    {
      "epoch": 2.9314565483476134,
      "grad_norm": 1.2491548110517527,
      "learning_rate": 4.54342853785178e-06,
      "loss": 0.579,
      "step": 2395
    },
    {
      "epoch": 2.9326805385556916,
      "grad_norm": 1.187906424336536,
      "learning_rate": 4.543059252777397e-06,
      "loss": 0.609,
      "step": 2396
    },
    {
      "epoch": 2.9339045287637697,
      "grad_norm": 1.2190365719655702,
      "learning_rate": 4.5426898334407184e-06,
      "loss": 0.9918,
      "step": 2397
    },
    {
      "epoch": 2.9351285189718483,
      "grad_norm": 1.5349285748240722,
      "learning_rate": 4.5423202798660184e-06,
      "loss": 1.0989,
      "step": 2398
    },
    {
      "epoch": 2.9363525091799265,
      "grad_norm": 1.574410951864215,
      "learning_rate": 4.541950592077584e-06,
      "loss": 0.6781,
      "step": 2399
    },
    {
      "epoch": 2.9375764993880047,
      "grad_norm": 0.9438098353460729,
      "learning_rate": 4.541580770099709e-06,
      "loss": 0.9502,
      "step": 2400
    },
    {
      "epoch": 2.9388004895960833,
      "grad_norm": 1.6018661666655294,
      "learning_rate": 4.5412108139566974e-06,
      "loss": 0.5958,
      "step": 2401
    },
    {
      "epoch": 2.9400244798041615,
      "grad_norm": 0.9403106439557926,
      "learning_rate": 4.540840723672861e-06,
      "loss": 0.416,
      "step": 2402
    },
    {
      "epoch": 2.9412484700122397,
      "grad_norm": 0.6277460192136504,
      "learning_rate": 4.54047049927252e-06,
      "loss": 0.432,
      "step": 2403
    },
    {
      "epoch": 2.9424724602203183,
      "grad_norm": 0.7025677598181642,
      "learning_rate": 4.540100140780006e-06,
      "loss": 0.6174,
      "step": 2404
    },
    {
      "epoch": 2.9436964504283964,
      "grad_norm": 1.0982753496696072,
      "learning_rate": 4.539729648219656e-06,
      "loss": 1.4724,
      "step": 2405
    },
    {
      "epoch": 2.944920440636475,
      "grad_norm": 1.550322998390007,
      "learning_rate": 4.539359021615819e-06,
      "loss": 0.8303,
      "step": 2406
    },
    {
      "epoch": 2.9461444308445532,
      "grad_norm": 1.3857986336331334,
      "learning_rate": 4.538988260992849e-06,
      "loss": 0.4802,
      "step": 2407
    },
    {
      "epoch": 2.9473684210526314,
      "grad_norm": 1.9021696068639997,
      "learning_rate": 4.538617366375112e-06,
      "loss": 0.7821,
      "step": 2408
    },
    {
      "epoch": 2.94859241126071,
      "grad_norm": 1.024451732034074,
      "learning_rate": 4.538246337786982e-06,
      "loss": 1.1471,
      "step": 2409
    },
    {
      "epoch": 2.949816401468788,
      "grad_norm": 1.0368732550294646,
      "learning_rate": 4.537875175252842e-06,
      "loss": 0.3588,
      "step": 2410
    },
    {
      "epoch": 2.951040391676867,
      "grad_norm": 1.6186569167940048,
      "learning_rate": 4.5375038787970835e-06,
      "loss": 1.0087,
      "step": 2411
    },
    {
      "epoch": 2.952264381884945,
      "grad_norm": 1.2576114415175097,
      "learning_rate": 4.537132448444106e-06,
      "loss": 0.8915,
      "step": 2412
    },
    {
      "epoch": 2.953488372093023,
      "grad_norm": 0.6530428496262475,
      "learning_rate": 4.536760884218319e-06,
      "loss": 0.5514,
      "step": 2413
    },
    {
      "epoch": 2.954712362301102,
      "grad_norm": 0.7735854861064247,
      "learning_rate": 4.53638918614414e-06,
      "loss": 0.5915,
      "step": 2414
    },
    {
      "epoch": 2.95593635250918,
      "grad_norm": 0.8801333205421741,
      "learning_rate": 4.5360173542459965e-06,
      "loss": 1.2085,
      "step": 2415
    },
    {
      "epoch": 2.957160342717258,
      "grad_norm": 0.8480761150488775,
      "learning_rate": 4.535645388548322e-06,
      "loss": 0.6868,
      "step": 2416
    },
    {
      "epoch": 2.9583843329253368,
      "grad_norm": 1.6718946179672758,
      "learning_rate": 4.5352732890755625e-06,
      "loss": 0.6837,
      "step": 2417
    },
    {
      "epoch": 2.959608323133415,
      "grad_norm": 1.721050113237689,
      "learning_rate": 4.53490105585217e-06,
      "loss": 0.5752,
      "step": 2418
    },
    {
      "epoch": 2.960832313341493,
      "grad_norm": 1.784178764002309,
      "learning_rate": 4.534528688902606e-06,
      "loss": 0.6267,
      "step": 2419
    },
    {
      "epoch": 2.9620563035495717,
      "grad_norm": 0.6186555727144681,
      "learning_rate": 4.534156188251343e-06,
      "loss": 0.6177,
      "step": 2420
    },
    {
      "epoch": 2.96328029375765,
      "grad_norm": 1.092749622573756,
      "learning_rate": 4.533783553922859e-06,
      "loss": 0.7537,
      "step": 2421
    },
    {
      "epoch": 2.964504283965728,
      "grad_norm": 0.7654709556096652,
      "learning_rate": 4.5334107859416424e-06,
      "loss": 0.6475,
      "step": 2422
    },
    {
      "epoch": 2.9657282741738067,
      "grad_norm": 0.8022229025155242,
      "learning_rate": 4.53303788433219e-06,
      "loss": 0.5166,
      "step": 2423
    },
    {
      "epoch": 2.966952264381885,
      "grad_norm": 2.1331389095439417,
      "learning_rate": 4.532664849119006e-06,
      "loss": 0.5149,
      "step": 2424
    },
    {
      "epoch": 2.968176254589963,
      "grad_norm": 0.6929950322415823,
      "learning_rate": 4.532291680326608e-06,
      "loss": 0.6014,
      "step": 2425
    },
    {
      "epoch": 2.9694002447980417,
      "grad_norm": 1.1747462017978718,
      "learning_rate": 4.531918377979517e-06,
      "loss": 0.5544,
      "step": 2426
    },
    {
      "epoch": 2.97062423500612,
      "grad_norm": 1.3660308124489196,
      "learning_rate": 4.531544942102267e-06,
      "loss": 0.6631,
      "step": 2427
    },
    {
      "epoch": 2.9718482252141984,
      "grad_norm": 0.4960179000828608,
      "learning_rate": 4.531171372719396e-06,
      "loss": 0.3815,
      "step": 2428
    },
    {
      "epoch": 2.9730722154222766,
      "grad_norm": 1.2339209623528586,
      "learning_rate": 4.530797669855456e-06,
      "loss": 0.6975,
      "step": 2429
    },
    {
      "epoch": 2.974296205630355,
      "grad_norm": 1.112689387111434,
      "learning_rate": 4.530423833535004e-06,
      "loss": 0.5287,
      "step": 2430
    },
    {
      "epoch": 2.9755201958384334,
      "grad_norm": 1.4224290995265993,
      "learning_rate": 4.5300498637826084e-06,
      "loss": 0.5761,
      "step": 2431
    },
    {
      "epoch": 2.9767441860465116,
      "grad_norm": 0.9889927076445122,
      "learning_rate": 4.529675760622844e-06,
      "loss": 0.5268,
      "step": 2432
    },
    {
      "epoch": 2.97796817625459,
      "grad_norm": 1.84058324507547,
      "learning_rate": 4.529301524080295e-06,
      "loss": 0.9563,
      "step": 2433
    },
    {
      "epoch": 2.9791921664626684,
      "grad_norm": 1.3064091011209367,
      "learning_rate": 4.528927154179557e-06,
      "loss": 0.7804,
      "step": 2434
    },
    {
      "epoch": 2.9804161566707466,
      "grad_norm": 1.287044932182841,
      "learning_rate": 4.5285526509452304e-06,
      "loss": 0.621,
      "step": 2435
    },
    {
      "epoch": 2.981640146878825,
      "grad_norm": 0.928779472641578,
      "learning_rate": 4.5281780144019275e-06,
      "loss": 0.8027,
      "step": 2436
    },
    {
      "epoch": 2.9828641370869033,
      "grad_norm": 1.0854465457516853,
      "learning_rate": 4.527803244574266e-06,
      "loss": 0.7787,
      "step": 2437
    },
    {
      "epoch": 2.9840881272949815,
      "grad_norm": 1.2416857524175433,
      "learning_rate": 4.527428341486876e-06,
      "loss": 0.5933,
      "step": 2438
    },
    {
      "epoch": 2.98531211750306,
      "grad_norm": 1.032605911127453,
      "learning_rate": 4.5270533051643944e-06,
      "loss": 1.303,
      "step": 2439
    },
    {
      "epoch": 2.9865361077111383,
      "grad_norm": 1.899655168161689,
      "learning_rate": 4.526678135631467e-06,
      "loss": 0.4806,
      "step": 2440
    },
    {
      "epoch": 2.9877600979192165,
      "grad_norm": 1.8746054501030678,
      "learning_rate": 4.526302832912749e-06,
      "loss": 0.6804,
      "step": 2441
    },
    {
      "epoch": 2.988984088127295,
      "grad_norm": 1.2445173536930647,
      "learning_rate": 4.525927397032903e-06,
      "loss": 0.5529,
      "step": 2442
    },
    {
      "epoch": 2.9902080783353733,
      "grad_norm": 0.4786375751365313,
      "learning_rate": 4.525551828016602e-06,
      "loss": 0.2736,
      "step": 2443
    },
    {
      "epoch": 2.9914320685434515,
      "grad_norm": 0.8028999296308257,
      "learning_rate": 4.525176125888528e-06,
      "loss": 0.4787,
      "step": 2444
    },
    {
      "epoch": 2.99265605875153,
      "grad_norm": 0.897776074380303,
      "learning_rate": 4.524800290673369e-06,
      "loss": 0.4897,
      "step": 2445
    },
    {
      "epoch": 2.9938800489596082,
      "grad_norm": 0.8093236665566064,
      "learning_rate": 4.524424322395824e-06,
      "loss": 0.5761,
      "step": 2446
    },
    {
      "epoch": 2.9951040391676864,
      "grad_norm": 0.9240487025444044,
      "learning_rate": 4.5240482210806e-06,
      "loss": 1.1636,
      "step": 2447
    },
    {
      "epoch": 2.996328029375765,
      "grad_norm": 0.9204495942975788,
      "learning_rate": 4.523671986752413e-06,
      "loss": 0.5325,
      "step": 2448
    },
    {
      "epoch": 2.997552019583843,
      "grad_norm": 1.3846168551586802,
      "learning_rate": 4.523295619435989e-06,
      "loss": 0.5384,
      "step": 2449
    },
    {
      "epoch": 2.998776009791922,
      "grad_norm": 1.1191577653609084,
      "learning_rate": 4.52291911915606e-06,
      "loss": 0.5962,
      "step": 2450
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.85819815943381,
      "learning_rate": 4.522542485937369e-06,
      "loss": 0.4979,
      "step": 2451
    },
    {
      "epoch": 3.001223990208078,
      "grad_norm": 1.5019726139100966,
      "learning_rate": 4.522165719804666e-06,
      "loss": 1.0324,
      "step": 2452
    },
    {
      "epoch": 3.002447980416157,
      "grad_norm": 0.7276967813393894,
      "learning_rate": 4.521788820782713e-06,
      "loss": 0.6414,
      "step": 2453
    },
    {
      "epoch": 3.003671970624235,
      "grad_norm": 1.2969745308777048,
      "learning_rate": 4.521411788896276e-06,
      "loss": 0.5061,
      "step": 2454
    },
    {
      "epoch": 3.004895960832313,
      "grad_norm": 0.6312288500058904,
      "learning_rate": 4.521034624170132e-06,
      "loss": 0.427,
      "step": 2455
    },
    {
      "epoch": 3.0061199510403918,
      "grad_norm": 1.0776205534979248,
      "learning_rate": 4.520657326629069e-06,
      "loss": 0.7401,
      "step": 2456
    },
    {
      "epoch": 3.00734394124847,
      "grad_norm": 0.6637683593237725,
      "learning_rate": 4.520279896297879e-06,
      "loss": 0.5784,
      "step": 2457
    },
    {
      "epoch": 3.0085679314565485,
      "grad_norm": 0.7162968128681916,
      "learning_rate": 4.519902333201368e-06,
      "loss": 0.5452,
      "step": 2458
    },
    {
      "epoch": 3.0097919216646267,
      "grad_norm": 0.9430975387522736,
      "learning_rate": 4.519524637364346e-06,
      "loss": 0.9417,
      "step": 2459
    },
    {
      "epoch": 3.011015911872705,
      "grad_norm": 0.6508567482485539,
      "learning_rate": 4.5191468088116355e-06,
      "loss": 0.4229,
      "step": 2460
    },
    {
      "epoch": 3.0122399020807835,
      "grad_norm": 0.984553097597724,
      "learning_rate": 4.5187688475680635e-06,
      "loss": 1.0287,
      "step": 2461
    },
    {
      "epoch": 3.0134638922888617,
      "grad_norm": 1.3459606202401266,
      "learning_rate": 4.518390753658471e-06,
      "loss": 1.0202,
      "step": 2462
    },
    {
      "epoch": 3.01468788249694,
      "grad_norm": 1.2906910635698514,
      "learning_rate": 4.518012527107704e-06,
      "loss": 0.6743,
      "step": 2463
    },
    {
      "epoch": 3.0159118727050185,
      "grad_norm": 0.9036588273704346,
      "learning_rate": 4.517634167940616e-06,
      "loss": 0.2634,
      "step": 2464
    },
    {
      "epoch": 3.0171358629130967,
      "grad_norm": 0.818827425477907,
      "learning_rate": 4.517255676182075e-06,
      "loss": 0.5027,
      "step": 2465
    },
    {
      "epoch": 3.018359853121175,
      "grad_norm": 1.1386835512152047,
      "learning_rate": 4.5168770518569515e-06,
      "loss": 0.691,
      "step": 2466
    },
    {
      "epoch": 3.0195838433292534,
      "grad_norm": 1.15605125838173,
      "learning_rate": 4.516498294990128e-06,
      "loss": 0.6235,
      "step": 2467
    },
    {
      "epoch": 3.0208078335373316,
      "grad_norm": 1.425243656840284,
      "learning_rate": 4.516119405606496e-06,
      "loss": 0.5339,
      "step": 2468
    },
    {
      "epoch": 3.0220318237454102,
      "grad_norm": 1.0460744036623735,
      "learning_rate": 4.5157403837309525e-06,
      "loss": 0.5516,
      "step": 2469
    },
    {
      "epoch": 3.0232558139534884,
      "grad_norm": 1.1877957005457382,
      "learning_rate": 4.515361229388408e-06,
      "loss": 0.409,
      "step": 2470
    },
    {
      "epoch": 3.0244798041615666,
      "grad_norm": 1.1464369534972503,
      "learning_rate": 4.514981942603777e-06,
      "loss": 0.6514,
      "step": 2471
    },
    {
      "epoch": 3.025703794369645,
      "grad_norm": 0.7981855813485291,
      "learning_rate": 4.514602523401986e-06,
      "loss": 0.5469,
      "step": 2472
    },
    {
      "epoch": 3.0269277845777234,
      "grad_norm": 1.1032676079549244,
      "learning_rate": 4.514222971807968e-06,
      "loss": 1.1575,
      "step": 2473
    },
    {
      "epoch": 3.0281517747858016,
      "grad_norm": 0.8141042808455591,
      "learning_rate": 4.513843287846666e-06,
      "loss": 0.6778,
      "step": 2474
    },
    {
      "epoch": 3.02937576499388,
      "grad_norm": 1.7813718926888147,
      "learning_rate": 4.513463471543032e-06,
      "loss": 1.022,
      "step": 2475
    },
    {
      "epoch": 3.0305997552019583,
      "grad_norm": 1.0129620810447315,
      "learning_rate": 4.513083522922027e-06,
      "loss": 1.0367,
      "step": 2476
    },
    {
      "epoch": 3.0318237454100365,
      "grad_norm": 0.616622859682934,
      "learning_rate": 4.512703442008618e-06,
      "loss": 0.4764,
      "step": 2477
    },
    {
      "epoch": 3.033047735618115,
      "grad_norm": 1.0490207416774018,
      "learning_rate": 4.512323228827784e-06,
      "loss": 1.3547,
      "step": 2478
    },
    {
      "epoch": 3.0342717258261933,
      "grad_norm": 1.1010133369367903,
      "learning_rate": 4.51194288340451e-06,
      "loss": 0.4812,
      "step": 2479
    },
    {
      "epoch": 3.035495716034272,
      "grad_norm": 1.1483409449005981,
      "learning_rate": 4.511562405763791e-06,
      "loss": 0.5555,
      "step": 2480
    },
    {
      "epoch": 3.03671970624235,
      "grad_norm": 1.198067024840011,
      "learning_rate": 4.511181795930631e-06,
      "loss": 0.598,
      "step": 2481
    },
    {
      "epoch": 3.0379436964504283,
      "grad_norm": 0.7862662538729204,
      "learning_rate": 4.510801053930043e-06,
      "loss": 1.0739,
      "step": 2482
    },
    {
      "epoch": 3.039167686658507,
      "grad_norm": 1.3339539002281275,
      "learning_rate": 4.510420179787047e-06,
      "loss": 1.214,
      "step": 2483
    },
    {
      "epoch": 3.040391676866585,
      "grad_norm": 0.8750390623351249,
      "learning_rate": 4.510039173526672e-06,
      "loss": 0.3703,
      "step": 2484
    },
    {
      "epoch": 3.0416156670746632,
      "grad_norm": 1.0945668105446344,
      "learning_rate": 4.509658035173957e-06,
      "loss": 0.5927,
      "step": 2485
    },
    {
      "epoch": 3.042839657282742,
      "grad_norm": 1.5574197083877737,
      "learning_rate": 4.50927676475395e-06,
      "loss": 0.5002,
      "step": 2486
    },
    {
      "epoch": 3.04406364749082,
      "grad_norm": 1.7709621078870266,
      "learning_rate": 4.5088953622917055e-06,
      "loss": 0.5305,
      "step": 2487
    },
    {
      "epoch": 3.045287637698898,
      "grad_norm": 0.6207389116696249,
      "learning_rate": 4.508513827812289e-06,
      "loss": 0.4737,
      "step": 2488
    },
    {
      "epoch": 3.046511627906977,
      "grad_norm": 0.6652657337315278,
      "learning_rate": 4.508132161340772e-06,
      "loss": 0.5493,
      "step": 2489
    },
    {
      "epoch": 3.047735618115055,
      "grad_norm": 0.795351113668382,
      "learning_rate": 4.507750362902236e-06,
      "loss": 0.5223,
      "step": 2490
    },
    {
      "epoch": 3.0489596083231336,
      "grad_norm": 0.8317659191339594,
      "learning_rate": 4.507368432521774e-06,
      "loss": 0.4068,
      "step": 2491
    },
    {
      "epoch": 3.050183598531212,
      "grad_norm": 1.079267609325809,
      "learning_rate": 4.506986370224482e-06,
      "loss": 0.5277,
      "step": 2492
    },
    {
      "epoch": 3.05140758873929,
      "grad_norm": 1.6943907060848709,
      "learning_rate": 4.50660417603547e-06,
      "loss": 1.0054,
      "step": 2493
    },
    {
      "epoch": 3.0526315789473686,
      "grad_norm": 1.0852775347509074,
      "learning_rate": 4.506221849979853e-06,
      "loss": 0.4409,
      "step": 2494
    },
    {
      "epoch": 3.0538555691554468,
      "grad_norm": 0.8319120033234779,
      "learning_rate": 4.505839392082757e-06,
      "loss": 0.485,
      "step": 2495
    },
    {
      "epoch": 3.055079559363525,
      "grad_norm": 0.9905877314093634,
      "learning_rate": 4.505456802369315e-06,
      "loss": 0.6807,
      "step": 2496
    },
    {
      "epoch": 3.0563035495716036,
      "grad_norm": 0.9877651012880011,
      "learning_rate": 4.5050740808646685e-06,
      "loss": 0.737,
      "step": 2497
    },
    {
      "epoch": 3.0575275397796817,
      "grad_norm": 1.1495843429704593,
      "learning_rate": 4.5046912275939716e-06,
      "loss": 0.6485,
      "step": 2498
    },
    {
      "epoch": 3.05875152998776,
      "grad_norm": 1.2532356027027147,
      "learning_rate": 4.50430824258238e-06,
      "loss": 0.9923,
      "step": 2499
    },
    {
      "epoch": 3.0599755201958385,
      "grad_norm": 1.8156530265529598,
      "learning_rate": 4.503925125855065e-06,
      "loss": 0.8953,
      "step": 2500
    },
    {
      "epoch": 3.0611995104039167,
      "grad_norm": 1.1340672614815723,
      "learning_rate": 4.503541877437204e-06,
      "loss": 1.1826,
      "step": 2501
    },
    {
      "epoch": 3.0624235006119953,
      "grad_norm": 0.6494141520443184,
      "learning_rate": 4.50315849735398e-06,
      "loss": 0.5602,
      "step": 2502
    },
    {
      "epoch": 3.0636474908200735,
      "grad_norm": 1.4889657288664015,
      "learning_rate": 4.5027749856305895e-06,
      "loss": 0.5483,
      "step": 2503
    },
    {
      "epoch": 3.0648714810281517,
      "grad_norm": 1.517082354325872,
      "learning_rate": 4.502391342292234e-06,
      "loss": 0.4983,
      "step": 2504
    },
    {
      "epoch": 3.0660954712362303,
      "grad_norm": 1.0916754944708404,
      "learning_rate": 4.502007567364126e-06,
      "loss": 1.001,
      "step": 2505
    },
    {
      "epoch": 3.0673194614443084,
      "grad_norm": 1.3034577494852235,
      "learning_rate": 4.501623660871486e-06,
      "loss": 0.5564,
      "step": 2506
    },
    {
      "epoch": 3.0685434516523866,
      "grad_norm": 1.3073526226808656,
      "learning_rate": 4.501239622839542e-06,
      "loss": 0.5465,
      "step": 2507
    },
    {
      "epoch": 3.0697674418604652,
      "grad_norm": 0.6824793409519839,
      "learning_rate": 4.500855453293532e-06,
      "loss": 0.6416,
      "step": 2508
    },
    {
      "epoch": 3.0709914320685434,
      "grad_norm": 0.9336633907729536,
      "learning_rate": 4.500471152258702e-06,
      "loss": 0.5839,
      "step": 2509
    },
    {
      "epoch": 3.0722154222766216,
      "grad_norm": 0.931630764176794,
      "learning_rate": 4.500086719760308e-06,
      "loss": 0.5225,
      "step": 2510
    },
    {
      "epoch": 3.0734394124847,
      "grad_norm": 0.8470899237529007,
      "learning_rate": 4.499702155823612e-06,
      "loss": 0.5813,
      "step": 2511
    },
    {
      "epoch": 3.0746634026927784,
      "grad_norm": 1.1289102134335578,
      "learning_rate": 4.4993174604738875e-06,
      "loss": 0.6649,
      "step": 2512
    },
    {
      "epoch": 3.075887392900857,
      "grad_norm": 1.7546668984096738,
      "learning_rate": 4.498932633736413e-06,
      "loss": 0.9952,
      "step": 2513
    },
    {
      "epoch": 3.077111383108935,
      "grad_norm": 0.926455409006507,
      "learning_rate": 4.4985476756364805e-06,
      "loss": 0.613,
      "step": 2514
    },
    {
      "epoch": 3.0783353733170133,
      "grad_norm": 1.737932678315694,
      "learning_rate": 4.4981625861993865e-06,
      "loss": 0.5932,
      "step": 2515
    },
    {
      "epoch": 3.079559363525092,
      "grad_norm": 1.074127298909647,
      "learning_rate": 4.497777365450439e-06,
      "loss": 0.8069,
      "step": 2516
    },
    {
      "epoch": 3.08078335373317,
      "grad_norm": 1.2119901735884306,
      "learning_rate": 4.4973920134149505e-06,
      "loss": 0.4537,
      "step": 2517
    },
    {
      "epoch": 3.0820073439412483,
      "grad_norm": 1.3889489466440683,
      "learning_rate": 4.497006530118248e-06,
      "loss": 1.1946,
      "step": 2518
    },
    {
      "epoch": 3.083231334149327,
      "grad_norm": 1.0553824916548724,
      "learning_rate": 4.496620915585662e-06,
      "loss": 0.6207,
      "step": 2519
    },
    {
      "epoch": 3.084455324357405,
      "grad_norm": 0.730695928479999,
      "learning_rate": 4.496235169842535e-06,
      "loss": 0.4471,
      "step": 2520
    },
    {
      "epoch": 3.0856793145654833,
      "grad_norm": 0.5939886189651445,
      "learning_rate": 4.495849292914216e-06,
      "loss": 0.4632,
      "step": 2521
    },
    {
      "epoch": 3.086903304773562,
      "grad_norm": 1.3752432860902977,
      "learning_rate": 4.495463284826064e-06,
      "loss": 0.5824,
      "step": 2522
    },
    {
      "epoch": 3.08812729498164,
      "grad_norm": 0.5920685368158534,
      "learning_rate": 4.495077145603445e-06,
      "loss": 0.414,
      "step": 2523
    },
    {
      "epoch": 3.0893512851897187,
      "grad_norm": 0.8224286588122312,
      "learning_rate": 4.494690875271735e-06,
      "loss": 0.3499,
      "step": 2524
    },
    {
      "epoch": 3.090575275397797,
      "grad_norm": 1.1287566408262688,
      "learning_rate": 4.49430447385632e-06,
      "loss": 0.9791,
      "step": 2525
    },
    {
      "epoch": 3.091799265605875,
      "grad_norm": 0.9541655559468333,
      "learning_rate": 4.4939179413825904e-06,
      "loss": 0.765,
      "step": 2526
    },
    {
      "epoch": 3.0930232558139537,
      "grad_norm": 1.2498854159339572,
      "learning_rate": 4.493531277875948e-06,
      "loss": 1.2396,
      "step": 2527
    },
    {
      "epoch": 3.094247246022032,
      "grad_norm": 1.5263142598187005,
      "learning_rate": 4.493144483361804e-06,
      "loss": 0.4109,
      "step": 2528
    },
    {
      "epoch": 3.09547123623011,
      "grad_norm": 1.6239053819590075,
      "learning_rate": 4.492757557865578e-06,
      "loss": 0.938,
      "step": 2529
    },
    {
      "epoch": 3.0966952264381886,
      "grad_norm": 1.1508525240457177,
      "learning_rate": 4.492370501412695e-06,
      "loss": 0.4311,
      "step": 2530
    },
    {
      "epoch": 3.097919216646267,
      "grad_norm": 1.4319509195017999,
      "learning_rate": 4.491983314028591e-06,
      "loss": 0.485,
      "step": 2531
    },
    {
      "epoch": 3.099143206854345,
      "grad_norm": 1.0943000248514514,
      "learning_rate": 4.491595995738713e-06,
      "loss": 0.6716,
      "step": 2532
    },
    {
      "epoch": 3.1003671970624236,
      "grad_norm": 1.3296752323361705,
      "learning_rate": 4.491208546568512e-06,
      "loss": 1.1891,
      "step": 2533
    },
    {
      "epoch": 3.1015911872705018,
      "grad_norm": 0.8724618743387204,
      "learning_rate": 4.49082096654345e-06,
      "loss": 0.5903,
      "step": 2534
    },
    {
      "epoch": 3.1028151774785804,
      "grad_norm": 1.8012886999445987,
      "learning_rate": 4.490433255688998e-06,
      "loss": 0.5288,
      "step": 2535
    },
    {
      "epoch": 3.1040391676866586,
      "grad_norm": 0.7326516374943186,
      "learning_rate": 4.490045414030634e-06,
      "loss": 0.6239,
      "step": 2536
    },
    {
      "epoch": 3.1052631578947367,
      "grad_norm": 1.1564490945426649,
      "learning_rate": 4.489657441593847e-06,
      "loss": 0.7635,
      "step": 2537
    },
    {
      "epoch": 3.1064871481028153,
      "grad_norm": 2.1475157268606053,
      "learning_rate": 4.489269338404131e-06,
      "loss": 0.3637,
      "step": 2538
    },
    {
      "epoch": 3.1077111383108935,
      "grad_norm": 0.9062416957861741,
      "learning_rate": 4.488881104486993e-06,
      "loss": 0.7106,
      "step": 2539
    },
    {
      "epoch": 3.1089351285189717,
      "grad_norm": 1.0294570052391228,
      "learning_rate": 4.488492739867944e-06,
      "loss": 0.5552,
      "step": 2540
    },
    {
      "epoch": 3.1101591187270503,
      "grad_norm": 0.9482646963895243,
      "learning_rate": 4.488104244572508e-06,
      "loss": 0.5673,
      "step": 2541
    },
    {
      "epoch": 3.1113831089351285,
      "grad_norm": 1.3036588073727717,
      "learning_rate": 4.487715618626214e-06,
      "loss": 0.6847,
      "step": 2542
    },
    {
      "epoch": 3.1126070991432067,
      "grad_norm": 0.7191065047045402,
      "learning_rate": 4.487326862054602e-06,
      "loss": 0.606,
      "step": 2543
    },
    {
      "epoch": 3.1138310893512853,
      "grad_norm": 0.8785554195337548,
      "learning_rate": 4.486937974883218e-06,
      "loss": 0.4905,
      "step": 2544
    },
    {
      "epoch": 3.1150550795593634,
      "grad_norm": 1.5823123717213547,
      "learning_rate": 4.4865489571376214e-06,
      "loss": 0.5697,
      "step": 2545
    },
    {
      "epoch": 3.116279069767442,
      "grad_norm": 0.9819448704483195,
      "learning_rate": 4.486159808843375e-06,
      "loss": 0.4906,
      "step": 2546
    },
    {
      "epoch": 3.1175030599755202,
      "grad_norm": 1.084161245638622,
      "learning_rate": 4.485770530026051e-06,
      "loss": 0.7904,
      "step": 2547
    },
    {
      "epoch": 3.1187270501835984,
      "grad_norm": 1.2624813146789096,
      "learning_rate": 4.485381120711233e-06,
      "loss": 0.486,
      "step": 2548
    },
    {
      "epoch": 3.119951040391677,
      "grad_norm": 1.1048840188370355,
      "learning_rate": 4.484991580924512e-06,
      "loss": 0.7681,
      "step": 2549
    },
    {
      "epoch": 3.121175030599755,
      "grad_norm": 1.0388582937901558,
      "learning_rate": 4.484601910691484e-06,
      "loss": 1.2213,
      "step": 2550
    },
    {
      "epoch": 3.1223990208078334,
      "grad_norm": 1.0212734142608066,
      "learning_rate": 4.484212110037761e-06,
      "loss": 0.9764,
      "step": 2551
    },
    {
      "epoch": 3.123623011015912,
      "grad_norm": 1.8855527405576038,
      "learning_rate": 4.483822178988957e-06,
      "loss": 0.8758,
      "step": 2552
    },
    {
      "epoch": 3.12484700122399,
      "grad_norm": 0.6890604543420231,
      "learning_rate": 4.4834321175706966e-06,
      "loss": 0.404,
      "step": 2553
    },
    {
      "epoch": 3.1260709914320683,
      "grad_norm": 0.824550200277482,
      "learning_rate": 4.483041925808614e-06,
      "loss": 0.6018,
      "step": 2554
    },
    {
      "epoch": 3.127294981640147,
      "grad_norm": 0.8285091645858446,
      "learning_rate": 4.482651603728351e-06,
      "loss": 0.5047,
      "step": 2555
    },
    {
      "epoch": 3.128518971848225,
      "grad_norm": 1.1437853456394,
      "learning_rate": 4.482261151355557e-06,
      "loss": 0.4827,
      "step": 2556
    },
    {
      "epoch": 3.1297429620563038,
      "grad_norm": 0.7360047354889917,
      "learning_rate": 4.481870568715893e-06,
      "loss": 0.613,
      "step": 2557
    },
    {
      "epoch": 3.130966952264382,
      "grad_norm": 1.0988782297806161,
      "learning_rate": 4.481479855835025e-06,
      "loss": 1.4744,
      "step": 2558
    },
    {
      "epoch": 3.13219094247246,
      "grad_norm": 0.8491742909974143,
      "learning_rate": 4.481089012738629e-06,
      "loss": 0.7704,
      "step": 2559
    },
    {
      "epoch": 3.1334149326805387,
      "grad_norm": 1.3485883571426112,
      "learning_rate": 4.480698039452392e-06,
      "loss": 0.6003,
      "step": 2560
    },
    {
      "epoch": 3.134638922888617,
      "grad_norm": 1.0576060740594164,
      "learning_rate": 4.480306936002006e-06,
      "loss": 0.7353,
      "step": 2561
    },
    {
      "epoch": 3.135862913096695,
      "grad_norm": 1.34443508004059,
      "learning_rate": 4.479915702413172e-06,
      "loss": 0.7046,
      "step": 2562
    },
    {
      "epoch": 3.1370869033047737,
      "grad_norm": 1.148993055606977,
      "learning_rate": 4.479524338711603e-06,
      "loss": 1.865,
      "step": 2563
    },
    {
      "epoch": 3.138310893512852,
      "grad_norm": 1.148211242345032,
      "learning_rate": 4.4791328449230145e-06,
      "loss": 1.6623,
      "step": 2564
    },
    {
      "epoch": 3.13953488372093,
      "grad_norm": 1.3501078667127324,
      "learning_rate": 4.478741221073136e-06,
      "loss": 1.1289,
      "step": 2565
    },
    {
      "epoch": 3.1407588739290087,
      "grad_norm": 0.8018485348368848,
      "learning_rate": 4.478349467187703e-06,
      "loss": 0.923,
      "step": 2566
    },
    {
      "epoch": 3.141982864137087,
      "grad_norm": 1.3806065413224533,
      "learning_rate": 4.477957583292461e-06,
      "loss": 0.6349,
      "step": 2567
    },
    {
      "epoch": 3.1432068543451654,
      "grad_norm": 0.9185231270917612,
      "learning_rate": 4.477565569413162e-06,
      "loss": 0.8237,
      "step": 2568
    },
    {
      "epoch": 3.1444308445532436,
      "grad_norm": 1.1068168042887472,
      "learning_rate": 4.47717342557557e-06,
      "loss": 0.5661,
      "step": 2569
    },
    {
      "epoch": 3.145654834761322,
      "grad_norm": 1.7243877722429901,
      "learning_rate": 4.476781151805451e-06,
      "loss": 0.4283,
      "step": 2570
    },
    {
      "epoch": 3.1468788249694004,
      "grad_norm": 0.7240379022317064,
      "learning_rate": 4.4763887481285875e-06,
      "loss": 0.6858,
      "step": 2571
    },
    {
      "epoch": 3.1481028151774786,
      "grad_norm": 0.7727291441214302,
      "learning_rate": 4.4759962145707656e-06,
      "loss": 0.4975,
      "step": 2572
    },
    {
      "epoch": 3.1493268053855568,
      "grad_norm": 1.635335405918015,
      "learning_rate": 4.475603551157783e-06,
      "loss": 0.5314,
      "step": 2573
    },
    {
      "epoch": 3.1505507955936354,
      "grad_norm": 0.8911762093944924,
      "learning_rate": 4.475210757915439e-06,
      "loss": 0.7568,
      "step": 2574
    },
    {
      "epoch": 3.1517747858017136,
      "grad_norm": 1.3399101721337259,
      "learning_rate": 4.474817834869551e-06,
      "loss": 0.6365,
      "step": 2575
    },
    {
      "epoch": 3.1529987760097917,
      "grad_norm": 1.1231964388897184,
      "learning_rate": 4.47442478204594e-06,
      "loss": 1.2456,
      "step": 2576
    },
    {
      "epoch": 3.1542227662178703,
      "grad_norm": 0.7503949169976555,
      "learning_rate": 4.474031599470435e-06,
      "loss": 0.7089,
      "step": 2577
    },
    {
      "epoch": 3.1554467564259485,
      "grad_norm": 0.8578697184501369,
      "learning_rate": 4.473638287168874e-06,
      "loss": 1.135,
      "step": 2578
    },
    {
      "epoch": 3.1566707466340267,
      "grad_norm": 1.3623970782300574,
      "learning_rate": 4.473244845167107e-06,
      "loss": 0.8811,
      "step": 2579
    },
    {
      "epoch": 3.1578947368421053,
      "grad_norm": 1.3097895022233739,
      "learning_rate": 4.472851273490985e-06,
      "loss": 0.7172,
      "step": 2580
    },
    {
      "epoch": 3.1591187270501835,
      "grad_norm": 0.9771684578550475,
      "learning_rate": 4.472457572166374e-06,
      "loss": 0.5275,
      "step": 2581
    },
    {
      "epoch": 3.160342717258262,
      "grad_norm": 0.9254007851239002,
      "learning_rate": 4.472063741219148e-06,
      "loss": 1.278,
      "step": 2582
    },
    {
      "epoch": 3.1615667074663403,
      "grad_norm": 1.2970624793838186,
      "learning_rate": 4.471669780675188e-06,
      "loss": 1.0132,
      "step": 2583
    },
    {
      "epoch": 3.1627906976744184,
      "grad_norm": 1.7273578058289327,
      "learning_rate": 4.471275690560381e-06,
      "loss": 0.6465,
      "step": 2584
    },
    {
      "epoch": 3.164014687882497,
      "grad_norm": 1.245479832060968,
      "learning_rate": 4.470881470900628e-06,
      "loss": 0.8348,
      "step": 2585
    },
    {
      "epoch": 3.1652386780905752,
      "grad_norm": 1.4199910371395013,
      "learning_rate": 4.470487121721834e-06,
      "loss": 0.5491,
      "step": 2586
    },
    {
      "epoch": 3.1664626682986534,
      "grad_norm": 1.1043075926980916,
      "learning_rate": 4.470092643049915e-06,
      "loss": 0.5834,
      "step": 2587
    },
    {
      "epoch": 3.167686658506732,
      "grad_norm": 1.0721091464719736,
      "learning_rate": 4.469698034910793e-06,
      "loss": 0.3474,
      "step": 2588
    },
    {
      "epoch": 3.16891064871481,
      "grad_norm": 0.8477306538676153,
      "learning_rate": 4.469303297330403e-06,
      "loss": 0.7799,
      "step": 2589
    },
    {
      "epoch": 3.170134638922889,
      "grad_norm": 1.1512127089742756,
      "learning_rate": 4.468908430334684e-06,
      "loss": 0.6803,
      "step": 2590
    },
    {
      "epoch": 3.171358629130967,
      "grad_norm": 1.2614828649850511,
      "learning_rate": 4.468513433949585e-06,
      "loss": 1.1614,
      "step": 2591
    },
    {
      "epoch": 3.172582619339045,
      "grad_norm": 1.5476053908647838,
      "learning_rate": 4.468118308201065e-06,
      "loss": 0.8099,
      "step": 2592
    },
    {
      "epoch": 3.173806609547124,
      "grad_norm": 0.8780273248879857,
      "learning_rate": 4.46772305311509e-06,
      "loss": 0.6551,
      "step": 2593
    },
    {
      "epoch": 3.175030599755202,
      "grad_norm": 0.7370991977249216,
      "learning_rate": 4.467327668717632e-06,
      "loss": 0.5215,
      "step": 2594
    },
    {
      "epoch": 3.17625458996328,
      "grad_norm": 0.9167905205632998,
      "learning_rate": 4.466932155034677e-06,
      "loss": 0.6079,
      "step": 2595
    },
    {
      "epoch": 3.1774785801713588,
      "grad_norm": 0.7912957149431679,
      "learning_rate": 4.466536512092216e-06,
      "loss": 0.5101,
      "step": 2596
    },
    {
      "epoch": 3.178702570379437,
      "grad_norm": 1.4423738512434698,
      "learning_rate": 4.46614073991625e-06,
      "loss": 0.566,
      "step": 2597
    },
    {
      "epoch": 3.179926560587515,
      "grad_norm": 0.8589861981648842,
      "learning_rate": 4.465744838532786e-06,
      "loss": 1.0072,
      "step": 2598
    },
    {
      "epoch": 3.1811505507955937,
      "grad_norm": 1.333058121826294,
      "learning_rate": 4.465348807967842e-06,
      "loss": 0.5097,
      "step": 2599
    },
    {
      "epoch": 3.182374541003672,
      "grad_norm": 1.1479632024002633,
      "learning_rate": 4.4649526482474435e-06,
      "loss": 0.3921,
      "step": 2600
    },
    {
      "epoch": 3.18359853121175,
      "grad_norm": 0.5494598044967577,
      "learning_rate": 4.464556359397625e-06,
      "loss": 0.4534,
      "step": 2601
    },
    {
      "epoch": 3.1848225214198287,
      "grad_norm": 0.9700017823564119,
      "learning_rate": 4.4641599414444295e-06,
      "loss": 1.3644,
      "step": 2602
    },
    {
      "epoch": 3.186046511627907,
      "grad_norm": 1.40526718621559,
      "learning_rate": 4.463763394413907e-06,
      "loss": 1.0614,
      "step": 2603
    },
    {
      "epoch": 3.1872705018359855,
      "grad_norm": 1.012231969705591,
      "learning_rate": 4.4633667183321175e-06,
      "loss": 0.6296,
      "step": 2604
    },
    {
      "epoch": 3.1884944920440637,
      "grad_norm": 1.913108551768777,
      "learning_rate": 4.462969913225129e-06,
      "loss": 0.8135,
      "step": 2605
    },
    {
      "epoch": 3.189718482252142,
      "grad_norm": 0.9463616946155443,
      "learning_rate": 4.46257297911902e-06,
      "loss": 0.6448,
      "step": 2606
    },
    {
      "epoch": 3.1909424724602204,
      "grad_norm": 1.001563362993544,
      "learning_rate": 4.462175916039873e-06,
      "loss": 0.6179,
      "step": 2607
    },
    {
      "epoch": 3.1921664626682986,
      "grad_norm": 1.1403831201551435,
      "learning_rate": 4.461778724013782e-06,
      "loss": 1.598,
      "step": 2608
    },
    {
      "epoch": 3.193390452876377,
      "grad_norm": 1.6825041551604836,
      "learning_rate": 4.461381403066849e-06,
      "loss": 0.667,
      "step": 2609
    },
    {
      "epoch": 3.1946144430844554,
      "grad_norm": 1.770864015869555,
      "learning_rate": 4.460983953225185e-06,
      "loss": 0.4891,
      "step": 2610
    },
    {
      "epoch": 3.1958384332925336,
      "grad_norm": 0.9044727007043715,
      "learning_rate": 4.460586374514909e-06,
      "loss": 0.6229,
      "step": 2611
    },
    {
      "epoch": 3.197062423500612,
      "grad_norm": 0.7637346254368684,
      "learning_rate": 4.460188666962148e-06,
      "loss": 0.5955,
      "step": 2612
    },
    {
      "epoch": 3.1982864137086904,
      "grad_norm": 0.7799998003264197,
      "learning_rate": 4.459790830593039e-06,
      "loss": 0.5398,
      "step": 2613
    },
    {
      "epoch": 3.1995104039167686,
      "grad_norm": 1.571755677225706,
      "learning_rate": 4.459392865433725e-06,
      "loss": 0.9315,
      "step": 2614
    },
    {
      "epoch": 3.200734394124847,
      "grad_norm": 0.9499668310293599,
      "learning_rate": 4.458994771510359e-06,
      "loss": 0.6063,
      "step": 2615
    },
    {
      "epoch": 3.2019583843329253,
      "grad_norm": 1.9911460683020177,
      "learning_rate": 4.458596548849102e-06,
      "loss": 0.6364,
      "step": 2616
    },
    {
      "epoch": 3.2031823745410035,
      "grad_norm": 1.7724146792601028,
      "learning_rate": 4.458198197476125e-06,
      "loss": 0.5533,
      "step": 2617
    },
    {
      "epoch": 3.204406364749082,
      "grad_norm": 1.169349072075869,
      "learning_rate": 4.457799717417605e-06,
      "loss": 1.1578,
      "step": 2618
    },
    {
      "epoch": 3.2056303549571603,
      "grad_norm": 1.2540338634203407,
      "learning_rate": 4.457401108699728e-06,
      "loss": 0.4386,
      "step": 2619
    },
    {
      "epoch": 3.2068543451652385,
      "grad_norm": 1.1851256631008218,
      "learning_rate": 4.457002371348691e-06,
      "loss": 0.6075,
      "step": 2620
    },
    {
      "epoch": 3.208078335373317,
      "grad_norm": 0.9322340878778848,
      "learning_rate": 4.456603505390697e-06,
      "loss": 1.0826,
      "step": 2621
    },
    {
      "epoch": 3.2093023255813953,
      "grad_norm": 0.9443083724456973,
      "learning_rate": 4.456204510851957e-06,
      "loss": 1.3784,
      "step": 2622
    },
    {
      "epoch": 3.2105263157894735,
      "grad_norm": 0.7320423469165174,
      "learning_rate": 4.4558053877586916e-06,
      "loss": 0.5414,
      "step": 2623
    },
    {
      "epoch": 3.211750305997552,
      "grad_norm": 1.0054551568023882,
      "learning_rate": 4.45540613613713e-06,
      "loss": 1.14,
      "step": 2624
    },
    {
      "epoch": 3.2129742962056302,
      "grad_norm": 1.0560265074825619,
      "learning_rate": 4.455006756013511e-06,
      "loss": 1.1931,
      "step": 2625
    },
    {
      "epoch": 3.214198286413709,
      "grad_norm": 2.1362992285726308,
      "learning_rate": 4.454607247414079e-06,
      "loss": 0.4561,
      "step": 2626
    },
    {
      "epoch": 3.215422276621787,
      "grad_norm": 1.4679701139238326,
      "learning_rate": 4.454207610365087e-06,
      "loss": 0.6557,
      "step": 2627
    },
    {
      "epoch": 3.216646266829865,
      "grad_norm": 1.462345635435965,
      "learning_rate": 4.4538078448928e-06,
      "loss": 0.4999,
      "step": 2628
    },
    {
      "epoch": 3.217870257037944,
      "grad_norm": 0.8687002546075069,
      "learning_rate": 4.4534079510234875e-06,
      "loss": 0.5528,
      "step": 2629
    },
    {
      "epoch": 3.219094247246022,
      "grad_norm": 1.1125521015557662,
      "learning_rate": 4.45300792878343e-06,
      "loss": 0.6672,
      "step": 2630
    },
    {
      "epoch": 3.2203182374541,
      "grad_norm": 1.1250931457548383,
      "learning_rate": 4.452607778198915e-06,
      "loss": 0.5062,
      "step": 2631
    },
    {
      "epoch": 3.221542227662179,
      "grad_norm": 1.021747761331091,
      "learning_rate": 4.4522074992962385e-06,
      "loss": 0.5198,
      "step": 2632
    },
    {
      "epoch": 3.222766217870257,
      "grad_norm": 1.0310077556661634,
      "learning_rate": 4.451807092101707e-06,
      "loss": 1.0103,
      "step": 2633
    },
    {
      "epoch": 3.2239902080783356,
      "grad_norm": 0.8196739162623282,
      "learning_rate": 4.451406556641632e-06,
      "loss": 0.5356,
      "step": 2634
    },
    {
      "epoch": 3.2252141982864138,
      "grad_norm": 0.7027200013796292,
      "learning_rate": 4.451005892942335e-06,
      "loss": 0.6986,
      "step": 2635
    },
    {
      "epoch": 3.226438188494492,
      "grad_norm": 1.4296481226181454,
      "learning_rate": 4.4506051010301476e-06,
      "loss": 0.4858,
      "step": 2636
    },
    {
      "epoch": 3.2276621787025706,
      "grad_norm": 1.2402432157775487,
      "learning_rate": 4.450204180931408e-06,
      "loss": 0.5568,
      "step": 2637
    },
    {
      "epoch": 3.2288861689106487,
      "grad_norm": 1.717963629582267,
      "learning_rate": 4.449803132672463e-06,
      "loss": 0.4306,
      "step": 2638
    },
    {
      "epoch": 3.230110159118727,
      "grad_norm": 0.985157708532004,
      "learning_rate": 4.449401956279668e-06,
      "loss": 0.544,
      "step": 2639
    },
    {
      "epoch": 3.2313341493268055,
      "grad_norm": 1.0390092615734745,
      "learning_rate": 4.449000651779386e-06,
      "loss": 1.1237,
      "step": 2640
    },
    {
      "epoch": 3.2325581395348837,
      "grad_norm": 0.9140381029303751,
      "learning_rate": 4.448599219197991e-06,
      "loss": 0.5715,
      "step": 2641
    },
    {
      "epoch": 3.233782129742962,
      "grad_norm": 1.1920371592866783,
      "learning_rate": 4.448197658561862e-06,
      "loss": 0.5904,
      "step": 2642
    },
    {
      "epoch": 3.2350061199510405,
      "grad_norm": 1.0112063502909725,
      "learning_rate": 4.447795969897389e-06,
      "loss": 0.5741,
      "step": 2643
    },
    {
      "epoch": 3.2362301101591187,
      "grad_norm": 1.1133376094433798,
      "learning_rate": 4.44739415323097e-06,
      "loss": 0.5129,
      "step": 2644
    },
    {
      "epoch": 3.237454100367197,
      "grad_norm": 2.0895401041569546,
      "learning_rate": 4.446992208589009e-06,
      "loss": 0.4258,
      "step": 2645
    },
    {
      "epoch": 3.2386780905752754,
      "grad_norm": 1.567868741342563,
      "learning_rate": 4.446590135997923e-06,
      "loss": 0.6882,
      "step": 2646
    },
    {
      "epoch": 3.2399020807833536,
      "grad_norm": 1.0914505646847985,
      "learning_rate": 4.446187935484132e-06,
      "loss": 0.9817,
      "step": 2647
    },
    {
      "epoch": 3.2411260709914322,
      "grad_norm": 1.4437849758416024,
      "learning_rate": 4.445785607074068e-06,
      "loss": 1.2419,
      "step": 2648
    },
    {
      "epoch": 3.2423500611995104,
      "grad_norm": 1.8514751747244613,
      "learning_rate": 4.445383150794171e-06,
      "loss": 0.4805,
      "step": 2649
    },
    {
      "epoch": 3.2435740514075886,
      "grad_norm": 0.9154754996574146,
      "learning_rate": 4.44498056667089e-06,
      "loss": 0.9878,
      "step": 2650
    },
    {
      "epoch": 3.244798041615667,
      "grad_norm": 0.8492533848844187,
      "learning_rate": 4.444577854730679e-06,
      "loss": 0.7193,
      "step": 2651
    },
    {
      "epoch": 3.2460220318237454,
      "grad_norm": 0.9369013577272924,
      "learning_rate": 4.444175015000004e-06,
      "loss": 0.52,
      "step": 2652
    },
    {
      "epoch": 3.2472460220318236,
      "grad_norm": 0.6864696330509883,
      "learning_rate": 4.443772047505338e-06,
      "loss": 0.5542,
      "step": 2653
    },
    {
      "epoch": 3.248470012239902,
      "grad_norm": 1.3931930514187927,
      "learning_rate": 4.443368952273163e-06,
      "loss": 0.4928,
      "step": 2654
    },
    {
      "epoch": 3.2496940024479803,
      "grad_norm": 0.868042442708177,
      "learning_rate": 4.442965729329968e-06,
      "loss": 0.5672,
      "step": 2655
    },
    {
      "epoch": 3.250917992656059,
      "grad_norm": 0.6890059002616634,
      "learning_rate": 4.442562378702252e-06,
      "loss": 0.7149,
      "step": 2656
    },
    {
      "epoch": 3.252141982864137,
      "grad_norm": 1.0487961905643408,
      "learning_rate": 4.442158900416522e-06,
      "loss": 1.1159,
      "step": 2657
    },
    {
      "epoch": 3.2533659730722153,
      "grad_norm": 1.052289384821845,
      "learning_rate": 4.441755294499291e-06,
      "loss": 0.6048,
      "step": 2658
    },
    {
      "epoch": 3.254589963280294,
      "grad_norm": 1.5612195175337875,
      "learning_rate": 4.441351560977086e-06,
      "loss": 0.6001,
      "step": 2659
    },
    {
      "epoch": 3.255813953488372,
      "grad_norm": 1.6370736612829806,
      "learning_rate": 4.4409476998764365e-06,
      "loss": 0.5807,
      "step": 2660
    },
    {
      "epoch": 3.2570379436964503,
      "grad_norm": 0.9275890148875694,
      "learning_rate": 4.440543711223883e-06,
      "loss": 0.7138,
      "step": 2661
    },
    {
      "epoch": 3.258261933904529,
      "grad_norm": 0.9665070840967706,
      "learning_rate": 4.440139595045974e-06,
      "loss": 0.52,
      "step": 2662
    },
    {
      "epoch": 3.259485924112607,
      "grad_norm": 0.7379025944041129,
      "learning_rate": 4.4397353513692674e-06,
      "loss": 0.6325,
      "step": 2663
    },
    {
      "epoch": 3.2607099143206852,
      "grad_norm": 0.8717083421995269,
      "learning_rate": 4.439330980220328e-06,
      "loss": 0.5528,
      "step": 2664
    },
    {
      "epoch": 3.261933904528764,
      "grad_norm": 1.2548864207990376,
      "learning_rate": 4.438926481625729e-06,
      "loss": 0.4705,
      "step": 2665
    },
    {
      "epoch": 3.263157894736842,
      "grad_norm": 1.2262493575653592,
      "learning_rate": 4.438521855612054e-06,
      "loss": 0.7954,
      "step": 2666
    },
    {
      "epoch": 3.26438188494492,
      "grad_norm": 1.1488454081167205,
      "learning_rate": 4.438117102205892e-06,
      "loss": 1.5043,
      "step": 2667
    },
    {
      "epoch": 3.265605875152999,
      "grad_norm": 0.7969860113308458,
      "learning_rate": 4.437712221433844e-06,
      "loss": 0.6872,
      "step": 2668
    },
    {
      "epoch": 3.266829865361077,
      "grad_norm": 0.6165026029884395,
      "learning_rate": 4.437307213322515e-06,
      "loss": 0.4368,
      "step": 2669
    },
    {
      "epoch": 3.2680538555691556,
      "grad_norm": 0.7365713676338234,
      "learning_rate": 4.436902077898522e-06,
      "loss": 0.6128,
      "step": 2670
    },
    {
      "epoch": 3.269277845777234,
      "grad_norm": 1.1803341188059886,
      "learning_rate": 4.436496815188488e-06,
      "loss": 0.6433,
      "step": 2671
    },
    {
      "epoch": 3.270501835985312,
      "grad_norm": 1.7173044344016908,
      "learning_rate": 4.4360914252190456e-06,
      "loss": 0.5093,
      "step": 2672
    },
    {
      "epoch": 3.2717258261933906,
      "grad_norm": 1.007549423758967,
      "learning_rate": 4.435685908016837e-06,
      "loss": 0.6425,
      "step": 2673
    },
    {
      "epoch": 3.2729498164014688,
      "grad_norm": 1.74256169945507,
      "learning_rate": 4.435280263608509e-06,
      "loss": 0.5804,
      "step": 2674
    },
    {
      "epoch": 3.274173806609547,
      "grad_norm": 0.8909658487128782,
      "learning_rate": 4.434874492020721e-06,
      "loss": 0.4306,
      "step": 2675
    },
    {
      "epoch": 3.2753977968176256,
      "grad_norm": 0.9439176644306071,
      "learning_rate": 4.434468593280138e-06,
      "loss": 0.5259,
      "step": 2676
    },
    {
      "epoch": 3.2766217870257037,
      "grad_norm": 1.711519333661347,
      "learning_rate": 4.434062567413433e-06,
      "loss": 0.4171,
      "step": 2677
    },
    {
      "epoch": 3.2778457772337823,
      "grad_norm": 1.8987673250280912,
      "learning_rate": 4.43365641444729e-06,
      "loss": 1.2462,
      "step": 2678
    },
    {
      "epoch": 3.2790697674418605,
      "grad_norm": 1.6680877692873155,
      "learning_rate": 4.433250134408401e-06,
      "loss": 1.2577,
      "step": 2679
    },
    {
      "epoch": 3.2802937576499387,
      "grad_norm": 0.9579754586159103,
      "learning_rate": 4.4328437273234625e-06,
      "loss": 0.6565,
      "step": 2680
    },
    {
      "epoch": 3.2815177478580173,
      "grad_norm": 0.9597824348685214,
      "learning_rate": 4.432437193219183e-06,
      "loss": 0.4802,
      "step": 2681
    },
    {
      "epoch": 3.2827417380660955,
      "grad_norm": 1.161809877233691,
      "learning_rate": 4.43203053212228e-06,
      "loss": 0.4529,
      "step": 2682
    },
    {
      "epoch": 3.2839657282741737,
      "grad_norm": 1.2952613916773414,
      "learning_rate": 4.431623744059476e-06,
      "loss": 0.9278,
      "step": 2683
    },
    {
      "epoch": 3.2851897184822523,
      "grad_norm": 1.3324884311604774,
      "learning_rate": 4.431216829057504e-06,
      "loss": 1.3681,
      "step": 2684
    },
    {
      "epoch": 3.2864137086903304,
      "grad_norm": 1.111229785629199,
      "learning_rate": 4.430809787143105e-06,
      "loss": 0.5966,
      "step": 2685
    },
    {
      "epoch": 3.2876376988984086,
      "grad_norm": 1.1208449653065697,
      "learning_rate": 4.430402618343028e-06,
      "loss": 0.5011,
      "step": 2686
    },
    {
      "epoch": 3.2888616891064872,
      "grad_norm": 1.7063354603515155,
      "learning_rate": 4.429995322684032e-06,
      "loss": 0.8115,
      "step": 2687
    },
    {
      "epoch": 3.2900856793145654,
      "grad_norm": 1.2262144475844399,
      "learning_rate": 4.429587900192881e-06,
      "loss": 0.4256,
      "step": 2688
    },
    {
      "epoch": 3.2913096695226436,
      "grad_norm": 1.626022521072038,
      "learning_rate": 4.4291803508963506e-06,
      "loss": 0.441,
      "step": 2689
    },
    {
      "epoch": 3.292533659730722,
      "grad_norm": 1.441735230277083,
      "learning_rate": 4.4287726748212235e-06,
      "loss": 0.6635,
      "step": 2690
    },
    {
      "epoch": 3.2937576499388004,
      "grad_norm": 1.3876374263586793,
      "learning_rate": 4.42836487199429e-06,
      "loss": 0.6625,
      "step": 2691
    },
    {
      "epoch": 3.294981640146879,
      "grad_norm": 1.482655370622302,
      "learning_rate": 4.427956942442349e-06,
      "loss": 0.516,
      "step": 2692
    },
    {
      "epoch": 3.296205630354957,
      "grad_norm": 2.1223864067081886,
      "learning_rate": 4.42754888619221e-06,
      "loss": 0.4424,
      "step": 2693
    },
    {
      "epoch": 3.2974296205630353,
      "grad_norm": 1.0482685391943547,
      "learning_rate": 4.427140703270687e-06,
      "loss": 0.5512,
      "step": 2694
    },
    {
      "epoch": 3.298653610771114,
      "grad_norm": 0.7887380587770276,
      "learning_rate": 4.426732393704605e-06,
      "loss": 0.5776,
      "step": 2695
    },
    {
      "epoch": 3.299877600979192,
      "grad_norm": 0.766690781713656,
      "learning_rate": 4.426323957520796e-06,
      "loss": 0.4764,
      "step": 2696
    },
    {
      "epoch": 3.3011015911872703,
      "grad_norm": 2.026563473934643,
      "learning_rate": 4.425915394746102e-06,
      "loss": 0.4985,
      "step": 2697
    },
    {
      "epoch": 3.302325581395349,
      "grad_norm": 1.4506597506064534,
      "learning_rate": 4.425506705407372e-06,
      "loss": 0.4462,
      "step": 2698
    },
    {
      "epoch": 3.303549571603427,
      "grad_norm": 1.1723034249916875,
      "learning_rate": 4.425097889531463e-06,
      "loss": 0.462,
      "step": 2699
    },
    {
      "epoch": 3.3047735618115057,
      "grad_norm": 1.2078255135433702,
      "learning_rate": 4.424688947145241e-06,
      "loss": 1.3588,
      "step": 2700
    },
    {
      "epoch": 3.305997552019584,
      "grad_norm": 1.4815255958965035,
      "learning_rate": 4.42427987827558e-06,
      "loss": 1.0733,
      "step": 2701
    },
    {
      "epoch": 3.307221542227662,
      "grad_norm": 1.132153635071585,
      "learning_rate": 4.423870682949364e-06,
      "loss": 0.5203,
      "step": 2702
    },
    {
      "epoch": 3.3084455324357407,
      "grad_norm": 0.6263112505062345,
      "learning_rate": 4.423461361193482e-06,
      "loss": 0.4892,
      "step": 2703
    },
    {
      "epoch": 3.309669522643819,
      "grad_norm": 0.889306256562072,
      "learning_rate": 4.423051913034834e-06,
      "loss": 0.4911,
      "step": 2704
    },
    {
      "epoch": 3.310893512851897,
      "grad_norm": 1.2987679984113512,
      "learning_rate": 4.422642338500326e-06,
      "loss": 0.65,
      "step": 2705
    },
    {
      "epoch": 3.3121175030599757,
      "grad_norm": 1.6312145481996354,
      "learning_rate": 4.422232637616876e-06,
      "loss": 0.4989,
      "step": 2706
    },
    {
      "epoch": 3.313341493268054,
      "grad_norm": 1.0070529103768027,
      "learning_rate": 4.4218228104114066e-06,
      "loss": 0.8547,
      "step": 2707
    },
    {
      "epoch": 3.314565483476132,
      "grad_norm": 1.2242933492081007,
      "learning_rate": 4.421412856910851e-06,
      "loss": 0.5457,
      "step": 2708
    },
    {
      "epoch": 3.3157894736842106,
      "grad_norm": 1.902067187385129,
      "learning_rate": 4.421002777142148e-06,
      "loss": 1.0382,
      "step": 2709
    },
    {
      "epoch": 3.317013463892289,
      "grad_norm": 1.264462272910692,
      "learning_rate": 4.420592571132248e-06,
      "loss": 0.4768,
      "step": 2710
    },
    {
      "epoch": 3.318237454100367,
      "grad_norm": 1.6804339570943123,
      "learning_rate": 4.420182238908109e-06,
      "loss": 0.7498,
      "step": 2711
    },
    {
      "epoch": 3.3194614443084456,
      "grad_norm": 1.0337061816433162,
      "learning_rate": 4.419771780496695e-06,
      "loss": 1.2718,
      "step": 2712
    },
    {
      "epoch": 3.3206854345165238,
      "grad_norm": 0.8650000890953032,
      "learning_rate": 4.419361195924981e-06,
      "loss": 0.5667,
      "step": 2713
    },
    {
      "epoch": 3.3219094247246024,
      "grad_norm": 0.8212654899883196,
      "learning_rate": 4.418950485219948e-06,
      "loss": 0.7553,
      "step": 2714
    },
    {
      "epoch": 3.3231334149326806,
      "grad_norm": 1.218013265399597,
      "learning_rate": 4.418539648408585e-06,
      "loss": 1.2433,
      "step": 2715
    },
    {
      "epoch": 3.3243574051407587,
      "grad_norm": 1.3577298519672396,
      "learning_rate": 4.4181286855178944e-06,
      "loss": 0.5406,
      "step": 2716
    },
    {
      "epoch": 3.3255813953488373,
      "grad_norm": 1.2480802355030498,
      "learning_rate": 4.417717596574881e-06,
      "loss": 1.4496,
      "step": 2717
    },
    {
      "epoch": 3.3268053855569155,
      "grad_norm": 0.8857464238732545,
      "learning_rate": 4.4173063816065605e-06,
      "loss": 0.689,
      "step": 2718
    },
    {
      "epoch": 3.3280293757649937,
      "grad_norm": 1.3435207876633417,
      "learning_rate": 4.416895040639956e-06,
      "loss": 0.3626,
      "step": 2719
    },
    {
      "epoch": 3.3292533659730723,
      "grad_norm": 0.5194498669147878,
      "learning_rate": 4.416483573702099e-06,
      "loss": 0.2386,
      "step": 2720
    },
    {
      "epoch": 3.3304773561811505,
      "grad_norm": 1.0070001710879286,
      "learning_rate": 4.416071980820031e-06,
      "loss": 0.8706,
      "step": 2721
    },
    {
      "epoch": 3.331701346389229,
      "grad_norm": 0.9347266902369363,
      "learning_rate": 4.415660262020799e-06,
      "loss": 1.1996,
      "step": 2722
    },
    {
      "epoch": 3.3329253365973073,
      "grad_norm": 1.5662282643601981,
      "learning_rate": 4.41524841733146e-06,
      "loss": 0.4973,
      "step": 2723
    },
    {
      "epoch": 3.3341493268053854,
      "grad_norm": 1.6341455588006524,
      "learning_rate": 4.41483644677908e-06,
      "loss": 0.8658,
      "step": 2724
    },
    {
      "epoch": 3.335373317013464,
      "grad_norm": 1.1362051047984356,
      "learning_rate": 4.41442435039073e-06,
      "loss": 0.6037,
      "step": 2725
    },
    {
      "epoch": 3.3365973072215422,
      "grad_norm": 0.9192416189778845,
      "learning_rate": 4.414012128193493e-06,
      "loss": 0.657,
      "step": 2726
    },
    {
      "epoch": 3.3378212974296204,
      "grad_norm": 1.7308278259054983,
      "learning_rate": 4.413599780214458e-06,
      "loss": 0.5298,
      "step": 2727
    },
    {
      "epoch": 3.339045287637699,
      "grad_norm": 1.8412556187973665,
      "learning_rate": 4.413187306480724e-06,
      "loss": 1.1041,
      "step": 2728
    },
    {
      "epoch": 3.340269277845777,
      "grad_norm": 1.4645183444670837,
      "learning_rate": 4.412774707019396e-06,
      "loss": 0.6633,
      "step": 2729
    },
    {
      "epoch": 3.3414932680538554,
      "grad_norm": 1.0312717594668688,
      "learning_rate": 4.41236198185759e-06,
      "loss": 0.5115,
      "step": 2730
    },
    {
      "epoch": 3.342717258261934,
      "grad_norm": 0.8122137838361809,
      "learning_rate": 4.411949131022427e-06,
      "loss": 0.8746,
      "step": 2731
    },
    {
      "epoch": 3.343941248470012,
      "grad_norm": 1.1773914063742836,
      "learning_rate": 4.4115361545410394e-06,
      "loss": 0.8711,
      "step": 2732
    },
    {
      "epoch": 3.3451652386780903,
      "grad_norm": 1.592330154498065,
      "learning_rate": 4.411123052440567e-06,
      "loss": 0.482,
      "step": 2733
    },
    {
      "epoch": 3.346389228886169,
      "grad_norm": 1.2787630443039923,
      "learning_rate": 4.4107098247481554e-06,
      "loss": 0.6868,
      "step": 2734
    },
    {
      "epoch": 3.347613219094247,
      "grad_norm": 0.7082083981743631,
      "learning_rate": 4.410296471490961e-06,
      "loss": 0.6662,
      "step": 2735
    },
    {
      "epoch": 3.3488372093023258,
      "grad_norm": 0.8093219688068988,
      "learning_rate": 4.4098829926961485e-06,
      "loss": 0.783,
      "step": 2736
    },
    {
      "epoch": 3.350061199510404,
      "grad_norm": 1.1772481129014791,
      "learning_rate": 4.409469388390889e-06,
      "loss": 0.9779,
      "step": 2737
    },
    {
      "epoch": 3.351285189718482,
      "grad_norm": 1.1056253636309834,
      "learning_rate": 4.4090556586023655e-06,
      "loss": 0.5179,
      "step": 2738
    },
    {
      "epoch": 3.3525091799265607,
      "grad_norm": 1.1931344120433756,
      "learning_rate": 4.4086418033577636e-06,
      "loss": 1.6174,
      "step": 2739
    },
    {
      "epoch": 3.353733170134639,
      "grad_norm": 0.8721041710742241,
      "learning_rate": 4.408227822684283e-06,
      "loss": 0.4448,
      "step": 2740
    },
    {
      "epoch": 3.354957160342717,
      "grad_norm": 1.251175103104469,
      "learning_rate": 4.407813716609128e-06,
      "loss": 0.5613,
      "step": 2741
    },
    {
      "epoch": 3.3561811505507957,
      "grad_norm": 0.9910594031298472,
      "learning_rate": 4.407399485159512e-06,
      "loss": 0.6944,
      "step": 2742
    },
    {
      "epoch": 3.357405140758874,
      "grad_norm": 0.9911667565221486,
      "learning_rate": 4.406985128362656e-06,
      "loss": 1.6074,
      "step": 2743
    },
    {
      "epoch": 3.3586291309669525,
      "grad_norm": 1.5959957600003385,
      "learning_rate": 4.406570646245793e-06,
      "loss": 0.6811,
      "step": 2744
    },
    {
      "epoch": 3.3598531211750307,
      "grad_norm": 0.7305389437577646,
      "learning_rate": 4.4061560388361564e-06,
      "loss": 0.6313,
      "step": 2745
    },
    {
      "epoch": 3.361077111383109,
      "grad_norm": 1.3163889709485193,
      "learning_rate": 4.405741306160997e-06,
      "loss": 1.0769,
      "step": 2746
    },
    {
      "epoch": 3.3623011015911874,
      "grad_norm": 0.7793767899228182,
      "learning_rate": 4.405326448247567e-06,
      "loss": 0.5971,
      "step": 2747
    },
    {
      "epoch": 3.3635250917992656,
      "grad_norm": 0.8966777003861152,
      "learning_rate": 4.404911465123131e-06,
      "loss": 0.5803,
      "step": 2748
    },
    {
      "epoch": 3.364749082007344,
      "grad_norm": 1.9126181154736117,
      "learning_rate": 4.404496356814958e-06,
      "loss": 0.479,
      "step": 2749
    },
    {
      "epoch": 3.3659730722154224,
      "grad_norm": 1.1276654648520013,
      "learning_rate": 4.4040811233503296e-06,
      "loss": 1.022,
      "step": 2750
    },
    {
      "epoch": 3.3671970624235006,
      "grad_norm": 1.0229723981976373,
      "learning_rate": 4.4036657647565315e-06,
      "loss": 0.7521,
      "step": 2751
    },
    {
      "epoch": 3.3684210526315788,
      "grad_norm": 1.151442856219347,
      "learning_rate": 4.403250281060862e-06,
      "loss": 0.6817,
      "step": 2752
    },
    {
      "epoch": 3.3696450428396574,
      "grad_norm": 1.1966727939792607,
      "learning_rate": 4.402834672290622e-06,
      "loss": 1.396,
      "step": 2753
    },
    {
      "epoch": 3.3708690330477356,
      "grad_norm": 0.9693675940728415,
      "learning_rate": 4.4024189384731275e-06,
      "loss": 1.3487,
      "step": 2754
    },
    {
      "epoch": 3.3720930232558137,
      "grad_norm": 2.0644468002142218,
      "learning_rate": 4.402003079635695e-06,
      "loss": 0.7661,
      "step": 2755
    },
    {
      "epoch": 3.3733170134638923,
      "grad_norm": 0.6825495044664927,
      "learning_rate": 4.401587095805656e-06,
      "loss": 0.5498,
      "step": 2756
    },
    {
      "epoch": 3.3745410036719705,
      "grad_norm": 1.021790621595063,
      "learning_rate": 4.401170987010347e-06,
      "loss": 0.5098,
      "step": 2757
    },
    {
      "epoch": 3.375764993880049,
      "grad_norm": 1.2775883023145036,
      "learning_rate": 4.400754753277112e-06,
      "loss": 0.454,
      "step": 2758
    },
    {
      "epoch": 3.3769889840881273,
      "grad_norm": 1.1441906126931822,
      "learning_rate": 4.400338394633305e-06,
      "loss": 0.7094,
      "step": 2759
    },
    {
      "epoch": 3.3782129742962055,
      "grad_norm": 1.2594700855639822,
      "learning_rate": 4.399921911106288e-06,
      "loss": 0.4317,
      "step": 2760
    },
    {
      "epoch": 3.379436964504284,
      "grad_norm": 1.5859558688348998,
      "learning_rate": 4.39950530272343e-06,
      "loss": 1.0807,
      "step": 2761
    },
    {
      "epoch": 3.3806609547123623,
      "grad_norm": 0.6769146465512795,
      "learning_rate": 4.399088569512109e-06,
      "loss": 0.5419,
      "step": 2762
    },
    {
      "epoch": 3.3818849449204405,
      "grad_norm": 0.9733351350375833,
      "learning_rate": 4.398671711499712e-06,
      "loss": 0.7202,
      "step": 2763
    },
    {
      "epoch": 3.383108935128519,
      "grad_norm": 1.1165237365821272,
      "learning_rate": 4.398254728713632e-06,
      "loss": 1.2179,
      "step": 2764
    },
    {
      "epoch": 3.3843329253365972,
      "grad_norm": 1.1976231747259047,
      "learning_rate": 4.397837621181273e-06,
      "loss": 0.8048,
      "step": 2765
    },
    {
      "epoch": 3.385556915544676,
      "grad_norm": 1.0921397616821866,
      "learning_rate": 4.397420388930046e-06,
      "loss": 0.5614,
      "step": 2766
    },
    {
      "epoch": 3.386780905752754,
      "grad_norm": 1.8336821917087471,
      "learning_rate": 4.397003031987368e-06,
      "loss": 0.8926,
      "step": 2767
    },
    {
      "epoch": 3.388004895960832,
      "grad_norm": 1.0981519979821863,
      "learning_rate": 4.3965855503806675e-06,
      "loss": 1.1344,
      "step": 2768
    },
    {
      "epoch": 3.389228886168911,
      "grad_norm": 1.4237473498939397,
      "learning_rate": 4.3961679441373795e-06,
      "loss": 0.6164,
      "step": 2769
    },
    {
      "epoch": 3.390452876376989,
      "grad_norm": 0.9575515860803475,
      "learning_rate": 4.395750213284948e-06,
      "loss": 0.9263,
      "step": 2770
    },
    {
      "epoch": 3.391676866585067,
      "grad_norm": 1.6718960381624643,
      "learning_rate": 4.395332357850824e-06,
      "loss": 0.513,
      "step": 2771
    },
    {
      "epoch": 3.392900856793146,
      "grad_norm": 0.9364567999550558,
      "learning_rate": 4.3949143778624684e-06,
      "loss": 1.6152,
      "step": 2772
    },
    {
      "epoch": 3.394124847001224,
      "grad_norm": 1.3155202728556288,
      "learning_rate": 4.394496273347347e-06,
      "loss": 0.5482,
      "step": 2773
    },
    {
      "epoch": 3.395348837209302,
      "grad_norm": 1.064892307261788,
      "learning_rate": 4.39407804433294e-06,
      "loss": 0.72,
      "step": 2774
    },
    {
      "epoch": 3.3965728274173808,
      "grad_norm": 1.6779136258700245,
      "learning_rate": 4.393659690846729e-06,
      "loss": 0.4574,
      "step": 2775
    },
    {
      "epoch": 3.397796817625459,
      "grad_norm": 0.8738043965238238,
      "learning_rate": 4.3932412129162064e-06,
      "loss": 0.4451,
      "step": 2776
    },
    {
      "epoch": 3.399020807833537,
      "grad_norm": 2.074419356457092,
      "learning_rate": 4.392822610568874e-06,
      "loss": 0.3974,
      "step": 2777
    },
    {
      "epoch": 3.4002447980416157,
      "grad_norm": 1.1979514605266683,
      "learning_rate": 4.3924038838322415e-06,
      "loss": 0.6566,
      "step": 2778
    },
    {
      "epoch": 3.401468788249694,
      "grad_norm": 1.6639487215398465,
      "learning_rate": 4.391985032733826e-06,
      "loss": 0.6592,
      "step": 2779
    },
    {
      "epoch": 3.4026927784577725,
      "grad_norm": 1.161894488077876,
      "learning_rate": 4.3915660573011495e-06,
      "loss": 0.4735,
      "step": 2780
    },
    {
      "epoch": 3.4039167686658507,
      "grad_norm": 1.0009624796843961,
      "learning_rate": 4.391146957561751e-06,
      "loss": 0.4476,
      "step": 2781
    },
    {
      "epoch": 3.405140758873929,
      "grad_norm": 1.4277098396592909,
      "learning_rate": 4.390727733543167e-06,
      "loss": 0.7394,
      "step": 2782
    },
    {
      "epoch": 3.4063647490820075,
      "grad_norm": 1.1029907752289099,
      "learning_rate": 4.390308385272951e-06,
      "loss": 1.303,
      "step": 2783
    },
    {
      "epoch": 3.4075887392900857,
      "grad_norm": 1.6145219953478458,
      "learning_rate": 4.389888912778659e-06,
      "loss": 0.607,
      "step": 2784
    },
    {
      "epoch": 3.408812729498164,
      "grad_norm": 1.413047487617001,
      "learning_rate": 4.389469316087859e-06,
      "loss": 0.9087,
      "step": 2785
    },
    {
      "epoch": 3.4100367197062424,
      "grad_norm": 0.7412197799281004,
      "learning_rate": 4.389049595228122e-06,
      "loss": 0.5373,
      "step": 2786
    },
    {
      "epoch": 3.4112607099143206,
      "grad_norm": 1.0073505450699676,
      "learning_rate": 4.388629750227035e-06,
      "loss": 0.4061,
      "step": 2787
    },
    {
      "epoch": 3.4124847001223992,
      "grad_norm": 1.0826847292925148,
      "learning_rate": 4.388209781112186e-06,
      "loss": 0.6302,
      "step": 2788
    },
    {
      "epoch": 3.4137086903304774,
      "grad_norm": 1.3320994310139884,
      "learning_rate": 4.387789687911173e-06,
      "loss": 1.0392,
      "step": 2789
    },
    {
      "epoch": 3.4149326805385556,
      "grad_norm": 1.4755602263112608,
      "learning_rate": 4.387369470651605e-06,
      "loss": 1.0751,
      "step": 2790
    },
    {
      "epoch": 3.416156670746634,
      "grad_norm": 0.7806862071470171,
      "learning_rate": 4.386949129361096e-06,
      "loss": 0.6239,
      "step": 2791
    },
    {
      "epoch": 3.4173806609547124,
      "grad_norm": 1.6963746588394495,
      "learning_rate": 4.38652866406727e-06,
      "loss": 0.5771,
      "step": 2792
    },
    {
      "epoch": 3.4186046511627906,
      "grad_norm": 1.4254390704624669,
      "learning_rate": 4.386108074797757e-06,
      "loss": 0.82,
      "step": 2793
    },
    {
      "epoch": 3.419828641370869,
      "grad_norm": 1.6593925469898378,
      "learning_rate": 4.3856873615801985e-06,
      "loss": 0.4679,
      "step": 2794
    },
    {
      "epoch": 3.4210526315789473,
      "grad_norm": 1.0413971224760616,
      "learning_rate": 4.385266524442241e-06,
      "loss": 1.1919,
      "step": 2795
    },
    {
      "epoch": 3.4222766217870255,
      "grad_norm": 0.9444471269865996,
      "learning_rate": 4.38484556341154e-06,
      "loss": 0.5425,
      "step": 2796
    },
    {
      "epoch": 3.423500611995104,
      "grad_norm": 0.8253637812964492,
      "learning_rate": 4.384424478515762e-06,
      "loss": 0.6636,
      "step": 2797
    },
    {
      "epoch": 3.4247246022031823,
      "grad_norm": 0.7165455207021285,
      "learning_rate": 4.384003269782575e-06,
      "loss": 0.4497,
      "step": 2798
    },
    {
      "epoch": 3.4259485924112605,
      "grad_norm": 1.1720862408597537,
      "learning_rate": 4.383581937239664e-06,
      "loss": 0.7829,
      "step": 2799
    },
    {
      "epoch": 3.427172582619339,
      "grad_norm": 1.592090900933828,
      "learning_rate": 4.383160480914713e-06,
      "loss": 0.3032,
      "step": 2800
    },
    {
      "epoch": 3.4283965728274173,
      "grad_norm": 0.8654987344695662,
      "learning_rate": 4.382738900835421e-06,
      "loss": 0.5874,
      "step": 2801
    },
    {
      "epoch": 3.429620563035496,
      "grad_norm": 1.2174178456780231,
      "learning_rate": 4.382317197029492e-06,
      "loss": 0.9788,
      "step": 2802
    },
    {
      "epoch": 3.430844553243574,
      "grad_norm": 1.0504898446920974,
      "learning_rate": 4.3818953695246395e-06,
      "loss": 1.2486,
      "step": 2803
    },
    {
      "epoch": 3.4320685434516522,
      "grad_norm": 1.3915311767448815,
      "learning_rate": 4.381473418348584e-06,
      "loss": 0.5994,
      "step": 2804
    },
    {
      "epoch": 3.433292533659731,
      "grad_norm": 1.5135759233079664,
      "learning_rate": 4.381051343529054e-06,
      "loss": 0.5812,
      "step": 2805
    },
    {
      "epoch": 3.434516523867809,
      "grad_norm": 1.0573431186018905,
      "learning_rate": 4.380629145093788e-06,
      "loss": 0.5489,
      "step": 2806
    },
    {
      "epoch": 3.435740514075887,
      "grad_norm": 1.1865854002744074,
      "learning_rate": 4.38020682307053e-06,
      "loss": 1.0484,
      "step": 2807
    },
    {
      "epoch": 3.436964504283966,
      "grad_norm": 1.0223057180080022,
      "learning_rate": 4.379784377487034e-06,
      "loss": 0.5096,
      "step": 2808
    },
    {
      "epoch": 3.438188494492044,
      "grad_norm": 1.2104581821304563,
      "learning_rate": 4.379361808371062e-06,
      "loss": 0.4524,
      "step": 2809
    },
    {
      "epoch": 3.4394124847001226,
      "grad_norm": 0.9935896933820176,
      "learning_rate": 4.3789391157503825e-06,
      "loss": 0.9077,
      "step": 2810
    },
    {
      "epoch": 3.440636474908201,
      "grad_norm": 0.7183126636169803,
      "learning_rate": 4.378516299652774e-06,
      "loss": 0.6056,
      "step": 2811
    },
    {
      "epoch": 3.441860465116279,
      "grad_norm": 1.2647044804634882,
      "learning_rate": 4.378093360106022e-06,
      "loss": 0.4443,
      "step": 2812
    },
    {
      "epoch": 3.4430844553243576,
      "grad_norm": 1.0727940380884586,
      "learning_rate": 4.377670297137923e-06,
      "loss": 0.4586,
      "step": 2813
    },
    {
      "epoch": 3.4443084455324358,
      "grad_norm": 0.7199060484475379,
      "learning_rate": 4.377247110776276e-06,
      "loss": 0.5686,
      "step": 2814
    },
    {
      "epoch": 3.445532435740514,
      "grad_norm": 1.0813027343577695,
      "learning_rate": 4.376823801048892e-06,
      "loss": 0.7906,
      "step": 2815
    },
    {
      "epoch": 3.4467564259485926,
      "grad_norm": 1.656179107573023,
      "learning_rate": 4.37640036798359e-06,
      "loss": 0.4882,
      "step": 2816
    },
    {
      "epoch": 3.4479804161566707,
      "grad_norm": 1.5862317341144805,
      "learning_rate": 4.375976811608196e-06,
      "loss": 0.7798,
      "step": 2817
    },
    {
      "epoch": 3.449204406364749,
      "grad_norm": 1.6050670819932364,
      "learning_rate": 4.375553131950544e-06,
      "loss": 0.6312,
      "step": 2818
    },
    {
      "epoch": 3.4504283965728275,
      "grad_norm": 1.9131052359791947,
      "learning_rate": 4.375129329038479e-06,
      "loss": 0.5434,
      "step": 2819
    },
    {
      "epoch": 3.4516523867809057,
      "grad_norm": 0.7085877702562351,
      "learning_rate": 4.374705402899849e-06,
      "loss": 0.5753,
      "step": 2820
    },
    {
      "epoch": 3.452876376988984,
      "grad_norm": 0.6630068421818639,
      "learning_rate": 4.374281353562514e-06,
      "loss": 0.5498,
      "step": 2821
    },
    {
      "epoch": 3.4541003671970625,
      "grad_norm": 0.7612810132680108,
      "learning_rate": 4.3738571810543416e-06,
      "loss": 0.6717,
      "step": 2822
    },
    {
      "epoch": 3.4553243574051407,
      "grad_norm": 0.905121133319078,
      "learning_rate": 4.3734328854032056e-06,
      "loss": 0.5816,
      "step": 2823
    },
    {
      "epoch": 3.4565483476132193,
      "grad_norm": 1.9777174846548027,
      "learning_rate": 4.37300846663699e-06,
      "loss": 0.9878,
      "step": 2824
    },
    {
      "epoch": 3.4577723378212974,
      "grad_norm": 1.3729450193411359,
      "learning_rate": 4.372583924783585e-06,
      "loss": 0.4283,
      "step": 2825
    },
    {
      "epoch": 3.4589963280293756,
      "grad_norm": 1.5775100830281101,
      "learning_rate": 4.372159259870892e-06,
      "loss": 0.4758,
      "step": 2826
    },
    {
      "epoch": 3.4602203182374542,
      "grad_norm": 1.155227440651814,
      "learning_rate": 4.371734471926816e-06,
      "loss": 0.9596,
      "step": 2827
    },
    {
      "epoch": 3.4614443084455324,
      "grad_norm": 0.7131379648292329,
      "learning_rate": 4.371309560979274e-06,
      "loss": 0.6291,
      "step": 2828
    },
    {
      "epoch": 3.4626682986536106,
      "grad_norm": 1.4634795383249997,
      "learning_rate": 4.370884527056189e-06,
      "loss": 1.0333,
      "step": 2829
    },
    {
      "epoch": 3.463892288861689,
      "grad_norm": 0.9407337259630113,
      "learning_rate": 4.370459370185492e-06,
      "loss": 0.7173,
      "step": 2830
    },
    {
      "epoch": 3.4651162790697674,
      "grad_norm": 1.6178966026633845,
      "learning_rate": 4.370034090395126e-06,
      "loss": 1.2697,
      "step": 2831
    },
    {
      "epoch": 3.466340269277846,
      "grad_norm": 1.4020711238365124,
      "learning_rate": 4.3696086877130335e-06,
      "loss": 0.3772,
      "step": 2832
    },
    {
      "epoch": 3.467564259485924,
      "grad_norm": 1.501711612900921,
      "learning_rate": 4.369183162167176e-06,
      "loss": 0.6024,
      "step": 2833
    },
    {
      "epoch": 3.4687882496940023,
      "grad_norm": 1.3929354173792095,
      "learning_rate": 4.368757513785512e-06,
      "loss": 1.0104,
      "step": 2834
    },
    {
      "epoch": 3.470012239902081,
      "grad_norm": 1.4588809591730025,
      "learning_rate": 4.368331742596018e-06,
      "loss": 0.649,
      "step": 2835
    },
    {
      "epoch": 3.471236230110159,
      "grad_norm": 1.1816098677361169,
      "learning_rate": 4.367905848626672e-06,
      "loss": 0.5384,
      "step": 2836
    },
    {
      "epoch": 3.4724602203182373,
      "grad_norm": 1.46101435663061,
      "learning_rate": 4.3674798319054625e-06,
      "loss": 0.5585,
      "step": 2837
    },
    {
      "epoch": 3.473684210526316,
      "grad_norm": 1.282884868140762,
      "learning_rate": 4.3670536924603855e-06,
      "loss": 0.4733,
      "step": 2838
    },
    {
      "epoch": 3.474908200734394,
      "grad_norm": 1.5235205881123561,
      "learning_rate": 4.366627430319445e-06,
      "loss": 0.5306,
      "step": 2839
    },
    {
      "epoch": 3.4761321909424723,
      "grad_norm": 0.7208413902971574,
      "learning_rate": 4.366201045510655e-06,
      "loss": 0.5024,
      "step": 2840
    },
    {
      "epoch": 3.477356181150551,
      "grad_norm": 1.5212634830713612,
      "learning_rate": 4.3657745380620345e-06,
      "loss": 0.5166,
      "step": 2841
    },
    {
      "epoch": 3.478580171358629,
      "grad_norm": 1.2630916907284164,
      "learning_rate": 4.365347908001613e-06,
      "loss": 1.0113,
      "step": 2842
    },
    {
      "epoch": 3.4798041615667072,
      "grad_norm": 1.1324915669238926,
      "learning_rate": 4.364921155357425e-06,
      "loss": 0.7533,
      "step": 2843
    },
    {
      "epoch": 3.481028151774786,
      "grad_norm": 0.829650568318722,
      "learning_rate": 4.3644942801575175e-06,
      "loss": 0.5995,
      "step": 2844
    },
    {
      "epoch": 3.482252141982864,
      "grad_norm": 0.8776309133059869,
      "learning_rate": 4.364067282429942e-06,
      "loss": 1.1716,
      "step": 2845
    },
    {
      "epoch": 3.4834761321909427,
      "grad_norm": 1.410099899447338,
      "learning_rate": 4.36364016220276e-06,
      "loss": 0.8629,
      "step": 2846
    },
    {
      "epoch": 3.484700122399021,
      "grad_norm": 0.7220109562801647,
      "learning_rate": 4.36321291950404e-06,
      "loss": 0.7451,
      "step": 2847
    },
    {
      "epoch": 3.485924112607099,
      "grad_norm": 0.8058746598568333,
      "learning_rate": 4.362785554361857e-06,
      "loss": 0.935,
      "step": 2848
    },
    {
      "epoch": 3.4871481028151776,
      "grad_norm": 1.0430948579037274,
      "learning_rate": 4.362358066804298e-06,
      "loss": 0.5357,
      "step": 2849
    },
    {
      "epoch": 3.488372093023256,
      "grad_norm": 1.410091925903012,
      "learning_rate": 4.361930456859455e-06,
      "loss": 0.5335,
      "step": 2850
    },
    {
      "epoch": 3.489596083231334,
      "grad_norm": 1.199296663083217,
      "learning_rate": 4.36150272455543e-06,
      "loss": 1.5411,
      "step": 2851
    },
    {
      "epoch": 3.4908200734394126,
      "grad_norm": 1.9081371690377402,
      "learning_rate": 4.36107486992033e-06,
      "loss": 0.6859,
      "step": 2852
    },
    {
      "epoch": 3.4920440636474908,
      "grad_norm": 1.0936317353552698,
      "learning_rate": 4.360646892982274e-06,
      "loss": 0.607,
      "step": 2853
    },
    {
      "epoch": 3.4932680538555694,
      "grad_norm": 0.9002175031295706,
      "learning_rate": 4.360218793769386e-06,
      "loss": 0.6728,
      "step": 2854
    },
    {
      "epoch": 3.4944920440636476,
      "grad_norm": 1.0280913199904282,
      "learning_rate": 4.359790572309799e-06,
      "loss": 0.4068,
      "step": 2855
    },
    {
      "epoch": 3.4957160342717257,
      "grad_norm": 0.9502640627615291,
      "learning_rate": 4.359362228631655e-06,
      "loss": 0.6132,
      "step": 2856
    },
    {
      "epoch": 3.4969400244798043,
      "grad_norm": 1.1774456243990288,
      "learning_rate": 4.358933762763102e-06,
      "loss": 0.4056,
      "step": 2857
    },
    {
      "epoch": 3.4981640146878825,
      "grad_norm": 1.0385350558629707,
      "learning_rate": 4.358505174732299e-06,
      "loss": 1.2151,
      "step": 2858
    },
    {
      "epoch": 3.4993880048959607,
      "grad_norm": 0.9093833920831976,
      "learning_rate": 4.358076464567409e-06,
      "loss": 0.4071,
      "step": 2859
    },
    {
      "epoch": 3.5006119951040393,
      "grad_norm": 0.7273023568297371,
      "learning_rate": 4.357647632296607e-06,
      "loss": 0.5659,
      "step": 2860
    },
    {
      "epoch": 3.5018359853121175,
      "grad_norm": 1.1432469956374232,
      "learning_rate": 4.357218677948073e-06,
      "loss": 0.6333,
      "step": 2861
    },
    {
      "epoch": 3.5030599755201957,
      "grad_norm": 0.7721303341088325,
      "learning_rate": 4.356789601549997e-06,
      "loss": 0.6548,
      "step": 2862
    },
    {
      "epoch": 3.5042839657282743,
      "grad_norm": 1.86207055388721,
      "learning_rate": 4.356360403130576e-06,
      "loss": 0.4994,
      "step": 2863
    },
    {
      "epoch": 3.5055079559363524,
      "grad_norm": 1.1082190705451813,
      "learning_rate": 4.355931082718016e-06,
      "loss": 1.3791,
      "step": 2864
    },
    {
      "epoch": 3.5067319461444306,
      "grad_norm": 1.4657520895397655,
      "learning_rate": 4.35550164034053e-06,
      "loss": 0.4633,
      "step": 2865
    },
    {
      "epoch": 3.5079559363525092,
      "grad_norm": 1.639356659161861,
      "learning_rate": 4.3550720760263384e-06,
      "loss": 0.7449,
      "step": 2866
    },
    {
      "epoch": 3.5091799265605874,
      "grad_norm": 0.8696338557828689,
      "learning_rate": 4.354642389803672e-06,
      "loss": 0.599,
      "step": 2867
    },
    {
      "epoch": 3.5104039167686656,
      "grad_norm": 1.2320731115894268,
      "learning_rate": 4.354212581700768e-06,
      "loss": 1.5121,
      "step": 2868
    },
    {
      "epoch": 3.511627906976744,
      "grad_norm": 1.0724019879786542,
      "learning_rate": 4.3537826517458705e-06,
      "loss": 2.3286,
      "step": 2869
    },
    {
      "epoch": 3.5128518971848224,
      "grad_norm": 1.424807700263062,
      "learning_rate": 4.353352599967235e-06,
      "loss": 1.257,
      "step": 2870
    },
    {
      "epoch": 3.514075887392901,
      "grad_norm": 1.734695150251853,
      "learning_rate": 4.35292242639312e-06,
      "loss": 0.9386,
      "step": 2871
    },
    {
      "epoch": 3.515299877600979,
      "grad_norm": 0.9493490477884802,
      "learning_rate": 4.3524921310517975e-06,
      "loss": 1.1547,
      "step": 2872
    },
    {
      "epoch": 3.516523867809058,
      "grad_norm": 1.0031853818344216,
      "learning_rate": 4.352061713971544e-06,
      "loss": 0.8201,
      "step": 2873
    },
    {
      "epoch": 3.517747858017136,
      "grad_norm": 0.8133102119444773,
      "learning_rate": 4.351631175180645e-06,
      "loss": 0.7188,
      "step": 2874
    },
    {
      "epoch": 3.518971848225214,
      "grad_norm": 1.333596407323286,
      "learning_rate": 4.3512005147073935e-06,
      "loss": 1.3437,
      "step": 2875
    },
    {
      "epoch": 3.5201958384332928,
      "grad_norm": 0.963409728127715,
      "learning_rate": 4.350769732580092e-06,
      "loss": 1.3209,
      "step": 2876
    },
    {
      "epoch": 3.521419828641371,
      "grad_norm": 0.7953622102156821,
      "learning_rate": 4.350338828827048e-06,
      "loss": 0.6409,
      "step": 2877
    },
    {
      "epoch": 3.522643818849449,
      "grad_norm": 2.0469246064562627,
      "learning_rate": 4.3499078034765815e-06,
      "loss": 0.4424,
      "step": 2878
    },
    {
      "epoch": 3.5238678090575277,
      "grad_norm": 1.1101438627335027,
      "learning_rate": 4.349476656557015e-06,
      "loss": 0.5345,
      "step": 2879
    },
    {
      "epoch": 3.525091799265606,
      "grad_norm": 1.8193174308253999,
      "learning_rate": 4.3490453880966855e-06,
      "loss": 0.9862,
      "step": 2880
    },
    {
      "epoch": 3.526315789473684,
      "grad_norm": 1.034196363468038,
      "learning_rate": 4.348613998123931e-06,
      "loss": 1.2781,
      "step": 2881
    },
    {
      "epoch": 3.5275397796817627,
      "grad_norm": 1.0523965277363985,
      "learning_rate": 4.3481824866671015e-06,
      "loss": 0.604,
      "step": 2882
    },
    {
      "epoch": 3.528763769889841,
      "grad_norm": 1.656567584339346,
      "learning_rate": 4.347750853754556e-06,
      "loss": 0.4674,
      "step": 2883
    },
    {
      "epoch": 3.529987760097919,
      "grad_norm": 1.2063002507104872,
      "learning_rate": 4.347319099414659e-06,
      "loss": 1.6032,
      "step": 2884
    },
    {
      "epoch": 3.5312117503059977,
      "grad_norm": 1.2588489507589293,
      "learning_rate": 4.346887223675782e-06,
      "loss": 0.5918,
      "step": 2885
    },
    {
      "epoch": 3.532435740514076,
      "grad_norm": 1.5753319132551966,
      "learning_rate": 4.346455226566308e-06,
      "loss": 0.5721,
      "step": 2886
    },
    {
      "epoch": 3.533659730722154,
      "grad_norm": 1.437677767836109,
      "learning_rate": 4.346023108114627e-06,
      "loss": 1.138,
      "step": 2887
    },
    {
      "epoch": 3.5348837209302326,
      "grad_norm": 1.6174159902513119,
      "learning_rate": 4.345590868349136e-06,
      "loss": 0.8475,
      "step": 2888
    },
    {
      "epoch": 3.536107711138311,
      "grad_norm": 1.9117625294983323,
      "learning_rate": 4.3451585072982385e-06,
      "loss": 0.7753,
      "step": 2889
    },
    {
      "epoch": 3.537331701346389,
      "grad_norm": 0.920941893558188,
      "learning_rate": 4.344726024990348e-06,
      "loss": 0.5894,
      "step": 2890
    },
    {
      "epoch": 3.5385556915544676,
      "grad_norm": 1.0737232633658507,
      "learning_rate": 4.3442934214538875e-06,
      "loss": 0.7515,
      "step": 2891
    },
    {
      "epoch": 3.5397796817625458,
      "grad_norm": 0.9068759220117667,
      "learning_rate": 4.343860696717284e-06,
      "loss": 1.0679,
      "step": 2892
    },
    {
      "epoch": 3.5410036719706244,
      "grad_norm": 1.1877016108546192,
      "learning_rate": 4.343427850808976e-06,
      "loss": 0.7803,
      "step": 2893
    },
    {
      "epoch": 3.5422276621787026,
      "grad_norm": 0.6994956168436867,
      "learning_rate": 4.342994883757408e-06,
      "loss": 0.6255,
      "step": 2894
    },
    {
      "epoch": 3.543451652386781,
      "grad_norm": 1.0149151345834697,
      "learning_rate": 4.342561795591033e-06,
      "loss": 0.9147,
      "step": 2895
    },
    {
      "epoch": 3.5446756425948593,
      "grad_norm": 1.5934095202750762,
      "learning_rate": 4.342128586338312e-06,
      "loss": 0.5539,
      "step": 2896
    },
    {
      "epoch": 3.5458996328029375,
      "grad_norm": 0.8753602322375681,
      "learning_rate": 4.341695256027714e-06,
      "loss": 0.7496,
      "step": 2897
    },
    {
      "epoch": 3.547123623011016,
      "grad_norm": 1.0438647113518067,
      "learning_rate": 4.3412618046877156e-06,
      "loss": 0.6661,
      "step": 2898
    },
    {
      "epoch": 3.5483476132190943,
      "grad_norm": 0.6356958795916889,
      "learning_rate": 4.3408282323468015e-06,
      "loss": 0.3716,
      "step": 2899
    },
    {
      "epoch": 3.5495716034271725,
      "grad_norm": 1.017960532421912,
      "learning_rate": 4.340394539033464e-06,
      "loss": 0.9629,
      "step": 2900
    },
    {
      "epoch": 3.550795593635251,
      "grad_norm": 1.3879733484722285,
      "learning_rate": 4.339960724776206e-06,
      "loss": 0.6111,
      "step": 2901
    },
    {
      "epoch": 3.5520195838433293,
      "grad_norm": 1.3610687652232283,
      "learning_rate": 4.339526789603534e-06,
      "loss": 0.6696,
      "step": 2902
    },
    {
      "epoch": 3.5532435740514074,
      "grad_norm": 1.4661209391882684,
      "learning_rate": 4.339092733543966e-06,
      "loss": 1.4012,
      "step": 2903
    },
    {
      "epoch": 3.554467564259486,
      "grad_norm": 2.0021521725087568,
      "learning_rate": 4.338658556626026e-06,
      "loss": 0.4631,
      "step": 2904
    },
    {
      "epoch": 3.5556915544675642,
      "grad_norm": 0.6959933730120821,
      "learning_rate": 4.338224258878244e-06,
      "loss": 0.3405,
      "step": 2905
    },
    {
      "epoch": 3.5569155446756424,
      "grad_norm": 0.9386699130500548,
      "learning_rate": 4.337789840329165e-06,
      "loss": 0.7559,
      "step": 2906
    },
    {
      "epoch": 3.558139534883721,
      "grad_norm": 0.902717120160523,
      "learning_rate": 4.337355301007336e-06,
      "loss": 0.8831,
      "step": 2907
    },
    {
      "epoch": 3.559363525091799,
      "grad_norm": 1.0899334819981827,
      "learning_rate": 4.33692064094131e-06,
      "loss": 0.9027,
      "step": 2908
    },
    {
      "epoch": 3.5605875152998774,
      "grad_norm": 1.0636529731664475,
      "learning_rate": 4.336485860159655e-06,
      "loss": 1.2134,
      "step": 2909
    },
    {
      "epoch": 3.561811505507956,
      "grad_norm": 1.4913058830162629,
      "learning_rate": 4.336050958690943e-06,
      "loss": 0.3932,
      "step": 2910
    },
    {
      "epoch": 3.563035495716034,
      "grad_norm": 1.5689633596339874,
      "learning_rate": 4.335615936563752e-06,
      "loss": 0.5574,
      "step": 2911
    },
    {
      "epoch": 3.5642594859241123,
      "grad_norm": 1.0337282029419634,
      "learning_rate": 4.335180793806671e-06,
      "loss": 0.5106,
      "step": 2912
    },
    {
      "epoch": 3.565483476132191,
      "grad_norm": 1.3621954638399791,
      "learning_rate": 4.334745530448296e-06,
      "loss": 0.539,
      "step": 2913
    },
    {
      "epoch": 3.566707466340269,
      "grad_norm": 0.6786718855947556,
      "learning_rate": 4.334310146517231e-06,
      "loss": 0.6264,
      "step": 2914
    },
    {
      "epoch": 3.5679314565483478,
      "grad_norm": 1.548475002732928,
      "learning_rate": 4.3338746420420885e-06,
      "loss": 0.6039,
      "step": 2915
    },
    {
      "epoch": 3.569155446756426,
      "grad_norm": 1.3093038227678264,
      "learning_rate": 4.333439017051487e-06,
      "loss": 0.5912,
      "step": 2916
    },
    {
      "epoch": 3.5703794369645045,
      "grad_norm": 0.8641330009975774,
      "learning_rate": 4.333003271574055e-06,
      "loss": 0.646,
      "step": 2917
    },
    {
      "epoch": 3.5716034271725827,
      "grad_norm": 1.2994748836516428,
      "learning_rate": 4.3325674056384274e-06,
      "loss": 0.7753,
      "step": 2918
    },
    {
      "epoch": 3.572827417380661,
      "grad_norm": 1.3311415003206768,
      "learning_rate": 4.332131419273249e-06,
      "loss": 0.589,
      "step": 2919
    },
    {
      "epoch": 3.5740514075887395,
      "grad_norm": 1.0597928957504616,
      "learning_rate": 4.331695312507169e-06,
      "loss": 1.4441,
      "step": 2920
    },
    {
      "epoch": 3.5752753977968177,
      "grad_norm": 0.7883088941387513,
      "learning_rate": 4.331259085368849e-06,
      "loss": 0.4757,
      "step": 2921
    },
    {
      "epoch": 3.576499388004896,
      "grad_norm": 2.0108915878429463,
      "learning_rate": 4.330822737886955e-06,
      "loss": 0.5491,
      "step": 2922
    },
    {
      "epoch": 3.5777233782129745,
      "grad_norm": 1.4615354812026373,
      "learning_rate": 4.330386270090162e-06,
      "loss": 0.4421,
      "step": 2923
    },
    {
      "epoch": 3.5789473684210527,
      "grad_norm": 1.2520819369621823,
      "learning_rate": 4.329949682007154e-06,
      "loss": 1.4592,
      "step": 2924
    },
    {
      "epoch": 3.580171358629131,
      "grad_norm": 1.7759958964803926,
      "learning_rate": 4.329512973666622e-06,
      "loss": 0.2783,
      "step": 2925
    },
    {
      "epoch": 3.5813953488372094,
      "grad_norm": 1.1180661635330762,
      "learning_rate": 4.329076145097263e-06,
      "loss": 0.6284,
      "step": 2926
    },
    {
      "epoch": 3.5826193390452876,
      "grad_norm": 0.7534276536189869,
      "learning_rate": 4.328639196327786e-06,
      "loss": 0.5784,
      "step": 2927
    },
    {
      "epoch": 3.583843329253366,
      "grad_norm": 0.9606622493202407,
      "learning_rate": 4.3282021273869044e-06,
      "loss": 0.6626,
      "step": 2928
    },
    {
      "epoch": 3.5850673194614444,
      "grad_norm": 1.3251032685439843,
      "learning_rate": 4.32776493830334e-06,
      "loss": 0.6655,
      "step": 2929
    },
    {
      "epoch": 3.5862913096695226,
      "grad_norm": 1.1428734678697354,
      "learning_rate": 4.327327629105826e-06,
      "loss": 0.9592,
      "step": 2930
    },
    {
      "epoch": 3.5875152998776008,
      "grad_norm": 1.517797832297763,
      "learning_rate": 4.326890199823098e-06,
      "loss": 0.9773,
      "step": 2931
    },
    {
      "epoch": 3.5887392900856794,
      "grad_norm": 1.3154842580167365,
      "learning_rate": 4.326452650483904e-06,
      "loss": 0.3502,
      "step": 2932
    },
    {
      "epoch": 3.5899632802937576,
      "grad_norm": 1.7968484425055977,
      "learning_rate": 4.326014981116996e-06,
      "loss": 0.4319,
      "step": 2933
    },
    {
      "epoch": 3.5911872705018357,
      "grad_norm": 1.1833902522404616,
      "learning_rate": 4.325577191751139e-06,
      "loss": 0.5145,
      "step": 2934
    },
    {
      "epoch": 3.5924112607099143,
      "grad_norm": 1.3018697964700894,
      "learning_rate": 4.325139282415101e-06,
      "loss": 0.564,
      "step": 2935
    },
    {
      "epoch": 3.5936352509179925,
      "grad_norm": 1.5766676791766687,
      "learning_rate": 4.3247012531376594e-06,
      "loss": 0.3882,
      "step": 2936
    },
    {
      "epoch": 3.594859241126071,
      "grad_norm": 1.596189690354281,
      "learning_rate": 4.324263103947601e-06,
      "loss": 1.1018,
      "step": 2937
    },
    {
      "epoch": 3.5960832313341493,
      "grad_norm": 1.0226506394508035,
      "learning_rate": 4.323824834873719e-06,
      "loss": 0.6443,
      "step": 2938
    },
    {
      "epoch": 3.597307221542228,
      "grad_norm": 1.1186633339195216,
      "learning_rate": 4.323386445944813e-06,
      "loss": 0.6768,
      "step": 2939
    },
    {
      "epoch": 3.598531211750306,
      "grad_norm": 1.1508859387295962,
      "learning_rate": 4.322947937189697e-06,
      "loss": 0.5746,
      "step": 2940
    },
    {
      "epoch": 3.5997552019583843,
      "grad_norm": 1.7215687685644296,
      "learning_rate": 4.322509308637183e-06,
      "loss": 1.0139,
      "step": 2941
    },
    {
      "epoch": 3.600979192166463,
      "grad_norm": 1.572727987584154,
      "learning_rate": 4.322070560316099e-06,
      "loss": 0.6197,
      "step": 2942
    },
    {
      "epoch": 3.602203182374541,
      "grad_norm": 1.151874912523627,
      "learning_rate": 4.321631692255277e-06,
      "loss": 0.5836,
      "step": 2943
    },
    {
      "epoch": 3.6034271725826192,
      "grad_norm": 1.5710696360548968,
      "learning_rate": 4.321192704483558e-06,
      "loss": 0.6482,
      "step": 2944
    },
    {
      "epoch": 3.604651162790698,
      "grad_norm": 0.8203846404613538,
      "learning_rate": 4.32075359702979e-06,
      "loss": 0.5793,
      "step": 2945
    },
    {
      "epoch": 3.605875152998776,
      "grad_norm": 0.9748441426848996,
      "learning_rate": 4.320314369922831e-06,
      "loss": 0.5789,
      "step": 2946
    },
    {
      "epoch": 3.607099143206854,
      "grad_norm": 1.3043319752950324,
      "learning_rate": 4.319875023191544e-06,
      "loss": 0.5848,
      "step": 2947
    },
    {
      "epoch": 3.608323133414933,
      "grad_norm": 1.5407516399105243,
      "learning_rate": 4.3194355568648015e-06,
      "loss": 0.4382,
      "step": 2948
    },
    {
      "epoch": 3.609547123623011,
      "grad_norm": 1.019025954754023,
      "learning_rate": 4.318995970971485e-06,
      "loss": 0.584,
      "step": 2949
    },
    {
      "epoch": 3.610771113831089,
      "grad_norm": 0.9691019674105982,
      "learning_rate": 4.31855626554048e-06,
      "loss": 0.8539,
      "step": 2950
    },
    {
      "epoch": 3.611995104039168,
      "grad_norm": 1.1101587502151693,
      "learning_rate": 4.318116440600683e-06,
      "loss": 0.3594,
      "step": 2951
    },
    {
      "epoch": 3.613219094247246,
      "grad_norm": 1.2982010373240704,
      "learning_rate": 4.317676496181e-06,
      "loss": 0.5932,
      "step": 2952
    },
    {
      "epoch": 3.614443084455324,
      "grad_norm": 0.817650242496108,
      "learning_rate": 4.317236432310341e-06,
      "loss": 0.7291,
      "step": 2953
    },
    {
      "epoch": 3.6156670746634028,
      "grad_norm": 1.2553094258715891,
      "learning_rate": 4.316796249017624e-06,
      "loss": 0.4662,
      "step": 2954
    },
    {
      "epoch": 3.616891064871481,
      "grad_norm": 0.8613309364649132,
      "learning_rate": 4.316355946331777e-06,
      "loss": 0.6126,
      "step": 2955
    },
    {
      "epoch": 3.618115055079559,
      "grad_norm": 0.898847023514368,
      "learning_rate": 4.315915524281736e-06,
      "loss": 0.5624,
      "step": 2956
    },
    {
      "epoch": 3.6193390452876377,
      "grad_norm": 0.7187855775506882,
      "learning_rate": 4.315474982896444e-06,
      "loss": 0.5673,
      "step": 2957
    },
    {
      "epoch": 3.620563035495716,
      "grad_norm": 1.397815487791166,
      "learning_rate": 4.315034322204851e-06,
      "loss": 1.0909,
      "step": 2958
    },
    {
      "epoch": 3.6217870257037945,
      "grad_norm": 0.7954250262460789,
      "learning_rate": 4.3145935422359154e-06,
      "loss": 0.5987,
      "step": 2959
    },
    {
      "epoch": 3.6230110159118727,
      "grad_norm": 1.6703271693458768,
      "learning_rate": 4.314152643018603e-06,
      "loss": 0.4089,
      "step": 2960
    },
    {
      "epoch": 3.6242350061199513,
      "grad_norm": 0.6712039056710571,
      "learning_rate": 4.31371162458189e-06,
      "loss": 0.5419,
      "step": 2961
    },
    {
      "epoch": 3.6254589963280295,
      "grad_norm": 2.5832252849684276,
      "learning_rate": 4.3132704869547575e-06,
      "loss": 0.4594,
      "step": 2962
    },
    {
      "epoch": 3.6266829865361077,
      "grad_norm": 0.7042627236230449,
      "learning_rate": 4.3128292301661955e-06,
      "loss": 0.5691,
      "step": 2963
    },
    {
      "epoch": 3.6279069767441863,
      "grad_norm": 1.1759148322568098,
      "learning_rate": 4.312387854245201e-06,
      "loss": 0.3473,
      "step": 2964
    },
    {
      "epoch": 3.6291309669522644,
      "grad_norm": 1.3652754673721526,
      "learning_rate": 4.31194635922078e-06,
      "loss": 0.5402,
      "step": 2965
    },
    {
      "epoch": 3.6303549571603426,
      "grad_norm": 1.0374557862926201,
      "learning_rate": 4.311504745121947e-06,
      "loss": 0.5431,
      "step": 2966
    },
    {
      "epoch": 3.6315789473684212,
      "grad_norm": 2.0238596964098923,
      "learning_rate": 4.311063011977723e-06,
      "loss": 0.6005,
      "step": 2967
    },
    {
      "epoch": 3.6328029375764994,
      "grad_norm": 0.970378465526723,
      "learning_rate": 4.310621159817135e-06,
      "loss": 0.3928,
      "step": 2968
    },
    {
      "epoch": 3.6340269277845776,
      "grad_norm": 1.20013594312382,
      "learning_rate": 4.310179188669222e-06,
      "loss": 1.4438,
      "step": 2969
    },
    {
      "epoch": 3.635250917992656,
      "grad_norm": 1.4745663220336436,
      "learning_rate": 4.309737098563029e-06,
      "loss": 0.4874,
      "step": 2970
    },
    {
      "epoch": 3.6364749082007344,
      "grad_norm": 1.2913192453657432,
      "learning_rate": 4.309294889527607e-06,
      "loss": 0.5406,
      "step": 2971
    },
    {
      "epoch": 3.6376988984088126,
      "grad_norm": 1.5984533050230296,
      "learning_rate": 4.308852561592017e-06,
      "loss": 0.5513,
      "step": 2972
    },
    {
      "epoch": 3.638922888616891,
      "grad_norm": 1.416615549400586,
      "learning_rate": 4.308410114785327e-06,
      "loss": 0.5753,
      "step": 2973
    },
    {
      "epoch": 3.6401468788249693,
      "grad_norm": 1.014033274623642,
      "learning_rate": 4.307967549136614e-06,
      "loss": 0.5336,
      "step": 2974
    },
    {
      "epoch": 3.6413708690330475,
      "grad_norm": 0.9574736851251069,
      "learning_rate": 4.307524864674959e-06,
      "loss": 0.7197,
      "step": 2975
    },
    {
      "epoch": 3.642594859241126,
      "grad_norm": 1.8108751206363718,
      "learning_rate": 4.3070820614294566e-06,
      "loss": 0.4868,
      "step": 2976
    },
    {
      "epoch": 3.6438188494492043,
      "grad_norm": 0.8819614192683026,
      "learning_rate": 4.306639139429205e-06,
      "loss": 0.4371,
      "step": 2977
    },
    {
      "epoch": 3.6450428396572825,
      "grad_norm": 1.1299782071341482,
      "learning_rate": 4.306196098703311e-06,
      "loss": 1.6064,
      "step": 2978
    },
    {
      "epoch": 3.646266829865361,
      "grad_norm": 1.5127683379167052,
      "learning_rate": 4.3057529392808905e-06,
      "loss": 0.7458,
      "step": 2979
    },
    {
      "epoch": 3.6474908200734393,
      "grad_norm": 0.9950284280157676,
      "learning_rate": 4.305309661191066e-06,
      "loss": 0.3716,
      "step": 2980
    },
    {
      "epoch": 3.648714810281518,
      "grad_norm": 1.7975230118444248,
      "learning_rate": 4.3048662644629675e-06,
      "loss": 0.5714,
      "step": 2981
    },
    {
      "epoch": 3.649938800489596,
      "grad_norm": 1.576473013604482,
      "learning_rate": 4.3044227491257345e-06,
      "loss": 0.7358,
      "step": 2982
    },
    {
      "epoch": 3.6511627906976747,
      "grad_norm": 1.6678818620435452,
      "learning_rate": 4.303979115208512e-06,
      "loss": 0.533,
      "step": 2983
    },
    {
      "epoch": 3.652386780905753,
      "grad_norm": 1.823383835929657,
      "learning_rate": 4.303535362740455e-06,
      "loss": 0.4796,
      "step": 2984
    },
    {
      "epoch": 3.653610771113831,
      "grad_norm": 1.621817227027596,
      "learning_rate": 4.303091491750724e-06,
      "loss": 0.6182,
      "step": 2985
    },
    {
      "epoch": 3.6548347613219097,
      "grad_norm": 1.0638994777171682,
      "learning_rate": 4.30264750226849e-06,
      "loss": 0.513,
      "step": 2986
    },
    {
      "epoch": 3.656058751529988,
      "grad_norm": 1.3565409743913774,
      "learning_rate": 4.302203394322929e-06,
      "loss": 1.1625,
      "step": 2987
    },
    {
      "epoch": 3.657282741738066,
      "grad_norm": 1.3004276559240062,
      "learning_rate": 4.301759167943228e-06,
      "loss": 1.0599,
      "step": 2988
    },
    {
      "epoch": 3.6585067319461446,
      "grad_norm": 1.4013600412580132,
      "learning_rate": 4.301314823158578e-06,
      "loss": 0.4961,
      "step": 2989
    },
    {
      "epoch": 3.659730722154223,
      "grad_norm": 1.1837034853762836,
      "learning_rate": 4.3008703599981795e-06,
      "loss": 1.5244,
      "step": 2990
    },
    {
      "epoch": 3.660954712362301,
      "grad_norm": 1.028262862475509,
      "learning_rate": 4.300425778491243e-06,
      "loss": 0.8406,
      "step": 2991
    },
    {
      "epoch": 3.6621787025703796,
      "grad_norm": 1.5651221614849686,
      "learning_rate": 4.299981078666984e-06,
      "loss": 0.9601,
      "step": 2992
    },
    {
      "epoch": 3.6634026927784578,
      "grad_norm": 1.4931347475636896,
      "learning_rate": 4.299536260554625e-06,
      "loss": 0.4835,
      "step": 2993
    },
    {
      "epoch": 3.664626682986536,
      "grad_norm": 1.0152398042063975,
      "learning_rate": 4.2990913241834e-06,
      "loss": 0.6631,
      "step": 2994
    },
    {
      "epoch": 3.6658506731946146,
      "grad_norm": 1.4191541057266053,
      "learning_rate": 4.298646269582547e-06,
      "loss": 0.9023,
      "step": 2995
    },
    {
      "epoch": 3.6670746634026927,
      "grad_norm": 1.3938596964146186,
      "learning_rate": 4.298201096781314e-06,
      "loss": 0.6048,
      "step": 2996
    },
    {
      "epoch": 3.668298653610771,
      "grad_norm": 1.5312461218899551,
      "learning_rate": 4.2977558058089565e-06,
      "loss": 0.4301,
      "step": 2997
    },
    {
      "epoch": 3.6695226438188495,
      "grad_norm": 0.8095875774354401,
      "learning_rate": 4.297310396694736e-06,
      "loss": 0.6156,
      "step": 2998
    },
    {
      "epoch": 3.6707466340269277,
      "grad_norm": 0.7539018615659618,
      "learning_rate": 4.2968648694679245e-06,
      "loss": 0.4204,
      "step": 2999
    },
    {
      "epoch": 3.671970624235006,
      "grad_norm": 1.278192535128054,
      "learning_rate": 4.2964192241578e-06,
      "loss": 0.6345,
      "step": 3000
    },
    {
      "epoch": 3.6731946144430845,
      "grad_norm": 1.27482935812848,
      "learning_rate": 4.295973460793649e-06,
      "loss": 0.5755,
      "step": 3001
    },
    {
      "epoch": 3.6744186046511627,
      "grad_norm": 1.1307060576979426,
      "learning_rate": 4.295527579404763e-06,
      "loss": 0.5661,
      "step": 3002
    },
    {
      "epoch": 3.6756425948592413,
      "grad_norm": 2.3173586365816856,
      "learning_rate": 4.2950815800204485e-06,
      "loss": 0.4903,
      "step": 3003
    },
    {
      "epoch": 3.6768665850673194,
      "grad_norm": 1.1565801089195158,
      "learning_rate": 4.29463546267001e-06,
      "loss": 1.5708,
      "step": 3004
    },
    {
      "epoch": 3.678090575275398,
      "grad_norm": 1.0704166461488451,
      "learning_rate": 4.294189227382768e-06,
      "loss": 0.7544,
      "step": 3005
    },
    {
      "epoch": 3.6793145654834762,
      "grad_norm": 2.51146969040047,
      "learning_rate": 4.293742874188045e-06,
      "loss": 0.4671,
      "step": 3006
    },
    {
      "epoch": 3.6805385556915544,
      "grad_norm": 0.8659487264990876,
      "learning_rate": 4.293296403115176e-06,
      "loss": 1.2332,
      "step": 3007
    },
    {
      "epoch": 3.681762545899633,
      "grad_norm": 1.2905523411602837,
      "learning_rate": 4.292849814193501e-06,
      "loss": 0.585,
      "step": 3008
    },
    {
      "epoch": 3.682986536107711,
      "grad_norm": 1.0507644364337232,
      "learning_rate": 4.292403107452366e-06,
      "loss": 0.6214,
      "step": 3009
    },
    {
      "epoch": 3.6842105263157894,
      "grad_norm": 1.5369018449907004,
      "learning_rate": 4.291956282921129e-06,
      "loss": 0.4722,
      "step": 3010
    },
    {
      "epoch": 3.685434516523868,
      "grad_norm": 1.2410666920877502,
      "learning_rate": 4.291509340629152e-06,
      "loss": 0.9595,
      "step": 3011
    },
    {
      "epoch": 3.686658506731946,
      "grad_norm": 1.7307243833399197,
      "learning_rate": 4.291062280605809e-06,
      "loss": 0.5883,
      "step": 3012
    },
    {
      "epoch": 3.6878824969400243,
      "grad_norm": 1.1819155125883472,
      "learning_rate": 4.290615102880477e-06,
      "loss": 1.0576,
      "step": 3013
    },
    {
      "epoch": 3.689106487148103,
      "grad_norm": 0.9915969395053581,
      "learning_rate": 4.2901678074825435e-06,
      "loss": 0.5243,
      "step": 3014
    },
    {
      "epoch": 3.690330477356181,
      "grad_norm": 2.4031169187436237,
      "learning_rate": 4.289720394441403e-06,
      "loss": 0.561,
      "step": 3015
    },
    {
      "epoch": 3.6915544675642593,
      "grad_norm": 0.8862365813641767,
      "learning_rate": 4.289272863786458e-06,
      "loss": 0.5075,
      "step": 3016
    },
    {
      "epoch": 3.692778457772338,
      "grad_norm": 0.8976684364126012,
      "learning_rate": 4.288825215547119e-06,
      "loss": 0.5736,
      "step": 3017
    },
    {
      "epoch": 3.694002447980416,
      "grad_norm": 0.9581946222731119,
      "learning_rate": 4.2883774497528015e-06,
      "loss": 0.4721,
      "step": 3018
    },
    {
      "epoch": 3.6952264381884943,
      "grad_norm": 1.823189190426306,
      "learning_rate": 4.287929566432935e-06,
      "loss": 0.4284,
      "step": 3019
    },
    {
      "epoch": 3.696450428396573,
      "grad_norm": 1.237808719279712,
      "learning_rate": 4.287481565616949e-06,
      "loss": 1.2473,
      "step": 3020
    },
    {
      "epoch": 3.697674418604651,
      "grad_norm": 1.3984379104864662,
      "learning_rate": 4.287033447334286e-06,
      "loss": 0.5109,
      "step": 3021
    },
    {
      "epoch": 3.6988984088127292,
      "grad_norm": 1.4322536349714843,
      "learning_rate": 4.286585211614396e-06,
      "loss": 0.5719,
      "step": 3022
    },
    {
      "epoch": 3.700122399020808,
      "grad_norm": 1.1723091285823055,
      "learning_rate": 4.286136858486732e-06,
      "loss": 1.4499,
      "step": 3023
    },
    {
      "epoch": 3.701346389228886,
      "grad_norm": 1.2395036197760911,
      "learning_rate": 4.285688387980762e-06,
      "loss": 0.5163,
      "step": 3024
    },
    {
      "epoch": 3.7025703794369647,
      "grad_norm": 1.0679567024286427,
      "learning_rate": 4.285239800125955e-06,
      "loss": 0.5721,
      "step": 3025
    },
    {
      "epoch": 3.703794369645043,
      "grad_norm": 0.8451870017028408,
      "learning_rate": 4.2847910949517916e-06,
      "loss": 0.6167,
      "step": 3026
    },
    {
      "epoch": 3.7050183598531214,
      "grad_norm": 1.2013410145122847,
      "learning_rate": 4.284342272487758e-06,
      "loss": 0.5995,
      "step": 3027
    },
    {
      "epoch": 3.7062423500611996,
      "grad_norm": 1.3689459113643188,
      "learning_rate": 4.283893332763352e-06,
      "loss": 0.8086,
      "step": 3028
    },
    {
      "epoch": 3.707466340269278,
      "grad_norm": 1.5005209860051842,
      "learning_rate": 4.283444275808073e-06,
      "loss": 1.0893,
      "step": 3029
    },
    {
      "epoch": 3.7086903304773564,
      "grad_norm": 1.830814939303419,
      "learning_rate": 4.282995101651433e-06,
      "loss": 0.8783,
      "step": 3030
    },
    {
      "epoch": 3.7099143206854346,
      "grad_norm": 1.0221729880332489,
      "learning_rate": 4.282545810322949e-06,
      "loss": 0.6164,
      "step": 3031
    },
    {
      "epoch": 3.7111383108935128,
      "grad_norm": 1.629024776985239,
      "learning_rate": 4.282096401852148e-06,
      "loss": 0.606,
      "step": 3032
    },
    {
      "epoch": 3.7123623011015914,
      "grad_norm": 0.9459133699899019,
      "learning_rate": 4.281646876268564e-06,
      "loss": 0.6673,
      "step": 3033
    },
    {
      "epoch": 3.7135862913096696,
      "grad_norm": 1.0718638885408114,
      "learning_rate": 4.281197233601736e-06,
      "loss": 0.5317,
      "step": 3034
    },
    {
      "epoch": 3.7148102815177477,
      "grad_norm": 2.0284136755809845,
      "learning_rate": 4.280747473881215e-06,
      "loss": 0.5971,
      "step": 3035
    },
    {
      "epoch": 3.7160342717258263,
      "grad_norm": 2.2875890473368616,
      "learning_rate": 4.280297597136556e-06,
      "loss": 0.8276,
      "step": 3036
    },
    {
      "epoch": 3.7172582619339045,
      "grad_norm": 1.2235306956113736,
      "learning_rate": 4.279847603397324e-06,
      "loss": 1.6525,
      "step": 3037
    },
    {
      "epoch": 3.7184822521419827,
      "grad_norm": 1.3195442949404703,
      "learning_rate": 4.27939749269309e-06,
      "loss": 0.9591,
      "step": 3038
    },
    {
      "epoch": 3.7197062423500613,
      "grad_norm": 1.4254870700214906,
      "learning_rate": 4.278947265053435e-06,
      "loss": 0.452,
      "step": 3039
    },
    {
      "epoch": 3.7209302325581395,
      "grad_norm": 1.088934581077748,
      "learning_rate": 4.278496920507945e-06,
      "loss": 0.9724,
      "step": 3040
    },
    {
      "epoch": 3.7221542227662177,
      "grad_norm": 1.129557276145332,
      "learning_rate": 4.278046459086216e-06,
      "loss": 1.3255,
      "step": 3041
    },
    {
      "epoch": 3.7233782129742963,
      "grad_norm": 1.6529284122332322,
      "learning_rate": 4.27759588081785e-06,
      "loss": 0.5384,
      "step": 3042
    },
    {
      "epoch": 3.7246022031823744,
      "grad_norm": 1.0272542785980685,
      "learning_rate": 4.277145185732458e-06,
      "loss": 0.6625,
      "step": 3043
    },
    {
      "epoch": 3.7258261933904526,
      "grad_norm": 0.8592320187266913,
      "learning_rate": 4.276694373859658e-06,
      "loss": 0.8152,
      "step": 3044
    },
    {
      "epoch": 3.7270501835985312,
      "grad_norm": 2.0377395768318363,
      "learning_rate": 4.276243445229074e-06,
      "loss": 0.6592,
      "step": 3045
    },
    {
      "epoch": 3.7282741738066094,
      "grad_norm": 1.0158832393233896,
      "learning_rate": 4.275792399870341e-06,
      "loss": 0.7923,
      "step": 3046
    },
    {
      "epoch": 3.729498164014688,
      "grad_norm": 0.8688083603342553,
      "learning_rate": 4.275341237813101e-06,
      "loss": 0.7977,
      "step": 3047
    },
    {
      "epoch": 3.730722154222766,
      "grad_norm": 0.8619724944432827,
      "learning_rate": 4.274889959087e-06,
      "loss": 1.0084,
      "step": 3048
    },
    {
      "epoch": 3.731946144430845,
      "grad_norm": 1.1469365954927044,
      "learning_rate": 4.274438563721695e-06,
      "loss": 0.9517,
      "step": 3049
    },
    {
      "epoch": 3.733170134638923,
      "grad_norm": 1.5293733074472167,
      "learning_rate": 4.273987051746853e-06,
      "loss": 0.6111,
      "step": 3050
    },
    {
      "epoch": 3.734394124847001,
      "grad_norm": 0.9550143177735964,
      "learning_rate": 4.273535423192142e-06,
      "loss": 0.5328,
      "step": 3051
    },
    {
      "epoch": 3.73561811505508,
      "grad_norm": 0.9252723806560965,
      "learning_rate": 4.273083678087243e-06,
      "loss": 0.798,
      "step": 3052
    },
    {
      "epoch": 3.736842105263158,
      "grad_norm": 1.6647000624629482,
      "learning_rate": 4.272631816461844e-06,
      "loss": 0.6605,
      "step": 3053
    },
    {
      "epoch": 3.738066095471236,
      "grad_norm": 1.3398399532864167,
      "learning_rate": 4.272179838345637e-06,
      "loss": 0.7667,
      "step": 3054
    },
    {
      "epoch": 3.7392900856793148,
      "grad_norm": 1.180749418641135,
      "learning_rate": 4.271727743768327e-06,
      "loss": 1.4093,
      "step": 3055
    },
    {
      "epoch": 3.740514075887393,
      "grad_norm": 1.2640635795032305,
      "learning_rate": 4.271275532759622e-06,
      "loss": 1.06,
      "step": 3056
    },
    {
      "epoch": 3.741738066095471,
      "grad_norm": 1.4668980873682436,
      "learning_rate": 4.27082320534924e-06,
      "loss": 1.1654,
      "step": 3057
    },
    {
      "epoch": 3.7429620563035497,
      "grad_norm": 0.882758695745506,
      "learning_rate": 4.270370761566909e-06,
      "loss": 0.4766,
      "step": 3058
    },
    {
      "epoch": 3.744186046511628,
      "grad_norm": 0.9396474940773939,
      "learning_rate": 4.269918201442358e-06,
      "loss": 0.743,
      "step": 3059
    },
    {
      "epoch": 3.745410036719706,
      "grad_norm": 1.0924411539815455,
      "learning_rate": 4.269465525005329e-06,
      "loss": 0.7783,
      "step": 3060
    },
    {
      "epoch": 3.7466340269277847,
      "grad_norm": 1.1767176894574323,
      "learning_rate": 4.269012732285571e-06,
      "loss": 1.6911,
      "step": 3061
    },
    {
      "epoch": 3.747858017135863,
      "grad_norm": 1.3303880340737395,
      "learning_rate": 4.26855982331284e-06,
      "loss": 1.9366,
      "step": 3062
    },
    {
      "epoch": 3.749082007343941,
      "grad_norm": 1.885643727072353,
      "learning_rate": 4.268106798116898e-06,
      "loss": 0.5759,
      "step": 3063
    },
    {
      "epoch": 3.7503059975520197,
      "grad_norm": 0.7288302125805673,
      "learning_rate": 4.267653656727518e-06,
      "loss": 0.6372,
      "step": 3064
    },
    {
      "epoch": 3.751529987760098,
      "grad_norm": 1.3233072866521771,
      "learning_rate": 4.267200399174477e-06,
      "loss": 0.5376,
      "step": 3065
    },
    {
      "epoch": 3.752753977968176,
      "grad_norm": 1.4556186154693775,
      "learning_rate": 4.266747025487562e-06,
      "loss": 0.4335,
      "step": 3066
    },
    {
      "epoch": 3.7539779681762546,
      "grad_norm": 1.1174895294330383,
      "learning_rate": 4.266293535696567e-06,
      "loss": 0.4233,
      "step": 3067
    },
    {
      "epoch": 3.755201958384333,
      "grad_norm": 1.562050801725391,
      "learning_rate": 4.2658399298312944e-06,
      "loss": 0.5354,
      "step": 3068
    },
    {
      "epoch": 3.7564259485924114,
      "grad_norm": 1.2175490096354604,
      "learning_rate": 4.2653862079215535e-06,
      "loss": 0.5298,
      "step": 3069
    },
    {
      "epoch": 3.7576499388004896,
      "grad_norm": 0.7783272560943433,
      "learning_rate": 4.26493236999716e-06,
      "loss": 0.4769,
      "step": 3070
    },
    {
      "epoch": 3.758873929008568,
      "grad_norm": 0.7801920692318581,
      "learning_rate": 4.264478416087939e-06,
      "loss": 0.6477,
      "step": 3071
    },
    {
      "epoch": 3.7600979192166464,
      "grad_norm": 1.2814011569091337,
      "learning_rate": 4.2640243462237245e-06,
      "loss": 0.5737,
      "step": 3072
    },
    {
      "epoch": 3.7613219094247246,
      "grad_norm": 1.2153143816438317,
      "learning_rate": 4.263570160434353e-06,
      "loss": 0.4961,
      "step": 3073
    },
    {
      "epoch": 3.762545899632803,
      "grad_norm": 1.5431838033877359,
      "learning_rate": 4.263115858749674e-06,
      "loss": 0.3468,
      "step": 3074
    },
    {
      "epoch": 3.7637698898408813,
      "grad_norm": 1.0150682561403608,
      "learning_rate": 4.2626614411995415e-06,
      "loss": 0.6306,
      "step": 3075
    },
    {
      "epoch": 3.7649938800489595,
      "grad_norm": 1.2765706301592676,
      "learning_rate": 4.262206907813819e-06,
      "loss": 0.6041,
      "step": 3076
    },
    {
      "epoch": 3.766217870257038,
      "grad_norm": 1.0389265367966636,
      "learning_rate": 4.261752258622377e-06,
      "loss": 1.0424,
      "step": 3077
    },
    {
      "epoch": 3.7674418604651163,
      "grad_norm": 1.361885743217014,
      "learning_rate": 4.261297493655092e-06,
      "loss": 0.6654,
      "step": 3078
    },
    {
      "epoch": 3.7686658506731945,
      "grad_norm": 1.4222532538635666,
      "learning_rate": 4.2608426129418514e-06,
      "loss": 0.7221,
      "step": 3079
    },
    {
      "epoch": 3.769889840881273,
      "grad_norm": 1.4784488106534912,
      "learning_rate": 4.260387616512547e-06,
      "loss": 0.6095,
      "step": 3080
    },
    {
      "epoch": 3.7711138310893513,
      "grad_norm": 1.0970894025974804,
      "learning_rate": 4.259932504397079e-06,
      "loss": 0.604,
      "step": 3081
    },
    {
      "epoch": 3.7723378212974294,
      "grad_norm": 2.214317374279887,
      "learning_rate": 4.259477276625357e-06,
      "loss": 0.5164,
      "step": 3082
    },
    {
      "epoch": 3.773561811505508,
      "grad_norm": 1.0525066742799187,
      "learning_rate": 4.259021933227295e-06,
      "loss": 0.572,
      "step": 3083
    },
    {
      "epoch": 3.7747858017135862,
      "grad_norm": 0.963528684407431,
      "learning_rate": 4.258566474232819e-06,
      "loss": 0.3119,
      "step": 3084
    },
    {
      "epoch": 3.7760097919216644,
      "grad_norm": 0.964290830606418,
      "learning_rate": 4.258110899671859e-06,
      "loss": 0.9058,
      "step": 3085
    },
    {
      "epoch": 3.777233782129743,
      "grad_norm": 0.761026398621415,
      "learning_rate": 4.257655209574352e-06,
      "loss": 0.5298,
      "step": 3086
    },
    {
      "epoch": 3.778457772337821,
      "grad_norm": 1.3548740377736872,
      "learning_rate": 4.257199403970247e-06,
      "loss": 1.0107,
      "step": 3087
    },
    {
      "epoch": 3.7796817625458994,
      "grad_norm": 1.5326219180775327,
      "learning_rate": 4.256743482889496e-06,
      "loss": 0.5027,
      "step": 3088
    },
    {
      "epoch": 3.780905752753978,
      "grad_norm": 1.7836578886929766,
      "learning_rate": 4.2562874463620615e-06,
      "loss": 1.0515,
      "step": 3089
    },
    {
      "epoch": 3.782129742962056,
      "grad_norm": 1.1636265243329595,
      "learning_rate": 4.255831294417912e-06,
      "loss": 0.9001,
      "step": 3090
    },
    {
      "epoch": 3.783353733170135,
      "grad_norm": 1.083848110618823,
      "learning_rate": 4.255375027087023e-06,
      "loss": 0.9743,
      "step": 3091
    },
    {
      "epoch": 3.784577723378213,
      "grad_norm": 1.1119584675723793,
      "learning_rate": 4.254918644399382e-06,
      "loss": 0.4747,
      "step": 3092
    },
    {
      "epoch": 3.7858017135862916,
      "grad_norm": 0.9505385868683223,
      "learning_rate": 4.2544621463849775e-06,
      "loss": 0.911,
      "step": 3093
    },
    {
      "epoch": 3.7870257037943698,
      "grad_norm": 0.7841506511896073,
      "learning_rate": 4.25400553307381e-06,
      "loss": 0.4816,
      "step": 3094
    },
    {
      "epoch": 3.788249694002448,
      "grad_norm": 0.8677801736542136,
      "learning_rate": 4.253548804495887e-06,
      "loss": 0.8706,
      "step": 3095
    },
    {
      "epoch": 3.7894736842105265,
      "grad_norm": 1.020093948489314,
      "learning_rate": 4.253091960681222e-06,
      "loss": 0.7422,
      "step": 3096
    },
    {
      "epoch": 3.7906976744186047,
      "grad_norm": 1.7951254514316166,
      "learning_rate": 4.252635001659838e-06,
      "loss": 0.4261,
      "step": 3097
    },
    {
      "epoch": 3.791921664626683,
      "grad_norm": 1.3405092704348531,
      "learning_rate": 4.252177927461763e-06,
      "loss": 0.4152,
      "step": 3098
    },
    {
      "epoch": 3.7931456548347615,
      "grad_norm": 1.829914946716124,
      "learning_rate": 4.251720738117037e-06,
      "loss": 0.4441,
      "step": 3099
    },
    {
      "epoch": 3.7943696450428397,
      "grad_norm": 1.0538893061229253,
      "learning_rate": 4.251263433655703e-06,
      "loss": 0.4538,
      "step": 3100
    },
    {
      "epoch": 3.795593635250918,
      "grad_norm": 1.7527708046737387,
      "learning_rate": 4.250806014107814e-06,
      "loss": 0.4698,
      "step": 3101
    },
    {
      "epoch": 3.7968176254589965,
      "grad_norm": 0.9086131355017543,
      "learning_rate": 4.250348479503428e-06,
      "loss": 0.4044,
      "step": 3102
    },
    {
      "epoch": 3.7980416156670747,
      "grad_norm": 1.177181158441611,
      "learning_rate": 4.249890829872615e-06,
      "loss": 1.034,
      "step": 3103
    },
    {
      "epoch": 3.799265605875153,
      "grad_norm": 1.7735560975448976,
      "learning_rate": 4.2494330652454485e-06,
      "loss": 0.5568,
      "step": 3104
    },
    {
      "epoch": 3.8004895960832314,
      "grad_norm": 1.2578239314121873,
      "learning_rate": 4.2489751856520125e-06,
      "loss": 0.5821,
      "step": 3105
    },
    {
      "epoch": 3.8017135862913096,
      "grad_norm": 0.9602259097736453,
      "learning_rate": 4.2485171911223965e-06,
      "loss": 0.7326,
      "step": 3106
    },
    {
      "epoch": 3.802937576499388,
      "grad_norm": 1.1038696015863398,
      "learning_rate": 4.2480590816866965e-06,
      "loss": 0.6491,
      "step": 3107
    },
    {
      "epoch": 3.8041615667074664,
      "grad_norm": 1.3982127460680442,
      "learning_rate": 4.24760085737502e-06,
      "loss": 0.7038,
      "step": 3108
    },
    {
      "epoch": 3.8053855569155446,
      "grad_norm": 1.6506495824943868,
      "learning_rate": 4.24714251821748e-06,
      "loss": 0.9469,
      "step": 3109
    },
    {
      "epoch": 3.8066095471236228,
      "grad_norm": 1.1975723789433264,
      "learning_rate": 4.2466840642441955e-06,
      "loss": 1.2197,
      "step": 3110
    },
    {
      "epoch": 3.8078335373317014,
      "grad_norm": 0.753948318331713,
      "learning_rate": 4.246225495485295e-06,
      "loss": 0.674,
      "step": 3111
    },
    {
      "epoch": 3.8090575275397796,
      "grad_norm": 1.3556576255140418,
      "learning_rate": 4.2457668119709124e-06,
      "loss": 0.6503,
      "step": 3112
    },
    {
      "epoch": 3.810281517747858,
      "grad_norm": 1.0101317078638816,
      "learning_rate": 4.245308013731193e-06,
      "loss": 0.6075,
      "step": 3113
    },
    {
      "epoch": 3.8115055079559363,
      "grad_norm": 1.908787870212063,
      "learning_rate": 4.244849100796287e-06,
      "loss": 1.1205,
      "step": 3114
    },
    {
      "epoch": 3.812729498164015,
      "grad_norm": 1.7264420296590912,
      "learning_rate": 4.24439007319635e-06,
      "loss": 0.5513,
      "step": 3115
    },
    {
      "epoch": 3.813953488372093,
      "grad_norm": 0.9618130130738143,
      "learning_rate": 4.243930930961551e-06,
      "loss": 0.6892,
      "step": 3116
    },
    {
      "epoch": 3.8151774785801713,
      "grad_norm": 1.358823937361697,
      "learning_rate": 4.243471674122061e-06,
      "loss": 1.1077,
      "step": 3117
    },
    {
      "epoch": 3.81640146878825,
      "grad_norm": 1.5417259239758323,
      "learning_rate": 4.243012302708063e-06,
      "loss": 0.5445,
      "step": 3118
    },
    {
      "epoch": 3.817625458996328,
      "grad_norm": 1.4080700278047578,
      "learning_rate": 4.242552816749741e-06,
      "loss": 1.1848,
      "step": 3119
    },
    {
      "epoch": 3.8188494492044063,
      "grad_norm": 1.2068833936777243,
      "learning_rate": 4.2420932162772936e-06,
      "loss": 0.5814,
      "step": 3120
    },
    {
      "epoch": 3.820073439412485,
      "grad_norm": 0.7787808034169793,
      "learning_rate": 4.241633501320924e-06,
      "loss": 0.5172,
      "step": 3121
    },
    {
      "epoch": 3.821297429620563,
      "grad_norm": 1.2689828702160677,
      "learning_rate": 4.241173671910843e-06,
      "loss": 0.53,
      "step": 3122
    },
    {
      "epoch": 3.8225214198286412,
      "grad_norm": 1.2070129118367148,
      "learning_rate": 4.240713728077268e-06,
      "loss": 0.5065,
      "step": 3123
    },
    {
      "epoch": 3.82374541003672,
      "grad_norm": 1.4371564958289553,
      "learning_rate": 4.240253669850425e-06,
      "loss": 1.1711,
      "step": 3124
    },
    {
      "epoch": 3.824969400244798,
      "grad_norm": 1.1051012433317404,
      "learning_rate": 4.239793497260548e-06,
      "loss": 0.6964,
      "step": 3125
    },
    {
      "epoch": 3.826193390452876,
      "grad_norm": 1.3210718792508302,
      "learning_rate": 4.239333210337877e-06,
      "loss": 0.7952,
      "step": 3126
    },
    {
      "epoch": 3.827417380660955,
      "grad_norm": 0.5405088512397118,
      "learning_rate": 4.2388728091126614e-06,
      "loss": 0.373,
      "step": 3127
    },
    {
      "epoch": 3.828641370869033,
      "grad_norm": 0.717990764227263,
      "learning_rate": 4.238412293615156e-06,
      "loss": 0.3954,
      "step": 3128
    },
    {
      "epoch": 3.829865361077111,
      "grad_norm": 1.567787365938041,
      "learning_rate": 4.237951663875625e-06,
      "loss": 1.4716,
      "step": 3129
    },
    {
      "epoch": 3.83108935128519,
      "grad_norm": 1.5672471837473472,
      "learning_rate": 4.237490919924338e-06,
      "loss": 0.4631,
      "step": 3130
    },
    {
      "epoch": 3.832313341493268,
      "grad_norm": 2.3391346785965244,
      "learning_rate": 4.237030061791575e-06,
      "loss": 0.5777,
      "step": 3131
    },
    {
      "epoch": 3.833537331701346,
      "grad_norm": 0.3845916090422017,
      "learning_rate": 4.23656908950762e-06,
      "loss": 0.1368,
      "step": 3132
    },
    {
      "epoch": 3.8347613219094248,
      "grad_norm": 1.2990141857626685,
      "learning_rate": 4.236108003102769e-06,
      "loss": 0.4642,
      "step": 3133
    },
    {
      "epoch": 3.835985312117503,
      "grad_norm": 1.3514626104621354,
      "learning_rate": 4.235646802607322e-06,
      "loss": 0.814,
      "step": 3134
    },
    {
      "epoch": 3.8372093023255816,
      "grad_norm": 0.8802330104009451,
      "learning_rate": 4.2351854880515856e-06,
      "loss": 0.7873,
      "step": 3135
    },
    {
      "epoch": 3.8384332925336597,
      "grad_norm": 1.1750498347401908,
      "learning_rate": 4.234724059465877e-06,
      "loss": 1.4327,
      "step": 3136
    },
    {
      "epoch": 3.8396572827417383,
      "grad_norm": 0.8456302170261832,
      "learning_rate": 4.23426251688052e-06,
      "loss": 0.4248,
      "step": 3137
    },
    {
      "epoch": 3.8408812729498165,
      "grad_norm": 0.9157836093366463,
      "learning_rate": 4.2338008603258444e-06,
      "loss": 0.5951,
      "step": 3138
    },
    {
      "epoch": 3.8421052631578947,
      "grad_norm": 0.7475062672582033,
      "learning_rate": 4.233339089832189e-06,
      "loss": 0.6687,
      "step": 3139
    },
    {
      "epoch": 3.8433292533659733,
      "grad_norm": 1.478957017449317,
      "learning_rate": 4.2328772054299005e-06,
      "loss": 1.3669,
      "step": 3140
    },
    {
      "epoch": 3.8445532435740515,
      "grad_norm": 1.1331979927799298,
      "learning_rate": 4.23241520714933e-06,
      "loss": 1.662,
      "step": 3141
    },
    {
      "epoch": 3.8457772337821297,
      "grad_norm": 1.6812337643420379,
      "learning_rate": 4.231953095020842e-06,
      "loss": 0.5876,
      "step": 3142
    },
    {
      "epoch": 3.8470012239902083,
      "grad_norm": 1.0211685297716826,
      "learning_rate": 4.231490869074802e-06,
      "loss": 0.4382,
      "step": 3143
    },
    {
      "epoch": 3.8482252141982864,
      "grad_norm": 1.5702306695523027,
      "learning_rate": 4.231028529341586e-06,
      "loss": 0.5609,
      "step": 3144
    },
    {
      "epoch": 3.8494492044063646,
      "grad_norm": 1.2739911686144465,
      "learning_rate": 4.230566075851578e-06,
      "loss": 0.5376,
      "step": 3145
    },
    {
      "epoch": 3.8506731946144432,
      "grad_norm": 1.3354310663560005,
      "learning_rate": 4.230103508635168e-06,
      "loss": 0.502,
      "step": 3146
    },
    {
      "epoch": 3.8518971848225214,
      "grad_norm": 1.3241972018157122,
      "learning_rate": 4.229640827722755e-06,
      "loss": 1.2592,
      "step": 3147
    },
    {
      "epoch": 3.8531211750305996,
      "grad_norm": 1.1432563835963918,
      "learning_rate": 4.229178033144744e-06,
      "loss": 0.4334,
      "step": 3148
    },
    {
      "epoch": 3.854345165238678,
      "grad_norm": 1.1284918225019118,
      "learning_rate": 4.228715124931548e-06,
      "loss": 0.8603,
      "step": 3149
    },
    {
      "epoch": 3.8555691554467564,
      "grad_norm": 2.0180471952193333,
      "learning_rate": 4.228252103113589e-06,
      "loss": 0.3848,
      "step": 3150
    },
    {
      "epoch": 3.8567931456548346,
      "grad_norm": 1.0328969637824374,
      "learning_rate": 4.227788967721294e-06,
      "loss": 0.8353,
      "step": 3151
    },
    {
      "epoch": 3.858017135862913,
      "grad_norm": 1.0923580492371523,
      "learning_rate": 4.227325718785098e-06,
      "loss": 0.737,
      "step": 3152
    },
    {
      "epoch": 3.8592411260709913,
      "grad_norm": 1.105930994172851,
      "learning_rate": 4.226862356335446e-06,
      "loss": 0.514,
      "step": 3153
    },
    {
      "epoch": 3.8604651162790695,
      "grad_norm": 0.976452990347767,
      "learning_rate": 4.226398880402786e-06,
      "loss": 0.6547,
      "step": 3154
    },
    {
      "epoch": 3.861689106487148,
      "grad_norm": 1.5399346960831135,
      "learning_rate": 4.2259352910175775e-06,
      "loss": 0.5066,
      "step": 3155
    },
    {
      "epoch": 3.8629130966952263,
      "grad_norm": 1.1738658945334446,
      "learning_rate": 4.2254715882102856e-06,
      "loss": 0.5902,
      "step": 3156
    },
    {
      "epoch": 3.864137086903305,
      "grad_norm": 1.367576524203623,
      "learning_rate": 4.2250077720113825e-06,
      "loss": 1.0509,
      "step": 3157
    },
    {
      "epoch": 3.865361077111383,
      "grad_norm": 1.119792873902574,
      "learning_rate": 4.2245438424513495e-06,
      "loss": 0.7073,
      "step": 3158
    },
    {
      "epoch": 3.8665850673194617,
      "grad_norm": 2.213799256156493,
      "learning_rate": 4.224079799560674e-06,
      "loss": 0.4268,
      "step": 3159
    },
    {
      "epoch": 3.86780905752754,
      "grad_norm": 1.2629838174437327,
      "learning_rate": 4.223615643369851e-06,
      "loss": 0.7086,
      "step": 3160
    },
    {
      "epoch": 3.869033047735618,
      "grad_norm": 0.9821557084779523,
      "learning_rate": 4.223151373909382e-06,
      "loss": 0.5463,
      "step": 3161
    },
    {
      "epoch": 3.8702570379436967,
      "grad_norm": 1.7348009311485701,
      "learning_rate": 4.222686991209779e-06,
      "loss": 0.4392,
      "step": 3162
    },
    {
      "epoch": 3.871481028151775,
      "grad_norm": 1.514410227098721,
      "learning_rate": 4.222222495301559e-06,
      "loss": 0.6918,
      "step": 3163
    },
    {
      "epoch": 3.872705018359853,
      "grad_norm": 0.9509436111495124,
      "learning_rate": 4.221757886215246e-06,
      "loss": 1.1195,
      "step": 3164
    },
    {
      "epoch": 3.8739290085679317,
      "grad_norm": 1.340753333376802,
      "learning_rate": 4.221293163981374e-06,
      "loss": 0.6885,
      "step": 3165
    },
    {
      "epoch": 3.87515299877601,
      "grad_norm": 1.6744785630659382,
      "learning_rate": 4.220828328630481e-06,
      "loss": 0.6695,
      "step": 3166
    },
    {
      "epoch": 3.876376988984088,
      "grad_norm": 1.280906820337949,
      "learning_rate": 4.220363380193115e-06,
      "loss": 0.4909,
      "step": 3167
    },
    {
      "epoch": 3.8776009791921666,
      "grad_norm": 1.407097955705677,
      "learning_rate": 4.219898318699831e-06,
      "loss": 0.4076,
      "step": 3168
    },
    {
      "epoch": 3.878824969400245,
      "grad_norm": 1.431948580842852,
      "learning_rate": 4.219433144181192e-06,
      "loss": 0.5589,
      "step": 3169
    },
    {
      "epoch": 3.880048959608323,
      "grad_norm": 1.1979153957375552,
      "learning_rate": 4.218967856667765e-06,
      "loss": 0.5197,
      "step": 3170
    },
    {
      "epoch": 3.8812729498164016,
      "grad_norm": 1.9553529273094514,
      "learning_rate": 4.21850245619013e-06,
      "loss": 0.4459,
      "step": 3171
    },
    {
      "epoch": 3.8824969400244798,
      "grad_norm": 0.5355398571508138,
      "learning_rate": 4.218036942778869e-06,
      "loss": 0.2704,
      "step": 3172
    },
    {
      "epoch": 3.883720930232558,
      "grad_norm": 1.1589045499130541,
      "learning_rate": 4.2175713164645745e-06,
      "loss": 0.6624,
      "step": 3173
    },
    {
      "epoch": 3.8849449204406366,
      "grad_norm": 1.2620656594994981,
      "learning_rate": 4.217105577277846e-06,
      "loss": 1.2049,
      "step": 3174
    },
    {
      "epoch": 3.8861689106487147,
      "grad_norm": 0.7687579689758555,
      "learning_rate": 4.216639725249291e-06,
      "loss": 0.6533,
      "step": 3175
    },
    {
      "epoch": 3.887392900856793,
      "grad_norm": 0.661068762510923,
      "learning_rate": 4.216173760409522e-06,
      "loss": 0.448,
      "step": 3176
    },
    {
      "epoch": 3.8886168910648715,
      "grad_norm": 0.6665320758772337,
      "learning_rate": 4.215707682789161e-06,
      "loss": 0.4667,
      "step": 3177
    },
    {
      "epoch": 3.8898408812729497,
      "grad_norm": 0.9241958251974215,
      "learning_rate": 4.215241492418837e-06,
      "loss": 0.474,
      "step": 3178
    },
    {
      "epoch": 3.8910648714810283,
      "grad_norm": 0.9987072549469569,
      "learning_rate": 4.214775189329186e-06,
      "loss": 0.8876,
      "step": 3179
    },
    {
      "epoch": 3.8922888616891065,
      "grad_norm": 0.9681594837587075,
      "learning_rate": 4.2143087735508535e-06,
      "loss": 0.8065,
      "step": 3180
    },
    {
      "epoch": 3.8935128518971847,
      "grad_norm": 0.898471855109452,
      "learning_rate": 4.2138422451144876e-06,
      "loss": 0.6737,
      "step": 3181
    },
    {
      "epoch": 3.8947368421052633,
      "grad_norm": 0.8988801462161548,
      "learning_rate": 4.21337560405075e-06,
      "loss": 0.4065,
      "step": 3182
    },
    {
      "epoch": 3.8959608323133414,
      "grad_norm": 1.807199486234844,
      "learning_rate": 4.212908850390304e-06,
      "loss": 1.3428,
      "step": 3183
    },
    {
      "epoch": 3.89718482252142,
      "grad_norm": 1.2205371714383537,
      "learning_rate": 4.212441984163824e-06,
      "loss": 1.0562,
      "step": 3184
    },
    {
      "epoch": 3.8984088127294982,
      "grad_norm": 0.7926346251388762,
      "learning_rate": 4.21197500540199e-06,
      "loss": 0.5134,
      "step": 3185
    },
    {
      "epoch": 3.8996328029375764,
      "grad_norm": 1.3974985082184015,
      "learning_rate": 4.211507914135492e-06,
      "loss": 0.7404,
      "step": 3186
    },
    {
      "epoch": 3.900856793145655,
      "grad_norm": 1.1849694837272244,
      "learning_rate": 4.2110407103950245e-06,
      "loss": 0.635,
      "step": 3187
    },
    {
      "epoch": 3.902080783353733,
      "grad_norm": 1.6772296568516902,
      "learning_rate": 4.21057339421129e-06,
      "loss": 1.0827,
      "step": 3188
    },
    {
      "epoch": 3.9033047735618114,
      "grad_norm": 1.195017656786451,
      "learning_rate": 4.210105965614998e-06,
      "loss": 1.14,
      "step": 3189
    },
    {
      "epoch": 3.90452876376989,
      "grad_norm": 1.1356266190465811,
      "learning_rate": 4.209638424636868e-06,
      "loss": 1.5805,
      "step": 3190
    },
    {
      "epoch": 3.905752753977968,
      "grad_norm": 1.9962716068031943,
      "learning_rate": 4.209170771307625e-06,
      "loss": 0.7587,
      "step": 3191
    },
    {
      "epoch": 3.9069767441860463,
      "grad_norm": 0.7672433662199944,
      "learning_rate": 4.208703005658e-06,
      "loss": 0.5909,
      "step": 3192
    },
    {
      "epoch": 3.908200734394125,
      "grad_norm": 1.0784443485965085,
      "learning_rate": 4.208235127718732e-06,
      "loss": 1.0902,
      "step": 3193
    },
    {
      "epoch": 3.909424724602203,
      "grad_norm": 1.3446926217017363,
      "learning_rate": 4.2077671375205715e-06,
      "loss": 0.7851,
      "step": 3194
    },
    {
      "epoch": 3.9106487148102813,
      "grad_norm": 1.4773187909919343,
      "learning_rate": 4.207299035094271e-06,
      "loss": 0.4784,
      "step": 3195
    },
    {
      "epoch": 3.91187270501836,
      "grad_norm": 1.1834761266468163,
      "learning_rate": 4.206830820470592e-06,
      "loss": 1.407,
      "step": 3196
    },
    {
      "epoch": 3.913096695226438,
      "grad_norm": 1.2889579712977055,
      "learning_rate": 4.206362493680306e-06,
      "loss": 1.2189,
      "step": 3197
    },
    {
      "epoch": 3.9143206854345163,
      "grad_norm": 1.3823076304267339,
      "learning_rate": 4.205894054754187e-06,
      "loss": 0.6096,
      "step": 3198
    },
    {
      "epoch": 3.915544675642595,
      "grad_norm": 1.3388542864008641,
      "learning_rate": 4.205425503723022e-06,
      "loss": 0.623,
      "step": 3199
    },
    {
      "epoch": 3.916768665850673,
      "grad_norm": 1.9933160584058467,
      "learning_rate": 4.2049568406175985e-06,
      "loss": 0.3982,
      "step": 3200
    },
    {
      "epoch": 3.9179926560587517,
      "grad_norm": 1.4292677960063531,
      "learning_rate": 4.20448806546872e-06,
      "loss": 0.6332,
      "step": 3201
    },
    {
      "epoch": 3.91921664626683,
      "grad_norm": 1.2451508826608653,
      "learning_rate": 4.204019178307188e-06,
      "loss": 1.1307,
      "step": 3202
    },
    {
      "epoch": 3.920440636474908,
      "grad_norm": 0.7806678914790082,
      "learning_rate": 4.20355017916382e-06,
      "loss": 0.6127,
      "step": 3203
    },
    {
      "epoch": 3.9216646266829867,
      "grad_norm": 1.1209116953406848,
      "learning_rate": 4.203081068069435e-06,
      "loss": 0.532,
      "step": 3204
    },
    {
      "epoch": 3.922888616891065,
      "grad_norm": 1.9833456060015826,
      "learning_rate": 4.202611845054861e-06,
      "loss": 0.6053,
      "step": 3205
    },
    {
      "epoch": 3.9241126070991434,
      "grad_norm": 1.3728766338253626,
      "learning_rate": 4.202142510150935e-06,
      "loss": 1.0235,
      "step": 3206
    },
    {
      "epoch": 3.9253365973072216,
      "grad_norm": 1.2202641772600984,
      "learning_rate": 4.201673063388498e-06,
      "loss": 0.4853,
      "step": 3207
    },
    {
      "epoch": 3.9265605875153,
      "grad_norm": 1.454628686766218,
      "learning_rate": 4.201203504798402e-06,
      "loss": 0.4859,
      "step": 3208
    },
    {
      "epoch": 3.9277845777233784,
      "grad_norm": 1.2449189979011324,
      "learning_rate": 4.200733834411503e-06,
      "loss": 0.5417,
      "step": 3209
    },
    {
      "epoch": 3.9290085679314566,
      "grad_norm": 0.9560082834304796,
      "learning_rate": 4.200264052258668e-06,
      "loss": 0.5944,
      "step": 3210
    },
    {
      "epoch": 3.9302325581395348,
      "grad_norm": 1.20536919265455,
      "learning_rate": 4.199794158370768e-06,
      "loss": 0.5717,
      "step": 3211
    },
    {
      "epoch": 3.9314565483476134,
      "grad_norm": 0.8158952700425665,
      "learning_rate": 4.1993241527786834e-06,
      "loss": 0.4123,
      "step": 3212
    },
    {
      "epoch": 3.9326805385556916,
      "grad_norm": 1.7207086748333111,
      "learning_rate": 4.1988540355133e-06,
      "loss": 0.3432,
      "step": 3213
    },
    {
      "epoch": 3.9339045287637697,
      "grad_norm": 0.988334551635925,
      "learning_rate": 4.198383806605514e-06,
      "loss": 0.6539,
      "step": 3214
    },
    {
      "epoch": 3.9351285189718483,
      "grad_norm": 1.856504578698825,
      "learning_rate": 4.197913466086224e-06,
      "loss": 0.5198,
      "step": 3215
    },
    {
      "epoch": 3.9363525091799265,
      "grad_norm": 0.8562463897578837,
      "learning_rate": 4.1974430139863435e-06,
      "loss": 0.6885,
      "step": 3216
    },
    {
      "epoch": 3.9375764993880047,
      "grad_norm": 1.1486449518367934,
      "learning_rate": 4.196972450336785e-06,
      "loss": 0.5681,
      "step": 3217
    },
    {
      "epoch": 3.9388004895960833,
      "grad_norm": 1.4599689079155813,
      "learning_rate": 4.196501775168475e-06,
      "loss": 0.886,
      "step": 3218
    },
    {
      "epoch": 3.9400244798041615,
      "grad_norm": 1.6047891731848427,
      "learning_rate": 4.196030988512343e-06,
      "loss": 0.8781,
      "step": 3219
    },
    {
      "epoch": 3.9412484700122397,
      "grad_norm": 1.1762603177850568,
      "learning_rate": 4.195560090399327e-06,
      "loss": 1.5444,
      "step": 3220
    },
    {
      "epoch": 3.9424724602203183,
      "grad_norm": 1.1126784995863812,
      "learning_rate": 4.195089080860374e-06,
      "loss": 0.4836,
      "step": 3221
    },
    {
      "epoch": 3.9436964504283964,
      "grad_norm": 1.5700165386306952,
      "learning_rate": 4.194617959926436e-06,
      "loss": 1.1205,
      "step": 3222
    },
    {
      "epoch": 3.944920440636475,
      "grad_norm": 0.9632983215117739,
      "learning_rate": 4.194146727628473e-06,
      "loss": 0.6148,
      "step": 3223
    },
    {
      "epoch": 3.9461444308445532,
      "grad_norm": 1.3082754592729957,
      "learning_rate": 4.1936753839974545e-06,
      "loss": 0.3689,
      "step": 3224
    },
    {
      "epoch": 3.9473684210526314,
      "grad_norm": 1.1079029818083657,
      "learning_rate": 4.1932039290643534e-06,
      "loss": 0.4782,
      "step": 3225
    },
    {
      "epoch": 3.94859241126071,
      "grad_norm": 1.0854158021281397,
      "learning_rate": 4.192732362860153e-06,
      "loss": 1.1578,
      "step": 3226
    },
    {
      "epoch": 3.949816401468788,
      "grad_norm": 1.150796134126557,
      "learning_rate": 4.1922606854158424e-06,
      "loss": 0.5303,
      "step": 3227
    },
    {
      "epoch": 3.951040391676867,
      "grad_norm": 1.0212608149482156,
      "learning_rate": 4.191788896762418e-06,
      "loss": 0.6171,
      "step": 3228
    },
    {
      "epoch": 3.952264381884945,
      "grad_norm": 1.0903885728486087,
      "learning_rate": 4.191316996930887e-06,
      "loss": 0.5023,
      "step": 3229
    },
    {
      "epoch": 3.953488372093023,
      "grad_norm": 1.0472816231174875,
      "learning_rate": 4.190844985952257e-06,
      "loss": 0.6424,
      "step": 3230
    },
    {
      "epoch": 3.954712362301102,
      "grad_norm": 0.9967431749239684,
      "learning_rate": 4.19037286385755e-06,
      "loss": 0.6363,
      "step": 3231
    },
    {
      "epoch": 3.95593635250918,
      "grad_norm": 1.1020261350821574,
      "learning_rate": 4.189900630677789e-06,
      "loss": 1.2871,
      "step": 3232
    },
    {
      "epoch": 3.957160342717258,
      "grad_norm": 1.0643312098170592,
      "learning_rate": 4.189428286444011e-06,
      "loss": 0.5426,
      "step": 3233
    },
    {
      "epoch": 3.9583843329253368,
      "grad_norm": 1.2425866321323995,
      "learning_rate": 4.188955831187253e-06,
      "loss": 0.512,
      "step": 3234
    },
    {
      "epoch": 3.959608323133415,
      "grad_norm": 2.1945178517922135,
      "learning_rate": 4.188483264938565e-06,
      "loss": 0.5272,
      "step": 3235
    },
    {
      "epoch": 3.960832313341493,
      "grad_norm": 0.9538433517694324,
      "learning_rate": 4.188010587729003e-06,
      "loss": 0.6053,
      "step": 3236
    },
    {
      "epoch": 3.9620563035495717,
      "grad_norm": 1.1130150621312103,
      "learning_rate": 4.187537799589628e-06,
      "loss": 0.5045,
      "step": 3237
    },
    {
      "epoch": 3.96328029375765,
      "grad_norm": 1.3838510638464083,
      "learning_rate": 4.187064900551511e-06,
      "loss": 0.6629,
      "step": 3238
    },
    {
      "epoch": 3.964504283965728,
      "grad_norm": 1.7891048767835043,
      "learning_rate": 4.186591890645728e-06,
      "loss": 0.7489,
      "step": 3239
    },
    {
      "epoch": 3.9657282741738067,
      "grad_norm": 0.9734314499524992,
      "learning_rate": 4.186118769903365e-06,
      "loss": 0.7779,
      "step": 3240
    },
    {
      "epoch": 3.966952264381885,
      "grad_norm": 1.1481597072537855,
      "learning_rate": 4.1856455383555124e-06,
      "loss": 0.7392,
      "step": 3241
    },
    {
      "epoch": 3.968176254589963,
      "grad_norm": 0.9156432422316886,
      "learning_rate": 4.18517219603327e-06,
      "loss": 0.8609,
      "step": 3242
    },
    {
      "epoch": 3.9694002447980417,
      "grad_norm": 0.8885352588113989,
      "learning_rate": 4.184698742967744e-06,
      "loss": 0.5541,
      "step": 3243
    },
    {
      "epoch": 3.97062423500612,
      "grad_norm": 1.1206194142053714,
      "learning_rate": 4.184225179190048e-06,
      "loss": 0.7709,
      "step": 3244
    },
    {
      "epoch": 3.9718482252141984,
      "grad_norm": 1.3753277018746342,
      "learning_rate": 4.183751504731303e-06,
      "loss": 0.6057,
      "step": 3245
    },
    {
      "epoch": 3.9730722154222766,
      "grad_norm": 1.543549493410082,
      "learning_rate": 4.183277719622636e-06,
      "loss": 0.7684,
      "step": 3246
    },
    {
      "epoch": 3.974296205630355,
      "grad_norm": 0.8460413132124633,
      "learning_rate": 4.182803823895184e-06,
      "loss": 0.5528,
      "step": 3247
    },
    {
      "epoch": 3.9755201958384334,
      "grad_norm": 1.054590519211611,
      "learning_rate": 4.182329817580089e-06,
      "loss": 1.3167,
      "step": 3248
    },
    {
      "epoch": 3.9767441860465116,
      "grad_norm": 0.6074308893167876,
      "learning_rate": 4.1818557007085e-06,
      "loss": 0.369,
      "step": 3249
    },
    {
      "epoch": 3.97796817625459,
      "grad_norm": 1.998817616009825,
      "learning_rate": 4.181381473311576e-06,
      "loss": 0.653,
      "step": 3250
    },
    {
      "epoch": 3.9791921664626684,
      "grad_norm": 1.1799043000883938,
      "learning_rate": 4.18090713542048e-06,
      "loss": 0.483,
      "step": 3251
    },
    {
      "epoch": 3.9804161566707466,
      "grad_norm": 0.6562459140235288,
      "learning_rate": 4.180432687066385e-06,
      "loss": 0.5185,
      "step": 3252
    },
    {
      "epoch": 3.981640146878825,
      "grad_norm": 1.6364731069240495,
      "learning_rate": 4.179958128280469e-06,
      "loss": 0.6471,
      "step": 3253
    },
    {
      "epoch": 3.9828641370869033,
      "grad_norm": 1.6597254854878378,
      "learning_rate": 4.179483459093918e-06,
      "loss": 0.5193,
      "step": 3254
    },
    {
      "epoch": 3.9840881272949815,
      "grad_norm": 1.183082950259907,
      "learning_rate": 4.179008679537927e-06,
      "loss": 0.3555,
      "step": 3255
    },
    {
      "epoch": 3.98531211750306,
      "grad_norm": 0.9294448612860027,
      "learning_rate": 4.1785337896436954e-06,
      "loss": 0.3776,
      "step": 3256
    },
    {
      "epoch": 3.9865361077111383,
      "grad_norm": 1.4145898583411525,
      "learning_rate": 4.178058789442431e-06,
      "loss": 1.046,
      "step": 3257
    },
    {
      "epoch": 3.9877600979192165,
      "grad_norm": 1.2453044262060387,
      "learning_rate": 4.17758367896535e-06,
      "loss": 0.509,
      "step": 3258
    },
    {
      "epoch": 3.988984088127295,
      "grad_norm": 1.0859080839856812,
      "learning_rate": 4.177108458243675e-06,
      "loss": 1.3496,
      "step": 3259
    },
    {
      "epoch": 3.9902080783353733,
      "grad_norm": 2.0012676872111226,
      "learning_rate": 4.1766331273086355e-06,
      "loss": 0.4217,
      "step": 3260
    },
    {
      "epoch": 3.9914320685434515,
      "grad_norm": 0.9975300610013733,
      "learning_rate": 4.1761576861914675e-06,
      "loss": 0.637,
      "step": 3261
    },
    {
      "epoch": 3.99265605875153,
      "grad_norm": 0.8582768205077455,
      "learning_rate": 4.175682134923417e-06,
      "loss": 0.6088,
      "step": 3262
    },
    {
      "epoch": 3.9938800489596082,
      "grad_norm": 1.3989164235659188,
      "learning_rate": 4.1752064735357335e-06,
      "loss": 0.5652,
      "step": 3263
    },
    {
      "epoch": 3.9951040391676864,
      "grad_norm": 1.8692129181327695,
      "learning_rate": 4.1747307020596775e-06,
      "loss": 0.9654,
      "step": 3264
    },
    {
      "epoch": 3.996328029375765,
      "grad_norm": 1.0531002565008702,
      "learning_rate": 4.174254820526513e-06,
      "loss": 0.332,
      "step": 3265
    },
    {
      "epoch": 3.997552019583843,
      "grad_norm": 1.4434190850273085,
      "learning_rate": 4.173778828967515e-06,
      "loss": 0.7892,
      "step": 3266
    },
    {
      "epoch": 3.998776009791922,
      "grad_norm": 1.1899233105511415,
      "learning_rate": 4.173302727413963e-06,
      "loss": 0.6891,
      "step": 3267
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.337153056515082,
      "learning_rate": 4.172826515897146e-06,
      "loss": 0.5197,
      "step": 3268
    },
    {
      "epoch": 4.001223990208079,
      "grad_norm": 1.0770682496225605,
      "learning_rate": 4.172350194448357e-06,
      "loss": 1.2424,
      "step": 3269
    },
    {
      "epoch": 4.002447980416156,
      "grad_norm": 0.7668975448193764,
      "learning_rate": 4.171873763098898e-06,
      "loss": 0.5314,
      "step": 3270
    },
    {
      "epoch": 4.003671970624235,
      "grad_norm": 0.7231309384458519,
      "learning_rate": 4.171397221880081e-06,
      "loss": 0.4413,
      "step": 3271
    },
    {
      "epoch": 4.004895960832314,
      "grad_norm": 0.8844410665672732,
      "learning_rate": 4.1709205708232195e-06,
      "loss": 0.8655,
      "step": 3272
    },
    {
      "epoch": 4.006119951040391,
      "grad_norm": 0.7040824748901879,
      "learning_rate": 4.170443809959638e-06,
      "loss": 0.5522,
      "step": 3273
    },
    {
      "epoch": 4.00734394124847,
      "grad_norm": 1.035018920890649,
      "learning_rate": 4.169966939320668e-06,
      "loss": 0.9305,
      "step": 3274
    },
    {
      "epoch": 4.0085679314565485,
      "grad_norm": 1.3547910207449279,
      "learning_rate": 4.169489958937648e-06,
      "loss": 0.6207,
      "step": 3275
    },
    {
      "epoch": 4.009791921664626,
      "grad_norm": 1.296706367308019,
      "learning_rate": 4.169012868841923e-06,
      "loss": 0.4518,
      "step": 3276
    },
    {
      "epoch": 4.011015911872705,
      "grad_norm": 0.9799395753157674,
      "learning_rate": 4.168535669064845e-06,
      "loss": 0.6466,
      "step": 3277
    },
    {
      "epoch": 4.0122399020807835,
      "grad_norm": 0.9831362619662715,
      "learning_rate": 4.168058359637775e-06,
      "loss": 0.5339,
      "step": 3278
    },
    {
      "epoch": 4.013463892288861,
      "grad_norm": 1.306770430794324,
      "learning_rate": 4.167580940592079e-06,
      "loss": 1.1229,
      "step": 3279
    },
    {
      "epoch": 4.01468788249694,
      "grad_norm": 1.6447021321167703,
      "learning_rate": 4.167103411959131e-06,
      "loss": 0.4751,
      "step": 3280
    },
    {
      "epoch": 4.0159118727050185,
      "grad_norm": 1.9017146324792293,
      "learning_rate": 4.166625773770314e-06,
      "loss": 0.4645,
      "step": 3281
    },
    {
      "epoch": 4.017135862913097,
      "grad_norm": 1.7414616995185106,
      "learning_rate": 4.166148026057014e-06,
      "loss": 0.6397,
      "step": 3282
    },
    {
      "epoch": 4.018359853121175,
      "grad_norm": 1.1133351967361695,
      "learning_rate": 4.1656701688506294e-06,
      "loss": 0.7075,
      "step": 3283
    },
    {
      "epoch": 4.0195838433292534,
      "grad_norm": 1.1917587436567887,
      "learning_rate": 4.165192202182563e-06,
      "loss": 0.5042,
      "step": 3284
    },
    {
      "epoch": 4.020807833537332,
      "grad_norm": 1.5181313013975306,
      "learning_rate": 4.164714126084223e-06,
      "loss": 0.7576,
      "step": 3285
    },
    {
      "epoch": 4.02203182374541,
      "grad_norm": 1.1233532461550566,
      "learning_rate": 4.164235940587029e-06,
      "loss": 0.4133,
      "step": 3286
    },
    {
      "epoch": 4.023255813953488,
      "grad_norm": 1.0653283514514762,
      "learning_rate": 4.163757645722403e-06,
      "loss": 1.4393,
      "step": 3287
    },
    {
      "epoch": 4.024479804161567,
      "grad_norm": 0.9800986218668981,
      "learning_rate": 4.163279241521779e-06,
      "loss": 0.4829,
      "step": 3288
    },
    {
      "epoch": 4.025703794369645,
      "grad_norm": 1.2482067350824637,
      "learning_rate": 4.162800728016596e-06,
      "loss": 0.7705,
      "step": 3289
    },
    {
      "epoch": 4.026927784577723,
      "grad_norm": 1.226151921127532,
      "learning_rate": 4.162322105238299e-06,
      "loss": 0.5877,
      "step": 3290
    },
    {
      "epoch": 4.028151774785802,
      "grad_norm": 0.8126059107480726,
      "learning_rate": 4.161843373218342e-06,
      "loss": 0.5718,
      "step": 3291
    },
    {
      "epoch": 4.02937576499388,
      "grad_norm": 1.232450665432227,
      "learning_rate": 4.161364531988185e-06,
      "loss": 0.4992,
      "step": 3292
    },
    {
      "epoch": 4.030599755201958,
      "grad_norm": 1.5638809795129538,
      "learning_rate": 4.1608855815792955e-06,
      "loss": 0.4745,
      "step": 3293
    },
    {
      "epoch": 4.031823745410037,
      "grad_norm": 1.1930217799431573,
      "learning_rate": 4.160406522023149e-06,
      "loss": 0.621,
      "step": 3294
    },
    {
      "epoch": 4.033047735618115,
      "grad_norm": 1.0652230139913195,
      "learning_rate": 4.1599273533512275e-06,
      "loss": 0.5234,
      "step": 3295
    },
    {
      "epoch": 4.034271725826193,
      "grad_norm": 0.9582027820409285,
      "learning_rate": 4.15944807559502e-06,
      "loss": 0.6783,
      "step": 3296
    },
    {
      "epoch": 4.035495716034272,
      "grad_norm": 0.9918953431957879,
      "learning_rate": 4.158968688786022e-06,
      "loss": 0.6542,
      "step": 3297
    },
    {
      "epoch": 4.03671970624235,
      "grad_norm": 1.2361849219416445,
      "learning_rate": 4.158489192955739e-06,
      "loss": 1.1945,
      "step": 3298
    },
    {
      "epoch": 4.037943696450428,
      "grad_norm": 1.3906176318514818,
      "learning_rate": 4.158009588135681e-06,
      "loss": 1.2236,
      "step": 3299
    },
    {
      "epoch": 4.039167686658507,
      "grad_norm": 1.463784422943233,
      "learning_rate": 4.157529874357364e-06,
      "loss": 1.3581,
      "step": 3300
    },
    {
      "epoch": 4.0403916768665855,
      "grad_norm": 2.9122546967199674,
      "learning_rate": 4.157050051652315e-06,
      "loss": 0.5205,
      "step": 3301
    },
    {
      "epoch": 4.041615667074663,
      "grad_norm": 2.2136999535922075,
      "learning_rate": 4.156570120052066e-06,
      "loss": 0.4829,
      "step": 3302
    },
    {
      "epoch": 4.042839657282742,
      "grad_norm": 1.2448173086035832,
      "learning_rate": 4.156090079588155e-06,
      "loss": 0.5189,
      "step": 3303
    },
    {
      "epoch": 4.0440636474908205,
      "grad_norm": 0.920344483403311,
      "learning_rate": 4.1556099302921305e-06,
      "loss": 0.4895,
      "step": 3304
    },
    {
      "epoch": 4.045287637698898,
      "grad_norm": 1.2827465444005968,
      "learning_rate": 4.155129672195545e-06,
      "loss": 0.6347,
      "step": 3305
    },
    {
      "epoch": 4.046511627906977,
      "grad_norm": 1.137513148026802,
      "learning_rate": 4.154649305329959e-06,
      "loss": 1.2929,
      "step": 3306
    },
    {
      "epoch": 4.0477356181150554,
      "grad_norm": 0.9476714409464848,
      "learning_rate": 4.15416882972694e-06,
      "loss": 1.1137,
      "step": 3307
    },
    {
      "epoch": 4.048959608323133,
      "grad_norm": 1.62143869083556,
      "learning_rate": 4.153688245418065e-06,
      "loss": 0.5147,
      "step": 3308
    },
    {
      "epoch": 4.050183598531212,
      "grad_norm": 1.4392643920216541,
      "learning_rate": 4.153207552434915e-06,
      "loss": 0.8744,
      "step": 3309
    },
    {
      "epoch": 4.05140758873929,
      "grad_norm": 0.979334342927425,
      "learning_rate": 4.152726750809078e-06,
      "loss": 0.727,
      "step": 3310
    },
    {
      "epoch": 4.052631578947368,
      "grad_norm": 1.22096457738444,
      "learning_rate": 4.152245840572153e-06,
      "loss": 1.4245,
      "step": 3311
    },
    {
      "epoch": 4.053855569155447,
      "grad_norm": 1.1480898082799866,
      "learning_rate": 4.151764821755743e-06,
      "loss": 0.5654,
      "step": 3312
    },
    {
      "epoch": 4.055079559363525,
      "grad_norm": 0.8148936352858921,
      "learning_rate": 4.151283694391458e-06,
      "loss": 0.6313,
      "step": 3313
    },
    {
      "epoch": 4.056303549571603,
      "grad_norm": 1.068359732549216,
      "learning_rate": 4.150802458510916e-06,
      "loss": 0.5159,
      "step": 3314
    },
    {
      "epoch": 4.057527539779682,
      "grad_norm": 0.9512149303346032,
      "learning_rate": 4.150321114145743e-06,
      "loss": 0.5571,
      "step": 3315
    },
    {
      "epoch": 4.05875152998776,
      "grad_norm": 1.36158379931695,
      "learning_rate": 4.14983966132757e-06,
      "loss": 1.0395,
      "step": 3316
    },
    {
      "epoch": 4.059975520195838,
      "grad_norm": 1.100842509140695,
      "learning_rate": 4.149358100088037e-06,
      "loss": 0.5251,
      "step": 3317
    },
    {
      "epoch": 4.061199510403917,
      "grad_norm": 1.5466030911978659,
      "learning_rate": 4.14887643045879e-06,
      "loss": 0.8692,
      "step": 3318
    },
    {
      "epoch": 4.062423500611995,
      "grad_norm": 1.6374620781358773,
      "learning_rate": 4.148394652471483e-06,
      "loss": 0.6388,
      "step": 3319
    },
    {
      "epoch": 4.063647490820073,
      "grad_norm": 1.7062142487924488,
      "learning_rate": 4.1479127661577765e-06,
      "loss": 0.4844,
      "step": 3320
    },
    {
      "epoch": 4.064871481028152,
      "grad_norm": 0.7840421361447805,
      "learning_rate": 4.1474307715493375e-06,
      "loss": 0.5974,
      "step": 3321
    },
    {
      "epoch": 4.06609547123623,
      "grad_norm": 0.9052106078110027,
      "learning_rate": 4.146948668677842e-06,
      "loss": 0.9155,
      "step": 3322
    },
    {
      "epoch": 4.067319461444308,
      "grad_norm": 0.7829094216802672,
      "learning_rate": 4.1464664575749725e-06,
      "loss": 0.5303,
      "step": 3323
    },
    {
      "epoch": 4.068543451652387,
      "grad_norm": 1.2059700042507526,
      "learning_rate": 4.145984138272416e-06,
      "loss": 0.5777,
      "step": 3324
    },
    {
      "epoch": 4.069767441860465,
      "grad_norm": 1.358936868433063,
      "learning_rate": 4.14550171080187e-06,
      "loss": 0.652,
      "step": 3325
    },
    {
      "epoch": 4.070991432068544,
      "grad_norm": 1.0507571847065702,
      "learning_rate": 4.1450191751950386e-06,
      "loss": 0.6348,
      "step": 3326
    },
    {
      "epoch": 4.072215422276622,
      "grad_norm": 1.0833986940428377,
      "learning_rate": 4.144536531483632e-06,
      "loss": 2.317,
      "step": 3327
    },
    {
      "epoch": 4.0734394124847,
      "grad_norm": 1.3233102874578704,
      "learning_rate": 4.144053779699366e-06,
      "loss": 1.2495,
      "step": 3328
    },
    {
      "epoch": 4.074663402692779,
      "grad_norm": 1.055034910513984,
      "learning_rate": 4.143570919873967e-06,
      "loss": 0.445,
      "step": 3329
    },
    {
      "epoch": 4.0758873929008566,
      "grad_norm": 1.067279350382245,
      "learning_rate": 4.1430879520391665e-06,
      "loss": 0.8586,
      "step": 3330
    },
    {
      "epoch": 4.077111383108935,
      "grad_norm": 1.149745433072691,
      "learning_rate": 4.1426048762267036e-06,
      "loss": 0.653,
      "step": 3331
    },
    {
      "epoch": 4.078335373317014,
      "grad_norm": 0.9326021614109036,
      "learning_rate": 4.142121692468323e-06,
      "loss": 0.5894,
      "step": 3332
    },
    {
      "epoch": 4.0795593635250915,
      "grad_norm": 1.254660614667816,
      "learning_rate": 4.141638400795779e-06,
      "loss": 0.6499,
      "step": 3333
    },
    {
      "epoch": 4.08078335373317,
      "grad_norm": 1.1316326945938204,
      "learning_rate": 4.141155001240831e-06,
      "loss": 1.106,
      "step": 3334
    },
    {
      "epoch": 4.082007343941249,
      "grad_norm": 1.4915646261237072,
      "learning_rate": 4.140671493835246e-06,
      "loss": 0.7232,
      "step": 3335
    },
    {
      "epoch": 4.0832313341493265,
      "grad_norm": 1.2514589338658708,
      "learning_rate": 4.140187878610801e-06,
      "loss": 0.3962,
      "step": 3336
    },
    {
      "epoch": 4.084455324357405,
      "grad_norm": 0.8525859797429366,
      "learning_rate": 4.139704155599273e-06,
      "loss": 0.5258,
      "step": 3337
    },
    {
      "epoch": 4.085679314565484,
      "grad_norm": 0.7490831009027872,
      "learning_rate": 4.1392203248324535e-06,
      "loss": 0.414,
      "step": 3338
    },
    {
      "epoch": 4.0869033047735615,
      "grad_norm": 1.4568370802378043,
      "learning_rate": 4.138736386342137e-06,
      "loss": 0.3498,
      "step": 3339
    },
    {
      "epoch": 4.08812729498164,
      "grad_norm": 1.2444327604985292,
      "learning_rate": 4.1382523401601275e-06,
      "loss": 0.4986,
      "step": 3340
    },
    {
      "epoch": 4.089351285189719,
      "grad_norm": 1.2341884029874994,
      "learning_rate": 4.137768186318232e-06,
      "loss": 0.4713,
      "step": 3341
    },
    {
      "epoch": 4.090575275397796,
      "grad_norm": 1.0583263663903184,
      "learning_rate": 4.1372839248482695e-06,
      "loss": 0.3938,
      "step": 3342
    },
    {
      "epoch": 4.091799265605875,
      "grad_norm": 1.0856302097144028,
      "learning_rate": 4.136799555782064e-06,
      "loss": 0.568,
      "step": 3343
    },
    {
      "epoch": 4.093023255813954,
      "grad_norm": 1.0056371362304188,
      "learning_rate": 4.136315079151446e-06,
      "loss": 0.6462,
      "step": 3344
    },
    {
      "epoch": 4.094247246022032,
      "grad_norm": 0.8803101472854242,
      "learning_rate": 4.135830494988252e-06,
      "loss": 0.6829,
      "step": 3345
    },
    {
      "epoch": 4.09547123623011,
      "grad_norm": 1.428710796666031,
      "learning_rate": 4.13534580332433e-06,
      "loss": 1.4356,
      "step": 3346
    },
    {
      "epoch": 4.096695226438189,
      "grad_norm": 1.287158535494212,
      "learning_rate": 4.13486100419153e-06,
      "loss": 0.569,
      "step": 3347
    },
    {
      "epoch": 4.097919216646267,
      "grad_norm": 1.2979646935441453,
      "learning_rate": 4.134376097621711e-06,
      "loss": 1.2301,
      "step": 3348
    },
    {
      "epoch": 4.099143206854345,
      "grad_norm": 0.8604568055411687,
      "learning_rate": 4.133891083646741e-06,
      "loss": 0.4159,
      "step": 3349
    },
    {
      "epoch": 4.100367197062424,
      "grad_norm": 1.4547603443147472,
      "learning_rate": 4.133405962298491e-06,
      "loss": 0.6685,
      "step": 3350
    },
    {
      "epoch": 4.101591187270502,
      "grad_norm": 1.1494986663139095,
      "learning_rate": 4.132920733608844e-06,
      "loss": 0.686,
      "step": 3351
    },
    {
      "epoch": 4.10281517747858,
      "grad_norm": 0.8567847094736317,
      "learning_rate": 4.132435397609686e-06,
      "loss": 0.9242,
      "step": 3352
    },
    {
      "epoch": 4.1040391676866586,
      "grad_norm": 0.6120191385649659,
      "learning_rate": 4.131949954332911e-06,
      "loss": 0.3667,
      "step": 3353
    },
    {
      "epoch": 4.105263157894737,
      "grad_norm": 1.100877630490201,
      "learning_rate": 4.1314644038104215e-06,
      "loss": 0.9622,
      "step": 3354
    },
    {
      "epoch": 4.106487148102815,
      "grad_norm": 0.8197685072738816,
      "learning_rate": 4.130978746074126e-06,
      "loss": 0.5108,
      "step": 3355
    },
    {
      "epoch": 4.1077111383108935,
      "grad_norm": 0.7073119506704658,
      "learning_rate": 4.13049298115594e-06,
      "loss": 0.5423,
      "step": 3356
    },
    {
      "epoch": 4.108935128518972,
      "grad_norm": 1.3971119906865381,
      "learning_rate": 4.130007109087786e-06,
      "loss": 0.6146,
      "step": 3357
    },
    {
      "epoch": 4.11015911872705,
      "grad_norm": 2.0341821009472136,
      "learning_rate": 4.129521129901593e-06,
      "loss": 1.0215,
      "step": 3358
    },
    {
      "epoch": 4.1113831089351285,
      "grad_norm": 0.7830734070449611,
      "learning_rate": 4.129035043629299e-06,
      "loss": 0.6681,
      "step": 3359
    },
    {
      "epoch": 4.112607099143207,
      "grad_norm": 1.0269977206113536,
      "learning_rate": 4.128548850302847e-06,
      "loss": 0.8317,
      "step": 3360
    },
    {
      "epoch": 4.113831089351285,
      "grad_norm": 1.0585622822675491,
      "learning_rate": 4.1280625499541884e-06,
      "loss": 0.3948,
      "step": 3361
    },
    {
      "epoch": 4.1150550795593634,
      "grad_norm": 1.2759517994277079,
      "learning_rate": 4.1275761426152815e-06,
      "loss": 0.47,
      "step": 3362
    },
    {
      "epoch": 4.116279069767442,
      "grad_norm": 0.8473812039034465,
      "learning_rate": 4.12708962831809e-06,
      "loss": 0.6084,
      "step": 3363
    },
    {
      "epoch": 4.11750305997552,
      "grad_norm": 1.930059082866513,
      "learning_rate": 4.126603007094586e-06,
      "loss": 0.972,
      "step": 3364
    },
    {
      "epoch": 4.118727050183598,
      "grad_norm": 0.4021353633336182,
      "learning_rate": 4.1261162789767486e-06,
      "loss": 0.1354,
      "step": 3365
    },
    {
      "epoch": 4.119951040391677,
      "grad_norm": 1.2254911663804537,
      "learning_rate": 4.125629443996564e-06,
      "loss": 0.5704,
      "step": 3366
    },
    {
      "epoch": 4.121175030599755,
      "grad_norm": 1.012895156297931,
      "learning_rate": 4.125142502186025e-06,
      "loss": 0.7482,
      "step": 3367
    },
    {
      "epoch": 4.122399020807833,
      "grad_norm": 1.4543991287337852,
      "learning_rate": 4.124655453577132e-06,
      "loss": 1.0568,
      "step": 3368
    },
    {
      "epoch": 4.123623011015912,
      "grad_norm": 1.0339246719405268,
      "learning_rate": 4.124168298201893e-06,
      "loss": 0.2529,
      "step": 3369
    },
    {
      "epoch": 4.124847001223991,
      "grad_norm": 1.5372965776420549,
      "learning_rate": 4.123681036092319e-06,
      "loss": 0.5435,
      "step": 3370
    },
    {
      "epoch": 4.126070991432068,
      "grad_norm": 1.2046307555173001,
      "learning_rate": 4.123193667280434e-06,
      "loss": 0.9659,
      "step": 3371
    },
    {
      "epoch": 4.127294981640147,
      "grad_norm": 1.2006451234436915,
      "learning_rate": 4.122706191798265e-06,
      "loss": 1.4606,
      "step": 3372
    },
    {
      "epoch": 4.128518971848226,
      "grad_norm": 1.489207150188012,
      "learning_rate": 4.122218609677846e-06,
      "loss": 0.528,
      "step": 3373
    },
    {
      "epoch": 4.129742962056303,
      "grad_norm": 1.2370766628438878,
      "learning_rate": 4.121730920951221e-06,
      "loss": 1.2332,
      "step": 3374
    },
    {
      "epoch": 4.130966952264382,
      "grad_norm": 0.8563189763469776,
      "learning_rate": 4.121243125650438e-06,
      "loss": 0.655,
      "step": 3375
    },
    {
      "epoch": 4.1321909424724605,
      "grad_norm": 0.8301835358113144,
      "learning_rate": 4.120755223807553e-06,
      "loss": 0.4709,
      "step": 3376
    },
    {
      "epoch": 4.133414932680538,
      "grad_norm": 1.398448233001719,
      "learning_rate": 4.12026721545463e-06,
      "loss": 0.8032,
      "step": 3377
    },
    {
      "epoch": 4.134638922888617,
      "grad_norm": 1.2871386715868687,
      "learning_rate": 4.119779100623739e-06,
      "loss": 1.2085,
      "step": 3378
    },
    {
      "epoch": 4.1358629130966955,
      "grad_norm": 0.7946313308724975,
      "learning_rate": 4.119290879346956e-06,
      "loss": 0.5912,
      "step": 3379
    },
    {
      "epoch": 4.137086903304773,
      "grad_norm": 1.4349278977233086,
      "learning_rate": 4.118802551656365e-06,
      "loss": 0.4259,
      "step": 3380
    },
    {
      "epoch": 4.138310893512852,
      "grad_norm": 1.3460252667720802,
      "learning_rate": 4.118314117584059e-06,
      "loss": 0.4049,
      "step": 3381
    },
    {
      "epoch": 4.1395348837209305,
      "grad_norm": 1.0551625784443917,
      "learning_rate": 4.1178255771621345e-06,
      "loss": 0.8448,
      "step": 3382
    },
    {
      "epoch": 4.140758873929008,
      "grad_norm": 1.8017906840485531,
      "learning_rate": 4.117336930422697e-06,
      "loss": 0.4565,
      "step": 3383
    },
    {
      "epoch": 4.141982864137087,
      "grad_norm": 1.7615574768450382,
      "learning_rate": 4.116848177397859e-06,
      "loss": 0.4135,
      "step": 3384
    },
    {
      "epoch": 4.1432068543451654,
      "grad_norm": 0.9597067988562306,
      "learning_rate": 4.116359318119738e-06,
      "loss": 1.1629,
      "step": 3385
    },
    {
      "epoch": 4.144430844553243,
      "grad_norm": 1.1481166203842035,
      "learning_rate": 4.1158703526204625e-06,
      "loss": 0.3485,
      "step": 3386
    },
    {
      "epoch": 4.145654834761322,
      "grad_norm": 1.0679904668142843,
      "learning_rate": 4.115381280932163e-06,
      "loss": 0.7444,
      "step": 3387
    },
    {
      "epoch": 4.1468788249694,
      "grad_norm": 1.1885405402643132,
      "learning_rate": 4.11489210308698e-06,
      "loss": 1.604,
      "step": 3388
    },
    {
      "epoch": 4.148102815177479,
      "grad_norm": 1.1840717553644868,
      "learning_rate": 4.114402819117063e-06,
      "loss": 0.595,
      "step": 3389
    },
    {
      "epoch": 4.149326805385557,
      "grad_norm": 0.9157845314182612,
      "learning_rate": 4.113913429054563e-06,
      "loss": 0.6104,
      "step": 3390
    },
    {
      "epoch": 4.150550795593635,
      "grad_norm": 1.105303835639854,
      "learning_rate": 4.113423932931642e-06,
      "loss": 1.4355,
      "step": 3391
    },
    {
      "epoch": 4.151774785801714,
      "grad_norm": 1.6727705382161635,
      "learning_rate": 4.112934330780468e-06,
      "loss": 0.9157,
      "step": 3392
    },
    {
      "epoch": 4.152998776009792,
      "grad_norm": 1.0389938439575055,
      "learning_rate": 4.112444622633215e-06,
      "loss": 1.0354,
      "step": 3393
    },
    {
      "epoch": 4.15422276621787,
      "grad_norm": 1.904912507546912,
      "learning_rate": 4.111954808522065e-06,
      "loss": 0.4729,
      "step": 3394
    },
    {
      "epoch": 4.155446756425949,
      "grad_norm": 1.2507277948029176,
      "learning_rate": 4.111464888479209e-06,
      "loss": 0.4968,
      "step": 3395
    },
    {
      "epoch": 4.156670746634027,
      "grad_norm": 1.4580196346799044,
      "learning_rate": 4.11097486253684e-06,
      "loss": 1.1504,
      "step": 3396
    },
    {
      "epoch": 4.157894736842105,
      "grad_norm": 1.974338990709747,
      "learning_rate": 4.110484730727161e-06,
      "loss": 0.483,
      "step": 3397
    },
    {
      "epoch": 4.159118727050184,
      "grad_norm": 1.416270405409402,
      "learning_rate": 4.109994493082382e-06,
      "loss": 1.3332,
      "step": 3398
    },
    {
      "epoch": 4.160342717258262,
      "grad_norm": 1.9593393937763,
      "learning_rate": 4.109504149634721e-06,
      "loss": 0.5095,
      "step": 3399
    },
    {
      "epoch": 4.16156670746634,
      "grad_norm": 1.0452628680081464,
      "learning_rate": 4.1090137004164e-06,
      "loss": 0.3635,
      "step": 3400
    },
    {
      "epoch": 4.162790697674419,
      "grad_norm": 1.7459279465470268,
      "learning_rate": 4.10852314545965e-06,
      "loss": 0.4535,
      "step": 3401
    },
    {
      "epoch": 4.164014687882497,
      "grad_norm": 1.87106817105612,
      "learning_rate": 4.108032484796708e-06,
      "loss": 0.984,
      "step": 3402
    },
    {
      "epoch": 4.165238678090575,
      "grad_norm": 2.1131099483590883,
      "learning_rate": 4.1075417184598194e-06,
      "loss": 0.6145,
      "step": 3403
    },
    {
      "epoch": 4.166462668298654,
      "grad_norm": 1.5980194817224995,
      "learning_rate": 4.107050846481235e-06,
      "loss": 0.6045,
      "step": 3404
    },
    {
      "epoch": 4.167686658506732,
      "grad_norm": 0.934322119336499,
      "learning_rate": 4.106559868893213e-06,
      "loss": 0.4692,
      "step": 3405
    },
    {
      "epoch": 4.16891064871481,
      "grad_norm": 1.2120398881253192,
      "learning_rate": 4.106068785728018e-06,
      "loss": 0.7773,
      "step": 3406
    },
    {
      "epoch": 4.170134638922889,
      "grad_norm": 1.3292982827600648,
      "learning_rate": 4.105577597017923e-06,
      "loss": 1.0246,
      "step": 3407
    },
    {
      "epoch": 4.1713586291309666,
      "grad_norm": 1.0567469719353078,
      "learning_rate": 4.105086302795208e-06,
      "loss": 0.5073,
      "step": 3408
    },
    {
      "epoch": 4.172582619339045,
      "grad_norm": 1.141817811442184,
      "learning_rate": 4.1045949030921575e-06,
      "loss": 1.2677,
      "step": 3409
    },
    {
      "epoch": 4.173806609547124,
      "grad_norm": 2.0563260994504913,
      "learning_rate": 4.1041033979410635e-06,
      "loss": 0.9712,
      "step": 3410
    },
    {
      "epoch": 4.1750305997552015,
      "grad_norm": 1.8415173855644855,
      "learning_rate": 4.103611787374229e-06,
      "loss": 0.5464,
      "step": 3411
    },
    {
      "epoch": 4.17625458996328,
      "grad_norm": 1.1679385086132568,
      "learning_rate": 4.103120071423959e-06,
      "loss": 0.539,
      "step": 3412
    },
    {
      "epoch": 4.177478580171359,
      "grad_norm": 0.7399973734100924,
      "learning_rate": 4.102628250122567e-06,
      "loss": 0.5363,
      "step": 3413
    },
    {
      "epoch": 4.178702570379437,
      "grad_norm": 1.189216194287651,
      "learning_rate": 4.102136323502374e-06,
      "loss": 0.4653,
      "step": 3414
    },
    {
      "epoch": 4.179926560587515,
      "grad_norm": 1.2016459422518375,
      "learning_rate": 4.101644291595708e-06,
      "loss": 1.21,
      "step": 3415
    },
    {
      "epoch": 4.181150550795594,
      "grad_norm": 1.4751014831600553,
      "learning_rate": 4.101152154434903e-06,
      "loss": 0.5211,
      "step": 3416
    },
    {
      "epoch": 4.182374541003672,
      "grad_norm": 1.4851188821896093,
      "learning_rate": 4.100659912052301e-06,
      "loss": 0.4395,
      "step": 3417
    },
    {
      "epoch": 4.18359853121175,
      "grad_norm": 0.9369889124385221,
      "learning_rate": 4.10016756448025e-06,
      "loss": 1.0003,
      "step": 3418
    },
    {
      "epoch": 4.184822521419829,
      "grad_norm": 0.7445031212957764,
      "learning_rate": 4.099675111751105e-06,
      "loss": 0.5392,
      "step": 3419
    },
    {
      "epoch": 4.186046511627907,
      "grad_norm": 1.0090092552096495,
      "learning_rate": 4.099182553897228e-06,
      "loss": 0.3028,
      "step": 3420
    },
    {
      "epoch": 4.187270501835985,
      "grad_norm": 1.311449400506681,
      "learning_rate": 4.09868989095099e-06,
      "loss": 1.8449,
      "step": 3421
    },
    {
      "epoch": 4.188494492044064,
      "grad_norm": 1.3641269734985664,
      "learning_rate": 4.098197122944764e-06,
      "loss": 0.5826,
      "step": 3422
    },
    {
      "epoch": 4.189718482252142,
      "grad_norm": 1.3044134823180682,
      "learning_rate": 4.097704249910935e-06,
      "loss": 1.3838,
      "step": 3423
    },
    {
      "epoch": 4.19094247246022,
      "grad_norm": 1.0260273025455398,
      "learning_rate": 4.097211271881892e-06,
      "loss": 0.5956,
      "step": 3424
    },
    {
      "epoch": 4.192166462668299,
      "grad_norm": 0.9843760542304437,
      "learning_rate": 4.096718188890033e-06,
      "loss": 0.5773,
      "step": 3425
    },
    {
      "epoch": 4.193390452876377,
      "grad_norm": 1.678136196159439,
      "learning_rate": 4.096225000967759e-06,
      "loss": 1.4589,
      "step": 3426
    },
    {
      "epoch": 4.194614443084455,
      "grad_norm": 1.2335648415404112,
      "learning_rate": 4.095731708147483e-06,
      "loss": 1.1685,
      "step": 3427
    },
    {
      "epoch": 4.195838433292534,
      "grad_norm": 1.0835358959382175,
      "learning_rate": 4.095238310461621e-06,
      "loss": 0.6322,
      "step": 3428
    },
    {
      "epoch": 4.197062423500612,
      "grad_norm": 1.5947622658822855,
      "learning_rate": 4.094744807942597e-06,
      "loss": 0.6331,
      "step": 3429
    },
    {
      "epoch": 4.19828641370869,
      "grad_norm": 0.7995320784791059,
      "learning_rate": 4.094251200622843e-06,
      "loss": 0.6132,
      "step": 3430
    },
    {
      "epoch": 4.1995104039167686,
      "grad_norm": 1.5087507718996735,
      "learning_rate": 4.093757488534798e-06,
      "loss": 0.6554,
      "step": 3431
    },
    {
      "epoch": 4.200734394124847,
      "grad_norm": 1.8157717691785868,
      "learning_rate": 4.093263671710904e-06,
      "loss": 0.4954,
      "step": 3432
    },
    {
      "epoch": 4.201958384332926,
      "grad_norm": 0.9672253014071925,
      "learning_rate": 4.092769750183616e-06,
      "loss": 0.3712,
      "step": 3433
    },
    {
      "epoch": 4.2031823745410035,
      "grad_norm": 0.8284608289432331,
      "learning_rate": 4.092275723985391e-06,
      "loss": 0.4072,
      "step": 3434
    },
    {
      "epoch": 4.204406364749082,
      "grad_norm": 1.5687642790434866,
      "learning_rate": 4.091781593148694e-06,
      "loss": 0.5856,
      "step": 3435
    },
    {
      "epoch": 4.205630354957161,
      "grad_norm": 1.369496131557744,
      "learning_rate": 4.0912873577059985e-06,
      "loss": 0.7798,
      "step": 3436
    },
    {
      "epoch": 4.2068543451652385,
      "grad_norm": 2.0075821243778784,
      "learning_rate": 4.090793017689784e-06,
      "loss": 0.4713,
      "step": 3437
    },
    {
      "epoch": 4.208078335373317,
      "grad_norm": 1.6768294033381532,
      "learning_rate": 4.090298573132536e-06,
      "loss": 1.02,
      "step": 3438
    },
    {
      "epoch": 4.209302325581396,
      "grad_norm": 1.6421993835755013,
      "learning_rate": 4.089804024066748e-06,
      "loss": 1.1279,
      "step": 3439
    },
    {
      "epoch": 4.2105263157894735,
      "grad_norm": 0.8451260739628348,
      "learning_rate": 4.089309370524921e-06,
      "loss": 0.5075,
      "step": 3440
    },
    {
      "epoch": 4.211750305997552,
      "grad_norm": 1.9472306033712028,
      "learning_rate": 4.088814612539559e-06,
      "loss": 0.4268,
      "step": 3441
    },
    {
      "epoch": 4.212974296205631,
      "grad_norm": 0.6777799799171045,
      "learning_rate": 4.088319750143178e-06,
      "loss": 0.3665,
      "step": 3442
    },
    {
      "epoch": 4.214198286413708,
      "grad_norm": 1.3530899950932693,
      "learning_rate": 4.087824783368297e-06,
      "loss": 0.5924,
      "step": 3443
    },
    {
      "epoch": 4.215422276621787,
      "grad_norm": 1.4369706998368568,
      "learning_rate": 4.0873297122474454e-06,
      "loss": 0.7302,
      "step": 3444
    },
    {
      "epoch": 4.216646266829866,
      "grad_norm": 1.7171545358324656,
      "learning_rate": 4.086834536813155e-06,
      "loss": 0.605,
      "step": 3445
    },
    {
      "epoch": 4.217870257037943,
      "grad_norm": 0.9464228740453237,
      "learning_rate": 4.086339257097969e-06,
      "loss": 0.608,
      "step": 3446
    },
    {
      "epoch": 4.219094247246022,
      "grad_norm": 1.2424965918270636,
      "learning_rate": 4.0858438731344344e-06,
      "loss": 0.9427,
      "step": 3447
    },
    {
      "epoch": 4.220318237454101,
      "grad_norm": 2.109091276648509,
      "learning_rate": 4.0853483849551066e-06,
      "loss": 0.3645,
      "step": 3448
    },
    {
      "epoch": 4.221542227662178,
      "grad_norm": 0.7792249912409837,
      "learning_rate": 4.0848527925925465e-06,
      "loss": 0.4755,
      "step": 3449
    },
    {
      "epoch": 4.222766217870257,
      "grad_norm": 1.6113886912176316,
      "learning_rate": 4.0843570960793225e-06,
      "loss": 0.9649,
      "step": 3450
    },
    {
      "epoch": 4.223990208078336,
      "grad_norm": 1.6331190714367771,
      "learning_rate": 4.083861295448009e-06,
      "loss": 0.535,
      "step": 3451
    },
    {
      "epoch": 4.225214198286413,
      "grad_norm": 1.3097383220102126,
      "learning_rate": 4.083365390731191e-06,
      "loss": 0.5008,
      "step": 3452
    },
    {
      "epoch": 4.226438188494492,
      "grad_norm": 1.3829409187392667,
      "learning_rate": 4.082869381961456e-06,
      "loss": 0.5825,
      "step": 3453
    },
    {
      "epoch": 4.2276621787025706,
      "grad_norm": 1.486947640461893,
      "learning_rate": 4.082373269171399e-06,
      "loss": 0.3586,
      "step": 3454
    },
    {
      "epoch": 4.228886168910648,
      "grad_norm": 1.411108028014503,
      "learning_rate": 4.081877052393625e-06,
      "loss": 0.949,
      "step": 3455
    },
    {
      "epoch": 4.230110159118727,
      "grad_norm": 1.1847688243642103,
      "learning_rate": 4.081380731660741e-06,
      "loss": 1.122,
      "step": 3456
    },
    {
      "epoch": 4.2313341493268055,
      "grad_norm": 1.387864209249282,
      "learning_rate": 4.0808843070053635e-06,
      "loss": 0.3576,
      "step": 3457
    },
    {
      "epoch": 4.232558139534884,
      "grad_norm": 1.4191590207061333,
      "learning_rate": 4.080387778460118e-06,
      "loss": 0.5776,
      "step": 3458
    },
    {
      "epoch": 4.233782129742962,
      "grad_norm": 2.0934770806531877,
      "learning_rate": 4.079891146057633e-06,
      "loss": 0.4599,
      "step": 3459
    },
    {
      "epoch": 4.2350061199510405,
      "grad_norm": 1.8574710841404967,
      "learning_rate": 4.079394409830546e-06,
      "loss": 0.5427,
      "step": 3460
    },
    {
      "epoch": 4.236230110159119,
      "grad_norm": 0.7244121582766999,
      "learning_rate": 4.078897569811499e-06,
      "loss": 0.4666,
      "step": 3461
    },
    {
      "epoch": 4.237454100367197,
      "grad_norm": 1.3386983665426126,
      "learning_rate": 4.0784006260331445e-06,
      "loss": 0.7447,
      "step": 3462
    },
    {
      "epoch": 4.2386780905752754,
      "grad_norm": 0.9920482941520884,
      "learning_rate": 4.077903578528138e-06,
      "loss": 0.6439,
      "step": 3463
    },
    {
      "epoch": 4.239902080783354,
      "grad_norm": 0.5596841280421857,
      "learning_rate": 4.077406427329146e-06,
      "loss": 0.2684,
      "step": 3464
    },
    {
      "epoch": 4.241126070991432,
      "grad_norm": 1.5672437161005648,
      "learning_rate": 4.076909172468837e-06,
      "loss": 0.5574,
      "step": 3465
    },
    {
      "epoch": 4.24235006119951,
      "grad_norm": 2.4427080534368777,
      "learning_rate": 4.07641181397989e-06,
      "loss": 0.4339,
      "step": 3466
    },
    {
      "epoch": 4.243574051407589,
      "grad_norm": 1.429304713530537,
      "learning_rate": 4.075914351894989e-06,
      "loss": 0.4597,
      "step": 3467
    },
    {
      "epoch": 4.244798041615667,
      "grad_norm": 1.1271992674598865,
      "learning_rate": 4.075416786246827e-06,
      "loss": 0.7863,
      "step": 3468
    },
    {
      "epoch": 4.246022031823745,
      "grad_norm": 1.2279787227145853,
      "learning_rate": 4.074919117068099e-06,
      "loss": 1.5338,
      "step": 3469
    },
    {
      "epoch": 4.247246022031824,
      "grad_norm": 1.0189829791884726,
      "learning_rate": 4.074421344391513e-06,
      "loss": 1.3667,
      "step": 3470
    },
    {
      "epoch": 4.248470012239902,
      "grad_norm": 1.3854844242203366,
      "learning_rate": 4.073923468249779e-06,
      "loss": 0.4376,
      "step": 3471
    },
    {
      "epoch": 4.24969400244798,
      "grad_norm": 1.047002386316789,
      "learning_rate": 4.073425488675616e-06,
      "loss": 1.6038,
      "step": 3472
    },
    {
      "epoch": 4.250917992656059,
      "grad_norm": 0.8297754860802374,
      "learning_rate": 4.072927405701751e-06,
      "loss": 0.6648,
      "step": 3473
    },
    {
      "epoch": 4.252141982864137,
      "grad_norm": 2.144567750764559,
      "learning_rate": 4.0724292193609135e-06,
      "loss": 0.5293,
      "step": 3474
    },
    {
      "epoch": 4.253365973072215,
      "grad_norm": 1.0833715498299052,
      "learning_rate": 4.071930929685844e-06,
      "loss": 0.507,
      "step": 3475
    },
    {
      "epoch": 4.254589963280294,
      "grad_norm": 0.8983121381958039,
      "learning_rate": 4.071432536709289e-06,
      "loss": 0.5549,
      "step": 3476
    },
    {
      "epoch": 4.2558139534883725,
      "grad_norm": 1.3759566348123105,
      "learning_rate": 4.070934040463999e-06,
      "loss": 0.6233,
      "step": 3477
    },
    {
      "epoch": 4.25703794369645,
      "grad_norm": 1.142307852223168,
      "learning_rate": 4.070435440982734e-06,
      "loss": 0.6538,
      "step": 3478
    },
    {
      "epoch": 4.258261933904529,
      "grad_norm": 0.766594137193178,
      "learning_rate": 4.069936738298261e-06,
      "loss": 0.4192,
      "step": 3479
    },
    {
      "epoch": 4.2594859241126075,
      "grad_norm": 0.7839139360070703,
      "learning_rate": 4.069437932443352e-06,
      "loss": 0.3339,
      "step": 3480
    },
    {
      "epoch": 4.260709914320685,
      "grad_norm": 1.5262836408918452,
      "learning_rate": 4.0689390234507875e-06,
      "loss": 0.5941,
      "step": 3481
    },
    {
      "epoch": 4.261933904528764,
      "grad_norm": 1.1269598666494727,
      "learning_rate": 4.068440011353353e-06,
      "loss": 0.7509,
      "step": 3482
    },
    {
      "epoch": 4.2631578947368425,
      "grad_norm": 0.9932017782192541,
      "learning_rate": 4.067940896183843e-06,
      "loss": 0.5659,
      "step": 3483
    },
    {
      "epoch": 4.26438188494492,
      "grad_norm": 0.9330861255005087,
      "learning_rate": 4.067441677975056e-06,
      "loss": 0.4281,
      "step": 3484
    },
    {
      "epoch": 4.265605875152999,
      "grad_norm": 1.014827730118032,
      "learning_rate": 4.0669423567597994e-06,
      "loss": 0.6279,
      "step": 3485
    },
    {
      "epoch": 4.2668298653610774,
      "grad_norm": 1.1683652570801795,
      "learning_rate": 4.0664429325708875e-06,
      "loss": 0.782,
      "step": 3486
    },
    {
      "epoch": 4.268053855569155,
      "grad_norm": 1.1855899801200867,
      "learning_rate": 4.065943405441139e-06,
      "loss": 0.5917,
      "step": 3487
    },
    {
      "epoch": 4.269277845777234,
      "grad_norm": 0.933580731843695,
      "learning_rate": 4.0654437754033835e-06,
      "loss": 0.5342,
      "step": 3488
    },
    {
      "epoch": 4.270501835985312,
      "grad_norm": 1.4792153490519848,
      "learning_rate": 4.064944042490452e-06,
      "loss": 0.6957,
      "step": 3489
    },
    {
      "epoch": 4.27172582619339,
      "grad_norm": 1.1206228670626632,
      "learning_rate": 4.0644442067351865e-06,
      "loss": 0.4982,
      "step": 3490
    },
    {
      "epoch": 4.272949816401469,
      "grad_norm": 0.9720975169060032,
      "learning_rate": 4.063944268170434e-06,
      "loss": 0.4759,
      "step": 3491
    },
    {
      "epoch": 4.274173806609547,
      "grad_norm": 1.1479688551990086,
      "learning_rate": 4.06344422682905e-06,
      "loss": 0.4945,
      "step": 3492
    },
    {
      "epoch": 4.275397796817625,
      "grad_norm": 1.725112556369631,
      "learning_rate": 4.062944082743892e-06,
      "loss": 0.6649,
      "step": 3493
    },
    {
      "epoch": 4.276621787025704,
      "grad_norm": 0.8131653167234579,
      "learning_rate": 4.062443835947832e-06,
      "loss": 0.6471,
      "step": 3494
    },
    {
      "epoch": 4.277845777233782,
      "grad_norm": 1.6194698837032113,
      "learning_rate": 4.061943486473741e-06,
      "loss": 0.6467,
      "step": 3495
    },
    {
      "epoch": 4.27906976744186,
      "grad_norm": 1.620543960602914,
      "learning_rate": 4.061443034354502e-06,
      "loss": 0.5865,
      "step": 3496
    },
    {
      "epoch": 4.280293757649939,
      "grad_norm": 1.0153986580439545,
      "learning_rate": 4.060942479623001e-06,
      "loss": 1.3097,
      "step": 3497
    },
    {
      "epoch": 4.281517747858017,
      "grad_norm": 1.0290774615034324,
      "learning_rate": 4.060441822312135e-06,
      "loss": 0.6023,
      "step": 3498
    },
    {
      "epoch": 4.282741738066095,
      "grad_norm": 0.961940242217773,
      "learning_rate": 4.059941062454804e-06,
      "loss": 0.7442,
      "step": 3499
    },
    {
      "epoch": 4.283965728274174,
      "grad_norm": 1.4534944664298968,
      "learning_rate": 4.059440200083915e-06,
      "loss": 0.3376,
      "step": 3500
    },
    {
      "epoch": 4.285189718482252,
      "grad_norm": 1.7349111096208065,
      "learning_rate": 4.0589392352323855e-06,
      "loss": 0.6714,
      "step": 3501
    },
    {
      "epoch": 4.286413708690331,
      "grad_norm": 1.8315760445316476,
      "learning_rate": 4.058438167933136e-06,
      "loss": 0.4227,
      "step": 3502
    },
    {
      "epoch": 4.287637698898409,
      "grad_norm": 1.5054194428691978,
      "learning_rate": 4.057936998219092e-06,
      "loss": 0.5368,
      "step": 3503
    },
    {
      "epoch": 4.288861689106487,
      "grad_norm": 1.106948346138757,
      "learning_rate": 4.057435726123192e-06,
      "loss": 0.4384,
      "step": 3504
    },
    {
      "epoch": 4.290085679314566,
      "grad_norm": 1.4769796416152725,
      "learning_rate": 4.0569343516783775e-06,
      "loss": 1.1122,
      "step": 3505
    },
    {
      "epoch": 4.291309669522644,
      "grad_norm": 1.3216681098571348,
      "learning_rate": 4.056432874917595e-06,
      "loss": 0.5014,
      "step": 3506
    },
    {
      "epoch": 4.292533659730722,
      "grad_norm": 1.7884435909896423,
      "learning_rate": 4.055931295873802e-06,
      "loss": 0.4698,
      "step": 3507
    },
    {
      "epoch": 4.293757649938801,
      "grad_norm": 0.8345925147903958,
      "learning_rate": 4.055429614579958e-06,
      "loss": 0.5824,
      "step": 3508
    },
    {
      "epoch": 4.2949816401468786,
      "grad_norm": 0.888216589134071,
      "learning_rate": 4.054927831069033e-06,
      "loss": 0.5089,
      "step": 3509
    },
    {
      "epoch": 4.296205630354957,
      "grad_norm": 1.5196138331292581,
      "learning_rate": 4.054425945374002e-06,
      "loss": 0.5286,
      "step": 3510
    },
    {
      "epoch": 4.297429620563036,
      "grad_norm": 1.2083757491180374,
      "learning_rate": 4.053923957527849e-06,
      "loss": 0.8929,
      "step": 3511
    },
    {
      "epoch": 4.2986536107711135,
      "grad_norm": 2.5132247395176486,
      "learning_rate": 4.05342186756356e-06,
      "loss": 0.3437,
      "step": 3512
    },
    {
      "epoch": 4.299877600979192,
      "grad_norm": 1.3451845996235703,
      "learning_rate": 4.052919675514131e-06,
      "loss": 0.6353,
      "step": 3513
    },
    {
      "epoch": 4.301101591187271,
      "grad_norm": 1.2599121127853568,
      "learning_rate": 4.052417381412566e-06,
      "loss": 1.5598,
      "step": 3514
    },
    {
      "epoch": 4.3023255813953485,
      "grad_norm": 1.1800492444399018,
      "learning_rate": 4.051914985291871e-06,
      "loss": 0.7138,
      "step": 3515
    },
    {
      "epoch": 4.303549571603427,
      "grad_norm": 0.7935073376706231,
      "learning_rate": 4.051412487185065e-06,
      "loss": 0.3953,
      "step": 3516
    },
    {
      "epoch": 4.304773561811506,
      "grad_norm": 1.608114098556723,
      "learning_rate": 4.050909887125167e-06,
      "loss": 0.468,
      "step": 3517
    },
    {
      "epoch": 4.3059975520195835,
      "grad_norm": 1.8404854643857707,
      "learning_rate": 4.05040718514521e-06,
      "loss": 0.457,
      "step": 3518
    },
    {
      "epoch": 4.307221542227662,
      "grad_norm": 2.0403252573404984,
      "learning_rate": 4.049904381278225e-06,
      "loss": 0.6587,
      "step": 3519
    },
    {
      "epoch": 4.308445532435741,
      "grad_norm": 1.2575446906565215,
      "learning_rate": 4.049401475557258e-06,
      "loss": 1.2075,
      "step": 3520
    },
    {
      "epoch": 4.309669522643819,
      "grad_norm": 1.3998647528768597,
      "learning_rate": 4.048898468015356e-06,
      "loss": 0.4598,
      "step": 3521
    },
    {
      "epoch": 4.310893512851897,
      "grad_norm": 1.592115232264391,
      "learning_rate": 4.048395358685576e-06,
      "loss": 1.0146,
      "step": 3522
    },
    {
      "epoch": 4.312117503059976,
      "grad_norm": 1.9566816219020757,
      "learning_rate": 4.047892147600981e-06,
      "loss": 0.4113,
      "step": 3523
    },
    {
      "epoch": 4.313341493268053,
      "grad_norm": 1.3201821075294098,
      "learning_rate": 4.047388834794639e-06,
      "loss": 0.5491,
      "step": 3524
    },
    {
      "epoch": 4.314565483476132,
      "grad_norm": 1.7372377854879704,
      "learning_rate": 4.046885420299625e-06,
      "loss": 0.557,
      "step": 3525
    },
    {
      "epoch": 4.315789473684211,
      "grad_norm": 0.6844512826514385,
      "learning_rate": 4.046381904149024e-06,
      "loss": 0.5141,
      "step": 3526
    },
    {
      "epoch": 4.317013463892289,
      "grad_norm": 1.2787514467415213,
      "learning_rate": 4.045878286375924e-06,
      "loss": 0.8583,
      "step": 3527
    },
    {
      "epoch": 4.318237454100367,
      "grad_norm": 1.3925710408764087,
      "learning_rate": 4.0453745670134196e-06,
      "loss": 0.6043,
      "step": 3528
    },
    {
      "epoch": 4.319461444308446,
      "grad_norm": 1.6281328859067759,
      "learning_rate": 4.044870746094616e-06,
      "loss": 0.6,
      "step": 3529
    },
    {
      "epoch": 4.320685434516524,
      "grad_norm": 1.437599956508346,
      "learning_rate": 4.04436682365262e-06,
      "loss": 0.5499,
      "step": 3530
    },
    {
      "epoch": 4.321909424724602,
      "grad_norm": 1.1954408594703136,
      "learning_rate": 4.043862799720548e-06,
      "loss": 0.4729,
      "step": 3531
    },
    {
      "epoch": 4.3231334149326806,
      "grad_norm": 0.8409391778799481,
      "learning_rate": 4.043358674331524e-06,
      "loss": 0.5872,
      "step": 3532
    },
    {
      "epoch": 4.324357405140759,
      "grad_norm": 1.426576184762066,
      "learning_rate": 4.042854447518677e-06,
      "loss": 0.6616,
      "step": 3533
    },
    {
      "epoch": 4.325581395348837,
      "grad_norm": 1.620620996265577,
      "learning_rate": 4.042350119315142e-06,
      "loss": 0.4998,
      "step": 3534
    },
    {
      "epoch": 4.3268053855569155,
      "grad_norm": 1.3354891493421994,
      "learning_rate": 4.041845689754062e-06,
      "loss": 0.6065,
      "step": 3535
    },
    {
      "epoch": 4.328029375764994,
      "grad_norm": 1.0723704400060392,
      "learning_rate": 4.041341158868586e-06,
      "loss": 1.1472,
      "step": 3536
    },
    {
      "epoch": 4.329253365973072,
      "grad_norm": 1.2664066978149722,
      "learning_rate": 4.040836526691869e-06,
      "loss": 0.9893,
      "step": 3537
    },
    {
      "epoch": 4.3304773561811505,
      "grad_norm": 1.0446982657659547,
      "learning_rate": 4.040331793257076e-06,
      "loss": 0.5972,
      "step": 3538
    },
    {
      "epoch": 4.331701346389229,
      "grad_norm": 1.0487017503243619,
      "learning_rate": 4.039826958597375e-06,
      "loss": 1.0585,
      "step": 3539
    },
    {
      "epoch": 4.332925336597307,
      "grad_norm": 1.6976021633629266,
      "learning_rate": 4.0393220227459405e-06,
      "loss": 0.4997,
      "step": 3540
    },
    {
      "epoch": 4.3341493268053854,
      "grad_norm": 1.5195620232837876,
      "learning_rate": 4.0388169857359576e-06,
      "loss": 0.7113,
      "step": 3541
    },
    {
      "epoch": 4.335373317013464,
      "grad_norm": 1.2494599757399052,
      "learning_rate": 4.038311847600613e-06,
      "loss": 0.5068,
      "step": 3542
    },
    {
      "epoch": 4.336597307221542,
      "grad_norm": 0.8985740602952051,
      "learning_rate": 4.0378066083731036e-06,
      "loss": 0.8038,
      "step": 3543
    },
    {
      "epoch": 4.33782129742962,
      "grad_norm": 1.285922981743088,
      "learning_rate": 4.037301268086632e-06,
      "loss": 1.4898,
      "step": 3544
    },
    {
      "epoch": 4.339045287637699,
      "grad_norm": 1.412920823417153,
      "learning_rate": 4.0367958267744074e-06,
      "loss": 0.4765,
      "step": 3545
    },
    {
      "epoch": 4.340269277845778,
      "grad_norm": 1.2796609608491059,
      "learning_rate": 4.036290284469646e-06,
      "loss": 0.4662,
      "step": 3546
    },
    {
      "epoch": 4.341493268053855,
      "grad_norm": 1.6953412521905944,
      "learning_rate": 4.035784641205568e-06,
      "loss": 1.0186,
      "step": 3547
    },
    {
      "epoch": 4.342717258261934,
      "grad_norm": 1.1040115500019598,
      "learning_rate": 4.035278897015404e-06,
      "loss": 1.0227,
      "step": 3548
    },
    {
      "epoch": 4.343941248470013,
      "grad_norm": 0.7755347096952091,
      "learning_rate": 4.034773051932391e-06,
      "loss": 0.6564,
      "step": 3549
    },
    {
      "epoch": 4.34516523867809,
      "grad_norm": 1.862216777053464,
      "learning_rate": 4.034267105989768e-06,
      "loss": 0.4379,
      "step": 3550
    },
    {
      "epoch": 4.346389228886169,
      "grad_norm": 1.5024608944510132,
      "learning_rate": 4.0337610592207865e-06,
      "loss": 0.5282,
      "step": 3551
    },
    {
      "epoch": 4.347613219094248,
      "grad_norm": 1.2395878399464617,
      "learning_rate": 4.0332549116587015e-06,
      "loss": 1.1418,
      "step": 3552
    },
    {
      "epoch": 4.348837209302325,
      "grad_norm": 0.7856914391942512,
      "learning_rate": 4.032748663336774e-06,
      "loss": 0.5583,
      "step": 3553
    },
    {
      "epoch": 4.350061199510404,
      "grad_norm": 1.4145804494652012,
      "learning_rate": 4.032242314288274e-06,
      "loss": 0.394,
      "step": 3554
    },
    {
      "epoch": 4.3512851897184825,
      "grad_norm": 0.7654873544894892,
      "learning_rate": 4.031735864546476e-06,
      "loss": 0.6182,
      "step": 3555
    },
    {
      "epoch": 4.35250917992656,
      "grad_norm": 1.2471136111198433,
      "learning_rate": 4.031229314144664e-06,
      "loss": 1.2098,
      "step": 3556
    },
    {
      "epoch": 4.353733170134639,
      "grad_norm": 1.3264828794325472,
      "learning_rate": 4.030722663116123e-06,
      "loss": 0.6279,
      "step": 3557
    },
    {
      "epoch": 4.3549571603427175,
      "grad_norm": 2.327451740179415,
      "learning_rate": 4.030215911494152e-06,
      "loss": 0.4116,
      "step": 3558
    },
    {
      "epoch": 4.356181150550795,
      "grad_norm": 0.8047543191391845,
      "learning_rate": 4.029709059312051e-06,
      "loss": 0.6042,
      "step": 3559
    },
    {
      "epoch": 4.357405140758874,
      "grad_norm": 1.827528540314403,
      "learning_rate": 4.029202106603127e-06,
      "loss": 0.4513,
      "step": 3560
    },
    {
      "epoch": 4.3586291309669525,
      "grad_norm": 1.2761373589330836,
      "learning_rate": 4.028695053400699e-06,
      "loss": 0.6697,
      "step": 3561
    },
    {
      "epoch": 4.35985312117503,
      "grad_norm": 1.4046801144966863,
      "learning_rate": 4.028187899738085e-06,
      "loss": 0.4542,
      "step": 3562
    },
    {
      "epoch": 4.361077111383109,
      "grad_norm": 1.651847627339744,
      "learning_rate": 4.027680645648615e-06,
      "loss": 0.5958,
      "step": 3563
    },
    {
      "epoch": 4.3623011015911874,
      "grad_norm": 1.418868096745927,
      "learning_rate": 4.027173291165624e-06,
      "loss": 0.4374,
      "step": 3564
    },
    {
      "epoch": 4.363525091799266,
      "grad_norm": 0.992365693141958,
      "learning_rate": 4.026665836322451e-06,
      "loss": 0.4212,
      "step": 3565
    },
    {
      "epoch": 4.364749082007344,
      "grad_norm": 1.6556920889971463,
      "learning_rate": 4.0261582811524476e-06,
      "loss": 0.7317,
      "step": 3566
    },
    {
      "epoch": 4.365973072215422,
      "grad_norm": 1.4573628988271081,
      "learning_rate": 4.025650625688966e-06,
      "loss": 1.0502,
      "step": 3567
    },
    {
      "epoch": 4.3671970624235,
      "grad_norm": 0.6629964440015367,
      "learning_rate": 4.0251428699653696e-06,
      "loss": 0.4811,
      "step": 3568
    },
    {
      "epoch": 4.368421052631579,
      "grad_norm": 1.390779731560753,
      "learning_rate": 4.024635014015023e-06,
      "loss": 1.448,
      "step": 3569
    },
    {
      "epoch": 4.369645042839657,
      "grad_norm": 1.4328647255739637,
      "learning_rate": 4.024127057871304e-06,
      "loss": 0.535,
      "step": 3570
    },
    {
      "epoch": 4.370869033047736,
      "grad_norm": 0.9627311275708256,
      "learning_rate": 4.023619001567591e-06,
      "loss": 0.3993,
      "step": 3571
    },
    {
      "epoch": 4.372093023255814,
      "grad_norm": 1.2725890055915594,
      "learning_rate": 4.0231108451372735e-06,
      "loss": 0.7272,
      "step": 3572
    },
    {
      "epoch": 4.373317013463892,
      "grad_norm": 1.7189231080733645,
      "learning_rate": 4.022602588613745e-06,
      "loss": 0.6308,
      "step": 3573
    },
    {
      "epoch": 4.374541003671971,
      "grad_norm": 0.8691569229991046,
      "learning_rate": 4.022094232030405e-06,
      "loss": 0.6139,
      "step": 3574
    },
    {
      "epoch": 4.375764993880049,
      "grad_norm": 0.9699646502486835,
      "learning_rate": 4.021585775420663e-06,
      "loss": 0.5393,
      "step": 3575
    },
    {
      "epoch": 4.376988984088127,
      "grad_norm": 2.3616833254253438,
      "learning_rate": 4.021077218817932e-06,
      "loss": 0.4239,
      "step": 3576
    },
    {
      "epoch": 4.378212974296206,
      "grad_norm": 1.0801248350464407,
      "learning_rate": 4.020568562255632e-06,
      "loss": 0.6838,
      "step": 3577
    },
    {
      "epoch": 4.379436964504284,
      "grad_norm": 1.553467545054019,
      "learning_rate": 4.0200598057671896e-06,
      "loss": 1.1725,
      "step": 3578
    },
    {
      "epoch": 4.380660954712362,
      "grad_norm": 1.7985942362402787,
      "learning_rate": 4.01955094938604e-06,
      "loss": 0.4595,
      "step": 3579
    },
    {
      "epoch": 4.381884944920441,
      "grad_norm": 1.1453951637626651,
      "learning_rate": 4.019041993145621e-06,
      "loss": 0.7387,
      "step": 3580
    },
    {
      "epoch": 4.383108935128519,
      "grad_norm": 0.9921812991799427,
      "learning_rate": 4.018532937079382e-06,
      "loss": 0.5395,
      "step": 3581
    },
    {
      "epoch": 4.384332925336597,
      "grad_norm": 1.1124886402031469,
      "learning_rate": 4.018023781220774e-06,
      "loss": 0.5109,
      "step": 3582
    },
    {
      "epoch": 4.385556915544676,
      "grad_norm": 2.3431915781992627,
      "learning_rate": 4.0175145256032595e-06,
      "loss": 0.4031,
      "step": 3583
    },
    {
      "epoch": 4.386780905752754,
      "grad_norm": 2.227003772680972,
      "learning_rate": 4.017005170260302e-06,
      "loss": 0.7452,
      "step": 3584
    },
    {
      "epoch": 4.388004895960832,
      "grad_norm": 1.6844306798499544,
      "learning_rate": 4.016495715225376e-06,
      "loss": 0.4593,
      "step": 3585
    },
    {
      "epoch": 4.389228886168911,
      "grad_norm": 1.2056995378456405,
      "learning_rate": 4.015986160531961e-06,
      "loss": 0.6979,
      "step": 3586
    },
    {
      "epoch": 4.3904528763769886,
      "grad_norm": 1.1730794659066093,
      "learning_rate": 4.015476506213543e-06,
      "loss": 0.7654,
      "step": 3587
    },
    {
      "epoch": 4.391676866585067,
      "grad_norm": 0.7788475080742777,
      "learning_rate": 4.014966752303614e-06,
      "loss": 0.5578,
      "step": 3588
    },
    {
      "epoch": 4.392900856793146,
      "grad_norm": 1.7458378660974556,
      "learning_rate": 4.014456898835674e-06,
      "loss": 0.7861,
      "step": 3589
    },
    {
      "epoch": 4.394124847001224,
      "grad_norm": 1.6898855550374514,
      "learning_rate": 4.0139469458432265e-06,
      "loss": 0.4956,
      "step": 3590
    },
    {
      "epoch": 4.395348837209302,
      "grad_norm": 1.2642617946827694,
      "learning_rate": 4.013436893359787e-06,
      "loss": 0.7211,
      "step": 3591
    },
    {
      "epoch": 4.396572827417381,
      "grad_norm": 0.8442796532347824,
      "learning_rate": 4.012926741418872e-06,
      "loss": 0.4974,
      "step": 3592
    },
    {
      "epoch": 4.397796817625459,
      "grad_norm": 0.9888909606818186,
      "learning_rate": 4.012416490054008e-06,
      "loss": 0.4604,
      "step": 3593
    },
    {
      "epoch": 4.399020807833537,
      "grad_norm": 1.0115358085385253,
      "learning_rate": 4.0119061392987265e-06,
      "loss": 0.5698,
      "step": 3594
    },
    {
      "epoch": 4.400244798041616,
      "grad_norm": 1.8113731223538332,
      "learning_rate": 4.011395689186564e-06,
      "loss": 0.7243,
      "step": 3595
    },
    {
      "epoch": 4.401468788249694,
      "grad_norm": 1.930101710005665,
      "learning_rate": 4.01088513975107e-06,
      "loss": 1.0018,
      "step": 3596
    },
    {
      "epoch": 4.402692778457772,
      "grad_norm": 1.1706397650678757,
      "learning_rate": 4.01037449102579e-06,
      "loss": 0.6053,
      "step": 3597
    },
    {
      "epoch": 4.403916768665851,
      "grad_norm": 1.2810271140811518,
      "learning_rate": 4.009863743044287e-06,
      "loss": 0.3338,
      "step": 3598
    },
    {
      "epoch": 4.405140758873929,
      "grad_norm": 0.9004097764031004,
      "learning_rate": 4.009352895840122e-06,
      "loss": 0.6275,
      "step": 3599
    },
    {
      "epoch": 4.406364749082007,
      "grad_norm": 0.7414546760690397,
      "learning_rate": 4.008841949446869e-06,
      "loss": 0.5377,
      "step": 3600
    },
    {
      "epoch": 4.407588739290086,
      "grad_norm": 1.1185530095191238,
      "learning_rate": 4.0083309038981025e-06,
      "loss": 0.3229,
      "step": 3601
    },
    {
      "epoch": 4.408812729498164,
      "grad_norm": 1.5488588709265267,
      "learning_rate": 4.007819759227408e-06,
      "loss": 0.4197,
      "step": 3602
    },
    {
      "epoch": 4.410036719706242,
      "grad_norm": 1.2784032862781616,
      "learning_rate": 4.007308515468376e-06,
      "loss": 0.6156,
      "step": 3603
    },
    {
      "epoch": 4.411260709914321,
      "grad_norm": 1.261201744643915,
      "learning_rate": 4.006797172654605e-06,
      "loss": 0.6823,
      "step": 3604
    },
    {
      "epoch": 4.412484700122399,
      "grad_norm": 1.3418589445815,
      "learning_rate": 4.006285730819696e-06,
      "loss": 1.6504,
      "step": 3605
    },
    {
      "epoch": 4.413708690330477,
      "grad_norm": 1.039389779978606,
      "learning_rate": 4.0057741899972594e-06,
      "loss": 0.596,
      "step": 3606
    },
    {
      "epoch": 4.414932680538556,
      "grad_norm": 1.886447070364013,
      "learning_rate": 4.005262550220913e-06,
      "loss": 0.5346,
      "step": 3607
    },
    {
      "epoch": 4.416156670746634,
      "grad_norm": 1.5473484636639312,
      "learning_rate": 4.00475081152428e-06,
      "loss": 0.4917,
      "step": 3608
    },
    {
      "epoch": 4.417380660954713,
      "grad_norm": 1.4444151557110954,
      "learning_rate": 4.00423897394099e-06,
      "loss": 0.5257,
      "step": 3609
    },
    {
      "epoch": 4.4186046511627906,
      "grad_norm": 1.661021306496762,
      "learning_rate": 4.003727037504676e-06,
      "loss": 0.8655,
      "step": 3610
    },
    {
      "epoch": 4.419828641370869,
      "grad_norm": 1.5116152651407901,
      "learning_rate": 4.0032150022489845e-06,
      "loss": 0.5933,
      "step": 3611
    },
    {
      "epoch": 4.421052631578947,
      "grad_norm": 1.3704728439465184,
      "learning_rate": 4.002702868207563e-06,
      "loss": 0.7896,
      "step": 3612
    },
    {
      "epoch": 4.4222766217870255,
      "grad_norm": 1.2737318254619006,
      "learning_rate": 4.002190635414067e-06,
      "loss": 1.588,
      "step": 3613
    },
    {
      "epoch": 4.423500611995104,
      "grad_norm": 1.9018429020882703,
      "learning_rate": 4.001678303902159e-06,
      "loss": 1.037,
      "step": 3614
    },
    {
      "epoch": 4.424724602203183,
      "grad_norm": 1.5761670495683102,
      "learning_rate": 4.001165873705507e-06,
      "loss": 0.6183,
      "step": 3615
    },
    {
      "epoch": 4.4259485924112605,
      "grad_norm": 1.5197640414387439,
      "learning_rate": 4.000653344857786e-06,
      "loss": 0.5781,
      "step": 3616
    },
    {
      "epoch": 4.427172582619339,
      "grad_norm": 1.4745453168520009,
      "learning_rate": 4.000140717392678e-06,
      "loss": 0.7998,
      "step": 3617
    },
    {
      "epoch": 4.428396572827418,
      "grad_norm": 1.817922589326754,
      "learning_rate": 3.99962799134387e-06,
      "loss": 1.0689,
      "step": 3618
    },
    {
      "epoch": 4.4296205630354955,
      "grad_norm": 1.968291880513057,
      "learning_rate": 3.9991151667450575e-06,
      "loss": 0.263,
      "step": 3619
    },
    {
      "epoch": 4.430844553243574,
      "grad_norm": 1.4127543324676335,
      "learning_rate": 3.998602243629942e-06,
      "loss": 1.1487,
      "step": 3620
    },
    {
      "epoch": 4.432068543451653,
      "grad_norm": 1.0295903161935245,
      "learning_rate": 3.998089222032229e-06,
      "loss": 0.6157,
      "step": 3621
    },
    {
      "epoch": 4.43329253365973,
      "grad_norm": 1.3148182471602252,
      "learning_rate": 3.997576101985633e-06,
      "loss": 1.2332,
      "step": 3622
    },
    {
      "epoch": 4.434516523867809,
      "grad_norm": 1.4585680798062761,
      "learning_rate": 3.997062883523876e-06,
      "loss": 0.6763,
      "step": 3623
    },
    {
      "epoch": 4.435740514075888,
      "grad_norm": 1.3542433363694415,
      "learning_rate": 3.996549566680682e-06,
      "loss": 0.5024,
      "step": 3624
    },
    {
      "epoch": 4.436964504283965,
      "grad_norm": 1.3869931546937466,
      "learning_rate": 3.996036151489787e-06,
      "loss": 0.786,
      "step": 3625
    },
    {
      "epoch": 4.438188494492044,
      "grad_norm": 1.770040399791114,
      "learning_rate": 3.995522637984929e-06,
      "loss": 0.3731,
      "step": 3626
    },
    {
      "epoch": 4.439412484700123,
      "grad_norm": 1.15622384786685,
      "learning_rate": 3.995009026199855e-06,
      "loss": 0.8507,
      "step": 3627
    },
    {
      "epoch": 4.4406364749082,
      "grad_norm": 0.8362963696211706,
      "learning_rate": 3.994495316168317e-06,
      "loss": 0.64,
      "step": 3628
    },
    {
      "epoch": 4.441860465116279,
      "grad_norm": 1.1124488127464958,
      "learning_rate": 3.993981507924076e-06,
      "loss": 1.1486,
      "step": 3629
    },
    {
      "epoch": 4.443084455324358,
      "grad_norm": 2.10012908019713,
      "learning_rate": 3.993467601500895e-06,
      "loss": 0.5542,
      "step": 3630
    },
    {
      "epoch": 4.444308445532435,
      "grad_norm": 1.1256433187938075,
      "learning_rate": 3.992953596932547e-06,
      "loss": 0.9638,
      "step": 3631
    },
    {
      "epoch": 4.445532435740514,
      "grad_norm": 1.9763330937865309,
      "learning_rate": 3.992439494252811e-06,
      "loss": 0.4089,
      "step": 3632
    },
    {
      "epoch": 4.4467564259485926,
      "grad_norm": 1.2676336801101187,
      "learning_rate": 3.991925293495472e-06,
      "loss": 0.5719,
      "step": 3633
    },
    {
      "epoch": 4.447980416156671,
      "grad_norm": 1.0898721623126888,
      "learning_rate": 3.991410994694321e-06,
      "loss": 0.7293,
      "step": 3634
    },
    {
      "epoch": 4.449204406364749,
      "grad_norm": 1.1974059705846383,
      "learning_rate": 3.990896597883155e-06,
      "loss": 0.494,
      "step": 3635
    },
    {
      "epoch": 4.4504283965728275,
      "grad_norm": 1.1514282781753624,
      "learning_rate": 3.99038210309578e-06,
      "loss": 0.4682,
      "step": 3636
    },
    {
      "epoch": 4.451652386780906,
      "grad_norm": 1.2575677823320714,
      "learning_rate": 3.989867510366005e-06,
      "loss": 0.3479,
      "step": 3637
    },
    {
      "epoch": 4.452876376988984,
      "grad_norm": 1.7847863323387223,
      "learning_rate": 3.989352819727648e-06,
      "loss": 1.0706,
      "step": 3638
    },
    {
      "epoch": 4.4541003671970625,
      "grad_norm": 1.1090017349857237,
      "learning_rate": 3.9888380312145325e-06,
      "loss": 0.6055,
      "step": 3639
    },
    {
      "epoch": 4.455324357405141,
      "grad_norm": 0.9335024202340734,
      "learning_rate": 3.988323144860489e-06,
      "loss": 0.5889,
      "step": 3640
    },
    {
      "epoch": 4.456548347613219,
      "grad_norm": 1.8505894156199396,
      "learning_rate": 3.987808160699352e-06,
      "loss": 0.9464,
      "step": 3641
    },
    {
      "epoch": 4.4577723378212974,
      "grad_norm": 1.3310560935409077,
      "learning_rate": 3.987293078764967e-06,
      "loss": 0.696,
      "step": 3642
    },
    {
      "epoch": 4.458996328029376,
      "grad_norm": 0.9083108131637816,
      "learning_rate": 3.986777899091181e-06,
      "loss": 0.6771,
      "step": 3643
    },
    {
      "epoch": 4.460220318237454,
      "grad_norm": 1.0905031357219217,
      "learning_rate": 3.986262621711851e-06,
      "loss": 0.7395,
      "step": 3644
    },
    {
      "epoch": 4.461444308445532,
      "grad_norm": 1.6027956660567728,
      "learning_rate": 3.9857472466608385e-06,
      "loss": 0.4268,
      "step": 3645
    },
    {
      "epoch": 4.462668298653611,
      "grad_norm": 0.9896148563635758,
      "learning_rate": 3.985231773972013e-06,
      "loss": 1.0058,
      "step": 3646
    },
    {
      "epoch": 4.463892288861689,
      "grad_norm": 1.141940114086837,
      "learning_rate": 3.984716203679249e-06,
      "loss": 0.6109,
      "step": 3647
    },
    {
      "epoch": 4.465116279069767,
      "grad_norm": 1.6822239158745074,
      "learning_rate": 3.984200535816427e-06,
      "loss": 0.4556,
      "step": 3648
    },
    {
      "epoch": 4.466340269277846,
      "grad_norm": 1.2718344533625352,
      "learning_rate": 3.983684770417435e-06,
      "loss": 1.3969,
      "step": 3649
    },
    {
      "epoch": 4.467564259485924,
      "grad_norm": 1.5064282310821562,
      "learning_rate": 3.9831689075161685e-06,
      "loss": 1.1701,
      "step": 3650
    },
    {
      "epoch": 4.468788249694002,
      "grad_norm": 1.9655783002905527,
      "learning_rate": 3.982652947146528e-06,
      "loss": 0.6239,
      "step": 3651
    },
    {
      "epoch": 4.470012239902081,
      "grad_norm": 0.7575334637802201,
      "learning_rate": 3.982136889342418e-06,
      "loss": 0.5664,
      "step": 3652
    },
    {
      "epoch": 4.47123623011016,
      "grad_norm": 1.105158793899495,
      "learning_rate": 3.981620734137754e-06,
      "loss": 0.5987,
      "step": 3653
    },
    {
      "epoch": 4.472460220318237,
      "grad_norm": 1.5281857008302024,
      "learning_rate": 3.981104481566457e-06,
      "loss": 0.4535,
      "step": 3654
    },
    {
      "epoch": 4.473684210526316,
      "grad_norm": 1.427785779713977,
      "learning_rate": 3.980588131662451e-06,
      "loss": 1.1497,
      "step": 3655
    },
    {
      "epoch": 4.474908200734394,
      "grad_norm": 1.1652391245463396,
      "learning_rate": 3.980071684459669e-06,
      "loss": 1.5117,
      "step": 3656
    },
    {
      "epoch": 4.476132190942472,
      "grad_norm": 1.747039309138808,
      "learning_rate": 3.979555139992049e-06,
      "loss": 0.7611,
      "step": 3657
    },
    {
      "epoch": 4.477356181150551,
      "grad_norm": 1.8283568115877316,
      "learning_rate": 3.97903849829354e-06,
      "loss": 0.5202,
      "step": 3658
    },
    {
      "epoch": 4.4785801713586295,
      "grad_norm": 0.8718840053301404,
      "learning_rate": 3.978521759398091e-06,
      "loss": 0.6008,
      "step": 3659
    },
    {
      "epoch": 4.479804161566707,
      "grad_norm": 2.0532145528278765,
      "learning_rate": 3.9780049233396586e-06,
      "loss": 0.5861,
      "step": 3660
    },
    {
      "epoch": 4.481028151774786,
      "grad_norm": 1.2603144348795186,
      "learning_rate": 3.977487990152212e-06,
      "loss": 0.6706,
      "step": 3661
    },
    {
      "epoch": 4.4822521419828645,
      "grad_norm": 1.33611713037046,
      "learning_rate": 3.976970959869718e-06,
      "loss": 0.4401,
      "step": 3662
    },
    {
      "epoch": 4.483476132190942,
      "grad_norm": 1.1746110037926547,
      "learning_rate": 3.976453832526156e-06,
      "loss": 0.6388,
      "step": 3663
    },
    {
      "epoch": 4.484700122399021,
      "grad_norm": 0.9883032434439264,
      "learning_rate": 3.9759366081555095e-06,
      "loss": 0.3938,
      "step": 3664
    },
    {
      "epoch": 4.4859241126070994,
      "grad_norm": 1.1445576484014797,
      "learning_rate": 3.975419286791767e-06,
      "loss": 0.5351,
      "step": 3665
    },
    {
      "epoch": 4.487148102815177,
      "grad_norm": 1.4285398907907503,
      "learning_rate": 3.974901868468928e-06,
      "loss": 0.7645,
      "step": 3666
    },
    {
      "epoch": 4.488372093023256,
      "grad_norm": 1.011930758837756,
      "learning_rate": 3.9743843532209925e-06,
      "loss": 0.435,
      "step": 3667
    },
    {
      "epoch": 4.489596083231334,
      "grad_norm": 0.8869831204927647,
      "learning_rate": 3.9738667410819705e-06,
      "loss": 0.5305,
      "step": 3668
    },
    {
      "epoch": 4.490820073439412,
      "grad_norm": 1.1461834333010872,
      "learning_rate": 3.973349032085878e-06,
      "loss": 1.2793,
      "step": 3669
    },
    {
      "epoch": 4.492044063647491,
      "grad_norm": 1.1088947583350852,
      "learning_rate": 3.9728312262667365e-06,
      "loss": 0.8117,
      "step": 3670
    },
    {
      "epoch": 4.493268053855569,
      "grad_norm": 0.8015518807299583,
      "learning_rate": 3.972313323658574e-06,
      "loss": 0.5926,
      "step": 3671
    },
    {
      "epoch": 4.494492044063647,
      "grad_norm": 1.390417370387761,
      "learning_rate": 3.971795324295427e-06,
      "loss": 0.5398,
      "step": 3672
    },
    {
      "epoch": 4.495716034271726,
      "grad_norm": 1.5118547577608668,
      "learning_rate": 3.971277228211333e-06,
      "loss": 1.2458,
      "step": 3673
    },
    {
      "epoch": 4.496940024479804,
      "grad_norm": 2.089139951291609,
      "learning_rate": 3.970759035440343e-06,
      "loss": 1.0851,
      "step": 3674
    },
    {
      "epoch": 4.498164014687882,
      "grad_norm": 1.2678083607450439,
      "learning_rate": 3.970240746016509e-06,
      "loss": 0.743,
      "step": 3675
    },
    {
      "epoch": 4.499388004895961,
      "grad_norm": 1.44325398664822,
      "learning_rate": 3.969722359973891e-06,
      "loss": 0.6487,
      "step": 3676
    },
    {
      "epoch": 4.500611995104039,
      "grad_norm": 1.3287037094770242,
      "learning_rate": 3.9692038773465555e-06,
      "loss": 0.4548,
      "step": 3677
    },
    {
      "epoch": 4.501835985312118,
      "grad_norm": 1.0799204115906897,
      "learning_rate": 3.968685298168575e-06,
      "loss": 0.61,
      "step": 3678
    },
    {
      "epoch": 4.503059975520196,
      "grad_norm": 2.172717161569549,
      "learning_rate": 3.96816662247403e-06,
      "loss": 0.7548,
      "step": 3679
    },
    {
      "epoch": 4.504283965728274,
      "grad_norm": 1.4032650146384935,
      "learning_rate": 3.967647850297004e-06,
      "loss": 0.5352,
      "step": 3680
    },
    {
      "epoch": 4.505507955936353,
      "grad_norm": 1.2932785544942034,
      "learning_rate": 3.96712898167159e-06,
      "loss": 1.035,
      "step": 3681
    },
    {
      "epoch": 4.506731946144431,
      "grad_norm": 1.5527720864591728,
      "learning_rate": 3.966610016631886e-06,
      "loss": 0.6481,
      "step": 3682
    },
    {
      "epoch": 4.507955936352509,
      "grad_norm": 0.7342824994726488,
      "learning_rate": 3.966090955211997e-06,
      "loss": 0.4623,
      "step": 3683
    },
    {
      "epoch": 4.509179926560588,
      "grad_norm": 1.2537774446187413,
      "learning_rate": 3.965571797446031e-06,
      "loss": 1.0461,
      "step": 3684
    },
    {
      "epoch": 4.510403916768666,
      "grad_norm": 1.642761420370547,
      "learning_rate": 3.965052543368109e-06,
      "loss": 0.4699,
      "step": 3685
    },
    {
      "epoch": 4.511627906976744,
      "grad_norm": 1.2744274405932199,
      "learning_rate": 3.9645331930123535e-06,
      "loss": 0.4274,
      "step": 3686
    },
    {
      "epoch": 4.512851897184823,
      "grad_norm": 2.2963281559593125,
      "learning_rate": 3.964013746412892e-06,
      "loss": 0.4186,
      "step": 3687
    },
    {
      "epoch": 4.5140758873929006,
      "grad_norm": 2.3260734224517488,
      "learning_rate": 3.963494203603863e-06,
      "loss": 0.7486,
      "step": 3688
    },
    {
      "epoch": 4.515299877600979,
      "grad_norm": 1.043860105498053,
      "learning_rate": 3.962974564619407e-06,
      "loss": 0.5111,
      "step": 3689
    },
    {
      "epoch": 4.516523867809058,
      "grad_norm": 0.847787523239775,
      "learning_rate": 3.962454829493674e-06,
      "loss": 0.5321,
      "step": 3690
    },
    {
      "epoch": 4.5177478580171355,
      "grad_norm": 1.3146116430504347,
      "learning_rate": 3.9619349982608195e-06,
      "loss": 0.7516,
      "step": 3691
    },
    {
      "epoch": 4.518971848225214,
      "grad_norm": 1.3034744796890867,
      "learning_rate": 3.961415070955004e-06,
      "loss": 0.6528,
      "step": 3692
    },
    {
      "epoch": 4.520195838433293,
      "grad_norm": 1.8146447921750692,
      "learning_rate": 3.960895047610395e-06,
      "loss": 0.4818,
      "step": 3693
    },
    {
      "epoch": 4.5214198286413705,
      "grad_norm": 0.8947358107939535,
      "learning_rate": 3.960374928261167e-06,
      "loss": 1.0634,
      "step": 3694
    },
    {
      "epoch": 4.522643818849449,
      "grad_norm": 1.935443718070234,
      "learning_rate": 3.959854712941501e-06,
      "loss": 0.9922,
      "step": 3695
    },
    {
      "epoch": 4.523867809057528,
      "grad_norm": 1.2932376877380862,
      "learning_rate": 3.959334401685583e-06,
      "loss": 0.45,
      "step": 3696
    },
    {
      "epoch": 4.525091799265606,
      "grad_norm": 1.0474783200071196,
      "learning_rate": 3.958813994527604e-06,
      "loss": 0.3834,
      "step": 3697
    },
    {
      "epoch": 4.526315789473684,
      "grad_norm": 1.2057585543058231,
      "learning_rate": 3.958293491501767e-06,
      "loss": 1.0805,
      "step": 3698
    },
    {
      "epoch": 4.527539779681763,
      "grad_norm": 1.663606794675035,
      "learning_rate": 3.9577728926422745e-06,
      "loss": 0.6885,
      "step": 3699
    },
    {
      "epoch": 4.52876376988984,
      "grad_norm": 1.3626904648788982,
      "learning_rate": 3.95725219798334e-06,
      "loss": 1.5006,
      "step": 3700
    },
    {
      "epoch": 4.529987760097919,
      "grad_norm": 2.0576813548332966,
      "learning_rate": 3.956731407559181e-06,
      "loss": 0.5351,
      "step": 3701
    },
    {
      "epoch": 4.531211750305998,
      "grad_norm": 1.6011797157249987,
      "learning_rate": 3.956210521404021e-06,
      "loss": 1.2377,
      "step": 3702
    },
    {
      "epoch": 4.532435740514076,
      "grad_norm": 2.5595426565274826,
      "learning_rate": 3.955689539552093e-06,
      "loss": 0.8136,
      "step": 3703
    },
    {
      "epoch": 4.533659730722154,
      "grad_norm": 1.077123640043357,
      "learning_rate": 3.955168462037632e-06,
      "loss": 1.3071,
      "step": 3704
    },
    {
      "epoch": 4.534883720930233,
      "grad_norm": 1.6483113457834127,
      "learning_rate": 3.9546472888948825e-06,
      "loss": 0.5648,
      "step": 3705
    },
    {
      "epoch": 4.536107711138311,
      "grad_norm": 1.1620760502047276,
      "learning_rate": 3.9541260201580935e-06,
      "loss": 1.0131,
      "step": 3706
    },
    {
      "epoch": 4.537331701346389,
      "grad_norm": 1.1735105288903187,
      "learning_rate": 3.953604655861521e-06,
      "loss": 0.7584,
      "step": 3707
    },
    {
      "epoch": 4.538555691554468,
      "grad_norm": 1.1327440399807838,
      "learning_rate": 3.953083196039426e-06,
      "loss": 0.4971,
      "step": 3708
    },
    {
      "epoch": 4.539779681762546,
      "grad_norm": 1.505653623243857,
      "learning_rate": 3.952561640726079e-06,
      "loss": 0.8068,
      "step": 3709
    },
    {
      "epoch": 4.541003671970624,
      "grad_norm": 1.7938754062355948,
      "learning_rate": 3.952039989955753e-06,
      "loss": 0.5212,
      "step": 3710
    },
    {
      "epoch": 4.5422276621787026,
      "grad_norm": 0.8886713087598161,
      "learning_rate": 3.95151824376273e-06,
      "loss": 0.5658,
      "step": 3711
    },
    {
      "epoch": 4.543451652386781,
      "grad_norm": 1.683199939295174,
      "learning_rate": 3.950996402181297e-06,
      "loss": 0.4225,
      "step": 3712
    },
    {
      "epoch": 4.544675642594859,
      "grad_norm": 1.0390369755137348,
      "learning_rate": 3.950474465245746e-06,
      "loss": 0.5781,
      "step": 3713
    },
    {
      "epoch": 4.5458996328029375,
      "grad_norm": 1.3481795463395334,
      "learning_rate": 3.949952432990379e-06,
      "loss": 0.5726,
      "step": 3714
    },
    {
      "epoch": 4.547123623011016,
      "grad_norm": 1.6910199693247343,
      "learning_rate": 3.949430305449501e-06,
      "loss": 0.5175,
      "step": 3715
    },
    {
      "epoch": 4.548347613219094,
      "grad_norm": 1.5392987081630878,
      "learning_rate": 3.9489080826574245e-06,
      "loss": 0.8514,
      "step": 3716
    },
    {
      "epoch": 4.5495716034271725,
      "grad_norm": 0.8336592507440458,
      "learning_rate": 3.948385764648468e-06,
      "loss": 0.5237,
      "step": 3717
    },
    {
      "epoch": 4.550795593635251,
      "grad_norm": 1.0636381659524086,
      "learning_rate": 3.947863351456956e-06,
      "loss": 0.3582,
      "step": 3718
    },
    {
      "epoch": 4.552019583843329,
      "grad_norm": 1.0992476842434735,
      "learning_rate": 3.94734084311722e-06,
      "loss": 1.1818,
      "step": 3719
    },
    {
      "epoch": 4.5532435740514074,
      "grad_norm": 1.431149950806469,
      "learning_rate": 3.946818239663598e-06,
      "loss": 0.4225,
      "step": 3720
    },
    {
      "epoch": 4.554467564259486,
      "grad_norm": 2.0632024548857557,
      "learning_rate": 3.946295541130432e-06,
      "loss": 0.8785,
      "step": 3721
    },
    {
      "epoch": 4.555691554467565,
      "grad_norm": 1.0451388694186405,
      "learning_rate": 3.945772747552072e-06,
      "loss": 1.1857,
      "step": 3722
    },
    {
      "epoch": 4.556915544675642,
      "grad_norm": 1.765309704737816,
      "learning_rate": 3.945249858962874e-06,
      "loss": 0.5794,
      "step": 3723
    },
    {
      "epoch": 4.558139534883721,
      "grad_norm": 1.0248451296827104,
      "learning_rate": 3.944726875397202e-06,
      "loss": 0.711,
      "step": 3724
    },
    {
      "epoch": 4.5593635250918,
      "grad_norm": 1.0354915690311854,
      "learning_rate": 3.944203796889424e-06,
      "loss": 0.9002,
      "step": 3725
    },
    {
      "epoch": 4.560587515299877,
      "grad_norm": 1.4392051955646605,
      "learning_rate": 3.943680623473913e-06,
      "loss": 0.6361,
      "step": 3726
    },
    {
      "epoch": 4.561811505507956,
      "grad_norm": 1.0365845466261265,
      "learning_rate": 3.943157355185052e-06,
      "loss": 0.4736,
      "step": 3727
    },
    {
      "epoch": 4.563035495716035,
      "grad_norm": 1.3884461773828494,
      "learning_rate": 3.942633992057227e-06,
      "loss": 0.6765,
      "step": 3728
    },
    {
      "epoch": 4.564259485924112,
      "grad_norm": 0.953546995569663,
      "learning_rate": 3.9421105341248326e-06,
      "loss": 0.5675,
      "step": 3729
    },
    {
      "epoch": 4.565483476132191,
      "grad_norm": 0.9598530907743082,
      "learning_rate": 3.941586981422268e-06,
      "loss": 0.799,
      "step": 3730
    },
    {
      "epoch": 4.56670746634027,
      "grad_norm": 2.5541100858255854,
      "learning_rate": 3.941063333983937e-06,
      "loss": 0.4791,
      "step": 3731
    },
    {
      "epoch": 4.567931456548347,
      "grad_norm": 0.9892782420166751,
      "learning_rate": 3.9405395918442555e-06,
      "loss": 0.3388,
      "step": 3732
    },
    {
      "epoch": 4.569155446756426,
      "grad_norm": 1.0580425809732712,
      "learning_rate": 3.9400157550376405e-06,
      "loss": 0.702,
      "step": 3733
    },
    {
      "epoch": 4.5703794369645045,
      "grad_norm": 0.9606595526163828,
      "learning_rate": 3.939491823598516e-06,
      "loss": 0.7053,
      "step": 3734
    },
    {
      "epoch": 4.571603427172582,
      "grad_norm": 1.901723517022399,
      "learning_rate": 3.9389677975613125e-06,
      "loss": 0.5637,
      "step": 3735
    },
    {
      "epoch": 4.572827417380661,
      "grad_norm": 2.0161764481894364,
      "learning_rate": 3.938443676960468e-06,
      "loss": 0.4125,
      "step": 3736
    },
    {
      "epoch": 4.5740514075887395,
      "grad_norm": 1.0267248598265035,
      "learning_rate": 3.937919461830426e-06,
      "loss": 0.7449,
      "step": 3737
    },
    {
      "epoch": 4.575275397796817,
      "grad_norm": 1.652689948333433,
      "learning_rate": 3.937395152205635e-06,
      "loss": 0.4145,
      "step": 3738
    },
    {
      "epoch": 4.576499388004896,
      "grad_norm": 2.6069325541103323,
      "learning_rate": 3.936870748120551e-06,
      "loss": 0.5529,
      "step": 3739
    },
    {
      "epoch": 4.5777233782129745,
      "grad_norm": 1.2912407266783907,
      "learning_rate": 3.936346249609637e-06,
      "loss": 0.4235,
      "step": 3740
    },
    {
      "epoch": 4.578947368421053,
      "grad_norm": 0.7648604980040093,
      "learning_rate": 3.935821656707359e-06,
      "loss": 0.6173,
      "step": 3741
    },
    {
      "epoch": 4.580171358629131,
      "grad_norm": 1.2345978671920412,
      "learning_rate": 3.935296969448194e-06,
      "loss": 0.4448,
      "step": 3742
    },
    {
      "epoch": 4.5813953488372094,
      "grad_norm": 0.8770647907634886,
      "learning_rate": 3.93477218786662e-06,
      "loss": 1.2236,
      "step": 3743
    },
    {
      "epoch": 4.582619339045287,
      "grad_norm": 1.1760844731339164,
      "learning_rate": 3.934247311997124e-06,
      "loss": 0.5542,
      "step": 3744
    },
    {
      "epoch": 4.583843329253366,
      "grad_norm": 1.071643161927744,
      "learning_rate": 3.9337223418742e-06,
      "loss": 0.5892,
      "step": 3745
    },
    {
      "epoch": 4.585067319461444,
      "grad_norm": 1.802753013838483,
      "learning_rate": 3.9331972775323476e-06,
      "loss": 0.6547,
      "step": 3746
    },
    {
      "epoch": 4.586291309669523,
      "grad_norm": 2.0056137507570395,
      "learning_rate": 3.93267211900607e-06,
      "loss": 0.5155,
      "step": 3747
    },
    {
      "epoch": 4.587515299877601,
      "grad_norm": 0.8851447932318773,
      "learning_rate": 3.93214686632988e-06,
      "loss": 0.6979,
      "step": 3748
    },
    {
      "epoch": 4.588739290085679,
      "grad_norm": 0.8159019297972027,
      "learning_rate": 3.931621519538296e-06,
      "loss": 0.621,
      "step": 3749
    },
    {
      "epoch": 4.589963280293758,
      "grad_norm": 1.0239211761741196,
      "learning_rate": 3.93109607866584e-06,
      "loss": 0.7306,
      "step": 3750
    },
    {
      "epoch": 4.591187270501836,
      "grad_norm": 1.4856012518209956,
      "learning_rate": 3.930570543747043e-06,
      "loss": 0.4634,
      "step": 3751
    },
    {
      "epoch": 4.592411260709914,
      "grad_norm": 1.1101546374249718,
      "learning_rate": 3.930044914816441e-06,
      "loss": 1.2635,
      "step": 3752
    },
    {
      "epoch": 4.593635250917993,
      "grad_norm": 1.1751828589349984,
      "learning_rate": 3.929519191908578e-06,
      "loss": 1.2011,
      "step": 3753
    },
    {
      "epoch": 4.594859241126071,
      "grad_norm": 1.3060102169577883,
      "learning_rate": 3.928993375057999e-06,
      "loss": 0.5617,
      "step": 3754
    },
    {
      "epoch": 4.596083231334149,
      "grad_norm": 2.203000550304208,
      "learning_rate": 3.928467464299263e-06,
      "loss": 0.8541,
      "step": 3755
    },
    {
      "epoch": 4.597307221542228,
      "grad_norm": 0.8978490628306895,
      "learning_rate": 3.927941459666927e-06,
      "loss": 0.6787,
      "step": 3756
    },
    {
      "epoch": 4.598531211750306,
      "grad_norm": 2.679472930667635,
      "learning_rate": 3.927415361195561e-06,
      "loss": 0.4427,
      "step": 3757
    },
    {
      "epoch": 4.599755201958384,
      "grad_norm": 1.5315774002188678,
      "learning_rate": 3.9268891689197365e-06,
      "loss": 0.7756,
      "step": 3758
    },
    {
      "epoch": 4.600979192166463,
      "grad_norm": 1.324761516294892,
      "learning_rate": 3.926362882874034e-06,
      "loss": 1.1423,
      "step": 3759
    },
    {
      "epoch": 4.602203182374541,
      "grad_norm": 1.0106329603589588,
      "learning_rate": 3.925836503093038e-06,
      "loss": 0.6599,
      "step": 3760
    },
    {
      "epoch": 4.603427172582619,
      "grad_norm": 1.981326883737026,
      "learning_rate": 3.925310029611341e-06,
      "loss": 0.8532,
      "step": 3761
    },
    {
      "epoch": 4.604651162790698,
      "grad_norm": 0.7405515538735531,
      "learning_rate": 3.924783462463541e-06,
      "loss": 0.6046,
      "step": 3762
    },
    {
      "epoch": 4.605875152998776,
      "grad_norm": 1.1404713669513582,
      "learning_rate": 3.924256801684242e-06,
      "loss": 0.5544,
      "step": 3763
    },
    {
      "epoch": 4.607099143206854,
      "grad_norm": 1.562275015438096,
      "learning_rate": 3.923730047308054e-06,
      "loss": 0.4066,
      "step": 3764
    },
    {
      "epoch": 4.608323133414933,
      "grad_norm": 0.9634286364947997,
      "learning_rate": 3.923203199369592e-06,
      "loss": 0.4353,
      "step": 3765
    },
    {
      "epoch": 4.6095471236230114,
      "grad_norm": 1.7607232278799965,
      "learning_rate": 3.922676257903481e-06,
      "loss": 0.2844,
      "step": 3766
    },
    {
      "epoch": 4.610771113831089,
      "grad_norm": 1.2184257451697964,
      "learning_rate": 3.922149222944348e-06,
      "loss": 0.6086,
      "step": 3767
    },
    {
      "epoch": 4.611995104039168,
      "grad_norm": 1.6866135450030748,
      "learning_rate": 3.921622094526829e-06,
      "loss": 0.5347,
      "step": 3768
    },
    {
      "epoch": 4.613219094247246,
      "grad_norm": 2.1251933094162347,
      "learning_rate": 3.921094872685563e-06,
      "loss": 0.4309,
      "step": 3769
    },
    {
      "epoch": 4.614443084455324,
      "grad_norm": 0.7802673236884407,
      "learning_rate": 3.920567557455199e-06,
      "loss": 0.6191,
      "step": 3770
    },
    {
      "epoch": 4.615667074663403,
      "grad_norm": 1.3465055840840237,
      "learning_rate": 3.920040148870389e-06,
      "loss": 1.4003,
      "step": 3771
    },
    {
      "epoch": 4.616891064871481,
      "grad_norm": 1.5936264741570534,
      "learning_rate": 3.919512646965793e-06,
      "loss": 0.4671,
      "step": 3772
    },
    {
      "epoch": 4.618115055079559,
      "grad_norm": 1.3388562485255457,
      "learning_rate": 3.9189850517760755e-06,
      "loss": 1.5271,
      "step": 3773
    },
    {
      "epoch": 4.619339045287638,
      "grad_norm": 1.41192148376063,
      "learning_rate": 3.91845736333591e-06,
      "loss": 0.5279,
      "step": 3774
    },
    {
      "epoch": 4.620563035495716,
      "grad_norm": 1.9685838160983313,
      "learning_rate": 3.9179295816799725e-06,
      "loss": 0.5057,
      "step": 3775
    },
    {
      "epoch": 4.621787025703794,
      "grad_norm": 1.8577213596355762,
      "learning_rate": 3.9174017068429486e-06,
      "loss": 0.9152,
      "step": 3776
    },
    {
      "epoch": 4.623011015911873,
      "grad_norm": 1.162891813606725,
      "learning_rate": 3.916873738859526e-06,
      "loss": 0.6304,
      "step": 3777
    },
    {
      "epoch": 4.624235006119951,
      "grad_norm": 1.3243308572738661,
      "learning_rate": 3.916345677764401e-06,
      "loss": 0.4883,
      "step": 3778
    },
    {
      "epoch": 4.625458996328029,
      "grad_norm": 0.8698987944725893,
      "learning_rate": 3.915817523592279e-06,
      "loss": 0.7715,
      "step": 3779
    },
    {
      "epoch": 4.626682986536108,
      "grad_norm": 0.8077517404600526,
      "learning_rate": 3.915289276377864e-06,
      "loss": 0.6298,
      "step": 3780
    },
    {
      "epoch": 4.627906976744186,
      "grad_norm": 0.9646245301985371,
      "learning_rate": 3.9147609361558735e-06,
      "loss": 0.8723,
      "step": 3781
    },
    {
      "epoch": 4.629130966952264,
      "grad_norm": 1.2027882582452825,
      "learning_rate": 3.914232502961028e-06,
      "loss": 1.2031,
      "step": 3782
    },
    {
      "epoch": 4.630354957160343,
      "grad_norm": 1.4166088322268275,
      "learning_rate": 3.913703976828052e-06,
      "loss": 0.4386,
      "step": 3783
    },
    {
      "epoch": 4.631578947368421,
      "grad_norm": 0.6544665548425483,
      "learning_rate": 3.91317535779168e-06,
      "loss": 0.369,
      "step": 3784
    },
    {
      "epoch": 4.6328029375765,
      "grad_norm": 1.7588369038288398,
      "learning_rate": 3.9126466458866505e-06,
      "loss": 0.9484,
      "step": 3785
    },
    {
      "epoch": 4.634026927784578,
      "grad_norm": 1.2416029736019707,
      "learning_rate": 3.912117841147708e-06,
      "loss": 0.6586,
      "step": 3786
    },
    {
      "epoch": 4.635250917992656,
      "grad_norm": 1.3089220838754938,
      "learning_rate": 3.911588943609605e-06,
      "loss": 0.5227,
      "step": 3787
    },
    {
      "epoch": 4.636474908200734,
      "grad_norm": 1.4641395116952016,
      "learning_rate": 3.911059953307097e-06,
      "loss": 0.7739,
      "step": 3788
    },
    {
      "epoch": 4.6376988984088126,
      "grad_norm": 1.1245421015045387,
      "learning_rate": 3.910530870274948e-06,
      "loss": 0.497,
      "step": 3789
    },
    {
      "epoch": 4.638922888616891,
      "grad_norm": 0.8102663763324813,
      "learning_rate": 3.910001694547928e-06,
      "loss": 0.661,
      "step": 3790
    },
    {
      "epoch": 4.64014687882497,
      "grad_norm": 1.692354711196083,
      "learning_rate": 3.909472426160811e-06,
      "loss": 0.5096,
      "step": 3791
    },
    {
      "epoch": 4.6413708690330475,
      "grad_norm": 0.9964318055290455,
      "learning_rate": 3.9089430651483805e-06,
      "loss": 0.5548,
      "step": 3792
    },
    {
      "epoch": 4.642594859241126,
      "grad_norm": 1.1782990203874486,
      "learning_rate": 3.9084136115454215e-06,
      "loss": 0.4722,
      "step": 3793
    },
    {
      "epoch": 4.643818849449205,
      "grad_norm": 1.8618496417329857,
      "learning_rate": 3.907884065386731e-06,
      "loss": 0.7328,
      "step": 3794
    },
    {
      "epoch": 4.6450428396572825,
      "grad_norm": 1.2565480472429427,
      "learning_rate": 3.907354426707106e-06,
      "loss": 1.185,
      "step": 3795
    },
    {
      "epoch": 4.646266829865361,
      "grad_norm": 1.4883315744333285,
      "learning_rate": 3.906824695541354e-06,
      "loss": 1.0775,
      "step": 3796
    },
    {
      "epoch": 4.64749082007344,
      "grad_norm": 1.8707060772347006,
      "learning_rate": 3.9062948719242865e-06,
      "loss": 0.7263,
      "step": 3797
    },
    {
      "epoch": 4.6487148102815175,
      "grad_norm": 1.5309523598488055,
      "learning_rate": 3.905764955890722e-06,
      "loss": 0.8897,
      "step": 3798
    },
    {
      "epoch": 4.649938800489596,
      "grad_norm": 0.997541393758724,
      "learning_rate": 3.905234947475483e-06,
      "loss": 1.1194,
      "step": 3799
    },
    {
      "epoch": 4.651162790697675,
      "grad_norm": 1.009438255607341,
      "learning_rate": 3.904704846713402e-06,
      "loss": 1.0694,
      "step": 3800
    },
    {
      "epoch": 4.652386780905752,
      "grad_norm": 2.09385381330016,
      "learning_rate": 3.9041746536393135e-06,
      "loss": 0.5636,
      "step": 3801
    },
    {
      "epoch": 4.653610771113831,
      "grad_norm": 0.7958581360915951,
      "learning_rate": 3.90364436828806e-06,
      "loss": 0.7355,
      "step": 3802
    },
    {
      "epoch": 4.65483476132191,
      "grad_norm": 1.77805467109313,
      "learning_rate": 3.903113990694492e-06,
      "loss": 0.5359,
      "step": 3803
    },
    {
      "epoch": 4.656058751529987,
      "grad_norm": 1.128435374071368,
      "learning_rate": 3.902583520893461e-06,
      "loss": 1.3505,
      "step": 3804
    },
    {
      "epoch": 4.657282741738066,
      "grad_norm": 0.9890765890485449,
      "learning_rate": 3.902052958919829e-06,
      "loss": 0.5789,
      "step": 3805
    },
    {
      "epoch": 4.658506731946145,
      "grad_norm": 1.3703466172787675,
      "learning_rate": 3.901522304808462e-06,
      "loss": 0.4104,
      "step": 3806
    },
    {
      "epoch": 4.659730722154222,
      "grad_norm": 1.538072819978701,
      "learning_rate": 3.900991558594232e-06,
      "loss": 0.5973,
      "step": 3807
    },
    {
      "epoch": 4.660954712362301,
      "grad_norm": 2.0969990605358197,
      "learning_rate": 3.900460720312021e-06,
      "loss": 1.1096,
      "step": 3808
    },
    {
      "epoch": 4.66217870257038,
      "grad_norm": 2.497257078687693,
      "learning_rate": 3.8999297899967105e-06,
      "loss": 0.5429,
      "step": 3809
    },
    {
      "epoch": 4.663402692778458,
      "grad_norm": 1.1208995632273182,
      "learning_rate": 3.899398767683191e-06,
      "loss": 1.3417,
      "step": 3810
    },
    {
      "epoch": 4.664626682986536,
      "grad_norm": 1.2771706805120748,
      "learning_rate": 3.898867653406363e-06,
      "loss": 0.5226,
      "step": 3811
    },
    {
      "epoch": 4.6658506731946146,
      "grad_norm": 1.2675047971271205,
      "learning_rate": 3.898336447201125e-06,
      "loss": 0.7905,
      "step": 3812
    },
    {
      "epoch": 4.667074663402692,
      "grad_norm": 1.7065910600520846,
      "learning_rate": 3.897805149102389e-06,
      "loss": 0.5508,
      "step": 3813
    },
    {
      "epoch": 4.668298653610771,
      "grad_norm": 0.835605501742259,
      "learning_rate": 3.897273759145067e-06,
      "loss": 0.8629,
      "step": 3814
    },
    {
      "epoch": 4.6695226438188495,
      "grad_norm": 0.6283904767990558,
      "learning_rate": 3.896742277364083e-06,
      "loss": 0.448,
      "step": 3815
    },
    {
      "epoch": 4.670746634026928,
      "grad_norm": 1.1716451379141146,
      "learning_rate": 3.896210703794362e-06,
      "loss": 1.336,
      "step": 3816
    },
    {
      "epoch": 4.671970624235006,
      "grad_norm": 1.4711902521517402,
      "learning_rate": 3.895679038470839e-06,
      "loss": 0.5619,
      "step": 3817
    },
    {
      "epoch": 4.6731946144430845,
      "grad_norm": 1.2724571075581546,
      "learning_rate": 3.8951472814284505e-06,
      "loss": 0.3338,
      "step": 3818
    },
    {
      "epoch": 4.674418604651163,
      "grad_norm": 1.3299251487765837,
      "learning_rate": 3.894615432702144e-06,
      "loss": 0.487,
      "step": 3819
    },
    {
      "epoch": 4.675642594859241,
      "grad_norm": 0.9860912540511926,
      "learning_rate": 3.894083492326869e-06,
      "loss": 0.5847,
      "step": 3820
    },
    {
      "epoch": 4.6768665850673194,
      "grad_norm": 1.1099945865516154,
      "learning_rate": 3.893551460337583e-06,
      "loss": 0.716,
      "step": 3821
    },
    {
      "epoch": 4.678090575275398,
      "grad_norm": 1.4025723920396533,
      "learning_rate": 3.8930193367692496e-06,
      "loss": 0.4805,
      "step": 3822
    },
    {
      "epoch": 4.679314565483476,
      "grad_norm": 1.4315753124567552,
      "learning_rate": 3.892487121656838e-06,
      "loss": 0.5485,
      "step": 3823
    },
    {
      "epoch": 4.680538555691554,
      "grad_norm": 1.3602505821971216,
      "learning_rate": 3.891954815035324e-06,
      "loss": 0.642,
      "step": 3824
    },
    {
      "epoch": 4.681762545899633,
      "grad_norm": 1.612372094925039,
      "learning_rate": 3.891422416939686e-06,
      "loss": 1.2251,
      "step": 3825
    },
    {
      "epoch": 4.682986536107711,
      "grad_norm": 0.9606976486392892,
      "learning_rate": 3.890889927404915e-06,
      "loss": 0.6647,
      "step": 3826
    },
    {
      "epoch": 4.684210526315789,
      "grad_norm": 0.9908076928923341,
      "learning_rate": 3.890357346466001e-06,
      "loss": 0.5837,
      "step": 3827
    },
    {
      "epoch": 4.685434516523868,
      "grad_norm": 1.9272653369729187,
      "learning_rate": 3.889824674157947e-06,
      "loss": 0.6451,
      "step": 3828
    },
    {
      "epoch": 4.686658506731947,
      "grad_norm": 0.8244525430795184,
      "learning_rate": 3.889291910515755e-06,
      "loss": 0.6205,
      "step": 3829
    },
    {
      "epoch": 4.687882496940024,
      "grad_norm": 2.2523050830263704,
      "learning_rate": 3.888759055574437e-06,
      "loss": 0.4641,
      "step": 3830
    },
    {
      "epoch": 4.689106487148103,
      "grad_norm": 0.8638109319487003,
      "learning_rate": 3.88822610936901e-06,
      "loss": 0.5066,
      "step": 3831
    },
    {
      "epoch": 4.690330477356181,
      "grad_norm": 2.4153918414921502,
      "learning_rate": 3.8876930719344984e-06,
      "loss": 0.4951,
      "step": 3832
    },
    {
      "epoch": 4.691554467564259,
      "grad_norm": 1.8146619878499939,
      "learning_rate": 3.887159943305931e-06,
      "loss": 0.4519,
      "step": 3833
    },
    {
      "epoch": 4.692778457772338,
      "grad_norm": 1.7342221894203216,
      "learning_rate": 3.886626723518343e-06,
      "loss": 0.651,
      "step": 3834
    },
    {
      "epoch": 4.6940024479804165,
      "grad_norm": 1.8249056818933473,
      "learning_rate": 3.886093412606775e-06,
      "loss": 1.0045,
      "step": 3835
    },
    {
      "epoch": 4.695226438188494,
      "grad_norm": 1.6911751232892338,
      "learning_rate": 3.885560010606275e-06,
      "loss": 0.4295,
      "step": 3836
    },
    {
      "epoch": 4.696450428396573,
      "grad_norm": 1.3772771561976032,
      "learning_rate": 3.885026517551895e-06,
      "loss": 1.6392,
      "step": 3837
    },
    {
      "epoch": 4.6976744186046515,
      "grad_norm": 1.563885197151202,
      "learning_rate": 3.884492933478696e-06,
      "loss": 0.3959,
      "step": 3838
    },
    {
      "epoch": 4.698898408812729,
      "grad_norm": 1.124725008174981,
      "learning_rate": 3.8839592584217434e-06,
      "loss": 0.6806,
      "step": 3839
    },
    {
      "epoch": 4.700122399020808,
      "grad_norm": 1.0399086892336007,
      "learning_rate": 3.8834254924161055e-06,
      "loss": 0.3974,
      "step": 3840
    },
    {
      "epoch": 4.7013463892288865,
      "grad_norm": 1.4774658943628078,
      "learning_rate": 3.882891635496862e-06,
      "loss": 0.43,
      "step": 3841
    },
    {
      "epoch": 4.702570379436964,
      "grad_norm": 1.510977323945999,
      "learning_rate": 3.882357687699095e-06,
      "loss": 0.6509,
      "step": 3842
    },
    {
      "epoch": 4.703794369645043,
      "grad_norm": 1.6431801473221397,
      "learning_rate": 3.881823649057894e-06,
      "loss": 0.6774,
      "step": 3843
    },
    {
      "epoch": 4.7050183598531214,
      "grad_norm": 0.787639944540383,
      "learning_rate": 3.881289519608354e-06,
      "loss": 0.415,
      "step": 3844
    },
    {
      "epoch": 4.706242350061199,
      "grad_norm": 1.664043546421963,
      "learning_rate": 3.880755299385574e-06,
      "loss": 1.1097,
      "step": 3845
    },
    {
      "epoch": 4.707466340269278,
      "grad_norm": 1.6891650363607682,
      "learning_rate": 3.880220988424664e-06,
      "loss": 0.4893,
      "step": 3846
    },
    {
      "epoch": 4.708690330477356,
      "grad_norm": 1.0682095676909167,
      "learning_rate": 3.879686586760736e-06,
      "loss": 0.6461,
      "step": 3847
    },
    {
      "epoch": 4.709914320685434,
      "grad_norm": 0.7082124683399725,
      "learning_rate": 3.87915209442891e-06,
      "loss": 0.4647,
      "step": 3848
    },
    {
      "epoch": 4.711138310893513,
      "grad_norm": 2.159525233397554,
      "learning_rate": 3.878617511464307e-06,
      "loss": 0.584,
      "step": 3849
    },
    {
      "epoch": 4.712362301101591,
      "grad_norm": 1.0553194564275217,
      "learning_rate": 3.878082837902061e-06,
      "loss": 0.7922,
      "step": 3850
    },
    {
      "epoch": 4.713586291309669,
      "grad_norm": 1.4233519059901172,
      "learning_rate": 3.8775480737773085e-06,
      "loss": 0.9968,
      "step": 3851
    },
    {
      "epoch": 4.714810281517748,
      "grad_norm": 1.3447389532978522,
      "learning_rate": 3.8770132191251916e-06,
      "loss": 0.7662,
      "step": 3852
    },
    {
      "epoch": 4.716034271725826,
      "grad_norm": 0.8573140192708791,
      "learning_rate": 3.876478273980859e-06,
      "loss": 0.6441,
      "step": 3853
    },
    {
      "epoch": 4.717258261933905,
      "grad_norm": 1.273784006199861,
      "learning_rate": 3.875943238379466e-06,
      "loss": 0.5343,
      "step": 3854
    },
    {
      "epoch": 4.718482252141983,
      "grad_norm": 1.4975323379275303,
      "learning_rate": 3.875408112356172e-06,
      "loss": 0.5526,
      "step": 3855
    },
    {
      "epoch": 4.719706242350061,
      "grad_norm": 1.4157188676822474,
      "learning_rate": 3.874872895946145e-06,
      "loss": 0.5627,
      "step": 3856
    },
    {
      "epoch": 4.720930232558139,
      "grad_norm": 1.4542202741856565,
      "learning_rate": 3.874337589184556e-06,
      "loss": 0.4886,
      "step": 3857
    },
    {
      "epoch": 4.722154222766218,
      "grad_norm": 2.2446102187818333,
      "learning_rate": 3.873802192106583e-06,
      "loss": 0.3806,
      "step": 3858
    },
    {
      "epoch": 4.723378212974296,
      "grad_norm": 0.859256029405725,
      "learning_rate": 3.873266704747413e-06,
      "loss": 0.7187,
      "step": 3859
    },
    {
      "epoch": 4.724602203182375,
      "grad_norm": 1.4745387803578551,
      "learning_rate": 3.872731127142234e-06,
      "loss": 0.6352,
      "step": 3860
    },
    {
      "epoch": 4.725826193390453,
      "grad_norm": 1.1267208366398835,
      "learning_rate": 3.872195459326242e-06,
      "loss": 0.5202,
      "step": 3861
    },
    {
      "epoch": 4.727050183598531,
      "grad_norm": 1.689831495441326,
      "learning_rate": 3.87165970133464e-06,
      "loss": 1.0777,
      "step": 3862
    },
    {
      "epoch": 4.72827417380661,
      "grad_norm": 1.2529103860082147,
      "learning_rate": 3.871123853202637e-06,
      "loss": 0.5174,
      "step": 3863
    },
    {
      "epoch": 4.729498164014688,
      "grad_norm": 0.6951985591814531,
      "learning_rate": 3.870587914965445e-06,
      "loss": 0.4068,
      "step": 3864
    },
    {
      "epoch": 4.730722154222766,
      "grad_norm": 0.9806651631674292,
      "learning_rate": 3.870051886658284e-06,
      "loss": 0.6633,
      "step": 3865
    },
    {
      "epoch": 4.731946144430845,
      "grad_norm": 1.7815474792867037,
      "learning_rate": 3.869515768316382e-06,
      "loss": 0.521,
      "step": 3866
    },
    {
      "epoch": 4.7331701346389226,
      "grad_norm": 0.9571825414754315,
      "learning_rate": 3.868979559974968e-06,
      "loss": 0.7796,
      "step": 3867
    },
    {
      "epoch": 4.734394124847001,
      "grad_norm": 1.3292045799176089,
      "learning_rate": 3.868443261669281e-06,
      "loss": 0.5652,
      "step": 3868
    },
    {
      "epoch": 4.73561811505508,
      "grad_norm": 0.7773568351021357,
      "learning_rate": 3.867906873434564e-06,
      "loss": 0.3879,
      "step": 3869
    },
    {
      "epoch": 4.7368421052631575,
      "grad_norm": 1.1509411372076819,
      "learning_rate": 3.8673703953060685e-06,
      "loss": 0.9985,
      "step": 3870
    },
    {
      "epoch": 4.738066095471236,
      "grad_norm": 1.3387846767700868,
      "learning_rate": 3.866833827319047e-06,
      "loss": 1.129,
      "step": 3871
    },
    {
      "epoch": 4.739290085679315,
      "grad_norm": 1.9065348346585043,
      "learning_rate": 3.866297169508762e-06,
      "loss": 0.65,
      "step": 3872
    },
    {
      "epoch": 4.740514075887393,
      "grad_norm": 1.4099242743255993,
      "learning_rate": 3.86576042191048e-06,
      "loss": 1.0476,
      "step": 3873
    },
    {
      "epoch": 4.741738066095471,
      "grad_norm": 1.1056517501451506,
      "learning_rate": 3.865223584559475e-06,
      "loss": 0.7662,
      "step": 3874
    },
    {
      "epoch": 4.74296205630355,
      "grad_norm": 1.3926385015866696,
      "learning_rate": 3.864686657491026e-06,
      "loss": 1.0619,
      "step": 3875
    },
    {
      "epoch": 4.7441860465116275,
      "grad_norm": 1.7544256475286069,
      "learning_rate": 3.8641496407404165e-06,
      "loss": 0.5399,
      "step": 3876
    },
    {
      "epoch": 4.745410036719706,
      "grad_norm": 1.1220500680339782,
      "learning_rate": 3.86361253434294e-06,
      "loss": 0.5673,
      "step": 3877
    },
    {
      "epoch": 4.746634026927785,
      "grad_norm": 1.163672484121933,
      "learning_rate": 3.86307533833389e-06,
      "loss": 0.6172,
      "step": 3878
    },
    {
      "epoch": 4.747858017135863,
      "grad_norm": 2.046618918161762,
      "learning_rate": 3.8625380527485716e-06,
      "loss": 0.4076,
      "step": 3879
    },
    {
      "epoch": 4.749082007343941,
      "grad_norm": 1.4977113031072897,
      "learning_rate": 3.862000677622291e-06,
      "loss": 0.5951,
      "step": 3880
    },
    {
      "epoch": 4.75030599755202,
      "grad_norm": 1.409934341838558,
      "learning_rate": 3.861463212990364e-06,
      "loss": 0.9983,
      "step": 3881
    },
    {
      "epoch": 4.751529987760098,
      "grad_norm": 1.1885927654938855,
      "learning_rate": 3.860925658888109e-06,
      "loss": 0.5043,
      "step": 3882
    },
    {
      "epoch": 4.752753977968176,
      "grad_norm": 1.274012822662529,
      "learning_rate": 3.860388015350855e-06,
      "loss": 0.5896,
      "step": 3883
    },
    {
      "epoch": 4.753977968176255,
      "grad_norm": 1.4847252849772758,
      "learning_rate": 3.859850282413933e-06,
      "loss": 1.1764,
      "step": 3884
    },
    {
      "epoch": 4.755201958384333,
      "grad_norm": 1.4089731217501331,
      "learning_rate": 3.859312460112679e-06,
      "loss": 0.5713,
      "step": 3885
    },
    {
      "epoch": 4.756425948592411,
      "grad_norm": 1.2257118490370726,
      "learning_rate": 3.858774548482438e-06,
      "loss": 0.5375,
      "step": 3886
    },
    {
      "epoch": 4.75764993880049,
      "grad_norm": 1.8533957962376275,
      "learning_rate": 3.858236547558559e-06,
      "loss": 0.4805,
      "step": 3887
    },
    {
      "epoch": 4.758873929008568,
      "grad_norm": 1.0257624226569373,
      "learning_rate": 3.857698457376399e-06,
      "loss": 0.5822,
      "step": 3888
    },
    {
      "epoch": 4.760097919216646,
      "grad_norm": 1.4579541814209134,
      "learning_rate": 3.857160277971317e-06,
      "loss": 1.0364,
      "step": 3889
    },
    {
      "epoch": 4.7613219094247246,
      "grad_norm": 1.5655760217243508,
      "learning_rate": 3.856622009378682e-06,
      "loss": 0.621,
      "step": 3890
    },
    {
      "epoch": 4.762545899632803,
      "grad_norm": 1.1390620555330881,
      "learning_rate": 3.856083651633866e-06,
      "loss": 0.8991,
      "step": 3891
    },
    {
      "epoch": 4.763769889840881,
      "grad_norm": 1.3243562133149331,
      "learning_rate": 3.855545204772248e-06,
      "loss": 0.615,
      "step": 3892
    },
    {
      "epoch": 4.7649938800489595,
      "grad_norm": 1.651045691244505,
      "learning_rate": 3.855006668829214e-06,
      "loss": 1.0869,
      "step": 3893
    },
    {
      "epoch": 4.766217870257038,
      "grad_norm": 1.2139920842516123,
      "learning_rate": 3.854468043840153e-06,
      "loss": 1.0115,
      "step": 3894
    },
    {
      "epoch": 4.767441860465116,
      "grad_norm": 1.1053117184750336,
      "learning_rate": 3.853929329840462e-06,
      "loss": 0.905,
      "step": 3895
    },
    {
      "epoch": 4.7686658506731945,
      "grad_norm": 1.0097624509665304,
      "learning_rate": 3.853390526865542e-06,
      "loss": 0.4634,
      "step": 3896
    },
    {
      "epoch": 4.769889840881273,
      "grad_norm": 2.083770270540241,
      "learning_rate": 3.852851634950804e-06,
      "loss": 0.5699,
      "step": 3897
    },
    {
      "epoch": 4.771113831089352,
      "grad_norm": 1.695993974895643,
      "learning_rate": 3.85231265413166e-06,
      "loss": 0.5282,
      "step": 3898
    },
    {
      "epoch": 4.7723378212974294,
      "grad_norm": 2.060608151187121,
      "learning_rate": 3.85177358444353e-06,
      "loss": 0.489,
      "step": 3899
    },
    {
      "epoch": 4.773561811505508,
      "grad_norm": 1.3797204391420028,
      "learning_rate": 3.851234425921841e-06,
      "loss": 0.561,
      "step": 3900
    },
    {
      "epoch": 4.774785801713586,
      "grad_norm": 1.2718566663716822,
      "learning_rate": 3.850695178602022e-06,
      "loss": 0.5047,
      "step": 3901
    },
    {
      "epoch": 4.776009791921664,
      "grad_norm": 1.8280578885409873,
      "learning_rate": 3.850155842519513e-06,
      "loss": 0.5683,
      "step": 3902
    },
    {
      "epoch": 4.777233782129743,
      "grad_norm": 1.1374211459357546,
      "learning_rate": 3.8496164177097544e-06,
      "loss": 0.5291,
      "step": 3903
    },
    {
      "epoch": 4.778457772337822,
      "grad_norm": 1.6697498705353433,
      "learning_rate": 3.849076904208198e-06,
      "loss": 1.385,
      "step": 3904
    },
    {
      "epoch": 4.779681762545899,
      "grad_norm": 2.0128462612031424,
      "learning_rate": 3.848537302050297e-06,
      "loss": 1.2275,
      "step": 3905
    },
    {
      "epoch": 4.780905752753978,
      "grad_norm": 1.8416172323402595,
      "learning_rate": 3.847997611271512e-06,
      "loss": 0.3913,
      "step": 3906
    },
    {
      "epoch": 4.782129742962057,
      "grad_norm": 1.3428582062374113,
      "learning_rate": 3.8474578319073095e-06,
      "loss": 0.965,
      "step": 3907
    },
    {
      "epoch": 4.783353733170134,
      "grad_norm": 1.2368517079472954,
      "learning_rate": 3.8469179639931635e-06,
      "loss": 1.365,
      "step": 3908
    },
    {
      "epoch": 4.784577723378213,
      "grad_norm": 1.2912882351347057,
      "learning_rate": 3.8463780075645505e-06,
      "loss": 0.5758,
      "step": 3909
    },
    {
      "epoch": 4.785801713586292,
      "grad_norm": 1.3707087802713092,
      "learning_rate": 3.8458379626569545e-06,
      "loss": 0.5549,
      "step": 3910
    },
    {
      "epoch": 4.787025703794369,
      "grad_norm": 1.2094011308054873,
      "learning_rate": 3.845297829305865e-06,
      "loss": 0.5537,
      "step": 3911
    },
    {
      "epoch": 4.788249694002448,
      "grad_norm": 1.7849196983501878,
      "learning_rate": 3.844757607546778e-06,
      "loss": 0.5693,
      "step": 3912
    },
    {
      "epoch": 4.7894736842105265,
      "grad_norm": 1.3578476158328867,
      "learning_rate": 3.844217297415196e-06,
      "loss": 0.5222,
      "step": 3913
    },
    {
      "epoch": 4.790697674418604,
      "grad_norm": 1.238331824857719,
      "learning_rate": 3.8436768989466245e-06,
      "loss": 1.3085,
      "step": 3914
    },
    {
      "epoch": 4.791921664626683,
      "grad_norm": 0.9282874643212091,
      "learning_rate": 3.843136412176577e-06,
      "loss": 0.4332,
      "step": 3915
    },
    {
      "epoch": 4.7931456548347615,
      "grad_norm": 0.9637192524883046,
      "learning_rate": 3.842595837140572e-06,
      "loss": 0.6014,
      "step": 3916
    },
    {
      "epoch": 4.79436964504284,
      "grad_norm": 2.2048676397048443,
      "learning_rate": 3.842055173874135e-06,
      "loss": 0.5097,
      "step": 3917
    },
    {
      "epoch": 4.795593635250918,
      "grad_norm": 1.4240705654445107,
      "learning_rate": 3.841514422412797e-06,
      "loss": 0.6704,
      "step": 3918
    },
    {
      "epoch": 4.7968176254589965,
      "grad_norm": 1.200234181878473,
      "learning_rate": 3.840973582792091e-06,
      "loss": 1.1215,
      "step": 3919
    },
    {
      "epoch": 4.798041615667074,
      "grad_norm": 0.7157132541242024,
      "learning_rate": 3.840432655047563e-06,
      "loss": 0.4425,
      "step": 3920
    },
    {
      "epoch": 4.799265605875153,
      "grad_norm": 1.444210785435878,
      "learning_rate": 3.839891639214757e-06,
      "loss": 0.9151,
      "step": 3921
    },
    {
      "epoch": 4.8004895960832314,
      "grad_norm": 0.891277643864051,
      "learning_rate": 3.8393505353292305e-06,
      "loss": 0.5597,
      "step": 3922
    },
    {
      "epoch": 4.80171358629131,
      "grad_norm": 2.0075120839738356,
      "learning_rate": 3.838809343426538e-06,
      "loss": 0.5559,
      "step": 3923
    },
    {
      "epoch": 4.802937576499388,
      "grad_norm": 2.21084791066355,
      "learning_rate": 3.83826806354225e-06,
      "loss": 0.6374,
      "step": 3924
    },
    {
      "epoch": 4.804161566707466,
      "grad_norm": 1.0919377509216344,
      "learning_rate": 3.837726695711933e-06,
      "loss": 0.8245,
      "step": 3925
    },
    {
      "epoch": 4.805385556915545,
      "grad_norm": 2.019896165377279,
      "learning_rate": 3.837185239971166e-06,
      "loss": 0.5066,
      "step": 3926
    },
    {
      "epoch": 4.806609547123623,
      "grad_norm": 2.3524757487844647,
      "learning_rate": 3.836643696355531e-06,
      "loss": 0.4446,
      "step": 3927
    },
    {
      "epoch": 4.807833537331701,
      "grad_norm": 1.3327706632966636,
      "learning_rate": 3.836102064900617e-06,
      "loss": 0.6236,
      "step": 3928
    },
    {
      "epoch": 4.80905752753978,
      "grad_norm": 1.2290677295235823,
      "learning_rate": 3.835560345642017e-06,
      "loss": 0.4761,
      "step": 3929
    },
    {
      "epoch": 4.810281517747858,
      "grad_norm": 1.2227336863670288,
      "learning_rate": 3.8350185386153294e-06,
      "loss": 0.9609,
      "step": 3930
    },
    {
      "epoch": 4.811505507955936,
      "grad_norm": 1.0804357760148826,
      "learning_rate": 3.8344766438561635e-06,
      "loss": 0.5642,
      "step": 3931
    },
    {
      "epoch": 4.812729498164015,
      "grad_norm": 1.272022159091841,
      "learning_rate": 3.833934661400126e-06,
      "loss": 1.5721,
      "step": 3932
    },
    {
      "epoch": 4.813953488372093,
      "grad_norm": 1.2554478297707332,
      "learning_rate": 3.833392591282838e-06,
      "loss": 0.5466,
      "step": 3933
    },
    {
      "epoch": 4.815177478580171,
      "grad_norm": 1.7788244770375572,
      "learning_rate": 3.832850433539921e-06,
      "loss": 0.9374,
      "step": 3934
    },
    {
      "epoch": 4.81640146878825,
      "grad_norm": 1.563687518903311,
      "learning_rate": 3.832308188207003e-06,
      "loss": 0.5469,
      "step": 3935
    },
    {
      "epoch": 4.817625458996328,
      "grad_norm": 1.2382492503926494,
      "learning_rate": 3.831765855319718e-06,
      "loss": 1.6768,
      "step": 3936
    },
    {
      "epoch": 4.818849449204406,
      "grad_norm": 2.2947354291407422,
      "learning_rate": 3.831223434913708e-06,
      "loss": 0.5784,
      "step": 3937
    },
    {
      "epoch": 4.820073439412485,
      "grad_norm": 1.4848037287341793,
      "learning_rate": 3.830680927024617e-06,
      "loss": 0.7542,
      "step": 3938
    },
    {
      "epoch": 4.821297429620563,
      "grad_norm": 1.2166180920349023,
      "learning_rate": 3.830138331688098e-06,
      "loss": 0.7277,
      "step": 3939
    },
    {
      "epoch": 4.822521419828641,
      "grad_norm": 1.1517721209729608,
      "learning_rate": 3.829595648939807e-06,
      "loss": 0.5598,
      "step": 3940
    },
    {
      "epoch": 4.82374541003672,
      "grad_norm": 0.594744248321611,
      "learning_rate": 3.829052878815408e-06,
      "loss": 0.2327,
      "step": 3941
    },
    {
      "epoch": 4.8249694002447985,
      "grad_norm": 0.8457872539264581,
      "learning_rate": 3.82851002135057e-06,
      "loss": 0.5691,
      "step": 3942
    },
    {
      "epoch": 4.826193390452876,
      "grad_norm": 0.7932153241107824,
      "learning_rate": 3.827967076580966e-06,
      "loss": 0.5602,
      "step": 3943
    },
    {
      "epoch": 4.827417380660955,
      "grad_norm": 1.2034220702369667,
      "learning_rate": 3.8274240445422785e-06,
      "loss": 0.6367,
      "step": 3944
    },
    {
      "epoch": 4.828641370869033,
      "grad_norm": 1.0789086954967142,
      "learning_rate": 3.826880925270192e-06,
      "loss": 0.5368,
      "step": 3945
    },
    {
      "epoch": 4.829865361077111,
      "grad_norm": 1.2751888726080984,
      "learning_rate": 3.8263377188004e-06,
      "loss": 0.5013,
      "step": 3946
    },
    {
      "epoch": 4.83108935128519,
      "grad_norm": 1.4040666526194419,
      "learning_rate": 3.825794425168598e-06,
      "loss": 1.9201,
      "step": 3947
    },
    {
      "epoch": 4.832313341493268,
      "grad_norm": 1.066444287680524,
      "learning_rate": 3.82525104441049e-06,
      "loss": 0.7052,
      "step": 3948
    },
    {
      "epoch": 4.833537331701346,
      "grad_norm": 1.2886722134821684,
      "learning_rate": 3.824707576561786e-06,
      "loss": 0.9453,
      "step": 3949
    },
    {
      "epoch": 4.834761321909425,
      "grad_norm": 1.0087804705338113,
      "learning_rate": 3.8241640216582e-06,
      "loss": 0.7667,
      "step": 3950
    },
    {
      "epoch": 4.835985312117503,
      "grad_norm": 1.1598181968857744,
      "learning_rate": 3.823620379735451e-06,
      "loss": 0.6595,
      "step": 3951
    },
    {
      "epoch": 4.837209302325581,
      "grad_norm": 1.3650876503101348,
      "learning_rate": 3.823076650829267e-06,
      "loss": 0.9479,
      "step": 3952
    },
    {
      "epoch": 4.83843329253366,
      "grad_norm": 1.5045762371737315,
      "learning_rate": 3.8225328349753785e-06,
      "loss": 0.9674,
      "step": 3953
    },
    {
      "epoch": 4.839657282741738,
      "grad_norm": 1.6010445721452395,
      "learning_rate": 3.821988932209525e-06,
      "loss": 0.5839,
      "step": 3954
    },
    {
      "epoch": 4.840881272949816,
      "grad_norm": 1.6965744893780248,
      "learning_rate": 3.821444942567449e-06,
      "loss": 0.493,
      "step": 3955
    },
    {
      "epoch": 4.842105263157895,
      "grad_norm": 1.3706340136681063,
      "learning_rate": 3.820900866084898e-06,
      "loss": 0.588,
      "step": 3956
    },
    {
      "epoch": 4.843329253365973,
      "grad_norm": 1.4051680512730989,
      "learning_rate": 3.820356702797628e-06,
      "loss": 0.3767,
      "step": 3957
    },
    {
      "epoch": 4.844553243574051,
      "grad_norm": 1.1698736613723644,
      "learning_rate": 3.819812452741398e-06,
      "loss": 1.2562,
      "step": 3958
    },
    {
      "epoch": 4.84577723378213,
      "grad_norm": 2.523830506395649,
      "learning_rate": 3.819268115951977e-06,
      "loss": 0.4927,
      "step": 3959
    },
    {
      "epoch": 4.847001223990208,
      "grad_norm": 1.065849043989781,
      "learning_rate": 3.818723692465134e-06,
      "loss": 0.9747,
      "step": 3960
    },
    {
      "epoch": 4.848225214198287,
      "grad_norm": 0.8407679129191928,
      "learning_rate": 3.818179182316648e-06,
      "loss": 0.6293,
      "step": 3961
    },
    {
      "epoch": 4.849449204406365,
      "grad_norm": 0.8311965534277637,
      "learning_rate": 3.817634585542301e-06,
      "loss": 0.6023,
      "step": 3962
    },
    {
      "epoch": 4.850673194614443,
      "grad_norm": 1.949280026175921,
      "learning_rate": 3.817089902177884e-06,
      "loss": 0.4228,
      "step": 3963
    },
    {
      "epoch": 4.851897184822521,
      "grad_norm": 1.1997873346518195,
      "learning_rate": 3.816545132259189e-06,
      "loss": 0.5893,
      "step": 3964
    },
    {
      "epoch": 4.8531211750306,
      "grad_norm": 1.4690326187518903,
      "learning_rate": 3.816000275822018e-06,
      "loss": 1.353,
      "step": 3965
    },
    {
      "epoch": 4.854345165238678,
      "grad_norm": 1.9143392940606023,
      "learning_rate": 3.815455332902177e-06,
      "loss": 0.5679,
      "step": 3966
    },
    {
      "epoch": 4.855569155446757,
      "grad_norm": 1.2756682061673152,
      "learning_rate": 3.814910303535476e-06,
      "loss": 0.5561,
      "step": 3967
    },
    {
      "epoch": 4.8567931456548346,
      "grad_norm": 2.139529447484724,
      "learning_rate": 3.8143651877577344e-06,
      "loss": 0.8778,
      "step": 3968
    },
    {
      "epoch": 4.858017135862913,
      "grad_norm": 1.21957366728038,
      "learning_rate": 3.8138199856047733e-06,
      "loss": 0.5235,
      "step": 3969
    },
    {
      "epoch": 4.859241126070992,
      "grad_norm": 1.888503528425335,
      "learning_rate": 3.8132746971124224e-06,
      "loss": 0.9168,
      "step": 3970
    },
    {
      "epoch": 4.8604651162790695,
      "grad_norm": 1.1720225828947206,
      "learning_rate": 3.812729322316516e-06,
      "loss": 0.4943,
      "step": 3971
    },
    {
      "epoch": 4.861689106487148,
      "grad_norm": 0.8362714506051125,
      "learning_rate": 3.8121838612528943e-06,
      "loss": 0.5627,
      "step": 3972
    },
    {
      "epoch": 4.862913096695227,
      "grad_norm": 1.243627080494953,
      "learning_rate": 3.8116383139574022e-06,
      "loss": 0.5049,
      "step": 3973
    },
    {
      "epoch": 4.8641370869033045,
      "grad_norm": 1.0634416646694753,
      "learning_rate": 3.8110926804658914e-06,
      "loss": 0.5563,
      "step": 3974
    },
    {
      "epoch": 4.865361077111383,
      "grad_norm": 1.8463541036261395,
      "learning_rate": 3.81054696081422e-06,
      "loss": 1.3296,
      "step": 3975
    },
    {
      "epoch": 4.866585067319462,
      "grad_norm": 1.2174062596067512,
      "learning_rate": 3.8100011550382486e-06,
      "loss": 1.3327,
      "step": 3976
    },
    {
      "epoch": 4.8678090575275395,
      "grad_norm": 1.0765050370263196,
      "learning_rate": 3.809455263173847e-06,
      "loss": 1.1051,
      "step": 3977
    },
    {
      "epoch": 4.869033047735618,
      "grad_norm": 1.936619438168795,
      "learning_rate": 3.8089092852568888e-06,
      "loss": 0.5894,
      "step": 3978
    },
    {
      "epoch": 4.870257037943697,
      "grad_norm": 1.4085095069385354,
      "learning_rate": 3.8083632213232537e-06,
      "loss": 1.343,
      "step": 3979
    },
    {
      "epoch": 4.871481028151774,
      "grad_norm": 1.7687083418227434,
      "learning_rate": 3.8078170714088268e-06,
      "loss": 0.5451,
      "step": 3980
    },
    {
      "epoch": 4.872705018359853,
      "grad_norm": 1.7002933979636856,
      "learning_rate": 3.8072708355495005e-06,
      "loss": 0.9938,
      "step": 3981
    },
    {
      "epoch": 4.873929008567932,
      "grad_norm": 1.2714081925551262,
      "learning_rate": 3.8067245137811683e-06,
      "loss": 1.6539,
      "step": 3982
    },
    {
      "epoch": 4.875152998776009,
      "grad_norm": 1.1090548968310594,
      "learning_rate": 3.806178106139735e-06,
      "loss": 1.5929,
      "step": 3983
    },
    {
      "epoch": 4.876376988984088,
      "grad_norm": 1.0995048240295118,
      "learning_rate": 3.8056316126611083e-06,
      "loss": 0.8757,
      "step": 3984
    },
    {
      "epoch": 4.877600979192167,
      "grad_norm": 1.3576843538700336,
      "learning_rate": 3.8050850333812013e-06,
      "loss": 0.6942,
      "step": 3985
    },
    {
      "epoch": 4.878824969400245,
      "grad_norm": 1.5748885387767562,
      "learning_rate": 3.8045383683359316e-06,
      "loss": 0.8926,
      "step": 3986
    },
    {
      "epoch": 4.880048959608323,
      "grad_norm": 1.1435382392174436,
      "learning_rate": 3.8039916175612274e-06,
      "loss": 0.5699,
      "step": 3987
    },
    {
      "epoch": 4.881272949816402,
      "grad_norm": 0.8300735069031168,
      "learning_rate": 3.8034447810930165e-06,
      "loss": 0.6866,
      "step": 3988
    },
    {
      "epoch": 4.882496940024479,
      "grad_norm": 1.0290272084837009,
      "learning_rate": 3.802897858967235e-06,
      "loss": 0.7894,
      "step": 3989
    },
    {
      "epoch": 4.883720930232558,
      "grad_norm": 1.3477011852888152,
      "learning_rate": 3.802350851219826e-06,
      "loss": 0.5085,
      "step": 3990
    },
    {
      "epoch": 4.8849449204406366,
      "grad_norm": 1.3971588342359387,
      "learning_rate": 3.801803757886736e-06,
      "loss": 0.5261,
      "step": 3991
    },
    {
      "epoch": 4.886168910648715,
      "grad_norm": 0.9807759244835675,
      "learning_rate": 3.801256579003918e-06,
      "loss": 0.8496,
      "step": 3992
    },
    {
      "epoch": 4.887392900856793,
      "grad_norm": 1.686594222226998,
      "learning_rate": 3.800709314607331e-06,
      "loss": 1.0612,
      "step": 3993
    },
    {
      "epoch": 4.8886168910648715,
      "grad_norm": 1.140738474425307,
      "learning_rate": 3.8001619647329386e-06,
      "loss": 0.4646,
      "step": 3994
    },
    {
      "epoch": 4.88984088127295,
      "grad_norm": 2.792143659579334,
      "learning_rate": 3.7996145294167102e-06,
      "loss": 0.4332,
      "step": 3995
    },
    {
      "epoch": 4.891064871481028,
      "grad_norm": 1.9653949758033364,
      "learning_rate": 3.799067008694623e-06,
      "loss": 0.7882,
      "step": 3996
    },
    {
      "epoch": 4.8922888616891065,
      "grad_norm": 1.0705272891708633,
      "learning_rate": 3.7985194026026563e-06,
      "loss": 0.8424,
      "step": 3997
    },
    {
      "epoch": 4.893512851897185,
      "grad_norm": 1.1114167521064,
      "learning_rate": 3.7979717111767972e-06,
      "loss": 0.5117,
      "step": 3998
    },
    {
      "epoch": 4.894736842105263,
      "grad_norm": 0.7056838581467734,
      "learning_rate": 3.797423934453038e-06,
      "loss": 0.4291,
      "step": 3999
    },
    {
      "epoch": 4.8959608323133414,
      "grad_norm": 1.1894898200360482,
      "learning_rate": 3.796876072467377e-06,
      "loss": 0.5197,
      "step": 4000
    },
    {
      "epoch": 4.89718482252142,
      "grad_norm": 2.376990553253133,
      "learning_rate": 3.7963281252558177e-06,
      "loss": 0.6507,
      "step": 4001
    },
    {
      "epoch": 4.898408812729498,
      "grad_norm": 1.6233673972836817,
      "learning_rate": 3.7957800928543686e-06,
      "loss": 0.5149,
      "step": 4002
    },
    {
      "epoch": 4.899632802937576,
      "grad_norm": 1.4097753879579535,
      "learning_rate": 3.7952319752990434e-06,
      "loss": 0.4807,
      "step": 4003
    },
    {
      "epoch": 4.900856793145655,
      "grad_norm": 1.8043290723005447,
      "learning_rate": 3.7946837726258645e-06,
      "loss": 0.6356,
      "step": 4004
    },
    {
      "epoch": 4.902080783353734,
      "grad_norm": 1.0988203550755848,
      "learning_rate": 3.7941354848708576e-06,
      "loss": 0.6945,
      "step": 4005
    },
    {
      "epoch": 4.903304773561811,
      "grad_norm": 1.3327654516188738,
      "learning_rate": 3.793587112070052e-06,
      "loss": 1.0456,
      "step": 4006
    },
    {
      "epoch": 4.90452876376989,
      "grad_norm": 2.3967307475244017,
      "learning_rate": 3.7930386542594866e-06,
      "loss": 0.3715,
      "step": 4007
    },
    {
      "epoch": 4.905752753977968,
      "grad_norm": 1.8542217101040182,
      "learning_rate": 3.792490111475203e-06,
      "loss": 0.3868,
      "step": 4008
    },
    {
      "epoch": 4.906976744186046,
      "grad_norm": 1.1357612820032272,
      "learning_rate": 3.791941483753251e-06,
      "loss": 0.6446,
      "step": 4009
    },
    {
      "epoch": 4.908200734394125,
      "grad_norm": 2.384612290663425,
      "learning_rate": 3.7913927711296815e-06,
      "loss": 0.796,
      "step": 4010
    },
    {
      "epoch": 4.909424724602204,
      "grad_norm": 0.8134532207235741,
      "learning_rate": 3.790843973640557e-06,
      "loss": 0.5313,
      "step": 4011
    },
    {
      "epoch": 4.910648714810281,
      "grad_norm": 1.3590158678039739,
      "learning_rate": 3.7902950913219403e-06,
      "loss": 0.8889,
      "step": 4012
    },
    {
      "epoch": 4.91187270501836,
      "grad_norm": 2.049161187157738,
      "learning_rate": 3.789746124209903e-06,
      "loss": 1.0015,
      "step": 4013
    },
    {
      "epoch": 4.9130966952264385,
      "grad_norm": 1.8265592203614736,
      "learning_rate": 3.7891970723405213e-06,
      "loss": 0.5882,
      "step": 4014
    },
    {
      "epoch": 4.914320685434516,
      "grad_norm": 1.8338230930272101,
      "learning_rate": 3.7886479357498757e-06,
      "loss": 1.2457,
      "step": 4015
    },
    {
      "epoch": 4.915544675642595,
      "grad_norm": 0.9513454296722619,
      "learning_rate": 3.7880987144740543e-06,
      "loss": 0.7097,
      "step": 4016
    },
    {
      "epoch": 4.9167686658506735,
      "grad_norm": 0.9947612670788142,
      "learning_rate": 3.7875494085491503e-06,
      "loss": 0.7452,
      "step": 4017
    },
    {
      "epoch": 4.917992656058751,
      "grad_norm": 1.6880385776084819,
      "learning_rate": 3.787000018011261e-06,
      "loss": 0.596,
      "step": 4018
    },
    {
      "epoch": 4.91921664626683,
      "grad_norm": 1.5856400415295477,
      "learning_rate": 3.7864505428964904e-06,
      "loss": 0.6951,
      "step": 4019
    },
    {
      "epoch": 4.9204406364749085,
      "grad_norm": 1.6654853768227722,
      "learning_rate": 3.7859009832409486e-06,
      "loss": 0.623,
      "step": 4020
    },
    {
      "epoch": 4.921664626682986,
      "grad_norm": 1.8437118208148826,
      "learning_rate": 3.7853513390807506e-06,
      "loss": 0.4773,
      "step": 4021
    },
    {
      "epoch": 4.922888616891065,
      "grad_norm": 1.5850634575645626,
      "learning_rate": 3.7848016104520173e-06,
      "loss": 1.0029,
      "step": 4022
    },
    {
      "epoch": 4.9241126070991434,
      "grad_norm": 0.996284996296396,
      "learning_rate": 3.7842517973908734e-06,
      "loss": 0.4957,
      "step": 4023
    },
    {
      "epoch": 4.925336597307221,
      "grad_norm": 1.5021426947182797,
      "learning_rate": 3.783701899933452e-06,
      "loss": 1.0912,
      "step": 4024
    },
    {
      "epoch": 4.9265605875153,
      "grad_norm": 1.6380761658031842,
      "learning_rate": 3.7831519181158897e-06,
      "loss": 0.3297,
      "step": 4025
    },
    {
      "epoch": 4.927784577723378,
      "grad_norm": 1.5925334332290995,
      "learning_rate": 3.7826018519743298e-06,
      "loss": 0.4794,
      "step": 4026
    },
    {
      "epoch": 4.929008567931456,
      "grad_norm": 1.676305339551745,
      "learning_rate": 3.782051701544919e-06,
      "loss": 1.0357,
      "step": 4027
    },
    {
      "epoch": 4.930232558139535,
      "grad_norm": 1.0165943256155465,
      "learning_rate": 3.781501466863814e-06,
      "loss": 0.3951,
      "step": 4028
    },
    {
      "epoch": 4.931456548347613,
      "grad_norm": 1.7310555124349274,
      "learning_rate": 3.7809511479671712e-06,
      "loss": 0.3738,
      "step": 4029
    },
    {
      "epoch": 4.932680538555692,
      "grad_norm": 0.8951008335490757,
      "learning_rate": 3.7804007448911574e-06,
      "loss": 0.4617,
      "step": 4030
    },
    {
      "epoch": 4.93390452876377,
      "grad_norm": 1.889247762461582,
      "learning_rate": 3.779850257671943e-06,
      "loss": 0.831,
      "step": 4031
    },
    {
      "epoch": 4.935128518971848,
      "grad_norm": 2.3886581051678903,
      "learning_rate": 3.779299686345702e-06,
      "loss": 0.5053,
      "step": 4032
    },
    {
      "epoch": 4.936352509179926,
      "grad_norm": 0.9000012224203673,
      "learning_rate": 3.7787490309486186e-06,
      "loss": 0.4659,
      "step": 4033
    },
    {
      "epoch": 4.937576499388005,
      "grad_norm": 1.1868639898759117,
      "learning_rate": 3.778198291516879e-06,
      "loss": 0.9462,
      "step": 4034
    },
    {
      "epoch": 4.938800489596083,
      "grad_norm": 1.317697712709867,
      "learning_rate": 3.7776474680866746e-06,
      "loss": 1.0201,
      "step": 4035
    },
    {
      "epoch": 4.940024479804162,
      "grad_norm": 2.2409038538264348,
      "learning_rate": 3.7770965606942045e-06,
      "loss": 0.4071,
      "step": 4036
    },
    {
      "epoch": 4.94124847001224,
      "grad_norm": 1.420891445048487,
      "learning_rate": 3.7765455693756724e-06,
      "loss": 0.5668,
      "step": 4037
    },
    {
      "epoch": 4.942472460220318,
      "grad_norm": 1.198143578888788,
      "learning_rate": 3.7759944941672865e-06,
      "loss": 1.589,
      "step": 4038
    },
    {
      "epoch": 4.943696450428397,
      "grad_norm": 1.2616164036216413,
      "learning_rate": 3.775443335105262e-06,
      "loss": 0.5881,
      "step": 4039
    },
    {
      "epoch": 4.944920440636475,
      "grad_norm": 1.5909513313853134,
      "learning_rate": 3.7748920922258196e-06,
      "loss": 1.1935,
      "step": 4040
    },
    {
      "epoch": 4.946144430844553,
      "grad_norm": 0.987817198283476,
      "learning_rate": 3.7743407655651844e-06,
      "loss": 0.5431,
      "step": 4041
    },
    {
      "epoch": 4.947368421052632,
      "grad_norm": 1.3793093774357976,
      "learning_rate": 3.773789355159587e-06,
      "loss": 0.4081,
      "step": 4042
    },
    {
      "epoch": 4.94859241126071,
      "grad_norm": 1.7640209703743965,
      "learning_rate": 3.7732378610452656e-06,
      "loss": 0.4732,
      "step": 4043
    },
    {
      "epoch": 4.949816401468788,
      "grad_norm": 2.034890475366439,
      "learning_rate": 3.772686283258461e-06,
      "loss": 0.3994,
      "step": 4044
    },
    {
      "epoch": 4.951040391676867,
      "grad_norm": 1.0971562469949732,
      "learning_rate": 3.7721346218354205e-06,
      "loss": 0.8991,
      "step": 4045
    },
    {
      "epoch": 4.9522643818849446,
      "grad_norm": 1.229640776022709,
      "learning_rate": 3.7715828768123984e-06,
      "loss": 0.5297,
      "step": 4046
    },
    {
      "epoch": 4.953488372093023,
      "grad_norm": 1.767418976903907,
      "learning_rate": 3.771031048225653e-06,
      "loss": 0.7204,
      "step": 4047
    },
    {
      "epoch": 4.954712362301102,
      "grad_norm": 1.4262609396107837,
      "learning_rate": 3.770479136111448e-06,
      "loss": 0.6621,
      "step": 4048
    },
    {
      "epoch": 4.95593635250918,
      "grad_norm": 2.0018096356307784,
      "learning_rate": 3.7699271405060534e-06,
      "loss": 0.5521,
      "step": 4049
    },
    {
      "epoch": 4.957160342717258,
      "grad_norm": 1.8239283581930785,
      "learning_rate": 3.7693750614457447e-06,
      "loss": 0.5785,
      "step": 4050
    },
    {
      "epoch": 4.958384332925337,
      "grad_norm": 1.0005215730288117,
      "learning_rate": 3.768822898966802e-06,
      "loss": 0.757,
      "step": 4051
    },
    {
      "epoch": 4.9596083231334145,
      "grad_norm": 0.993603048927035,
      "learning_rate": 3.7682706531055112e-06,
      "loss": 0.4965,
      "step": 4052
    },
    {
      "epoch": 4.960832313341493,
      "grad_norm": 1.1707175819987075,
      "learning_rate": 3.7677183238981637e-06,
      "loss": 0.4237,
      "step": 4053
    },
    {
      "epoch": 4.962056303549572,
      "grad_norm": 1.195842628875113,
      "learning_rate": 3.767165911381057e-06,
      "loss": 0.9499,
      "step": 4054
    },
    {
      "epoch": 4.96328029375765,
      "grad_norm": 1.516191082446433,
      "learning_rate": 3.766613415590494e-06,
      "loss": 0.8149,
      "step": 4055
    },
    {
      "epoch": 4.964504283965728,
      "grad_norm": 1.4816079796836763,
      "learning_rate": 3.7660608365627823e-06,
      "loss": 0.5304,
      "step": 4056
    },
    {
      "epoch": 4.965728274173807,
      "grad_norm": 0.8321710354434264,
      "learning_rate": 3.765508174334235e-06,
      "loss": 0.7032,
      "step": 4057
    },
    {
      "epoch": 4.966952264381885,
      "grad_norm": 0.949275790272603,
      "learning_rate": 3.764955428941171e-06,
      "loss": 0.5452,
      "step": 4058
    },
    {
      "epoch": 4.968176254589963,
      "grad_norm": 0.7330766998440373,
      "learning_rate": 3.7644026004199153e-06,
      "loss": 0.4548,
      "step": 4059
    },
    {
      "epoch": 4.969400244798042,
      "grad_norm": 1.4541933343538465,
      "learning_rate": 3.7638496888067977e-06,
      "loss": 1.5921,
      "step": 4060
    },
    {
      "epoch": 4.97062423500612,
      "grad_norm": 1.25206838591234,
      "learning_rate": 3.763296694138152e-06,
      "loss": 0.9662,
      "step": 4061
    },
    {
      "epoch": 4.971848225214198,
      "grad_norm": 1.4262675088837862,
      "learning_rate": 3.762743616450321e-06,
      "loss": 0.9633,
      "step": 4062
    },
    {
      "epoch": 4.973072215422277,
      "grad_norm": 0.7185696974661676,
      "learning_rate": 3.76219045577965e-06,
      "loss": 0.5432,
      "step": 4063
    },
    {
      "epoch": 4.974296205630355,
      "grad_norm": 0.9837601685795316,
      "learning_rate": 3.761637212162491e-06,
      "loss": 0.6323,
      "step": 4064
    },
    {
      "epoch": 4.975520195838433,
      "grad_norm": 1.0909951142774328,
      "learning_rate": 3.761083885635201e-06,
      "loss": 0.9187,
      "step": 4065
    },
    {
      "epoch": 4.976744186046512,
      "grad_norm": 0.853095821941795,
      "learning_rate": 3.760530476234142e-06,
      "loss": 0.5877,
      "step": 4066
    },
    {
      "epoch": 4.97796817625459,
      "grad_norm": 1.2293647125410725,
      "learning_rate": 3.759976983995683e-06,
      "loss": 0.7243,
      "step": 4067
    },
    {
      "epoch": 4.979192166462668,
      "grad_norm": 1.4547359687504466,
      "learning_rate": 3.7594234089561966e-06,
      "loss": 0.5647,
      "step": 4068
    },
    {
      "epoch": 4.9804161566707466,
      "grad_norm": 1.5669472031218648,
      "learning_rate": 3.758869751152062e-06,
      "loss": 1.1577,
      "step": 4069
    },
    {
      "epoch": 4.981640146878825,
      "grad_norm": 0.955509897662767,
      "learning_rate": 3.758316010619664e-06,
      "loss": 0.4811,
      "step": 4070
    },
    {
      "epoch": 4.982864137086903,
      "grad_norm": 1.0703371762364289,
      "learning_rate": 3.757762187395391e-06,
      "loss": 0.6559,
      "step": 4071
    },
    {
      "epoch": 4.9840881272949815,
      "grad_norm": 1.1282513397309308,
      "learning_rate": 3.7572082815156407e-06,
      "loss": 0.6514,
      "step": 4072
    },
    {
      "epoch": 4.98531211750306,
      "grad_norm": 1.2305169584881515,
      "learning_rate": 3.7566542930168103e-06,
      "loss": 0.5804,
      "step": 4073
    },
    {
      "epoch": 4.986536107711139,
      "grad_norm": 1.1097513332192956,
      "learning_rate": 3.7561002219353092e-06,
      "loss": 0.6243,
      "step": 4074
    },
    {
      "epoch": 4.9877600979192165,
      "grad_norm": 1.3417008359444649,
      "learning_rate": 3.755546068307547e-06,
      "loss": 0.4562,
      "step": 4075
    },
    {
      "epoch": 4.988984088127295,
      "grad_norm": 1.6089805662552246,
      "learning_rate": 3.754991832169942e-06,
      "loss": 0.4963,
      "step": 4076
    },
    {
      "epoch": 4.990208078335373,
      "grad_norm": 1.3370186700907496,
      "learning_rate": 3.754437513558914e-06,
      "loss": 0.54,
      "step": 4077
    },
    {
      "epoch": 4.9914320685434515,
      "grad_norm": 1.1741392124636132,
      "learning_rate": 3.753883112510894e-06,
      "loss": 0.6535,
      "step": 4078
    },
    {
      "epoch": 4.99265605875153,
      "grad_norm": 1.0396880202292036,
      "learning_rate": 3.753328629062313e-06,
      "loss": 0.4897,
      "step": 4079
    },
    {
      "epoch": 4.993880048959609,
      "grad_norm": 1.2740686738281883,
      "learning_rate": 3.75277406324961e-06,
      "loss": 1.4296,
      "step": 4080
    },
    {
      "epoch": 4.995104039167686,
      "grad_norm": 1.065512737727095,
      "learning_rate": 3.7522194151092305e-06,
      "loss": 0.8055,
      "step": 4081
    },
    {
      "epoch": 4.996328029375765,
      "grad_norm": 2.0131264303419667,
      "learning_rate": 3.751664684677621e-06,
      "loss": 0.8663,
      "step": 4082
    },
    {
      "epoch": 4.997552019583844,
      "grad_norm": 1.9110052523783156,
      "learning_rate": 3.7511098719912385e-06,
      "loss": 0.3287,
      "step": 4083
    },
    {
      "epoch": 4.998776009791921,
      "grad_norm": 1.4078873348147225,
      "learning_rate": 3.7505549770865437e-06,
      "loss": 0.5291,
      "step": 4084
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.2426745584441603,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.5178,
      "step": 4085
    },
    {
      "epoch": 5.001223990208079,
      "grad_norm": 0.879063109954564,
      "learning_rate": 3.7494449407680806e-06,
      "loss": 0.5334,
      "step": 4086
    },
    {
      "epoch": 5.002447980416156,
      "grad_norm": 1.2929674394615043,
      "learning_rate": 3.748889799427261e-06,
      "loss": 1.1984,
      "step": 4087
    },
    {
      "epoch": 5.003671970624235,
      "grad_norm": 0.7850779611533373,
      "learning_rate": 3.748334576014023e-06,
      "loss": 0.4127,
      "step": 4088
    },
    {
      "epoch": 5.004895960832314,
      "grad_norm": 1.37050649233921,
      "learning_rate": 3.7477792705648536e-06,
      "loss": 0.4912,
      "step": 4089
    },
    {
      "epoch": 5.006119951040391,
      "grad_norm": 1.3461191383221272,
      "learning_rate": 3.747223883116246e-06,
      "loss": 0.4866,
      "step": 4090
    },
    {
      "epoch": 5.00734394124847,
      "grad_norm": 1.439936680629212,
      "learning_rate": 3.746668413704698e-06,
      "loss": 0.5285,
      "step": 4091
    },
    {
      "epoch": 5.0085679314565485,
      "grad_norm": 1.2370517721579402,
      "learning_rate": 3.7461128623667126e-06,
      "loss": 1.1801,
      "step": 4092
    },
    {
      "epoch": 5.009791921664626,
      "grad_norm": 0.9368490407702835,
      "learning_rate": 3.7455572291388e-06,
      "loss": 0.7959,
      "step": 4093
    },
    {
      "epoch": 5.011015911872705,
      "grad_norm": 2.1715713162912396,
      "learning_rate": 3.7450015140574724e-06,
      "loss": 0.3894,
      "step": 4094
    },
    {
      "epoch": 5.0122399020807835,
      "grad_norm": 1.1576250475191492,
      "learning_rate": 3.7444457171592496e-06,
      "loss": 1.4326,
      "step": 4095
    },
    {
      "epoch": 5.013463892288861,
      "grad_norm": 1.785604686964227,
      "learning_rate": 3.7438898384806586e-06,
      "loss": 0.9932,
      "step": 4096
    },
    {
      "epoch": 5.01468788249694,
      "grad_norm": 1.8536028120383583,
      "learning_rate": 3.7433338780582272e-06,
      "loss": 0.6609,
      "step": 4097
    },
    {
      "epoch": 5.0159118727050185,
      "grad_norm": 1.0548000609628008,
      "learning_rate": 3.7427778359284927e-06,
      "loss": 0.9104,
      "step": 4098
    },
    {
      "epoch": 5.017135862913097,
      "grad_norm": 1.0968286145308728,
      "learning_rate": 3.742221712127995e-06,
      "loss": 1.3288,
      "step": 4099
    },
    {
      "epoch": 5.018359853121175,
      "grad_norm": 0.7925730418467919,
      "learning_rate": 3.7416655066932818e-06,
      "loss": 0.6108,
      "step": 4100
    },
    {
      "epoch": 5.0195838433292534,
      "grad_norm": 0.8513827638255641,
      "learning_rate": 3.7411092196609035e-06,
      "loss": 0.6235,
      "step": 4101
    },
    {
      "epoch": 5.020807833537332,
      "grad_norm": 1.0823604876770776,
      "learning_rate": 3.7405528510674183e-06,
      "loss": 0.3498,
      "step": 4102
    },
    {
      "epoch": 5.02203182374541,
      "grad_norm": 1.3650857619678705,
      "learning_rate": 3.7399964009493874e-06,
      "loss": 0.4554,
      "step": 4103
    },
    {
      "epoch": 5.023255813953488,
      "grad_norm": 1.629904401306993,
      "learning_rate": 3.7394398693433798e-06,
      "loss": 0.4298,
      "step": 4104
    },
    {
      "epoch": 5.024479804161567,
      "grad_norm": 1.0138147831073214,
      "learning_rate": 3.738883256285969e-06,
      "loss": 0.7369,
      "step": 4105
    },
    {
      "epoch": 5.025703794369645,
      "grad_norm": 1.3697913775691386,
      "learning_rate": 3.7383265618137322e-06,
      "loss": 0.5079,
      "step": 4106
    },
    {
      "epoch": 5.026927784577723,
      "grad_norm": 1.1908476921551605,
      "learning_rate": 3.7377697859632543e-06,
      "loss": 0.9422,
      "step": 4107
    },
    {
      "epoch": 5.028151774785802,
      "grad_norm": 1.0493736635141853,
      "learning_rate": 3.7372129287711247e-06,
      "loss": 1.1541,
      "step": 4108
    },
    {
      "epoch": 5.02937576499388,
      "grad_norm": 1.7049165168200764,
      "learning_rate": 3.7366559902739368e-06,
      "loss": 0.9547,
      "step": 4109
    },
    {
      "epoch": 5.030599755201958,
      "grad_norm": 1.3537115414480767,
      "learning_rate": 3.7360989705082917e-06,
      "loss": 1.4369,
      "step": 4110
    },
    {
      "epoch": 5.031823745410037,
      "grad_norm": 0.6896965918692632,
      "learning_rate": 3.735541869510794e-06,
      "loss": 0.4776,
      "step": 4111
    },
    {
      "epoch": 5.033047735618115,
      "grad_norm": 0.9936970077908196,
      "learning_rate": 3.7349846873180544e-06,
      "loss": 0.3902,
      "step": 4112
    },
    {
      "epoch": 5.034271725826193,
      "grad_norm": 0.9055234887371305,
      "learning_rate": 3.7344274239666893e-06,
      "loss": 0.623,
      "step": 4113
    },
    {
      "epoch": 5.035495716034272,
      "grad_norm": 0.9707554280565841,
      "learning_rate": 3.7338700794933204e-06,
      "loss": 0.5349,
      "step": 4114
    },
    {
      "epoch": 5.03671970624235,
      "grad_norm": 1.6356064623884405,
      "learning_rate": 3.7333126539345725e-06,
      "loss": 1.0474,
      "step": 4115
    },
    {
      "epoch": 5.037943696450428,
      "grad_norm": 1.9145207426846733,
      "learning_rate": 3.7327551473270798e-06,
      "loss": 0.7126,
      "step": 4116
    },
    {
      "epoch": 5.039167686658507,
      "grad_norm": 1.3860322026856795,
      "learning_rate": 3.7321975597074777e-06,
      "loss": 0.5156,
      "step": 4117
    },
    {
      "epoch": 5.0403916768665855,
      "grad_norm": 1.3506605803598457,
      "learning_rate": 3.7316398911124097e-06,
      "loss": 0.4198,
      "step": 4118
    },
    {
      "epoch": 5.041615667074663,
      "grad_norm": 1.7236380888291132,
      "learning_rate": 3.7310821415785236e-06,
      "loss": 0.4653,
      "step": 4119
    },
    {
      "epoch": 5.042839657282742,
      "grad_norm": 1.4444066149651857,
      "learning_rate": 3.7305243111424725e-06,
      "loss": 1.4217,
      "step": 4120
    },
    {
      "epoch": 5.0440636474908205,
      "grad_norm": 1.9240292627805193,
      "learning_rate": 3.7299663998409153e-06,
      "loss": 0.4382,
      "step": 4121
    },
    {
      "epoch": 5.045287637698898,
      "grad_norm": 1.1667152231757323,
      "learning_rate": 3.7294084077105153e-06,
      "loss": 0.3866,
      "step": 4122
    },
    {
      "epoch": 5.046511627906977,
      "grad_norm": 1.137813148904686,
      "learning_rate": 3.7288503347879423e-06,
      "loss": 0.5611,
      "step": 4123
    },
    {
      "epoch": 5.0477356181150554,
      "grad_norm": 1.3882770754921245,
      "learning_rate": 3.7282921811098706e-06,
      "loss": 0.4719,
      "step": 4124
    },
    {
      "epoch": 5.048959608323133,
      "grad_norm": 1.1697571681947387,
      "learning_rate": 3.72773394671298e-06,
      "loss": 0.5961,
      "step": 4125
    },
    {
      "epoch": 5.050183598531212,
      "grad_norm": 1.356974391428421,
      "learning_rate": 3.7271756316339553e-06,
      "loss": 0.6464,
      "step": 4126
    },
    {
      "epoch": 5.05140758873929,
      "grad_norm": 1.4844829647516093,
      "learning_rate": 3.7266172359094864e-06,
      "loss": 1.0278,
      "step": 4127
    },
    {
      "epoch": 5.052631578947368,
      "grad_norm": 2.0216594710684337,
      "learning_rate": 3.726058759576271e-06,
      "loss": 0.5017,
      "step": 4128
    },
    {
      "epoch": 5.053855569155447,
      "grad_norm": 1.0691567618640823,
      "learning_rate": 3.7255002026710075e-06,
      "loss": 0.7224,
      "step": 4129
    },
    {
      "epoch": 5.055079559363525,
      "grad_norm": 2.701848169656287,
      "learning_rate": 3.7249415652304044e-06,
      "loss": 0.8036,
      "step": 4130
    },
    {
      "epoch": 5.056303549571603,
      "grad_norm": 1.131681151158039,
      "learning_rate": 3.7243828472911727e-06,
      "loss": 1.0979,
      "step": 4131
    },
    {
      "epoch": 5.057527539779682,
      "grad_norm": 1.4617588409866498,
      "learning_rate": 3.7238240488900277e-06,
      "loss": 1.9094,
      "step": 4132
    },
    {
      "epoch": 5.05875152998776,
      "grad_norm": 1.2104573817211184,
      "learning_rate": 3.7232651700636936e-06,
      "loss": 1.1141,
      "step": 4133
    },
    {
      "epoch": 5.059975520195838,
      "grad_norm": 1.495578480473626,
      "learning_rate": 3.722706210848897e-06,
      "loss": 0.4783,
      "step": 4134
    },
    {
      "epoch": 5.061199510403917,
      "grad_norm": 1.263405021383673,
      "learning_rate": 3.7221471712823707e-06,
      "loss": 1.2573,
      "step": 4135
    },
    {
      "epoch": 5.062423500611995,
      "grad_norm": 1.8459275100418115,
      "learning_rate": 3.7215880514008525e-06,
      "loss": 1.234,
      "step": 4136
    },
    {
      "epoch": 5.063647490820073,
      "grad_norm": 1.3739367706553296,
      "learning_rate": 3.7210288512410865e-06,
      "loss": 0.4993,
      "step": 4137
    },
    {
      "epoch": 5.064871481028152,
      "grad_norm": 1.1289127087500335,
      "learning_rate": 3.7204695708398198e-06,
      "loss": 0.9894,
      "step": 4138
    },
    {
      "epoch": 5.06609547123623,
      "grad_norm": 2.113310230898397,
      "learning_rate": 3.7199102102338075e-06,
      "loss": 0.4552,
      "step": 4139
    },
    {
      "epoch": 5.067319461444308,
      "grad_norm": 1.1999709750403422,
      "learning_rate": 3.7193507694598084e-06,
      "loss": 0.5266,
      "step": 4140
    },
    {
      "epoch": 5.068543451652387,
      "grad_norm": 1.1937346031405207,
      "learning_rate": 3.7187912485545867e-06,
      "loss": 0.318,
      "step": 4141
    },
    {
      "epoch": 5.069767441860465,
      "grad_norm": 2.0589744143496036,
      "learning_rate": 3.718231647554912e-06,
      "loss": 0.4725,
      "step": 4142
    },
    {
      "epoch": 5.070991432068544,
      "grad_norm": 1.799782457581629,
      "learning_rate": 3.7176719664975604e-06,
      "loss": 0.6213,
      "step": 4143
    },
    {
      "epoch": 5.072215422276622,
      "grad_norm": 1.3056023423748775,
      "learning_rate": 3.7171122054193093e-06,
      "loss": 1.0377,
      "step": 4144
    },
    {
      "epoch": 5.0734394124847,
      "grad_norm": 0.8847452985521517,
      "learning_rate": 3.7165523643569466e-06,
      "loss": 0.7149,
      "step": 4145
    },
    {
      "epoch": 5.074663402692779,
      "grad_norm": 1.0880395155092983,
      "learning_rate": 3.7159924433472635e-06,
      "loss": 1.5937,
      "step": 4146
    },
    {
      "epoch": 5.0758873929008566,
      "grad_norm": 0.8108060198005063,
      "learning_rate": 3.7154324424270526e-06,
      "loss": 0.5517,
      "step": 4147
    },
    {
      "epoch": 5.077111383108935,
      "grad_norm": 1.3199667032488156,
      "learning_rate": 3.7148723616331184e-06,
      "loss": 0.5656,
      "step": 4148
    },
    {
      "epoch": 5.078335373317014,
      "grad_norm": 0.7923682615015675,
      "learning_rate": 3.7143122010022657e-06,
      "loss": 0.6491,
      "step": 4149
    },
    {
      "epoch": 5.0795593635250915,
      "grad_norm": 1.1042858860232514,
      "learning_rate": 3.713751960571307e-06,
      "loss": 1.2991,
      "step": 4150
    },
    {
      "epoch": 5.08078335373317,
      "grad_norm": 1.6440592353873784,
      "learning_rate": 3.713191640377058e-06,
      "loss": 0.5887,
      "step": 4151
    },
    {
      "epoch": 5.082007343941249,
      "grad_norm": 1.1036556716091726,
      "learning_rate": 3.7126312404563425e-06,
      "loss": 0.4735,
      "step": 4152
    },
    {
      "epoch": 5.0832313341493265,
      "grad_norm": 0.9546618077624657,
      "learning_rate": 3.7120707608459862e-06,
      "loss": 0.5163,
      "step": 4153
    },
    {
      "epoch": 5.084455324357405,
      "grad_norm": 1.4016475397757442,
      "learning_rate": 3.711510201582823e-06,
      "loss": 0.9069,
      "step": 4154
    },
    {
      "epoch": 5.085679314565484,
      "grad_norm": 1.9474226357838167,
      "learning_rate": 3.7109495627036907e-06,
      "loss": 0.5026,
      "step": 4155
    },
    {
      "epoch": 5.0869033047735615,
      "grad_norm": 1.531924578440938,
      "learning_rate": 3.7103888442454323e-06,
      "loss": 0.5163,
      "step": 4156
    },
    {
      "epoch": 5.08812729498164,
      "grad_norm": 1.2275641519446574,
      "learning_rate": 3.709828046244896e-06,
      "loss": 1.0355,
      "step": 4157
    },
    {
      "epoch": 5.089351285189719,
      "grad_norm": 1.3145639856708806,
      "learning_rate": 3.7092671687389347e-06,
      "loss": 0.8858,
      "step": 4158
    },
    {
      "epoch": 5.090575275397796,
      "grad_norm": 1.720542482863626,
      "learning_rate": 3.7087062117644086e-06,
      "loss": 0.4566,
      "step": 4159
    },
    {
      "epoch": 5.091799265605875,
      "grad_norm": 1.618040203535799,
      "learning_rate": 3.7081451753581804e-06,
      "loss": 1.2262,
      "step": 4160
    },
    {
      "epoch": 5.093023255813954,
      "grad_norm": 1.602327120837178,
      "learning_rate": 3.70758405955712e-06,
      "loss": 0.6385,
      "step": 4161
    },
    {
      "epoch": 5.094247246022032,
      "grad_norm": 1.4034583226792785,
      "learning_rate": 3.7070228643981014e-06,
      "loss": 1.6267,
      "step": 4162
    },
    {
      "epoch": 5.09547123623011,
      "grad_norm": 1.0862418430091398,
      "learning_rate": 3.706461589918005e-06,
      "loss": 0.5545,
      "step": 4163
    },
    {
      "epoch": 5.096695226438189,
      "grad_norm": 1.2184890344342532,
      "learning_rate": 3.7059002361537157e-06,
      "loss": 0.9522,
      "step": 4164
    },
    {
      "epoch": 5.097919216646267,
      "grad_norm": 0.6731944428744475,
      "learning_rate": 3.705338803142123e-06,
      "loss": 0.3624,
      "step": 4165
    },
    {
      "epoch": 5.099143206854345,
      "grad_norm": 0.9512192731237781,
      "learning_rate": 3.704777290920122e-06,
      "loss": 0.676,
      "step": 4166
    },
    {
      "epoch": 5.100367197062424,
      "grad_norm": 2.041530024402415,
      "learning_rate": 3.7042156995246136e-06,
      "loss": 0.8525,
      "step": 4167
    },
    {
      "epoch": 5.101591187270502,
      "grad_norm": 1.5954274062483031,
      "learning_rate": 3.703654028992503e-06,
      "loss": 0.8653,
      "step": 4168
    },
    {
      "epoch": 5.10281517747858,
      "grad_norm": 1.8890975149014362,
      "learning_rate": 3.7030922793607033e-06,
      "loss": 0.457,
      "step": 4169
    },
    {
      "epoch": 5.1040391676866586,
      "grad_norm": 0.9512379699183214,
      "learning_rate": 3.702530450666128e-06,
      "loss": 0.7673,
      "step": 4170
    },
    {
      "epoch": 5.105263157894737,
      "grad_norm": 1.1872806363991375,
      "learning_rate": 3.7019685429456986e-06,
      "loss": 0.7139,
      "step": 4171
    },
    {
      "epoch": 5.106487148102815,
      "grad_norm": 1.1928107148642928,
      "learning_rate": 3.7014065562363433e-06,
      "loss": 1.0118,
      "step": 4172
    },
    {
      "epoch": 5.1077111383108935,
      "grad_norm": 1.56796414804517,
      "learning_rate": 3.700844490574993e-06,
      "loss": 1.3419,
      "step": 4173
    },
    {
      "epoch": 5.108935128518972,
      "grad_norm": 1.8714973853575212,
      "learning_rate": 3.700282345998584e-06,
      "loss": 0.9021,
      "step": 4174
    },
    {
      "epoch": 5.11015911872705,
      "grad_norm": 0.8545078042749233,
      "learning_rate": 3.699720122544058e-06,
      "loss": 0.5572,
      "step": 4175
    },
    {
      "epoch": 5.1113831089351285,
      "grad_norm": 1.2603126544688865,
      "learning_rate": 3.699157820248364e-06,
      "loss": 0.4862,
      "step": 4176
    },
    {
      "epoch": 5.112607099143207,
      "grad_norm": 1.2421682439620025,
      "learning_rate": 3.698595439148453e-06,
      "loss": 0.6311,
      "step": 4177
    },
    {
      "epoch": 5.113831089351285,
      "grad_norm": 1.045275274719113,
      "learning_rate": 3.6980329792812834e-06,
      "loss": 1.3553,
      "step": 4178
    },
    {
      "epoch": 5.1150550795593634,
      "grad_norm": 0.764871336470395,
      "learning_rate": 3.697470440683817e-06,
      "loss": 0.4603,
      "step": 4179
    },
    {
      "epoch": 5.116279069767442,
      "grad_norm": 1.3572763086511919,
      "learning_rate": 3.6969078233930235e-06,
      "loss": 0.324,
      "step": 4180
    },
    {
      "epoch": 5.11750305997552,
      "grad_norm": 1.8504411927839788,
      "learning_rate": 3.696345127445874e-06,
      "loss": 0.5875,
      "step": 4181
    },
    {
      "epoch": 5.118727050183598,
      "grad_norm": 0.8957635388106361,
      "learning_rate": 3.695782352879348e-06,
      "loss": 0.6044,
      "step": 4182
    },
    {
      "epoch": 5.119951040391677,
      "grad_norm": 1.5018970518248727,
      "learning_rate": 3.695219499730428e-06,
      "loss": 1.2084,
      "step": 4183
    },
    {
      "epoch": 5.121175030599755,
      "grad_norm": 1.7669966369496808,
      "learning_rate": 3.694656568036104e-06,
      "loss": 0.5161,
      "step": 4184
    },
    {
      "epoch": 5.122399020807833,
      "grad_norm": 1.9853933456709052,
      "learning_rate": 3.694093557833369e-06,
      "loss": 0.4,
      "step": 4185
    },
    {
      "epoch": 5.123623011015912,
      "grad_norm": 1.1833097307579932,
      "learning_rate": 3.6935304691592216e-06,
      "loss": 1.1906,
      "step": 4186
    },
    {
      "epoch": 5.124847001223991,
      "grad_norm": 2.104227960818154,
      "learning_rate": 3.692967302050667e-06,
      "loss": 0.4266,
      "step": 4187
    },
    {
      "epoch": 5.126070991432068,
      "grad_norm": 1.2816751454893844,
      "learning_rate": 3.6924040565447132e-06,
      "loss": 1.2014,
      "step": 4188
    },
    {
      "epoch": 5.127294981640147,
      "grad_norm": 1.59963374214599,
      "learning_rate": 3.6918407326783755e-06,
      "loss": 0.3379,
      "step": 4189
    },
    {
      "epoch": 5.128518971848226,
      "grad_norm": 0.8847622790177622,
      "learning_rate": 3.6912773304886735e-06,
      "loss": 0.4638,
      "step": 4190
    },
    {
      "epoch": 5.129742962056303,
      "grad_norm": 1.117305048085905,
      "learning_rate": 3.690713850012631e-06,
      "loss": 0.5604,
      "step": 4191
    },
    {
      "epoch": 5.130966952264382,
      "grad_norm": 0.9019646797406172,
      "learning_rate": 3.6901502912872785e-06,
      "loss": 0.4997,
      "step": 4192
    },
    {
      "epoch": 5.1321909424724605,
      "grad_norm": 1.9949322959785634,
      "learning_rate": 3.6895866543496513e-06,
      "loss": 0.7736,
      "step": 4193
    },
    {
      "epoch": 5.133414932680538,
      "grad_norm": 2.392832768104058,
      "learning_rate": 3.689022939236788e-06,
      "loss": 0.6324,
      "step": 4194
    },
    {
      "epoch": 5.134638922888617,
      "grad_norm": 2.3504975349535093,
      "learning_rate": 3.6884591459857366e-06,
      "loss": 0.7423,
      "step": 4195
    },
    {
      "epoch": 5.1358629130966955,
      "grad_norm": 1.424858648740238,
      "learning_rate": 3.6878952746335454e-06,
      "loss": 1.2217,
      "step": 4196
    },
    {
      "epoch": 5.137086903304773,
      "grad_norm": 1.0722400936032446,
      "learning_rate": 3.6873313252172697e-06,
      "loss": 0.5489,
      "step": 4197
    },
    {
      "epoch": 5.138310893512852,
      "grad_norm": 1.1277997100219441,
      "learning_rate": 3.686767297773972e-06,
      "loss": 0.8362,
      "step": 4198
    },
    {
      "epoch": 5.1395348837209305,
      "grad_norm": 1.6709588311305699,
      "learning_rate": 3.6862031923407166e-06,
      "loss": 0.8448,
      "step": 4199
    },
    {
      "epoch": 5.140758873929008,
      "grad_norm": 1.9040741481170438,
      "learning_rate": 3.6856390089545748e-06,
      "loss": 0.4416,
      "step": 4200
    },
    {
      "epoch": 5.141982864137087,
      "grad_norm": 1.902631704634107,
      "learning_rate": 3.685074747652622e-06,
      "loss": 0.8998,
      "step": 4201
    },
    {
      "epoch": 5.1432068543451654,
      "grad_norm": 1.3327495043594968,
      "learning_rate": 3.684510408471942e-06,
      "loss": 0.5968,
      "step": 4202
    },
    {
      "epoch": 5.144430844553243,
      "grad_norm": 2.2138819261694858,
      "learning_rate": 3.6839459914496174e-06,
      "loss": 0.4656,
      "step": 4203
    },
    {
      "epoch": 5.145654834761322,
      "grad_norm": 1.3383867155755353,
      "learning_rate": 3.683381496622742e-06,
      "loss": 0.9334,
      "step": 4204
    },
    {
      "epoch": 5.1468788249694,
      "grad_norm": 1.8122387182328838,
      "learning_rate": 3.6828169240284117e-06,
      "loss": 0.4576,
      "step": 4205
    },
    {
      "epoch": 5.148102815177479,
      "grad_norm": 1.1587037698078475,
      "learning_rate": 3.682252273703728e-06,
      "loss": 0.7065,
      "step": 4206
    },
    {
      "epoch": 5.149326805385557,
      "grad_norm": 1.2836000263235592,
      "learning_rate": 3.6816875456857975e-06,
      "loss": 1.5624,
      "step": 4207
    },
    {
      "epoch": 5.150550795593635,
      "grad_norm": 0.868628635143518,
      "learning_rate": 3.681122740011733e-06,
      "loss": 0.6546,
      "step": 4208
    },
    {
      "epoch": 5.151774785801714,
      "grad_norm": 2.274036871394697,
      "learning_rate": 3.6805578567186494e-06,
      "loss": 0.7296,
      "step": 4209
    },
    {
      "epoch": 5.152998776009792,
      "grad_norm": 1.4012855292464763,
      "learning_rate": 3.6799928958436707e-06,
      "loss": 0.5361,
      "step": 4210
    },
    {
      "epoch": 5.15422276621787,
      "grad_norm": 0.6474501364057795,
      "learning_rate": 3.679427857423924e-06,
      "loss": 0.4437,
      "step": 4211
    },
    {
      "epoch": 5.155446756425949,
      "grad_norm": 1.652837255842577,
      "learning_rate": 3.67886274149654e-06,
      "loss": 0.5843,
      "step": 4212
    },
    {
      "epoch": 5.156670746634027,
      "grad_norm": 1.1994377999079722,
      "learning_rate": 3.6782975480986573e-06,
      "loss": 0.4336,
      "step": 4213
    },
    {
      "epoch": 5.157894736842105,
      "grad_norm": 1.9037169410052552,
      "learning_rate": 3.6777322772674185e-06,
      "loss": 0.3775,
      "step": 4214
    },
    {
      "epoch": 5.159118727050184,
      "grad_norm": 1.8120169682393892,
      "learning_rate": 3.6771669290399697e-06,
      "loss": 0.9274,
      "step": 4215
    },
    {
      "epoch": 5.160342717258262,
      "grad_norm": 1.2996121844962045,
      "learning_rate": 3.676601503453465e-06,
      "loss": 0.7781,
      "step": 4216
    },
    {
      "epoch": 5.16156670746634,
      "grad_norm": 2.2817268015283916,
      "learning_rate": 3.6760360005450612e-06,
      "loss": 0.4193,
      "step": 4217
    },
    {
      "epoch": 5.162790697674419,
      "grad_norm": 0.9105964926527785,
      "learning_rate": 3.675470420351921e-06,
      "loss": 0.5987,
      "step": 4218
    },
    {
      "epoch": 5.164014687882497,
      "grad_norm": 0.9348753065922383,
      "learning_rate": 3.6749047629112133e-06,
      "loss": 0.646,
      "step": 4219
    },
    {
      "epoch": 5.165238678090575,
      "grad_norm": 1.1132848755892766,
      "learning_rate": 3.674339028260109e-06,
      "loss": 1.0234,
      "step": 4220
    },
    {
      "epoch": 5.166462668298654,
      "grad_norm": 0.9630361764337361,
      "learning_rate": 3.673773216435788e-06,
      "loss": 1.058,
      "step": 4221
    },
    {
      "epoch": 5.167686658506732,
      "grad_norm": 1.9118891677009413,
      "learning_rate": 3.673207327475433e-06,
      "loss": 0.5549,
      "step": 4222
    },
    {
      "epoch": 5.16891064871481,
      "grad_norm": 1.7222204406699753,
      "learning_rate": 3.6726413614162314e-06,
      "loss": 0.8573,
      "step": 4223
    },
    {
      "epoch": 5.170134638922889,
      "grad_norm": 1.9478035758937027,
      "learning_rate": 3.6720753182953773e-06,
      "loss": 0.5718,
      "step": 4224
    },
    {
      "epoch": 5.1713586291309666,
      "grad_norm": 1.3528751275392972,
      "learning_rate": 3.6715091981500684e-06,
      "loss": 1.4802,
      "step": 4225
    },
    {
      "epoch": 5.172582619339045,
      "grad_norm": 1.7613979802160893,
      "learning_rate": 3.6709430010175085e-06,
      "loss": 0.4615,
      "step": 4226
    },
    {
      "epoch": 5.173806609547124,
      "grad_norm": 1.1047008845395279,
      "learning_rate": 3.6703767269349043e-06,
      "loss": 0.7411,
      "step": 4227
    },
    {
      "epoch": 5.1750305997552015,
      "grad_norm": 1.2672379226662693,
      "learning_rate": 3.6698103759394722e-06,
      "loss": 0.9371,
      "step": 4228
    },
    {
      "epoch": 5.17625458996328,
      "grad_norm": 0.9645125998270491,
      "learning_rate": 3.6692439480684273e-06,
      "loss": 0.7038,
      "step": 4229
    },
    {
      "epoch": 5.177478580171359,
      "grad_norm": 1.395446972639668,
      "learning_rate": 3.6686774433589956e-06,
      "loss": 0.9791,
      "step": 4230
    },
    {
      "epoch": 5.178702570379437,
      "grad_norm": 1.4963112653346184,
      "learning_rate": 3.668110861848405e-06,
      "loss": 0.4498,
      "step": 4231
    },
    {
      "epoch": 5.179926560587515,
      "grad_norm": 1.3687443724500308,
      "learning_rate": 3.6675442035738894e-06,
      "loss": 0.5696,
      "step": 4232
    },
    {
      "epoch": 5.181150550795594,
      "grad_norm": 1.0323074139480561,
      "learning_rate": 3.666977468572687e-06,
      "loss": 0.5446,
      "step": 4233
    },
    {
      "epoch": 5.182374541003672,
      "grad_norm": 1.416248451570722,
      "learning_rate": 3.666410656882042e-06,
      "loss": 0.7775,
      "step": 4234
    },
    {
      "epoch": 5.18359853121175,
      "grad_norm": 0.9383900752641997,
      "learning_rate": 3.665843768539202e-06,
      "loss": 0.526,
      "step": 4235
    },
    {
      "epoch": 5.184822521419829,
      "grad_norm": 0.8362777184511306,
      "learning_rate": 3.665276803581422e-06,
      "loss": 0.5577,
      "step": 4236
    },
    {
      "epoch": 5.186046511627907,
      "grad_norm": 1.8664563613432958,
      "learning_rate": 3.6647097620459616e-06,
      "loss": 0.5218,
      "step": 4237
    },
    {
      "epoch": 5.187270501835985,
      "grad_norm": 0.9510034781743811,
      "learning_rate": 3.6641426439700823e-06,
      "loss": 0.5608,
      "step": 4238
    },
    {
      "epoch": 5.188494492044064,
      "grad_norm": 1.1113972467260913,
      "learning_rate": 3.6635754493910545e-06,
      "loss": 0.7994,
      "step": 4239
    },
    {
      "epoch": 5.189718482252142,
      "grad_norm": 1.0703422473750874,
      "learning_rate": 3.6630081783461524e-06,
      "loss": 0.4643,
      "step": 4240
    },
    {
      "epoch": 5.19094247246022,
      "grad_norm": 2.236755590282444,
      "learning_rate": 3.6624408308726544e-06,
      "loss": 0.5429,
      "step": 4241
    },
    {
      "epoch": 5.192166462668299,
      "grad_norm": 2.758617405891889,
      "learning_rate": 3.6618734070078438e-06,
      "loss": 0.4853,
      "step": 4242
    },
    {
      "epoch": 5.193390452876377,
      "grad_norm": 0.872513061045636,
      "learning_rate": 3.6613059067890116e-06,
      "loss": 0.5527,
      "step": 4243
    },
    {
      "epoch": 5.194614443084455,
      "grad_norm": 0.8795404940423862,
      "learning_rate": 3.6607383302534495e-06,
      "loss": 0.5245,
      "step": 4244
    },
    {
      "epoch": 5.195838433292534,
      "grad_norm": 1.141350221848387,
      "learning_rate": 3.660170677438458e-06,
      "loss": 0.6369,
      "step": 4245
    },
    {
      "epoch": 5.197062423500612,
      "grad_norm": 0.8425276661498231,
      "learning_rate": 3.6596029483813416e-06,
      "loss": 0.6145,
      "step": 4246
    },
    {
      "epoch": 5.19828641370869,
      "grad_norm": 1.8482680757987349,
      "learning_rate": 3.6590351431194072e-06,
      "loss": 0.5768,
      "step": 4247
    },
    {
      "epoch": 5.1995104039167686,
      "grad_norm": 0.7710801091320874,
      "learning_rate": 3.658467261689971e-06,
      "loss": 0.4572,
      "step": 4248
    },
    {
      "epoch": 5.200734394124847,
      "grad_norm": 1.6895540235547482,
      "learning_rate": 3.6578993041303505e-06,
      "loss": 0.6684,
      "step": 4249
    },
    {
      "epoch": 5.201958384332926,
      "grad_norm": 1.0556606630181802,
      "learning_rate": 3.6573312704778713e-06,
      "loss": 0.892,
      "step": 4250
    },
    {
      "epoch": 5.2031823745410035,
      "grad_norm": 2.345209956522388,
      "learning_rate": 3.6567631607698607e-06,
      "loss": 1.2143,
      "step": 4251
    },
    {
      "epoch": 5.204406364749082,
      "grad_norm": 1.2435432924020606,
      "learning_rate": 3.6561949750436556e-06,
      "loss": 0.4869,
      "step": 4252
    },
    {
      "epoch": 5.205630354957161,
      "grad_norm": 2.203321861233657,
      "learning_rate": 3.655626713336591e-06,
      "loss": 0.8624,
      "step": 4253
    },
    {
      "epoch": 5.2068543451652385,
      "grad_norm": 2.151907305915236,
      "learning_rate": 3.6550583756860144e-06,
      "loss": 0.456,
      "step": 4254
    },
    {
      "epoch": 5.208078335373317,
      "grad_norm": 1.4895870012533947,
      "learning_rate": 3.6544899621292735e-06,
      "loss": 0.5795,
      "step": 4255
    },
    {
      "epoch": 5.209302325581396,
      "grad_norm": 1.0134358082535861,
      "learning_rate": 3.653921472703722e-06,
      "loss": 0.4887,
      "step": 4256
    },
    {
      "epoch": 5.2105263157894735,
      "grad_norm": 2.0343314285458987,
      "learning_rate": 3.65335290744672e-06,
      "loss": 0.647,
      "step": 4257
    },
    {
      "epoch": 5.211750305997552,
      "grad_norm": 1.8262710790604717,
      "learning_rate": 3.6527842663956303e-06,
      "loss": 0.5659,
      "step": 4258
    },
    {
      "epoch": 5.212974296205631,
      "grad_norm": 0.9453135998552632,
      "learning_rate": 3.652215549587823e-06,
      "loss": 0.8411,
      "step": 4259
    },
    {
      "epoch": 5.214198286413708,
      "grad_norm": 1.8262681854862421,
      "learning_rate": 3.6516467570606704e-06,
      "loss": 1.1156,
      "step": 4260
    },
    {
      "epoch": 5.215422276621787,
      "grad_norm": 2.106782380211626,
      "learning_rate": 3.6510778888515547e-06,
      "loss": 0.4956,
      "step": 4261
    },
    {
      "epoch": 5.216646266829866,
      "grad_norm": 1.6541828641626706,
      "learning_rate": 3.650508944997856e-06,
      "loss": 0.5122,
      "step": 4262
    },
    {
      "epoch": 5.217870257037943,
      "grad_norm": 1.5530828907858478,
      "learning_rate": 3.649939925536965e-06,
      "loss": 0.5839,
      "step": 4263
    },
    {
      "epoch": 5.219094247246022,
      "grad_norm": 1.2533263589874262,
      "learning_rate": 3.649370830506276e-06,
      "loss": 1.6655,
      "step": 4264
    },
    {
      "epoch": 5.220318237454101,
      "grad_norm": 1.3848748863640334,
      "learning_rate": 3.6488016599431867e-06,
      "loss": 0.717,
      "step": 4265
    },
    {
      "epoch": 5.221542227662178,
      "grad_norm": 1.5625080246024787,
      "learning_rate": 3.6482324138851022e-06,
      "loss": 0.7896,
      "step": 4266
    },
    {
      "epoch": 5.222766217870257,
      "grad_norm": 0.8230990358383853,
      "learning_rate": 3.6476630923694302e-06,
      "loss": 0.7295,
      "step": 4267
    },
    {
      "epoch": 5.223990208078336,
      "grad_norm": 1.522168515146073,
      "learning_rate": 3.6470936954335833e-06,
      "loss": 0.9592,
      "step": 4268
    },
    {
      "epoch": 5.225214198286413,
      "grad_norm": 1.341722139791942,
      "learning_rate": 3.6465242231149835e-06,
      "loss": 0.7645,
      "step": 4269
    },
    {
      "epoch": 5.226438188494492,
      "grad_norm": 1.255611653051533,
      "learning_rate": 3.645954675451051e-06,
      "loss": 0.5132,
      "step": 4270
    },
    {
      "epoch": 5.2276621787025706,
      "grad_norm": 1.9476168524834452,
      "learning_rate": 3.6453850524792168e-06,
      "loss": 0.7537,
      "step": 4271
    },
    {
      "epoch": 5.228886168910648,
      "grad_norm": 1.9694017941230038,
      "learning_rate": 3.6448153542369137e-06,
      "loss": 0.3814,
      "step": 4272
    },
    {
      "epoch": 5.230110159118727,
      "grad_norm": 1.0873351913771432,
      "learning_rate": 3.644245580761579e-06,
      "loss": 0.5667,
      "step": 4273
    },
    {
      "epoch": 5.2313341493268055,
      "grad_norm": 1.4755616377331195,
      "learning_rate": 3.6436757320906575e-06,
      "loss": 0.6715,
      "step": 4274
    },
    {
      "epoch": 5.232558139534884,
      "grad_norm": 0.8574292158653031,
      "learning_rate": 3.6431058082615966e-06,
      "loss": 0.5548,
      "step": 4275
    },
    {
      "epoch": 5.233782129742962,
      "grad_norm": 1.4267907787767884,
      "learning_rate": 3.642535809311851e-06,
      "loss": 1.5856,
      "step": 4276
    },
    {
      "epoch": 5.2350061199510405,
      "grad_norm": 1.6404010650653775,
      "learning_rate": 3.641965735278876e-06,
      "loss": 1.162,
      "step": 4277
    },
    {
      "epoch": 5.236230110159119,
      "grad_norm": 1.3438574165722783,
      "learning_rate": 3.6413955862001393e-06,
      "loss": 0.9572,
      "step": 4278
    },
    {
      "epoch": 5.237454100367197,
      "grad_norm": 1.3896228728937945,
      "learning_rate": 3.6408253621131046e-06,
      "loss": 0.5608,
      "step": 4279
    },
    {
      "epoch": 5.2386780905752754,
      "grad_norm": 1.2877063442489123,
      "learning_rate": 3.6402550630552467e-06,
      "loss": 0.7576,
      "step": 4280
    },
    {
      "epoch": 5.239902080783354,
      "grad_norm": 0.8593877803908888,
      "learning_rate": 3.639684689064045e-06,
      "loss": 0.6595,
      "step": 4281
    },
    {
      "epoch": 5.241126070991432,
      "grad_norm": 0.9903624087764138,
      "learning_rate": 3.6391142401769795e-06,
      "loss": 0.7714,
      "step": 4282
    },
    {
      "epoch": 5.24235006119951,
      "grad_norm": 1.017372684148912,
      "learning_rate": 3.6385437164315397e-06,
      "loss": 0.6592,
      "step": 4283
    },
    {
      "epoch": 5.243574051407589,
      "grad_norm": 1.4417815363737445,
      "learning_rate": 3.637973117865219e-06,
      "loss": 0.5447,
      "step": 4284
    },
    {
      "epoch": 5.244798041615667,
      "grad_norm": 1.5823713893278137,
      "learning_rate": 3.6374024445155132e-06,
      "loss": 0.7489,
      "step": 4285
    },
    {
      "epoch": 5.246022031823745,
      "grad_norm": 1.4653821487486918,
      "learning_rate": 3.6368316964199253e-06,
      "loss": 1.0406,
      "step": 4286
    },
    {
      "epoch": 5.247246022031824,
      "grad_norm": 0.7469820993382222,
      "learning_rate": 3.6362608736159646e-06,
      "loss": 0.3605,
      "step": 4287
    },
    {
      "epoch": 5.248470012239902,
      "grad_norm": 1.1421261119741495,
      "learning_rate": 3.63568997614114e-06,
      "loss": 0.5748,
      "step": 4288
    },
    {
      "epoch": 5.24969400244798,
      "grad_norm": 1.2515724347401114,
      "learning_rate": 3.6351190040329725e-06,
      "loss": 0.6293,
      "step": 4289
    },
    {
      "epoch": 5.250917992656059,
      "grad_norm": 1.741300229539501,
      "learning_rate": 3.634547957328982e-06,
      "loss": 1.3733,
      "step": 4290
    },
    {
      "epoch": 5.252141982864137,
      "grad_norm": 1.9757614712840033,
      "learning_rate": 3.633976836066696e-06,
      "loss": 1.3193,
      "step": 4291
    },
    {
      "epoch": 5.253365973072215,
      "grad_norm": 1.927187384614435,
      "learning_rate": 3.6334056402836467e-06,
      "loss": 0.6998,
      "step": 4292
    },
    {
      "epoch": 5.254589963280294,
      "grad_norm": 1.0603840141672665,
      "learning_rate": 3.632834370017371e-06,
      "loss": 0.7028,
      "step": 4293
    },
    {
      "epoch": 5.2558139534883725,
      "grad_norm": 1.9318601824678152,
      "learning_rate": 3.6322630253054104e-06,
      "loss": 1.0189,
      "step": 4294
    },
    {
      "epoch": 5.25703794369645,
      "grad_norm": 1.1057170345620964,
      "learning_rate": 3.631691606185312e-06,
      "loss": 0.5487,
      "step": 4295
    },
    {
      "epoch": 5.258261933904529,
      "grad_norm": 1.672547013443651,
      "learning_rate": 3.631120112694627e-06,
      "loss": 0.7487,
      "step": 4296
    },
    {
      "epoch": 5.2594859241126075,
      "grad_norm": 1.4242594766587162,
      "learning_rate": 3.630548544870912e-06,
      "loss": 1.2183,
      "step": 4297
    },
    {
      "epoch": 5.260709914320685,
      "grad_norm": 1.6617565266105125,
      "learning_rate": 3.6299769027517284e-06,
      "loss": 0.8531,
      "step": 4298
    },
    {
      "epoch": 5.261933904528764,
      "grad_norm": 1.3713296595350237,
      "learning_rate": 3.6294051863746417e-06,
      "loss": 1.2205,
      "step": 4299
    },
    {
      "epoch": 5.2631578947368425,
      "grad_norm": 1.237369473220036,
      "learning_rate": 3.628833395777224e-06,
      "loss": 1.2713,
      "step": 4300
    },
    {
      "epoch": 5.26438188494492,
      "grad_norm": 0.8752038254204331,
      "learning_rate": 3.62826153099705e-06,
      "loss": 0.584,
      "step": 4301
    },
    {
      "epoch": 5.265605875152999,
      "grad_norm": 1.1589345523404555,
      "learning_rate": 3.6276895920717026e-06,
      "loss": 1.3414,
      "step": 4302
    },
    {
      "epoch": 5.2668298653610774,
      "grad_norm": 0.9981508254046779,
      "learning_rate": 3.627117579038765e-06,
      "loss": 0.6779,
      "step": 4303
    },
    {
      "epoch": 5.268053855569155,
      "grad_norm": 1.0932820397484142,
      "learning_rate": 3.6265454919358303e-06,
      "loss": 0.6691,
      "step": 4304
    },
    {
      "epoch": 5.269277845777234,
      "grad_norm": 1.449495574383733,
      "learning_rate": 3.625973330800491e-06,
      "loss": 0.6458,
      "step": 4305
    },
    {
      "epoch": 5.270501835985312,
      "grad_norm": 1.799563036915311,
      "learning_rate": 3.6254010956703507e-06,
      "loss": 0.4373,
      "step": 4306
    },
    {
      "epoch": 5.27172582619339,
      "grad_norm": 1.8928072195161407,
      "learning_rate": 3.6248287865830124e-06,
      "loss": 0.4449,
      "step": 4307
    },
    {
      "epoch": 5.272949816401469,
      "grad_norm": 1.6948485079208713,
      "learning_rate": 3.6242564035760863e-06,
      "loss": 0.395,
      "step": 4308
    },
    {
      "epoch": 5.274173806609547,
      "grad_norm": 1.0147783171711569,
      "learning_rate": 3.6236839466871882e-06,
      "loss": 0.553,
      "step": 4309
    },
    {
      "epoch": 5.275397796817625,
      "grad_norm": 1.2931642657680622,
      "learning_rate": 3.6231114159539367e-06,
      "loss": 0.7733,
      "step": 4310
    },
    {
      "epoch": 5.276621787025704,
      "grad_norm": 1.6768857578002503,
      "learning_rate": 3.622538811413958e-06,
      "loss": 0.5098,
      "step": 4311
    },
    {
      "epoch": 5.277845777233782,
      "grad_norm": 0.9122733049889713,
      "learning_rate": 3.6219661331048793e-06,
      "loss": 0.8604,
      "step": 4312
    },
    {
      "epoch": 5.27906976744186,
      "grad_norm": 1.8750814128719118,
      "learning_rate": 3.6213933810643374e-06,
      "loss": 0.507,
      "step": 4313
    },
    {
      "epoch": 5.280293757649939,
      "grad_norm": 1.7671061774966188,
      "learning_rate": 3.6208205553299704e-06,
      "loss": 0.4821,
      "step": 4314
    },
    {
      "epoch": 5.281517747858017,
      "grad_norm": 1.1726248470822982,
      "learning_rate": 3.6202476559394212e-06,
      "loss": 0.3834,
      "step": 4315
    },
    {
      "epoch": 5.282741738066095,
      "grad_norm": 1.456481473442461,
      "learning_rate": 3.6196746829303396e-06,
      "loss": 0.5542,
      "step": 4316
    },
    {
      "epoch": 5.283965728274174,
      "grad_norm": 1.8393266753312836,
      "learning_rate": 3.6191016363403796e-06,
      "loss": 0.7752,
      "step": 4317
    },
    {
      "epoch": 5.285189718482252,
      "grad_norm": 0.6288523591449686,
      "learning_rate": 3.6185285162071994e-06,
      "loss": 0.3653,
      "step": 4318
    },
    {
      "epoch": 5.286413708690331,
      "grad_norm": 1.7753837001902582,
      "learning_rate": 3.6179553225684623e-06,
      "loss": 0.3622,
      "step": 4319
    },
    {
      "epoch": 5.287637698898409,
      "grad_norm": 1.371734070150983,
      "learning_rate": 3.617382055461836e-06,
      "loss": 0.4289,
      "step": 4320
    },
    {
      "epoch": 5.288861689106487,
      "grad_norm": 1.3392432312160503,
      "learning_rate": 3.616808714924994e-06,
      "loss": 1.1126,
      "step": 4321
    },
    {
      "epoch": 5.290085679314566,
      "grad_norm": 2.0862647165762853,
      "learning_rate": 3.6162353009956147e-06,
      "loss": 0.8235,
      "step": 4322
    },
    {
      "epoch": 5.291309669522644,
      "grad_norm": 1.49136270423915,
      "learning_rate": 3.6156618137113796e-06,
      "loss": 0.6126,
      "step": 4323
    },
    {
      "epoch": 5.292533659730722,
      "grad_norm": 1.487436385736882,
      "learning_rate": 3.6150882531099766e-06,
      "loss": 0.4424,
      "step": 4324
    },
    {
      "epoch": 5.293757649938801,
      "grad_norm": 0.8562819828701981,
      "learning_rate": 3.614514619229098e-06,
      "loss": 0.6231,
      "step": 4325
    },
    {
      "epoch": 5.2949816401468786,
      "grad_norm": 0.8443544891622166,
      "learning_rate": 3.6139409121064415e-06,
      "loss": 0.5357,
      "step": 4326
    },
    {
      "epoch": 5.296205630354957,
      "grad_norm": 1.434583532710112,
      "learning_rate": 3.6133671317797077e-06,
      "loss": 0.6337,
      "step": 4327
    },
    {
      "epoch": 5.297429620563036,
      "grad_norm": 0.8520367352551844,
      "learning_rate": 3.612793278286605e-06,
      "loss": 0.5868,
      "step": 4328
    },
    {
      "epoch": 5.2986536107711135,
      "grad_norm": 0.8348702934201354,
      "learning_rate": 3.612219351664843e-06,
      "loss": 0.5629,
      "step": 4329
    },
    {
      "epoch": 5.299877600979192,
      "grad_norm": 1.350359829851054,
      "learning_rate": 3.611645351952139e-06,
      "loss": 0.8494,
      "step": 4330
    },
    {
      "epoch": 5.301101591187271,
      "grad_norm": 1.103332734325432,
      "learning_rate": 3.6110712791862144e-06,
      "loss": 0.6079,
      "step": 4331
    },
    {
      "epoch": 5.3023255813953485,
      "grad_norm": 1.6788274533826235,
      "learning_rate": 3.6104971334047954e-06,
      "loss": 0.799,
      "step": 4332
    },
    {
      "epoch": 5.303549571603427,
      "grad_norm": 2.3286078102087324,
      "learning_rate": 3.6099229146456117e-06,
      "loss": 0.5926,
      "step": 4333
    },
    {
      "epoch": 5.304773561811506,
      "grad_norm": 1.7727867096881784,
      "learning_rate": 3.6093486229464e-06,
      "loss": 0.4852,
      "step": 4334
    },
    {
      "epoch": 5.3059975520195835,
      "grad_norm": 1.0179037168070049,
      "learning_rate": 3.608774258344899e-06,
      "loss": 0.6,
      "step": 4335
    },
    {
      "epoch": 5.307221542227662,
      "grad_norm": 0.9358668039847722,
      "learning_rate": 3.608199820878855e-06,
      "loss": 0.4602,
      "step": 4336
    },
    {
      "epoch": 5.308445532435741,
      "grad_norm": 1.5551217655873437,
      "learning_rate": 3.6076253105860186e-06,
      "loss": 0.604,
      "step": 4337
    },
    {
      "epoch": 5.309669522643819,
      "grad_norm": 1.5577739875215777,
      "learning_rate": 3.607050727504143e-06,
      "loss": 0.6656,
      "step": 4338
    },
    {
      "epoch": 5.310893512851897,
      "grad_norm": 1.1324409520924996,
      "learning_rate": 3.606476071670988e-06,
      "loss": 0.588,
      "step": 4339
    },
    {
      "epoch": 5.312117503059976,
      "grad_norm": 2.447626100294901,
      "learning_rate": 3.6059013431243186e-06,
      "loss": 0.4034,
      "step": 4340
    },
    {
      "epoch": 5.313341493268053,
      "grad_norm": 1.6662919707250525,
      "learning_rate": 3.6053265419019034e-06,
      "loss": 0.4454,
      "step": 4341
    },
    {
      "epoch": 5.314565483476132,
      "grad_norm": 1.7237407926263766,
      "learning_rate": 3.604751668041516e-06,
      "loss": 0.4165,
      "step": 4342
    },
    {
      "epoch": 5.315789473684211,
      "grad_norm": 1.1196518810755984,
      "learning_rate": 3.6041767215809353e-06,
      "loss": 0.6398,
      "step": 4343
    },
    {
      "epoch": 5.317013463892289,
      "grad_norm": 1.7148925863675162,
      "learning_rate": 3.603601702557944e-06,
      "loss": 0.5829,
      "step": 4344
    },
    {
      "epoch": 5.318237454100367,
      "grad_norm": 1.5169354463987696,
      "learning_rate": 3.6030266110103314e-06,
      "loss": 0.5551,
      "step": 4345
    },
    {
      "epoch": 5.319461444308446,
      "grad_norm": 1.1835446323880072,
      "learning_rate": 3.6024514469758896e-06,
      "loss": 0.7367,
      "step": 4346
    },
    {
      "epoch": 5.320685434516524,
      "grad_norm": 1.3457978780760955,
      "learning_rate": 3.6018762104924163e-06,
      "loss": 0.5528,
      "step": 4347
    },
    {
      "epoch": 5.321909424724602,
      "grad_norm": 2.2841990181883745,
      "learning_rate": 3.6013009015977142e-06,
      "loss": 0.5573,
      "step": 4348
    },
    {
      "epoch": 5.3231334149326806,
      "grad_norm": 1.813954282864968,
      "learning_rate": 3.6007255203295898e-06,
      "loss": 0.4407,
      "step": 4349
    },
    {
      "epoch": 5.324357405140759,
      "grad_norm": 2.1636875493002488,
      "learning_rate": 3.600150066725856e-06,
      "loss": 0.6115,
      "step": 4350
    },
    {
      "epoch": 5.325581395348837,
      "grad_norm": 2.1184485394445547,
      "learning_rate": 3.599574540824329e-06,
      "loss": 0.9931,
      "step": 4351
    },
    {
      "epoch": 5.3268053855569155,
      "grad_norm": 1.4186937201400487,
      "learning_rate": 3.5989989426628303e-06,
      "loss": 0.5756,
      "step": 4352
    },
    {
      "epoch": 5.328029375764994,
      "grad_norm": 1.4965356021641048,
      "learning_rate": 3.598423272279185e-06,
      "loss": 0.5241,
      "step": 4353
    },
    {
      "epoch": 5.329253365973072,
      "grad_norm": 0.7342273336844484,
      "learning_rate": 3.597847529711226e-06,
      "loss": 0.5378,
      "step": 4354
    },
    {
      "epoch": 5.3304773561811505,
      "grad_norm": 1.2931859231517615,
      "learning_rate": 3.597271714996788e-06,
      "loss": 0.4018,
      "step": 4355
    },
    {
      "epoch": 5.331701346389229,
      "grad_norm": 1.421579682372198,
      "learning_rate": 3.5966958281737107e-06,
      "loss": 0.7393,
      "step": 4356
    },
    {
      "epoch": 5.332925336597307,
      "grad_norm": 1.544927793795705,
      "learning_rate": 3.596119869279841e-06,
      "loss": 0.524,
      "step": 4357
    },
    {
      "epoch": 5.3341493268053854,
      "grad_norm": 0.8784407626313382,
      "learning_rate": 3.5955438383530274e-06,
      "loss": 0.6812,
      "step": 4358
    },
    {
      "epoch": 5.335373317013464,
      "grad_norm": 1.4182692983300471,
      "learning_rate": 3.594967735431124e-06,
      "loss": 0.461,
      "step": 4359
    },
    {
      "epoch": 5.336597307221542,
      "grad_norm": 1.2005036187314102,
      "learning_rate": 3.594391560551992e-06,
      "loss": 0.7082,
      "step": 4360
    },
    {
      "epoch": 5.33782129742962,
      "grad_norm": 1.417995903756882,
      "learning_rate": 3.593815313753495e-06,
      "loss": 0.7294,
      "step": 4361
    },
    {
      "epoch": 5.339045287637699,
      "grad_norm": 1.2313356296347129,
      "learning_rate": 3.5932389950735e-06,
      "loss": 0.5965,
      "step": 4362
    },
    {
      "epoch": 5.340269277845778,
      "grad_norm": 1.295394005446654,
      "learning_rate": 3.5926626045498817e-06,
      "loss": 1.5788,
      "step": 4363
    },
    {
      "epoch": 5.341493268053855,
      "grad_norm": 1.0024655238931273,
      "learning_rate": 3.59208614222052e-06,
      "loss": 0.5816,
      "step": 4364
    },
    {
      "epoch": 5.342717258261934,
      "grad_norm": 1.6121054145141585,
      "learning_rate": 3.5915096081232953e-06,
      "loss": 0.7949,
      "step": 4365
    },
    {
      "epoch": 5.343941248470013,
      "grad_norm": 1.0022150458059447,
      "learning_rate": 3.5909330022960963e-06,
      "loss": 0.4188,
      "step": 4366
    },
    {
      "epoch": 5.34516523867809,
      "grad_norm": 1.531882937102185,
      "learning_rate": 3.5903563247768164e-06,
      "loss": 0.4195,
      "step": 4367
    },
    {
      "epoch": 5.346389228886169,
      "grad_norm": 0.7274114367092167,
      "learning_rate": 3.5897795756033505e-06,
      "loss": 0.4024,
      "step": 4368
    },
    {
      "epoch": 5.347613219094248,
      "grad_norm": 1.6480890964032626,
      "learning_rate": 3.5892027548136026e-06,
      "loss": 0.5233,
      "step": 4369
    },
    {
      "epoch": 5.348837209302325,
      "grad_norm": 1.2695711896280573,
      "learning_rate": 3.5886258624454786e-06,
      "loss": 0.4865,
      "step": 4370
    },
    {
      "epoch": 5.350061199510404,
      "grad_norm": 1.2774384495371223,
      "learning_rate": 3.5880488985368888e-06,
      "loss": 0.5326,
      "step": 4371
    },
    {
      "epoch": 5.3512851897184825,
      "grad_norm": 1.4553646779688971,
      "learning_rate": 3.5874718631257497e-06,
      "loss": 1.0149,
      "step": 4372
    },
    {
      "epoch": 5.35250917992656,
      "grad_norm": 1.040897274729759,
      "learning_rate": 3.5868947562499827e-06,
      "loss": 0.5777,
      "step": 4373
    },
    {
      "epoch": 5.353733170134639,
      "grad_norm": 2.073585803648856,
      "learning_rate": 3.5863175779475122e-06,
      "loss": 0.54,
      "step": 4374
    },
    {
      "epoch": 5.3549571603427175,
      "grad_norm": 1.4962406860238233,
      "learning_rate": 3.5857403282562685e-06,
      "loss": 0.9841,
      "step": 4375
    },
    {
      "epoch": 5.356181150550795,
      "grad_norm": 1.6731887449972997,
      "learning_rate": 3.5851630072141864e-06,
      "loss": 0.4813,
      "step": 4376
    },
    {
      "epoch": 5.357405140758874,
      "grad_norm": 1.6684988591503531,
      "learning_rate": 3.5845856148592045e-06,
      "loss": 0.5189,
      "step": 4377
    },
    {
      "epoch": 5.3586291309669525,
      "grad_norm": 1.039459441022642,
      "learning_rate": 3.5840081512292685e-06,
      "loss": 0.4275,
      "step": 4378
    },
    {
      "epoch": 5.35985312117503,
      "grad_norm": 1.5289440300994905,
      "learning_rate": 3.5834306163623255e-06,
      "loss": 0.6543,
      "step": 4379
    },
    {
      "epoch": 5.361077111383109,
      "grad_norm": 1.8506014753877247,
      "learning_rate": 3.5828530102963303e-06,
      "loss": 0.5273,
      "step": 4380
    },
    {
      "epoch": 5.3623011015911874,
      "grad_norm": 1.2196314516550142,
      "learning_rate": 3.5822753330692405e-06,
      "loss": 0.4288,
      "step": 4381
    },
    {
      "epoch": 5.363525091799266,
      "grad_norm": 1.244475218336885,
      "learning_rate": 3.581697584719019e-06,
      "loss": 1.1792,
      "step": 4382
    },
    {
      "epoch": 5.364749082007344,
      "grad_norm": 1.3275825352143475,
      "learning_rate": 3.5811197652836323e-06,
      "loss": 0.5469,
      "step": 4383
    },
    {
      "epoch": 5.365973072215422,
      "grad_norm": 1.3352502350714146,
      "learning_rate": 3.5805418748010546e-06,
      "loss": 0.6708,
      "step": 4384
    },
    {
      "epoch": 5.3671970624235,
      "grad_norm": 0.8769138602754324,
      "learning_rate": 3.5799639133092605e-06,
      "loss": 0.5639,
      "step": 4385
    },
    {
      "epoch": 5.368421052631579,
      "grad_norm": 1.2300068842946812,
      "learning_rate": 3.579385880846232e-06,
      "loss": 0.5449,
      "step": 4386
    },
    {
      "epoch": 5.369645042839657,
      "grad_norm": 0.7561712132703756,
      "learning_rate": 3.578807777449958e-06,
      "loss": 0.461,
      "step": 4387
    },
    {
      "epoch": 5.370869033047736,
      "grad_norm": 1.1448192298490794,
      "learning_rate": 3.578229603158425e-06,
      "loss": 0.3551,
      "step": 4388
    },
    {
      "epoch": 5.372093023255814,
      "grad_norm": 0.943902370000432,
      "learning_rate": 3.5776513580096316e-06,
      "loss": 0.6919,
      "step": 4389
    },
    {
      "epoch": 5.373317013463892,
      "grad_norm": 2.761148219866134,
      "learning_rate": 3.577073042041577e-06,
      "loss": 0.4607,
      "step": 4390
    },
    {
      "epoch": 5.374541003671971,
      "grad_norm": 1.669255683766952,
      "learning_rate": 3.576494655292265e-06,
      "loss": 0.4865,
      "step": 4391
    },
    {
      "epoch": 5.375764993880049,
      "grad_norm": 2.172366589679472,
      "learning_rate": 3.5759161977997064e-06,
      "loss": 0.3981,
      "step": 4392
    },
    {
      "epoch": 5.376988984088127,
      "grad_norm": 1.2561067029599369,
      "learning_rate": 3.5753376696019152e-06,
      "loss": 0.5513,
      "step": 4393
    },
    {
      "epoch": 5.378212974296206,
      "grad_norm": 2.309337118355062,
      "learning_rate": 3.574759070736909e-06,
      "loss": 0.8409,
      "step": 4394
    },
    {
      "epoch": 5.379436964504284,
      "grad_norm": 1.292257419420401,
      "learning_rate": 3.574180401242713e-06,
      "loss": 1.1427,
      "step": 4395
    },
    {
      "epoch": 5.380660954712362,
      "grad_norm": 1.6713161383724755,
      "learning_rate": 3.5736016611573533e-06,
      "loss": 1.1017,
      "step": 4396
    },
    {
      "epoch": 5.381884944920441,
      "grad_norm": 1.5276971637072985,
      "learning_rate": 3.5730228505188636e-06,
      "loss": 0.6594,
      "step": 4397
    },
    {
      "epoch": 5.383108935128519,
      "grad_norm": 1.6342347931594354,
      "learning_rate": 3.5724439693652813e-06,
      "loss": 0.8769,
      "step": 4398
    },
    {
      "epoch": 5.384332925336597,
      "grad_norm": 1.659845353935088,
      "learning_rate": 3.5718650177346474e-06,
      "loss": 0.6093,
      "step": 4399
    },
    {
      "epoch": 5.385556915544676,
      "grad_norm": 1.1175759137492824,
      "learning_rate": 3.57128599566501e-06,
      "loss": 0.7613,
      "step": 4400
    },
    {
      "epoch": 5.386780905752754,
      "grad_norm": 0.9472263410683649,
      "learning_rate": 3.570706903194418e-06,
      "loss": 0.5482,
      "step": 4401
    },
    {
      "epoch": 5.388004895960832,
      "grad_norm": 2.076163846179155,
      "learning_rate": 3.5701277403609302e-06,
      "loss": 0.4703,
      "step": 4402
    },
    {
      "epoch": 5.389228886168911,
      "grad_norm": 1.1767031420288685,
      "learning_rate": 3.569548507202604e-06,
      "loss": 0.6435,
      "step": 4403
    },
    {
      "epoch": 5.3904528763769886,
      "grad_norm": 1.3456553461663245,
      "learning_rate": 3.568969203757507e-06,
      "loss": 0.9361,
      "step": 4404
    },
    {
      "epoch": 5.391676866585067,
      "grad_norm": 1.2101204159734256,
      "learning_rate": 3.5683898300637072e-06,
      "loss": 0.5544,
      "step": 4405
    },
    {
      "epoch": 5.392900856793146,
      "grad_norm": 1.8957763865826212,
      "learning_rate": 3.5678103861592796e-06,
      "loss": 0.5426,
      "step": 4406
    },
    {
      "epoch": 5.394124847001224,
      "grad_norm": 1.827879309881904,
      "learning_rate": 3.5672308720823025e-06,
      "loss": 0.5043,
      "step": 4407
    },
    {
      "epoch": 5.395348837209302,
      "grad_norm": 2.013835020713998,
      "learning_rate": 3.566651287870861e-06,
      "loss": 0.4399,
      "step": 4408
    },
    {
      "epoch": 5.396572827417381,
      "grad_norm": 1.368392801812708,
      "learning_rate": 3.566071633563041e-06,
      "loss": 0.5246,
      "step": 4409
    },
    {
      "epoch": 5.397796817625459,
      "grad_norm": 2.7655456968316634,
      "learning_rate": 3.565491909196937e-06,
      "loss": 0.5281,
      "step": 4410
    },
    {
      "epoch": 5.399020807833537,
      "grad_norm": 1.5812899297411138,
      "learning_rate": 3.5649121148106457e-06,
      "loss": 0.4281,
      "step": 4411
    },
    {
      "epoch": 5.400244798041616,
      "grad_norm": 1.0883904580828152,
      "learning_rate": 3.5643322504422687e-06,
      "loss": 1.0638,
      "step": 4412
    },
    {
      "epoch": 5.401468788249694,
      "grad_norm": 1.5130573201321875,
      "learning_rate": 3.563752316129913e-06,
      "loss": 0.7632,
      "step": 4413
    },
    {
      "epoch": 5.402692778457772,
      "grad_norm": 1.4248786018512765,
      "learning_rate": 3.56317231191169e-06,
      "loss": 1.8325,
      "step": 4414
    },
    {
      "epoch": 5.403916768665851,
      "grad_norm": 1.3123055747480625,
      "learning_rate": 3.562592237825715e-06,
      "loss": 1.6467,
      "step": 4415
    },
    {
      "epoch": 5.405140758873929,
      "grad_norm": 1.094530621434538,
      "learning_rate": 3.562012093910108e-06,
      "loss": 0.4814,
      "step": 4416
    },
    {
      "epoch": 5.406364749082007,
      "grad_norm": 1.2376879944018688,
      "learning_rate": 3.5614318802029948e-06,
      "loss": 0.4881,
      "step": 4417
    },
    {
      "epoch": 5.407588739290086,
      "grad_norm": 1.0975471044741623,
      "learning_rate": 3.560851596742504e-06,
      "loss": 0.651,
      "step": 4418
    },
    {
      "epoch": 5.408812729498164,
      "grad_norm": 1.7829496385327708,
      "learning_rate": 3.5602712435667697e-06,
      "loss": 0.6199,
      "step": 4419
    },
    {
      "epoch": 5.410036719706242,
      "grad_norm": 1.4183840963326084,
      "learning_rate": 3.559690820713931e-06,
      "loss": 0.3967,
      "step": 4420
    },
    {
      "epoch": 5.411260709914321,
      "grad_norm": 1.285576769175522,
      "learning_rate": 3.5591103282221324e-06,
      "loss": 0.6752,
      "step": 4421
    },
    {
      "epoch": 5.412484700122399,
      "grad_norm": 2.665806644536347,
      "learning_rate": 3.5585297661295195e-06,
      "loss": 0.3878,
      "step": 4422
    },
    {
      "epoch": 5.413708690330477,
      "grad_norm": 1.2172779038728183,
      "learning_rate": 3.5579491344742457e-06,
      "loss": 0.6,
      "step": 4423
    },
    {
      "epoch": 5.414932680538556,
      "grad_norm": 1.3592611363098044,
      "learning_rate": 3.5573684332944682e-06,
      "loss": 0.6826,
      "step": 4424
    },
    {
      "epoch": 5.416156670746634,
      "grad_norm": 1.3068103378677902,
      "learning_rate": 3.556787662628347e-06,
      "loss": 1.0711,
      "step": 4425
    },
    {
      "epoch": 5.417380660954713,
      "grad_norm": 0.9874115035146913,
      "learning_rate": 3.5562068225140506e-06,
      "loss": 0.6689,
      "step": 4426
    },
    {
      "epoch": 5.4186046511627906,
      "grad_norm": 2.6423205667022467,
      "learning_rate": 3.5556259129897473e-06,
      "loss": 0.4806,
      "step": 4427
    },
    {
      "epoch": 5.419828641370869,
      "grad_norm": 2.1654432229767075,
      "learning_rate": 3.5550449340936145e-06,
      "loss": 0.9048,
      "step": 4428
    },
    {
      "epoch": 5.421052631578947,
      "grad_norm": 1.4871514220920214,
      "learning_rate": 3.5544638858638307e-06,
      "loss": 0.5555,
      "step": 4429
    },
    {
      "epoch": 5.4222766217870255,
      "grad_norm": 1.2779830647419022,
      "learning_rate": 3.553882768338581e-06,
      "loss": 0.4138,
      "step": 4430
    },
    {
      "epoch": 5.423500611995104,
      "grad_norm": 2.329771727033092,
      "learning_rate": 3.5533015815560535e-06,
      "loss": 0.6225,
      "step": 4431
    },
    {
      "epoch": 5.424724602203183,
      "grad_norm": 1.2696143709153584,
      "learning_rate": 3.5527203255544417e-06,
      "loss": 0.5451,
      "step": 4432
    },
    {
      "epoch": 5.4259485924112605,
      "grad_norm": 1.7945268530866494,
      "learning_rate": 3.5521390003719435e-06,
      "loss": 1.0239,
      "step": 4433
    },
    {
      "epoch": 5.427172582619339,
      "grad_norm": 0.9050532356083686,
      "learning_rate": 3.5515576060467627e-06,
      "loss": 0.4669,
      "step": 4434
    },
    {
      "epoch": 5.428396572827418,
      "grad_norm": 1.4446034012616487,
      "learning_rate": 3.5509761426171046e-06,
      "loss": 1.4875,
      "step": 4435
    },
    {
      "epoch": 5.4296205630354955,
      "grad_norm": 1.8655077353402538,
      "learning_rate": 3.5503946101211817e-06,
      "loss": 1.4466,
      "step": 4436
    },
    {
      "epoch": 5.430844553243574,
      "grad_norm": 1.4126364288108972,
      "learning_rate": 3.5498130085972116e-06,
      "loss": 1.4489,
      "step": 4437
    },
    {
      "epoch": 5.432068543451653,
      "grad_norm": 1.121588472853089,
      "learning_rate": 3.5492313380834122e-06,
      "loss": 0.6161,
      "step": 4438
    },
    {
      "epoch": 5.43329253365973,
      "grad_norm": 0.8348305130734763,
      "learning_rate": 3.5486495986180104e-06,
      "loss": 0.5258,
      "step": 4439
    },
    {
      "epoch": 5.434516523867809,
      "grad_norm": 1.1786803552119138,
      "learning_rate": 3.5480677902392358e-06,
      "loss": 0.9186,
      "step": 4440
    },
    {
      "epoch": 5.435740514075888,
      "grad_norm": 1.4870723160456336,
      "learning_rate": 3.547485912985322e-06,
      "loss": 1.3734,
      "step": 4441
    },
    {
      "epoch": 5.436964504283965,
      "grad_norm": 1.3264472504564593,
      "learning_rate": 3.546903966894509e-06,
      "loss": 0.4975,
      "step": 4442
    },
    {
      "epoch": 5.438188494492044,
      "grad_norm": 1.0958306880271862,
      "learning_rate": 3.5463219520050397e-06,
      "loss": 0.4605,
      "step": 4443
    },
    {
      "epoch": 5.439412484700123,
      "grad_norm": 0.95532951324432,
      "learning_rate": 3.545739868355161e-06,
      "loss": 0.5823,
      "step": 4444
    },
    {
      "epoch": 5.4406364749082,
      "grad_norm": 1.5419797894091691,
      "learning_rate": 3.545157715983126e-06,
      "loss": 0.6243,
      "step": 4445
    },
    {
      "epoch": 5.441860465116279,
      "grad_norm": 2.253804782729319,
      "learning_rate": 3.5445754949271925e-06,
      "loss": 0.5726,
      "step": 4446
    },
    {
      "epoch": 5.443084455324358,
      "grad_norm": 1.3059517478683114,
      "learning_rate": 3.543993205225621e-06,
      "loss": 0.8414,
      "step": 4447
    },
    {
      "epoch": 5.444308445532435,
      "grad_norm": 1.1433580973522437,
      "learning_rate": 3.543410846916677e-06,
      "loss": 1.5826,
      "step": 4448
    },
    {
      "epoch": 5.445532435740514,
      "grad_norm": 1.298884011113135,
      "learning_rate": 3.542828420038632e-06,
      "loss": 0.9525,
      "step": 4449
    },
    {
      "epoch": 5.4467564259485926,
      "grad_norm": 1.675102851311704,
      "learning_rate": 3.5422459246297604e-06,
      "loss": 0.6862,
      "step": 4450
    },
    {
      "epoch": 5.447980416156671,
      "grad_norm": 1.3758286999998326,
      "learning_rate": 3.5416633607283407e-06,
      "loss": 1.6382,
      "step": 4451
    },
    {
      "epoch": 5.449204406364749,
      "grad_norm": 1.156106062911457,
      "learning_rate": 3.5410807283726588e-06,
      "loss": 0.5904,
      "step": 4452
    },
    {
      "epoch": 5.4504283965728275,
      "grad_norm": 2.0648972959905123,
      "learning_rate": 3.5404980276010016e-06,
      "loss": 0.9716,
      "step": 4453
    },
    {
      "epoch": 5.451652386780906,
      "grad_norm": 1.5032976171187695,
      "learning_rate": 3.539915258451663e-06,
      "loss": 1.2002,
      "step": 4454
    },
    {
      "epoch": 5.452876376988984,
      "grad_norm": 1.3585662371557394,
      "learning_rate": 3.5393324209629397e-06,
      "loss": 1.3855,
      "step": 4455
    },
    {
      "epoch": 5.4541003671970625,
      "grad_norm": 2.3938038548610474,
      "learning_rate": 3.538749515173134e-06,
      "loss": 0.3594,
      "step": 4456
    },
    {
      "epoch": 5.455324357405141,
      "grad_norm": 1.6420610582189552,
      "learning_rate": 3.5381665411205518e-06,
      "loss": 0.9922,
      "step": 4457
    },
    {
      "epoch": 5.456548347613219,
      "grad_norm": 1.497521706436548,
      "learning_rate": 3.5375834988435054e-06,
      "loss": 0.7744,
      "step": 4458
    },
    {
      "epoch": 5.4577723378212974,
      "grad_norm": 0.9389741706771537,
      "learning_rate": 3.5370003883803084e-06,
      "loss": 0.4543,
      "step": 4459
    },
    {
      "epoch": 5.458996328029376,
      "grad_norm": 1.1314869313911855,
      "learning_rate": 3.536417209769282e-06,
      "loss": 0.8904,
      "step": 4460
    },
    {
      "epoch": 5.460220318237454,
      "grad_norm": 1.2938274180867675,
      "learning_rate": 3.53583396304875e-06,
      "loss": 1.1266,
      "step": 4461
    },
    {
      "epoch": 5.461444308445532,
      "grad_norm": 1.8206280604236806,
      "learning_rate": 3.53525064825704e-06,
      "loss": 0.5222,
      "step": 4462
    },
    {
      "epoch": 5.462668298653611,
      "grad_norm": 0.837120156868484,
      "learning_rate": 3.5346672654324878e-06,
      "loss": 0.5438,
      "step": 4463
    },
    {
      "epoch": 5.463892288861689,
      "grad_norm": 2.315343636747441,
      "learning_rate": 3.5340838146134294e-06,
      "loss": 0.9579,
      "step": 4464
    },
    {
      "epoch": 5.465116279069767,
      "grad_norm": 0.8944383081308822,
      "learning_rate": 3.533500295838208e-06,
      "loss": 0.3887,
      "step": 4465
    },
    {
      "epoch": 5.466340269277846,
      "grad_norm": 1.5159104253022815,
      "learning_rate": 3.5329167091451688e-06,
      "loss": 0.9374,
      "step": 4466
    },
    {
      "epoch": 5.467564259485924,
      "grad_norm": 1.9135547680589282,
      "learning_rate": 3.5323330545726653e-06,
      "loss": 0.6202,
      "step": 4467
    },
    {
      "epoch": 5.468788249694002,
      "grad_norm": 1.641453736780936,
      "learning_rate": 3.5317493321590506e-06,
      "loss": 1.1588,
      "step": 4468
    },
    {
      "epoch": 5.470012239902081,
      "grad_norm": 1.5606503208951517,
      "learning_rate": 3.531165541942687e-06,
      "loss": 0.3811,
      "step": 4469
    },
    {
      "epoch": 5.47123623011016,
      "grad_norm": 1.2165694999762116,
      "learning_rate": 3.530581683961937e-06,
      "loss": 0.502,
      "step": 4470
    },
    {
      "epoch": 5.472460220318237,
      "grad_norm": 0.9421442425340472,
      "learning_rate": 3.5299977582551714e-06,
      "loss": 0.593,
      "step": 4471
    },
    {
      "epoch": 5.473684210526316,
      "grad_norm": 1.4103693642167372,
      "learning_rate": 3.529413764860763e-06,
      "loss": 0.6103,
      "step": 4472
    },
    {
      "epoch": 5.474908200734394,
      "grad_norm": 1.3541626570393468,
      "learning_rate": 3.528829703817089e-06,
      "loss": 0.3395,
      "step": 4473
    },
    {
      "epoch": 5.476132190942472,
      "grad_norm": 1.4573118240729364,
      "learning_rate": 3.5282455751625323e-06,
      "loss": 0.4572,
      "step": 4474
    },
    {
      "epoch": 5.477356181150551,
      "grad_norm": 1.7545645227151554,
      "learning_rate": 3.5276613789354797e-06,
      "loss": 0.4783,
      "step": 4475
    },
    {
      "epoch": 5.4785801713586295,
      "grad_norm": 1.1364945592281195,
      "learning_rate": 3.527077115174322e-06,
      "loss": 0.6716,
      "step": 4476
    },
    {
      "epoch": 5.479804161566707,
      "grad_norm": 1.0705828996674054,
      "learning_rate": 3.526492783917455e-06,
      "loss": 1.107,
      "step": 4477
    },
    {
      "epoch": 5.481028151774786,
      "grad_norm": 1.600817079073567,
      "learning_rate": 3.5259083852032792e-06,
      "loss": 1.0661,
      "step": 4478
    },
    {
      "epoch": 5.4822521419828645,
      "grad_norm": 1.415594517386211,
      "learning_rate": 3.525323919070199e-06,
      "loss": 0.5137,
      "step": 4479
    },
    {
      "epoch": 5.483476132190942,
      "grad_norm": 1.4831808276103746,
      "learning_rate": 3.524739385556624e-06,
      "loss": 0.6839,
      "step": 4480
    },
    {
      "epoch": 5.484700122399021,
      "grad_norm": 1.9325781951383452,
      "learning_rate": 3.524154784700966e-06,
      "loss": 0.2723,
      "step": 4481
    },
    {
      "epoch": 5.4859241126070994,
      "grad_norm": 1.5150412033523168,
      "learning_rate": 3.5235701165416437e-06,
      "loss": 0.5677,
      "step": 4482
    },
    {
      "epoch": 5.487148102815177,
      "grad_norm": 0.7992915043471042,
      "learning_rate": 3.522985381117079e-06,
      "loss": 0.5277,
      "step": 4483
    },
    {
      "epoch": 5.488372093023256,
      "grad_norm": 1.2047839286459492,
      "learning_rate": 3.5224005784657e-06,
      "loss": 0.4958,
      "step": 4484
    },
    {
      "epoch": 5.489596083231334,
      "grad_norm": 1.3030182592149477,
      "learning_rate": 3.5218157086259354e-06,
      "loss": 0.4493,
      "step": 4485
    },
    {
      "epoch": 5.490820073439412,
      "grad_norm": 1.1242606118363498,
      "learning_rate": 3.521230771636222e-06,
      "loss": 1.1372,
      "step": 4486
    },
    {
      "epoch": 5.492044063647491,
      "grad_norm": 2.0637961772962443,
      "learning_rate": 3.520645767535e-06,
      "loss": 0.9867,
      "step": 4487
    },
    {
      "epoch": 5.493268053855569,
      "grad_norm": 1.0007247923835334,
      "learning_rate": 3.520060696360713e-06,
      "loss": 0.859,
      "step": 4488
    },
    {
      "epoch": 5.494492044063647,
      "grad_norm": 2.227523235346055,
      "learning_rate": 3.5194755581518103e-06,
      "loss": 0.8645,
      "step": 4489
    },
    {
      "epoch": 5.495716034271726,
      "grad_norm": 0.860968763681482,
      "learning_rate": 3.5188903529467444e-06,
      "loss": 0.435,
      "step": 4490
    },
    {
      "epoch": 5.496940024479804,
      "grad_norm": 1.915214990178703,
      "learning_rate": 3.518305080783973e-06,
      "loss": 0.505,
      "step": 4491
    },
    {
      "epoch": 5.498164014687882,
      "grad_norm": 1.3260308026172227,
      "learning_rate": 3.5177197417019584e-06,
      "loss": 0.4357,
      "step": 4492
    },
    {
      "epoch": 5.499388004895961,
      "grad_norm": 2.2339140649262683,
      "learning_rate": 3.517134335739167e-06,
      "loss": 1.0042,
      "step": 4493
    },
    {
      "epoch": 5.500611995104039,
      "grad_norm": 0.8261312558054136,
      "learning_rate": 3.516548862934068e-06,
      "loss": 0.3816,
      "step": 4494
    },
    {
      "epoch": 5.501835985312118,
      "grad_norm": 1.6155576516107917,
      "learning_rate": 3.515963323325138e-06,
      "loss": 0.4677,
      "step": 4495
    },
    {
      "epoch": 5.503059975520196,
      "grad_norm": 1.0322846134830173,
      "learning_rate": 3.5153777169508566e-06,
      "loss": 0.654,
      "step": 4496
    },
    {
      "epoch": 5.504283965728274,
      "grad_norm": 1.253312162175109,
      "learning_rate": 3.5147920438497073e-06,
      "loss": 0.9562,
      "step": 4497
    },
    {
      "epoch": 5.505507955936353,
      "grad_norm": 1.3018463718875648,
      "learning_rate": 3.514206304060178e-06,
      "loss": 0.6039,
      "step": 4498
    },
    {
      "epoch": 5.506731946144431,
      "grad_norm": 1.950458710917427,
      "learning_rate": 3.5136204976207627e-06,
      "loss": 0.4033,
      "step": 4499
    },
    {
      "epoch": 5.507955936352509,
      "grad_norm": 1.4261986831975177,
      "learning_rate": 3.513034624569957e-06,
      "loss": 0.3832,
      "step": 4500
    },
    {
      "epoch": 5.509179926560588,
      "grad_norm": 1.72178934908833,
      "learning_rate": 3.512448684946262e-06,
      "loss": 0.9836,
      "step": 4501
    },
    {
      "epoch": 5.510403916768666,
      "grad_norm": 1.3706483216317793,
      "learning_rate": 3.511862678788185e-06,
      "loss": 0.5799,
      "step": 4502
    },
    {
      "epoch": 5.511627906976744,
      "grad_norm": 1.3550558367786212,
      "learning_rate": 3.5112766061342346e-06,
      "loss": 0.5826,
      "step": 4503
    },
    {
      "epoch": 5.512851897184823,
      "grad_norm": 1.786526091837288,
      "learning_rate": 3.510690467022926e-06,
      "loss": 0.3205,
      "step": 4504
    },
    {
      "epoch": 5.5140758873929006,
      "grad_norm": 0.9536819554236164,
      "learning_rate": 3.5101042614927794e-06,
      "loss": 0.9104,
      "step": 4505
    },
    {
      "epoch": 5.515299877600979,
      "grad_norm": 1.514253489767651,
      "learning_rate": 3.509517989582316e-06,
      "loss": 0.393,
      "step": 4506
    },
    {
      "epoch": 5.516523867809058,
      "grad_norm": 2.2099752109057014,
      "learning_rate": 3.5089316513300643e-06,
      "loss": 0.5326,
      "step": 4507
    },
    {
      "epoch": 5.5177478580171355,
      "grad_norm": 1.089544589394896,
      "learning_rate": 3.5083452467745565e-06,
      "loss": 0.4528,
      "step": 4508
    },
    {
      "epoch": 5.518971848225214,
      "grad_norm": 1.5448205511476936,
      "learning_rate": 3.5077587759543285e-06,
      "loss": 0.4419,
      "step": 4509
    },
    {
      "epoch": 5.520195838433293,
      "grad_norm": 1.4246716310894927,
      "learning_rate": 3.507172238907921e-06,
      "loss": 0.4876,
      "step": 4510
    },
    {
      "epoch": 5.5214198286413705,
      "grad_norm": 1.450941148201937,
      "learning_rate": 3.5065856356738796e-06,
      "loss": 0.5778,
      "step": 4511
    },
    {
      "epoch": 5.522643818849449,
      "grad_norm": 1.337003392726757,
      "learning_rate": 3.5059989662907536e-06,
      "loss": 0.5287,
      "step": 4512
    },
    {
      "epoch": 5.523867809057528,
      "grad_norm": 1.5392359607983546,
      "learning_rate": 3.505412230797096e-06,
      "loss": 1.1108,
      "step": 4513
    },
    {
      "epoch": 5.525091799265606,
      "grad_norm": 1.6185275318676418,
      "learning_rate": 3.5048254292314654e-06,
      "loss": 0.8057,
      "step": 4514
    },
    {
      "epoch": 5.526315789473684,
      "grad_norm": 0.9544119587209946,
      "learning_rate": 3.5042385616324243e-06,
      "loss": 0.5634,
      "step": 4515
    },
    {
      "epoch": 5.527539779681763,
      "grad_norm": 1.1277508237321088,
      "learning_rate": 3.5036516280385386e-06,
      "loss": 0.9998,
      "step": 4516
    },
    {
      "epoch": 5.52876376988984,
      "grad_norm": 1.249757566077675,
      "learning_rate": 3.503064628488382e-06,
      "loss": 0.4953,
      "step": 4517
    },
    {
      "epoch": 5.529987760097919,
      "grad_norm": 1.5072032371191788,
      "learning_rate": 3.502477563020526e-06,
      "loss": 0.9879,
      "step": 4518
    },
    {
      "epoch": 5.531211750305998,
      "grad_norm": 1.553006396826202,
      "learning_rate": 3.5018904316735534e-06,
      "loss": 0.6048,
      "step": 4519
    },
    {
      "epoch": 5.532435740514076,
      "grad_norm": 1.2126353497554916,
      "learning_rate": 3.5013032344860478e-06,
      "loss": 0.4888,
      "step": 4520
    },
    {
      "epoch": 5.533659730722154,
      "grad_norm": 1.5962911281496284,
      "learning_rate": 3.5007159714965966e-06,
      "loss": 0.6533,
      "step": 4521
    },
    {
      "epoch": 5.534883720930233,
      "grad_norm": 1.4617366764802464,
      "learning_rate": 3.500128642743793e-06,
      "loss": 0.5218,
      "step": 4522
    },
    {
      "epoch": 5.536107711138311,
      "grad_norm": 1.17772115383515,
      "learning_rate": 3.499541248266235e-06,
      "loss": 0.2442,
      "step": 4523
    },
    {
      "epoch": 5.537331701346389,
      "grad_norm": 1.9983894678419472,
      "learning_rate": 3.498953788102522e-06,
      "loss": 0.9809,
      "step": 4524
    },
    {
      "epoch": 5.538555691554468,
      "grad_norm": 1.5609606154458777,
      "learning_rate": 3.4983662622912625e-06,
      "loss": 0.6433,
      "step": 4525
    },
    {
      "epoch": 5.539779681762546,
      "grad_norm": 1.2342106761869924,
      "learning_rate": 3.497778670871064e-06,
      "loss": 0.5034,
      "step": 4526
    },
    {
      "epoch": 5.541003671970624,
      "grad_norm": 0.9643959081184507,
      "learning_rate": 3.497191013880541e-06,
      "loss": 0.4743,
      "step": 4527
    },
    {
      "epoch": 5.5422276621787026,
      "grad_norm": 1.8629260894428068,
      "learning_rate": 3.4966032913583143e-06,
      "loss": 0.6313,
      "step": 4528
    },
    {
      "epoch": 5.543451652386781,
      "grad_norm": 1.2991335125703563,
      "learning_rate": 3.496015503343005e-06,
      "loss": 0.4618,
      "step": 4529
    },
    {
      "epoch": 5.544675642594859,
      "grad_norm": 1.0519949448564376,
      "learning_rate": 3.495427649873241e-06,
      "loss": 0.572,
      "step": 4530
    },
    {
      "epoch": 5.5458996328029375,
      "grad_norm": 2.9771598374108885,
      "learning_rate": 3.494839730987653e-06,
      "loss": 0.4326,
      "step": 4531
    },
    {
      "epoch": 5.547123623011016,
      "grad_norm": 1.2611873490767895,
      "learning_rate": 3.4942517467248783e-06,
      "loss": 0.5791,
      "step": 4532
    },
    {
      "epoch": 5.548347613219094,
      "grad_norm": 1.5593917488608595,
      "learning_rate": 3.4936636971235555e-06,
      "loss": 0.4135,
      "step": 4533
    },
    {
      "epoch": 5.5495716034271725,
      "grad_norm": 1.8629401784231996,
      "learning_rate": 3.4930755822223307e-06,
      "loss": 0.4985,
      "step": 4534
    },
    {
      "epoch": 5.550795593635251,
      "grad_norm": 1.0629718580952152,
      "learning_rate": 3.4924874020598508e-06,
      "loss": 0.6978,
      "step": 4535
    },
    {
      "epoch": 5.552019583843329,
      "grad_norm": 1.3587721925172065,
      "learning_rate": 3.4918991566747705e-06,
      "loss": 1.5476,
      "step": 4536
    },
    {
      "epoch": 5.5532435740514074,
      "grad_norm": 1.063876480974483,
      "learning_rate": 3.4913108461057467e-06,
      "loss": 1.1123,
      "step": 4537
    },
    {
      "epoch": 5.554467564259486,
      "grad_norm": 1.38080796330293,
      "learning_rate": 3.49072247039144e-06,
      "loss": 1.0129,
      "step": 4538
    },
    {
      "epoch": 5.555691554467565,
      "grad_norm": 1.4123067297672984,
      "learning_rate": 3.4901340295705177e-06,
      "loss": 0.9388,
      "step": 4539
    },
    {
      "epoch": 5.556915544675642,
      "grad_norm": 1.3049772487053188,
      "learning_rate": 3.4895455236816488e-06,
      "loss": 1.0035,
      "step": 4540
    },
    {
      "epoch": 5.558139534883721,
      "grad_norm": 0.9942072851418005,
      "learning_rate": 3.488956952763508e-06,
      "loss": 0.4092,
      "step": 4541
    },
    {
      "epoch": 5.5593635250918,
      "grad_norm": 1.1756568507021177,
      "learning_rate": 3.4883683168547737e-06,
      "loss": 0.5028,
      "step": 4542
    },
    {
      "epoch": 5.560587515299877,
      "grad_norm": 1.8163524327934293,
      "learning_rate": 3.4877796159941305e-06,
      "loss": 0.5363,
      "step": 4543
    },
    {
      "epoch": 5.561811505507956,
      "grad_norm": 1.568205123004877,
      "learning_rate": 3.487190850220263e-06,
      "loss": 0.5727,
      "step": 4544
    },
    {
      "epoch": 5.563035495716035,
      "grad_norm": 1.2316535949039265,
      "learning_rate": 3.486602019571866e-06,
      "loss": 0.5216,
      "step": 4545
    },
    {
      "epoch": 5.564259485924112,
      "grad_norm": 1.3494579919185705,
      "learning_rate": 3.4860131240876327e-06,
      "loss": 1.2005,
      "step": 4546
    },
    {
      "epoch": 5.565483476132191,
      "grad_norm": 1.656995356552588,
      "learning_rate": 3.4854241638062638e-06,
      "loss": 1.1487,
      "step": 4547
    },
    {
      "epoch": 5.56670746634027,
      "grad_norm": 1.01722414791791,
      "learning_rate": 3.484835138766463e-06,
      "loss": 0.6018,
      "step": 4548
    },
    {
      "epoch": 5.567931456548347,
      "grad_norm": 0.7345505918581199,
      "learning_rate": 3.4842460490069407e-06,
      "loss": 0.5082,
      "step": 4549
    },
    {
      "epoch": 5.569155446756426,
      "grad_norm": 0.8716065633225716,
      "learning_rate": 3.4836568945664083e-06,
      "loss": 0.6125,
      "step": 4550
    },
    {
      "epoch": 5.5703794369645045,
      "grad_norm": 1.1275261427508716,
      "learning_rate": 3.4830676754835824e-06,
      "loss": 0.5231,
      "step": 4551
    },
    {
      "epoch": 5.571603427172582,
      "grad_norm": 1.067519169162483,
      "learning_rate": 3.482478391797186e-06,
      "loss": 0.5672,
      "step": 4552
    },
    {
      "epoch": 5.572827417380661,
      "grad_norm": 2.0099138059333566,
      "learning_rate": 3.481889043545942e-06,
      "loss": 0.5202,
      "step": 4553
    },
    {
      "epoch": 5.5740514075887395,
      "grad_norm": 1.372440849791154,
      "learning_rate": 3.481299630768583e-06,
      "loss": 0.3228,
      "step": 4554
    },
    {
      "epoch": 5.575275397796817,
      "grad_norm": 2.2194235204547543,
      "learning_rate": 3.480710153503842e-06,
      "loss": 1.067,
      "step": 4555
    },
    {
      "epoch": 5.576499388004896,
      "grad_norm": 1.1103151661363448,
      "learning_rate": 3.480120611790457e-06,
      "loss": 0.3834,
      "step": 4556
    },
    {
      "epoch": 5.5777233782129745,
      "grad_norm": 1.0673240103273345,
      "learning_rate": 3.4795310056671698e-06,
      "loss": 0.5736,
      "step": 4557
    },
    {
      "epoch": 5.578947368421053,
      "grad_norm": 1.5624231178227903,
      "learning_rate": 3.478941335172729e-06,
      "loss": 0.5652,
      "step": 4558
    },
    {
      "epoch": 5.580171358629131,
      "grad_norm": 0.907406388327696,
      "learning_rate": 3.478351600345884e-06,
      "loss": 0.6157,
      "step": 4559
    },
    {
      "epoch": 5.5813953488372094,
      "grad_norm": 1.8265788302024,
      "learning_rate": 3.47776180122539e-06,
      "loss": 0.4437,
      "step": 4560
    },
    {
      "epoch": 5.582619339045287,
      "grad_norm": 1.4905920444912402,
      "learning_rate": 3.477171937850008e-06,
      "loss": 0.6384,
      "step": 4561
    },
    {
      "epoch": 5.583843329253366,
      "grad_norm": 1.748326430907189,
      "learning_rate": 3.4765820102585e-06,
      "loss": 0.6734,
      "step": 4562
    },
    {
      "epoch": 5.585067319461444,
      "grad_norm": 1.4264402857974654,
      "learning_rate": 3.4759920184896345e-06,
      "loss": 0.7347,
      "step": 4563
    },
    {
      "epoch": 5.586291309669523,
      "grad_norm": 1.279186249412405,
      "learning_rate": 3.4754019625821835e-06,
      "loss": 0.7486,
      "step": 4564
    },
    {
      "epoch": 5.587515299877601,
      "grad_norm": 2.0643576089605333,
      "learning_rate": 3.4748118425749233e-06,
      "loss": 0.592,
      "step": 4565
    },
    {
      "epoch": 5.588739290085679,
      "grad_norm": 1.3065774460443254,
      "learning_rate": 3.4742216585066334e-06,
      "loss": 0.4953,
      "step": 4566
    },
    {
      "epoch": 5.589963280293758,
      "grad_norm": 1.5553452122542701,
      "learning_rate": 3.473631410416101e-06,
      "loss": 0.5601,
      "step": 4567
    },
    {
      "epoch": 5.591187270501836,
      "grad_norm": 1.7524647803016193,
      "learning_rate": 3.4730410983421126e-06,
      "loss": 1.1008,
      "step": 4568
    },
    {
      "epoch": 5.592411260709914,
      "grad_norm": 1.6409793517090185,
      "learning_rate": 3.472450722323463e-06,
      "loss": 0.3449,
      "step": 4569
    },
    {
      "epoch": 5.593635250917993,
      "grad_norm": 0.9743554345885426,
      "learning_rate": 3.4718602823989477e-06,
      "loss": 0.6053,
      "step": 4570
    },
    {
      "epoch": 5.594859241126071,
      "grad_norm": 2.3670113050007964,
      "learning_rate": 3.47126977860737e-06,
      "loss": 0.4496,
      "step": 4571
    },
    {
      "epoch": 5.596083231334149,
      "grad_norm": 1.8880610440514325,
      "learning_rate": 3.4706792109875352e-06,
      "loss": 1.0053,
      "step": 4572
    },
    {
      "epoch": 5.597307221542228,
      "grad_norm": 1.0882213285905011,
      "learning_rate": 3.4700885795782522e-06,
      "loss": 0.525,
      "step": 4573
    },
    {
      "epoch": 5.598531211750306,
      "grad_norm": 1.1630271625421398,
      "learning_rate": 3.4694978844183364e-06,
      "loss": 0.5295,
      "step": 4574
    },
    {
      "epoch": 5.599755201958384,
      "grad_norm": 1.7083406927578428,
      "learning_rate": 3.4689071255466055e-06,
      "loss": 0.502,
      "step": 4575
    },
    {
      "epoch": 5.600979192166463,
      "grad_norm": 1.623912515892955,
      "learning_rate": 3.4683163030018824e-06,
      "loss": 0.4658,
      "step": 4576
    },
    {
      "epoch": 5.602203182374541,
      "grad_norm": 0.8674286008863695,
      "learning_rate": 3.467725416822992e-06,
      "loss": 0.6969,
      "step": 4577
    },
    {
      "epoch": 5.603427172582619,
      "grad_norm": 1.1899056452051293,
      "learning_rate": 3.4671344670487683e-06,
      "loss": 0.6244,
      "step": 4578
    },
    {
      "epoch": 5.604651162790698,
      "grad_norm": 1.1966907964581717,
      "learning_rate": 3.4665434537180435e-06,
      "loss": 0.4543,
      "step": 4579
    },
    {
      "epoch": 5.605875152998776,
      "grad_norm": 1.170519748499361,
      "learning_rate": 3.4659523768696584e-06,
      "loss": 0.5991,
      "step": 4580
    },
    {
      "epoch": 5.607099143206854,
      "grad_norm": 1.8201086862446105,
      "learning_rate": 3.465361236542456e-06,
      "loss": 0.5818,
      "step": 4581
    },
    {
      "epoch": 5.608323133414933,
      "grad_norm": 1.5690091666775403,
      "learning_rate": 3.4647700327752837e-06,
      "loss": 0.4453,
      "step": 4582
    },
    {
      "epoch": 5.6095471236230114,
      "grad_norm": 1.324426782562437,
      "learning_rate": 3.4641787656069924e-06,
      "loss": 0.8771,
      "step": 4583
    },
    {
      "epoch": 5.610771113831089,
      "grad_norm": 1.7607524492803233,
      "learning_rate": 3.4635874350764405e-06,
      "loss": 0.5569,
      "step": 4584
    },
    {
      "epoch": 5.611995104039168,
      "grad_norm": 1.2795171772223004,
      "learning_rate": 3.462996041222485e-06,
      "loss": 0.5091,
      "step": 4585
    },
    {
      "epoch": 5.613219094247246,
      "grad_norm": 1.3263327073565494,
      "learning_rate": 3.462404584083992e-06,
      "loss": 1.3228,
      "step": 4586
    },
    {
      "epoch": 5.614443084455324,
      "grad_norm": 1.5055461648632704,
      "learning_rate": 3.4618130636998297e-06,
      "loss": 0.7649,
      "step": 4587
    },
    {
      "epoch": 5.615667074663403,
      "grad_norm": 1.2761071582071668,
      "learning_rate": 3.46122148010887e-06,
      "loss": 0.4929,
      "step": 4588
    },
    {
      "epoch": 5.616891064871481,
      "grad_norm": 1.0761960680953506,
      "learning_rate": 3.4606298333499903e-06,
      "loss": 0.9121,
      "step": 4589
    },
    {
      "epoch": 5.618115055079559,
      "grad_norm": 1.2676981090094064,
      "learning_rate": 3.460038123462071e-06,
      "loss": 0.4842,
      "step": 4590
    },
    {
      "epoch": 5.619339045287638,
      "grad_norm": 1.3910693142435633,
      "learning_rate": 3.459446350483997e-06,
      "loss": 0.955,
      "step": 4591
    },
    {
      "epoch": 5.620563035495716,
      "grad_norm": 1.3070360986105605,
      "learning_rate": 3.4588545144546572e-06,
      "loss": 1.2493,
      "step": 4592
    },
    {
      "epoch": 5.621787025703794,
      "grad_norm": 1.7186688561289396,
      "learning_rate": 3.4582626154129464e-06,
      "loss": 0.6047,
      "step": 4593
    },
    {
      "epoch": 5.623011015911873,
      "grad_norm": 1.2985435660791929,
      "learning_rate": 3.4576706533977596e-06,
      "loss": 0.9606,
      "step": 4594
    },
    {
      "epoch": 5.624235006119951,
      "grad_norm": 1.4281199858853284,
      "learning_rate": 3.4570786284480003e-06,
      "loss": 0.5541,
      "step": 4595
    },
    {
      "epoch": 5.625458996328029,
      "grad_norm": 1.117426426037937,
      "learning_rate": 3.456486540602574e-06,
      "loss": 1.1758,
      "step": 4596
    },
    {
      "epoch": 5.626682986536108,
      "grad_norm": 0.7490830079084273,
      "learning_rate": 3.4558943899003893e-06,
      "loss": 0.4505,
      "step": 4597
    },
    {
      "epoch": 5.627906976744186,
      "grad_norm": 2.5929595914830044,
      "learning_rate": 3.4553021763803606e-06,
      "loss": 0.4078,
      "step": 4598
    },
    {
      "epoch": 5.629130966952264,
      "grad_norm": 1.1322187200254483,
      "learning_rate": 3.454709900081407e-06,
      "loss": 0.8675,
      "step": 4599
    },
    {
      "epoch": 5.630354957160343,
      "grad_norm": 1.737086604315628,
      "learning_rate": 3.4541175610424492e-06,
      "loss": 0.3854,
      "step": 4600
    },
    {
      "epoch": 5.631578947368421,
      "grad_norm": 1.54428185800753,
      "learning_rate": 3.4535251593024157e-06,
      "loss": 0.7541,
      "step": 4601
    },
    {
      "epoch": 5.6328029375765,
      "grad_norm": 1.075339172706889,
      "learning_rate": 3.452932694900235e-06,
      "loss": 0.7363,
      "step": 4602
    },
    {
      "epoch": 5.634026927784578,
      "grad_norm": 0.975111545127296,
      "learning_rate": 3.4523401678748415e-06,
      "loss": 0.4899,
      "step": 4603
    },
    {
      "epoch": 5.635250917992656,
      "grad_norm": 1.1000692358903483,
      "learning_rate": 3.4517475782651754e-06,
      "loss": 0.7838,
      "step": 4604
    },
    {
      "epoch": 5.636474908200734,
      "grad_norm": 1.889902040337147,
      "learning_rate": 3.4511549261101785e-06,
      "loss": 0.4901,
      "step": 4605
    },
    {
      "epoch": 5.6376988984088126,
      "grad_norm": 1.4802673699327256,
      "learning_rate": 3.450562211448798e-06,
      "loss": 0.3677,
      "step": 4606
    },
    {
      "epoch": 5.638922888616891,
      "grad_norm": 1.2260124900267313,
      "learning_rate": 3.449969434319984e-06,
      "loss": 0.5895,
      "step": 4607
    },
    {
      "epoch": 5.64014687882497,
      "grad_norm": 0.9098568360986651,
      "learning_rate": 3.449376594762694e-06,
      "loss": 0.5157,
      "step": 4608
    },
    {
      "epoch": 5.6413708690330475,
      "grad_norm": 1.1417728705232,
      "learning_rate": 3.4487836928158846e-06,
      "loss": 0.3867,
      "step": 4609
    },
    {
      "epoch": 5.642594859241126,
      "grad_norm": 2.9712287069535215,
      "learning_rate": 3.4481907285185202e-06,
      "loss": 0.4147,
      "step": 4610
    },
    {
      "epoch": 5.643818849449205,
      "grad_norm": 1.0968067296264468,
      "learning_rate": 3.4475977019095684e-06,
      "loss": 0.3892,
      "step": 4611
    },
    {
      "epoch": 5.6450428396572825,
      "grad_norm": 1.3901764591758,
      "learning_rate": 3.447004613028001e-06,
      "loss": 1.3865,
      "step": 4612
    },
    {
      "epoch": 5.646266829865361,
      "grad_norm": 1.1254571113320158,
      "learning_rate": 3.446411461912793e-06,
      "loss": 0.4247,
      "step": 4613
    },
    {
      "epoch": 5.64749082007344,
      "grad_norm": 1.8125857761007533,
      "learning_rate": 3.4458182486029246e-06,
      "loss": 0.6097,
      "step": 4614
    },
    {
      "epoch": 5.6487148102815175,
      "grad_norm": 1.3558703377104802,
      "learning_rate": 3.445224973137379e-06,
      "loss": 1.5069,
      "step": 4615
    },
    {
      "epoch": 5.649938800489596,
      "grad_norm": 1.5200198098608584,
      "learning_rate": 3.4446316355551434e-06,
      "loss": 0.3993,
      "step": 4616
    },
    {
      "epoch": 5.651162790697675,
      "grad_norm": 1.2329440308529962,
      "learning_rate": 3.4440382358952116e-06,
      "loss": 0.5412,
      "step": 4617
    },
    {
      "epoch": 5.652386780905752,
      "grad_norm": 2.2481245129144267,
      "learning_rate": 3.443444774196578e-06,
      "loss": 0.4948,
      "step": 4618
    },
    {
      "epoch": 5.653610771113831,
      "grad_norm": 2.547123657267804,
      "learning_rate": 3.4428512504982443e-06,
      "loss": 0.4296,
      "step": 4619
    },
    {
      "epoch": 5.65483476132191,
      "grad_norm": 1.1031783944530436,
      "learning_rate": 3.4422576648392138e-06,
      "loss": 0.759,
      "step": 4620
    },
    {
      "epoch": 5.656058751529987,
      "grad_norm": 1.9856031480028187,
      "learning_rate": 3.4416640172584935e-06,
      "loss": 0.357,
      "step": 4621
    },
    {
      "epoch": 5.657282741738066,
      "grad_norm": 1.653930678874258,
      "learning_rate": 3.441070307795098e-06,
      "loss": 0.7674,
      "step": 4622
    },
    {
      "epoch": 5.658506731946145,
      "grad_norm": 1.3191015223249565,
      "learning_rate": 3.4404765364880425e-06,
      "loss": 0.5788,
      "step": 4623
    },
    {
      "epoch": 5.659730722154222,
      "grad_norm": 1.2407103634413703,
      "learning_rate": 3.4398827033763472e-06,
      "loss": 0.8192,
      "step": 4624
    },
    {
      "epoch": 5.660954712362301,
      "grad_norm": 1.4905111320663105,
      "learning_rate": 3.4392888084990374e-06,
      "loss": 0.7559,
      "step": 4625
    },
    {
      "epoch": 5.66217870257038,
      "grad_norm": 1.3731509503347776,
      "learning_rate": 3.4386948518951417e-06,
      "loss": 0.5523,
      "step": 4626
    },
    {
      "epoch": 5.663402692778458,
      "grad_norm": 1.0241204448351604,
      "learning_rate": 3.4381008336036915e-06,
      "loss": 0.9907,
      "step": 4627
    },
    {
      "epoch": 5.664626682986536,
      "grad_norm": 2.732782533476069,
      "learning_rate": 3.437506753663725e-06,
      "loss": 0.3258,
      "step": 4628
    },
    {
      "epoch": 5.6658506731946146,
      "grad_norm": 1.5004625474668927,
      "learning_rate": 3.4369126121142813e-06,
      "loss": 0.6204,
      "step": 4629
    },
    {
      "epoch": 5.667074663402692,
      "grad_norm": 0.9614849258595926,
      "learning_rate": 3.436318408994407e-06,
      "loss": 0.4963,
      "step": 4630
    },
    {
      "epoch": 5.668298653610771,
      "grad_norm": 1.1046101698094355,
      "learning_rate": 3.43572414434315e-06,
      "loss": 0.6314,
      "step": 4631
    },
    {
      "epoch": 5.6695226438188495,
      "grad_norm": 1.4943283012927704,
      "learning_rate": 3.435129818199563e-06,
      "loss": 0.5128,
      "step": 4632
    },
    {
      "epoch": 5.670746634026928,
      "grad_norm": 1.060113954302605,
      "learning_rate": 3.4345354306027024e-06,
      "loss": 0.476,
      "step": 4633
    },
    {
      "epoch": 5.671970624235006,
      "grad_norm": 1.1587033681445946,
      "learning_rate": 3.4339409815916315e-06,
      "loss": 0.697,
      "step": 4634
    },
    {
      "epoch": 5.6731946144430845,
      "grad_norm": 1.3348774079725048,
      "learning_rate": 3.433346471205412e-06,
      "loss": 0.5998,
      "step": 4635
    },
    {
      "epoch": 5.674418604651163,
      "grad_norm": 1.3691465511880456,
      "learning_rate": 3.432751899483116e-06,
      "loss": 1.525,
      "step": 4636
    },
    {
      "epoch": 5.675642594859241,
      "grad_norm": 2.4489984485203,
      "learning_rate": 3.4321572664638146e-06,
      "loss": 0.4627,
      "step": 4637
    },
    {
      "epoch": 5.6768665850673194,
      "grad_norm": 1.98024984745998,
      "learning_rate": 3.4315625721865858e-06,
      "loss": 0.5053,
      "step": 4638
    },
    {
      "epoch": 5.678090575275398,
      "grad_norm": 1.533127171216978,
      "learning_rate": 3.43096781669051e-06,
      "loss": 0.347,
      "step": 4639
    },
    {
      "epoch": 5.679314565483476,
      "grad_norm": 1.5791045294283645,
      "learning_rate": 3.430373000014673e-06,
      "loss": 0.6269,
      "step": 4640
    },
    {
      "epoch": 5.680538555691554,
      "grad_norm": 1.7890505384000008,
      "learning_rate": 3.4297781221981636e-06,
      "loss": 0.4124,
      "step": 4641
    },
    {
      "epoch": 5.681762545899633,
      "grad_norm": 1.895416453156362,
      "learning_rate": 3.429183183280074e-06,
      "loss": 0.5745,
      "step": 4642
    },
    {
      "epoch": 5.682986536107711,
      "grad_norm": 1.6628795879579383,
      "learning_rate": 3.4285881832995038e-06,
      "loss": 0.4124,
      "step": 4643
    },
    {
      "epoch": 5.684210526315789,
      "grad_norm": 1.344043380610681,
      "learning_rate": 3.427993122295552e-06,
      "loss": 1.1957,
      "step": 4644
    },
    {
      "epoch": 5.685434516523868,
      "grad_norm": 0.9310119804813397,
      "learning_rate": 3.4273980003073247e-06,
      "loss": 0.6573,
      "step": 4645
    },
    {
      "epoch": 5.686658506731947,
      "grad_norm": 1.3531528628695146,
      "learning_rate": 3.4268028173739315e-06,
      "loss": 0.458,
      "step": 4646
    },
    {
      "epoch": 5.687882496940024,
      "grad_norm": 1.3338578802989087,
      "learning_rate": 3.426207573534485e-06,
      "loss": 0.3363,
      "step": 4647
    },
    {
      "epoch": 5.689106487148103,
      "grad_norm": 2.0264481881053777,
      "learning_rate": 3.4256122688281025e-06,
      "loss": 0.469,
      "step": 4648
    },
    {
      "epoch": 5.690330477356181,
      "grad_norm": 1.1473377351043126,
      "learning_rate": 3.425016903293905e-06,
      "loss": 0.6171,
      "step": 4649
    },
    {
      "epoch": 5.691554467564259,
      "grad_norm": 0.4516739472672605,
      "learning_rate": 3.4244214769710184e-06,
      "loss": 0.134,
      "step": 4650
    },
    {
      "epoch": 5.692778457772338,
      "grad_norm": 0.8960516175496366,
      "learning_rate": 3.4238259898985714e-06,
      "loss": 0.5846,
      "step": 4651
    },
    {
      "epoch": 5.6940024479804165,
      "grad_norm": 2.0299269483928155,
      "learning_rate": 3.423230442115698e-06,
      "loss": 0.5479,
      "step": 4652
    },
    {
      "epoch": 5.695226438188494,
      "grad_norm": 1.547111122503179,
      "learning_rate": 3.4226348336615333e-06,
      "loss": 0.4259,
      "step": 4653
    },
    {
      "epoch": 5.696450428396573,
      "grad_norm": 2.105076560413822,
      "learning_rate": 3.422039164575221e-06,
      "loss": 0.4107,
      "step": 4654
    },
    {
      "epoch": 5.6976744186046515,
      "grad_norm": 2.1452855613864816,
      "learning_rate": 3.421443434895905e-06,
      "loss": 0.54,
      "step": 4655
    },
    {
      "epoch": 5.698898408812729,
      "grad_norm": 1.3391254131083528,
      "learning_rate": 3.420847644662734e-06,
      "loss": 0.7212,
      "step": 4656
    },
    {
      "epoch": 5.700122399020808,
      "grad_norm": 2.1127671042318883,
      "learning_rate": 3.420251793914862e-06,
      "loss": 0.6241,
      "step": 4657
    },
    {
      "epoch": 5.7013463892288865,
      "grad_norm": 2.27567421981427,
      "learning_rate": 3.4196558826914473e-06,
      "loss": 0.4435,
      "step": 4658
    },
    {
      "epoch": 5.702570379436964,
      "grad_norm": 0.9006584892840007,
      "learning_rate": 3.4190599110316477e-06,
      "loss": 0.6349,
      "step": 4659
    },
    {
      "epoch": 5.703794369645043,
      "grad_norm": 1.5657420356876341,
      "learning_rate": 3.4184638789746304e-06,
      "loss": 1.1387,
      "step": 4660
    },
    {
      "epoch": 5.7050183598531214,
      "grad_norm": 1.316867985665217,
      "learning_rate": 3.417867786559565e-06,
      "loss": 0.5947,
      "step": 4661
    },
    {
      "epoch": 5.706242350061199,
      "grad_norm": 1.1973451356468439,
      "learning_rate": 3.4172716338256235e-06,
      "loss": 0.673,
      "step": 4662
    },
    {
      "epoch": 5.707466340269278,
      "grad_norm": 1.4193805676585383,
      "learning_rate": 3.4166754208119823e-06,
      "loss": 0.6139,
      "step": 4663
    },
    {
      "epoch": 5.708690330477356,
      "grad_norm": 1.0717095457665797,
      "learning_rate": 3.4160791475578238e-06,
      "loss": 0.7904,
      "step": 4664
    },
    {
      "epoch": 5.709914320685434,
      "grad_norm": 1.3178983950945542,
      "learning_rate": 3.4154828141023316e-06,
      "loss": 0.6209,
      "step": 4665
    },
    {
      "epoch": 5.711138310893513,
      "grad_norm": 1.7210678312548422,
      "learning_rate": 3.4148864204846943e-06,
      "loss": 0.6536,
      "step": 4666
    },
    {
      "epoch": 5.712362301101591,
      "grad_norm": 1.0651184938867828,
      "learning_rate": 3.414289966744107e-06,
      "loss": 0.53,
      "step": 4667
    },
    {
      "epoch": 5.713586291309669,
      "grad_norm": 1.7110208110587926,
      "learning_rate": 3.4136934529197634e-06,
      "loss": 0.5429,
      "step": 4668
    },
    {
      "epoch": 5.714810281517748,
      "grad_norm": 1.655381376865821,
      "learning_rate": 3.4130968790508662e-06,
      "loss": 0.4472,
      "step": 4669
    },
    {
      "epoch": 5.716034271725826,
      "grad_norm": 1.1897691459913609,
      "learning_rate": 3.4125002451766194e-06,
      "loss": 0.6334,
      "step": 4670
    },
    {
      "epoch": 5.717258261933905,
      "grad_norm": 0.7636724248855992,
      "learning_rate": 3.411903551336232e-06,
      "loss": 0.4386,
      "step": 4671
    },
    {
      "epoch": 5.718482252141983,
      "grad_norm": 2.0847589220841662,
      "learning_rate": 3.411306797568915e-06,
      "loss": 0.557,
      "step": 4672
    },
    {
      "epoch": 5.719706242350061,
      "grad_norm": 1.4033157477760756,
      "learning_rate": 3.410709983913887e-06,
      "loss": 0.5595,
      "step": 4673
    },
    {
      "epoch": 5.720930232558139,
      "grad_norm": 1.8597527760853743,
      "learning_rate": 3.4101131104103664e-06,
      "loss": 1.0679,
      "step": 4674
    },
    {
      "epoch": 5.722154222766218,
      "grad_norm": 1.3787498956955657,
      "learning_rate": 3.4095161770975793e-06,
      "loss": 0.5715,
      "step": 4675
    },
    {
      "epoch": 5.723378212974296,
      "grad_norm": 1.3684012927345333,
      "learning_rate": 3.408919184014753e-06,
      "loss": 0.7072,
      "step": 4676
    },
    {
      "epoch": 5.724602203182375,
      "grad_norm": 1.6562503709412002,
      "learning_rate": 3.4083221312011195e-06,
      "loss": 0.682,
      "step": 4677
    },
    {
      "epoch": 5.725826193390453,
      "grad_norm": 1.463340150596859,
      "learning_rate": 3.4077250186959154e-06,
      "loss": 1.1864,
      "step": 4678
    },
    {
      "epoch": 5.727050183598531,
      "grad_norm": 1.6926720080950746,
      "learning_rate": 3.40712784653838e-06,
      "loss": 1.0315,
      "step": 4679
    },
    {
      "epoch": 5.72827417380661,
      "grad_norm": 1.4405817192828536,
      "learning_rate": 3.4065306147677585e-06,
      "loss": 0.4984,
      "step": 4680
    },
    {
      "epoch": 5.729498164014688,
      "grad_norm": 2.5677780088546895,
      "learning_rate": 3.4059333234232977e-06,
      "loss": 0.3911,
      "step": 4681
    },
    {
      "epoch": 5.730722154222766,
      "grad_norm": 2.193469501658793,
      "learning_rate": 3.40533597254425e-06,
      "loss": 0.5512,
      "step": 4682
    },
    {
      "epoch": 5.731946144430845,
      "grad_norm": 1.054097147279582,
      "learning_rate": 3.4047385621698696e-06,
      "loss": 0.7385,
      "step": 4683
    },
    {
      "epoch": 5.7331701346389226,
      "grad_norm": 1.2041869839717803,
      "learning_rate": 3.404141092339418e-06,
      "loss": 0.8034,
      "step": 4684
    },
    {
      "epoch": 5.734394124847001,
      "grad_norm": 1.1678821944741715,
      "learning_rate": 3.4035435630921587e-06,
      "loss": 0.5579,
      "step": 4685
    },
    {
      "epoch": 5.73561811505508,
      "grad_norm": 1.1511041760116236,
      "learning_rate": 3.402945974467358e-06,
      "loss": 0.3594,
      "step": 4686
    },
    {
      "epoch": 5.7368421052631575,
      "grad_norm": 1.3294651142489309,
      "learning_rate": 3.402348326504287e-06,
      "loss": 1.0031,
      "step": 4687
    },
    {
      "epoch": 5.738066095471236,
      "grad_norm": 1.223895806399522,
      "learning_rate": 3.401750619242222e-06,
      "loss": 0.4107,
      "step": 4688
    },
    {
      "epoch": 5.739290085679315,
      "grad_norm": 1.1826858359428178,
      "learning_rate": 3.401152852720441e-06,
      "loss": 1.252,
      "step": 4689
    },
    {
      "epoch": 5.740514075887393,
      "grad_norm": 1.7217059342451206,
      "learning_rate": 3.4005550269782297e-06,
      "loss": 1.14,
      "step": 4690
    },
    {
      "epoch": 5.741738066095471,
      "grad_norm": 2.2147249206878517,
      "learning_rate": 3.3999571420548715e-06,
      "loss": 0.2498,
      "step": 4691
    },
    {
      "epoch": 5.74296205630355,
      "grad_norm": 1.3405545892404183,
      "learning_rate": 3.3993591979896578e-06,
      "loss": 0.6869,
      "step": 4692
    },
    {
      "epoch": 5.7441860465116275,
      "grad_norm": 1.3885961932353337,
      "learning_rate": 3.398761194821886e-06,
      "loss": 0.6568,
      "step": 4693
    },
    {
      "epoch": 5.745410036719706,
      "grad_norm": 1.006290292257215,
      "learning_rate": 3.398163132590851e-06,
      "loss": 0.6168,
      "step": 4694
    },
    {
      "epoch": 5.746634026927785,
      "grad_norm": 1.655474878102463,
      "learning_rate": 3.3975650113358584e-06,
      "loss": 0.4252,
      "step": 4695
    },
    {
      "epoch": 5.747858017135863,
      "grad_norm": 1.114559452741794,
      "learning_rate": 3.3969668310962123e-06,
      "loss": 0.5799,
      "step": 4696
    },
    {
      "epoch": 5.749082007343941,
      "grad_norm": 1.835562648527684,
      "learning_rate": 3.396368591911224e-06,
      "loss": 0.4962,
      "step": 4697
    },
    {
      "epoch": 5.75030599755202,
      "grad_norm": 1.4550619060394339,
      "learning_rate": 3.3957702938202064e-06,
      "loss": 1.1312,
      "step": 4698
    },
    {
      "epoch": 5.751529987760098,
      "grad_norm": 1.399794925451687,
      "learning_rate": 3.39517193686248e-06,
      "loss": 0.4921,
      "step": 4699
    },
    {
      "epoch": 5.752753977968176,
      "grad_norm": 1.3425441513821108,
      "learning_rate": 3.3945735210773634e-06,
      "loss": 0.5276,
      "step": 4700
    },
    {
      "epoch": 5.753977968176255,
      "grad_norm": 1.6814308181579505,
      "learning_rate": 3.3939750465041844e-06,
      "loss": 0.5667,
      "step": 4701
    },
    {
      "epoch": 5.755201958384333,
      "grad_norm": 0.9799731042801744,
      "learning_rate": 3.393376513182273e-06,
      "loss": 0.5751,
      "step": 4702
    },
    {
      "epoch": 5.756425948592411,
      "grad_norm": 1.1311701212683511,
      "learning_rate": 3.3927779211509593e-06,
      "loss": 0.7161,
      "step": 4703
    },
    {
      "epoch": 5.75764993880049,
      "grad_norm": 0.937795753745444,
      "learning_rate": 3.392179270449584e-06,
      "loss": 0.6315,
      "step": 4704
    },
    {
      "epoch": 5.758873929008568,
      "grad_norm": 1.605199697768647,
      "learning_rate": 3.3915805611174872e-06,
      "loss": 0.6255,
      "step": 4705
    },
    {
      "epoch": 5.760097919216646,
      "grad_norm": 1.819755120798666,
      "learning_rate": 3.3909817931940135e-06,
      "loss": 0.4025,
      "step": 4706
    },
    {
      "epoch": 5.7613219094247246,
      "grad_norm": 1.7233900962708872,
      "learning_rate": 3.3903829667185114e-06,
      "loss": 0.9989,
      "step": 4707
    },
    {
      "epoch": 5.762545899632803,
      "grad_norm": 1.874320020158142,
      "learning_rate": 3.389784081730335e-06,
      "loss": 0.4815,
      "step": 4708
    },
    {
      "epoch": 5.763769889840881,
      "grad_norm": 1.4948401422322481,
      "learning_rate": 3.3891851382688383e-06,
      "loss": 1.3339,
      "step": 4709
    },
    {
      "epoch": 5.7649938800489595,
      "grad_norm": 1.1027776295305411,
      "learning_rate": 3.388586136373384e-06,
      "loss": 0.5302,
      "step": 4710
    },
    {
      "epoch": 5.766217870257038,
      "grad_norm": 1.6723292291859584,
      "learning_rate": 3.387987076083336e-06,
      "loss": 0.5435,
      "step": 4711
    },
    {
      "epoch": 5.767441860465116,
      "grad_norm": 1.6483645009063461,
      "learning_rate": 3.3873879574380615e-06,
      "loss": 1.167,
      "step": 4712
    },
    {
      "epoch": 5.7686658506731945,
      "grad_norm": 1.7467049923132079,
      "learning_rate": 3.3867887804769315e-06,
      "loss": 0.6351,
      "step": 4713
    },
    {
      "epoch": 5.769889840881273,
      "grad_norm": 1.0669768437911418,
      "learning_rate": 3.386189545239325e-06,
      "loss": 0.7506,
      "step": 4714
    },
    {
      "epoch": 5.771113831089352,
      "grad_norm": 2.005357207935525,
      "learning_rate": 3.3855902517646185e-06,
      "loss": 0.5641,
      "step": 4715
    },
    {
      "epoch": 5.7723378212974294,
      "grad_norm": 1.2044479059765083,
      "learning_rate": 3.3849909000921955e-06,
      "loss": 0.8958,
      "step": 4716
    },
    {
      "epoch": 5.773561811505508,
      "grad_norm": 1.8752794311572092,
      "learning_rate": 3.384391490261445e-06,
      "loss": 0.6233,
      "step": 4717
    },
    {
      "epoch": 5.774785801713586,
      "grad_norm": 1.534799460904561,
      "learning_rate": 3.3837920223117553e-06,
      "loss": 0.4714,
      "step": 4718
    },
    {
      "epoch": 5.776009791921664,
      "grad_norm": 1.2564495992480567,
      "learning_rate": 3.383192496282524e-06,
      "loss": 0.5146,
      "step": 4719
    },
    {
      "epoch": 5.777233782129743,
      "grad_norm": 2.4046941607296413,
      "learning_rate": 3.3825929122131484e-06,
      "loss": 0.494,
      "step": 4720
    },
    {
      "epoch": 5.778457772337822,
      "grad_norm": 0.6508267653326365,
      "learning_rate": 3.381993270143031e-06,
      "loss": 0.23,
      "step": 4721
    },
    {
      "epoch": 5.779681762545899,
      "grad_norm": 0.9355567998644042,
      "learning_rate": 3.3813935701115785e-06,
      "loss": 0.5759,
      "step": 4722
    },
    {
      "epoch": 5.780905752753978,
      "grad_norm": 1.3413944913774398,
      "learning_rate": 3.3807938121582002e-06,
      "loss": 0.6963,
      "step": 4723
    },
    {
      "epoch": 5.782129742962057,
      "grad_norm": 1.617985887241337,
      "learning_rate": 3.38019399632231e-06,
      "loss": 0.4439,
      "step": 4724
    },
    {
      "epoch": 5.783353733170134,
      "grad_norm": 1.0246479107500495,
      "learning_rate": 3.3795941226433265e-06,
      "loss": 0.424,
      "step": 4725
    },
    {
      "epoch": 5.784577723378213,
      "grad_norm": 1.1001922253770324,
      "learning_rate": 3.3789941911606707e-06,
      "loss": 0.3301,
      "step": 4726
    },
    {
      "epoch": 5.785801713586292,
      "grad_norm": 1.8860657180826064,
      "learning_rate": 3.378394201913768e-06,
      "loss": 1.0741,
      "step": 4727
    },
    {
      "epoch": 5.787025703794369,
      "grad_norm": 2.112050689122249,
      "learning_rate": 3.3777941549420466e-06,
      "loss": 0.7086,
      "step": 4728
    },
    {
      "epoch": 5.788249694002448,
      "grad_norm": 1.4431526108850723,
      "learning_rate": 3.377194050284941e-06,
      "loss": 1.0263,
      "step": 4729
    },
    {
      "epoch": 5.7894736842105265,
      "grad_norm": 1.1048324208513078,
      "learning_rate": 3.3765938879818865e-06,
      "loss": 0.4645,
      "step": 4730
    },
    {
      "epoch": 5.790697674418604,
      "grad_norm": 1.6978917551806825,
      "learning_rate": 3.3759936680723238e-06,
      "loss": 0.6393,
      "step": 4731
    },
    {
      "epoch": 5.791921664626683,
      "grad_norm": 1.8030446848071437,
      "learning_rate": 3.375393390595698e-06,
      "loss": 0.4515,
      "step": 4732
    },
    {
      "epoch": 5.7931456548347615,
      "grad_norm": 2.2061038025235593,
      "learning_rate": 3.374793055591455e-06,
      "loss": 0.4412,
      "step": 4733
    },
    {
      "epoch": 5.79436964504284,
      "grad_norm": 1.26932723582523,
      "learning_rate": 3.374192663099049e-06,
      "loss": 0.8314,
      "step": 4734
    },
    {
      "epoch": 5.795593635250918,
      "grad_norm": 1.2526051109845684,
      "learning_rate": 3.3735922131579346e-06,
      "loss": 0.6063,
      "step": 4735
    },
    {
      "epoch": 5.7968176254589965,
      "grad_norm": 1.1583103137090285,
      "learning_rate": 3.372991705807571e-06,
      "loss": 0.6459,
      "step": 4736
    },
    {
      "epoch": 5.798041615667074,
      "grad_norm": 0.9071214656185663,
      "learning_rate": 3.3723911410874215e-06,
      "loss": 0.4053,
      "step": 4737
    },
    {
      "epoch": 5.799265605875153,
      "grad_norm": 1.289727845311229,
      "learning_rate": 3.371790519036953e-06,
      "loss": 0.7286,
      "step": 4738
    },
    {
      "epoch": 5.8004895960832314,
      "grad_norm": 2.4306760695029586,
      "learning_rate": 3.3711898396956357e-06,
      "loss": 0.7795,
      "step": 4739
    },
    {
      "epoch": 5.80171358629131,
      "grad_norm": 1.6536154427922443,
      "learning_rate": 3.3705891031029454e-06,
      "loss": 1.1838,
      "step": 4740
    },
    {
      "epoch": 5.802937576499388,
      "grad_norm": 1.3944987042177612,
      "learning_rate": 3.3699883092983586e-06,
      "loss": 1.355,
      "step": 4741
    },
    {
      "epoch": 5.804161566707466,
      "grad_norm": 1.0704019704817036,
      "learning_rate": 3.3693874583213583e-06,
      "loss": 0.558,
      "step": 4742
    },
    {
      "epoch": 5.805385556915545,
      "grad_norm": 0.9670669082458496,
      "learning_rate": 3.368786550211429e-06,
      "loss": 0.5248,
      "step": 4743
    },
    {
      "epoch": 5.806609547123623,
      "grad_norm": 1.2832717294643772,
      "learning_rate": 3.3681855850080624e-06,
      "loss": 0.8455,
      "step": 4744
    },
    {
      "epoch": 5.807833537331701,
      "grad_norm": 1.4170675010619873,
      "learning_rate": 3.3675845627507497e-06,
      "loss": 0.4414,
      "step": 4745
    },
    {
      "epoch": 5.80905752753978,
      "grad_norm": 1.8026649235831094,
      "learning_rate": 3.3669834834789883e-06,
      "loss": 0.7087,
      "step": 4746
    },
    {
      "epoch": 5.810281517747858,
      "grad_norm": 1.2473497493158057,
      "learning_rate": 3.36638234723228e-06,
      "loss": 0.8896,
      "step": 4747
    },
    {
      "epoch": 5.811505507955936,
      "grad_norm": 1.4492645513702238,
      "learning_rate": 3.3657811540501266e-06,
      "loss": 1.117,
      "step": 4748
    },
    {
      "epoch": 5.812729498164015,
      "grad_norm": 1.7462706317995693,
      "learning_rate": 3.3651799039720402e-06,
      "loss": 0.5727,
      "step": 4749
    },
    {
      "epoch": 5.813953488372093,
      "grad_norm": 2.3965266258956226,
      "learning_rate": 3.3645785970375288e-06,
      "loss": 0.505,
      "step": 4750
    },
    {
      "epoch": 5.815177478580171,
      "grad_norm": 1.5802749973379855,
      "learning_rate": 3.363977233286111e-06,
      "loss": 0.5154,
      "step": 4751
    },
    {
      "epoch": 5.81640146878825,
      "grad_norm": 1.0419001467404796,
      "learning_rate": 3.363375812757305e-06,
      "loss": 0.6241,
      "step": 4752
    },
    {
      "epoch": 5.817625458996328,
      "grad_norm": 1.4216324596669725,
      "learning_rate": 3.362774335490633e-06,
      "loss": 1.4236,
      "step": 4753
    },
    {
      "epoch": 5.818849449204406,
      "grad_norm": 1.9718734130970683,
      "learning_rate": 3.362172801525624e-06,
      "loss": 0.5916,
      "step": 4754
    },
    {
      "epoch": 5.820073439412485,
      "grad_norm": 1.4076777735912656,
      "learning_rate": 3.361571210901807e-06,
      "loss": 0.4754,
      "step": 4755
    },
    {
      "epoch": 5.821297429620563,
      "grad_norm": 2.223873939331688,
      "learning_rate": 3.3609695636587162e-06,
      "loss": 0.6374,
      "step": 4756
    },
    {
      "epoch": 5.822521419828641,
      "grad_norm": 1.7048840538523715,
      "learning_rate": 3.3603678598358903e-06,
      "loss": 0.4058,
      "step": 4757
    },
    {
      "epoch": 5.82374541003672,
      "grad_norm": 1.5960632445842935,
      "learning_rate": 3.359766099472872e-06,
      "loss": 0.5354,
      "step": 4758
    },
    {
      "epoch": 5.8249694002447985,
      "grad_norm": 1.2041645383826338,
      "learning_rate": 3.3591642826092045e-06,
      "loss": 0.6388,
      "step": 4759
    },
    {
      "epoch": 5.826193390452876,
      "grad_norm": 1.8675812023442606,
      "learning_rate": 3.3585624092844386e-06,
      "loss": 0.5307,
      "step": 4760
    },
    {
      "epoch": 5.827417380660955,
      "grad_norm": 2.0466984037844513,
      "learning_rate": 3.357960479538127e-06,
      "loss": 0.3121,
      "step": 4761
    },
    {
      "epoch": 5.828641370869033,
      "grad_norm": 1.2850509807147308,
      "learning_rate": 3.357358493409826e-06,
      "loss": 1.3011,
      "step": 4762
    },
    {
      "epoch": 5.829865361077111,
      "grad_norm": 2.1222696427252017,
      "learning_rate": 3.3567564509390952e-06,
      "loss": 0.3855,
      "step": 4763
    },
    {
      "epoch": 5.83108935128519,
      "grad_norm": 1.0402785241550008,
      "learning_rate": 3.356154352165501e-06,
      "loss": 0.6002,
      "step": 4764
    },
    {
      "epoch": 5.832313341493268,
      "grad_norm": 1.6420742956173897,
      "learning_rate": 3.3555521971286082e-06,
      "loss": 0.5382,
      "step": 4765
    },
    {
      "epoch": 5.833537331701346,
      "grad_norm": 1.1345272892712217,
      "learning_rate": 3.3549499858679896e-06,
      "loss": 0.4553,
      "step": 4766
    },
    {
      "epoch": 5.834761321909425,
      "grad_norm": 1.2505532668733663,
      "learning_rate": 3.3543477184232208e-06,
      "loss": 0.6424,
      "step": 4767
    },
    {
      "epoch": 5.835985312117503,
      "grad_norm": 1.044059822245221,
      "learning_rate": 3.3537453948338793e-06,
      "loss": 0.5391,
      "step": 4768
    },
    {
      "epoch": 5.837209302325581,
      "grad_norm": 1.502179118890304,
      "learning_rate": 3.3531430151395483e-06,
      "loss": 0.4848,
      "step": 4769
    },
    {
      "epoch": 5.83843329253366,
      "grad_norm": 1.3236058369666388,
      "learning_rate": 3.3525405793798142e-06,
      "loss": 1.0939,
      "step": 4770
    },
    {
      "epoch": 5.839657282741738,
      "grad_norm": 1.0708368979179848,
      "learning_rate": 3.3519380875942663e-06,
      "loss": 0.8642,
      "step": 4771
    },
    {
      "epoch": 5.840881272949816,
      "grad_norm": 1.422177584055925,
      "learning_rate": 3.3513355398224985e-06,
      "loss": 0.55,
      "step": 4772
    },
    {
      "epoch": 5.842105263157895,
      "grad_norm": 1.5147739122249333,
      "learning_rate": 3.350732936104108e-06,
      "loss": 1.5179,
      "step": 4773
    },
    {
      "epoch": 5.843329253365973,
      "grad_norm": 1.0916718929203444,
      "learning_rate": 3.350130276478695e-06,
      "loss": 0.4118,
      "step": 4774
    },
    {
      "epoch": 5.844553243574051,
      "grad_norm": 1.1436822933028665,
      "learning_rate": 3.3495275609858653e-06,
      "loss": 0.9678,
      "step": 4775
    },
    {
      "epoch": 5.84577723378213,
      "grad_norm": 1.994620528801175,
      "learning_rate": 3.3489247896652265e-06,
      "loss": 1.0603,
      "step": 4776
    },
    {
      "epoch": 5.847001223990208,
      "grad_norm": 2.2830398121084854,
      "learning_rate": 3.34832196255639e-06,
      "loss": 0.4111,
      "step": 4777
    },
    {
      "epoch": 5.848225214198287,
      "grad_norm": 1.9855414359761454,
      "learning_rate": 3.3477190796989724e-06,
      "loss": 1.0512,
      "step": 4778
    },
    {
      "epoch": 5.849449204406365,
      "grad_norm": 1.051053101122059,
      "learning_rate": 3.347116141132592e-06,
      "loss": 0.4982,
      "step": 4779
    },
    {
      "epoch": 5.850673194614443,
      "grad_norm": 2.1695379138526576,
      "learning_rate": 3.3465131468968725e-06,
      "loss": 0.3987,
      "step": 4780
    },
    {
      "epoch": 5.851897184822521,
      "grad_norm": 0.6491680049106968,
      "learning_rate": 3.3459100970314394e-06,
      "loss": 0.2649,
      "step": 4781
    },
    {
      "epoch": 5.8531211750306,
      "grad_norm": 2.08306071318415,
      "learning_rate": 3.345306991575924e-06,
      "loss": 0.4668,
      "step": 4782
    },
    {
      "epoch": 5.854345165238678,
      "grad_norm": 1.4774349095876746,
      "learning_rate": 3.344703830569959e-06,
      "loss": 1.1575,
      "step": 4783
    },
    {
      "epoch": 5.855569155446757,
      "grad_norm": 1.8226289803551823,
      "learning_rate": 3.3441006140531833e-06,
      "loss": 0.9366,
      "step": 4784
    },
    {
      "epoch": 5.8567931456548346,
      "grad_norm": 1.5921354923097943,
      "learning_rate": 3.3434973420652374e-06,
      "loss": 0.7641,
      "step": 4785
    },
    {
      "epoch": 5.858017135862913,
      "grad_norm": 2.4448581852209497,
      "learning_rate": 3.3428940146457655e-06,
      "loss": 0.3528,
      "step": 4786
    },
    {
      "epoch": 5.859241126070992,
      "grad_norm": 1.5062296284080245,
      "learning_rate": 3.3422906318344174e-06,
      "loss": 1.1355,
      "step": 4787
    },
    {
      "epoch": 5.8604651162790695,
      "grad_norm": 2.3889292932573323,
      "learning_rate": 3.341687193670844e-06,
      "loss": 0.4435,
      "step": 4788
    },
    {
      "epoch": 5.861689106487148,
      "grad_norm": 1.1725768666417697,
      "learning_rate": 3.3410837001947005e-06,
      "loss": 0.2948,
      "step": 4789
    },
    {
      "epoch": 5.862913096695227,
      "grad_norm": 2.1597600679037243,
      "learning_rate": 3.3404801514456487e-06,
      "loss": 0.6345,
      "step": 4790
    },
    {
      "epoch": 5.8641370869033045,
      "grad_norm": 2.191558437015007,
      "learning_rate": 3.339876547463349e-06,
      "loss": 0.9573,
      "step": 4791
    },
    {
      "epoch": 5.865361077111383,
      "grad_norm": 1.4296110615214401,
      "learning_rate": 3.3392728882874693e-06,
      "loss": 0.6574,
      "step": 4792
    },
    {
      "epoch": 5.866585067319462,
      "grad_norm": 1.3774336177067132,
      "learning_rate": 3.33866917395768e-06,
      "loss": 0.5055,
      "step": 4793
    },
    {
      "epoch": 5.8678090575275395,
      "grad_norm": 1.51907597486823,
      "learning_rate": 3.3380654045136545e-06,
      "loss": 0.5175,
      "step": 4794
    },
    {
      "epoch": 5.869033047735618,
      "grad_norm": 2.69226392410121,
      "learning_rate": 3.3374615799950705e-06,
      "loss": 0.473,
      "step": 4795
    },
    {
      "epoch": 5.870257037943697,
      "grad_norm": 2.7343609671756055,
      "learning_rate": 3.3368577004416087e-06,
      "loss": 0.4069,
      "step": 4796
    },
    {
      "epoch": 5.871481028151774,
      "grad_norm": 1.1948536082153638,
      "learning_rate": 3.3362537658929544e-06,
      "loss": 0.8166,
      "step": 4797
    },
    {
      "epoch": 5.872705018359853,
      "grad_norm": 1.648627077244817,
      "learning_rate": 3.3356497763887953e-06,
      "loss": 1.3205,
      "step": 4798
    },
    {
      "epoch": 5.873929008567932,
      "grad_norm": 1.5330271205918367,
      "learning_rate": 3.3350457319688247e-06,
      "loss": 1.4144,
      "step": 4799
    },
    {
      "epoch": 5.875152998776009,
      "grad_norm": 2.2029791979788356,
      "learning_rate": 3.3344416326727356e-06,
      "loss": 0.9373,
      "step": 4800
    },
    {
      "epoch": 5.876376988984088,
      "grad_norm": 0.9937856804874036,
      "learning_rate": 3.33383747854023e-06,
      "loss": 0.4005,
      "step": 4801
    },
    {
      "epoch": 5.877600979192167,
      "grad_norm": 1.6279576069493515,
      "learning_rate": 3.3332332696110093e-06,
      "loss": 0.7163,
      "step": 4802
    },
    {
      "epoch": 5.878824969400245,
      "grad_norm": 1.7994527430915137,
      "learning_rate": 3.33262900592478e-06,
      "loss": 1.0051,
      "step": 4803
    },
    {
      "epoch": 5.880048959608323,
      "grad_norm": 1.9282381378389652,
      "learning_rate": 3.332024687521252e-06,
      "loss": 0.5695,
      "step": 4804
    },
    {
      "epoch": 5.881272949816402,
      "grad_norm": 0.8705454495556584,
      "learning_rate": 3.3314203144401393e-06,
      "loss": 0.6115,
      "step": 4805
    },
    {
      "epoch": 5.882496940024479,
      "grad_norm": 1.2747471371300139,
      "learning_rate": 3.330815886721158e-06,
      "loss": 0.7341,
      "step": 4806
    },
    {
      "epoch": 5.883720930232558,
      "grad_norm": 1.1914805131666637,
      "learning_rate": 3.3302114044040302e-06,
      "loss": 1.0482,
      "step": 4807
    },
    {
      "epoch": 5.8849449204406366,
      "grad_norm": 2.1334463695290706,
      "learning_rate": 3.32960686752848e-06,
      "loss": 0.8339,
      "step": 4808
    },
    {
      "epoch": 5.886168910648715,
      "grad_norm": 1.242129280596423,
      "learning_rate": 3.329002276134234e-06,
      "loss": 0.5023,
      "step": 4809
    },
    {
      "epoch": 5.887392900856793,
      "grad_norm": 1.290554537585815,
      "learning_rate": 3.328397630261025e-06,
      "loss": 1.2332,
      "step": 4810
    },
    {
      "epoch": 5.8886168910648715,
      "grad_norm": 1.4237047347952605,
      "learning_rate": 3.3277929299485878e-06,
      "loss": 0.4477,
      "step": 4811
    },
    {
      "epoch": 5.88984088127295,
      "grad_norm": 1.4926738762301228,
      "learning_rate": 3.3271881752366615e-06,
      "loss": 0.5606,
      "step": 4812
    },
    {
      "epoch": 5.891064871481028,
      "grad_norm": 1.5377588137706495,
      "learning_rate": 3.3265833661649873e-06,
      "loss": 0.6141,
      "step": 4813
    },
    {
      "epoch": 5.8922888616891065,
      "grad_norm": 2.132515545635948,
      "learning_rate": 3.325978502773312e-06,
      "loss": 0.5254,
      "step": 4814
    },
    {
      "epoch": 5.893512851897185,
      "grad_norm": 1.6499524698659518,
      "learning_rate": 3.325373585101384e-06,
      "loss": 1.2327,
      "step": 4815
    },
    {
      "epoch": 5.894736842105263,
      "grad_norm": 1.48053152388336,
      "learning_rate": 3.3247686131889573e-06,
      "loss": 1.0495,
      "step": 4816
    },
    {
      "epoch": 5.8959608323133414,
      "grad_norm": 1.3174300776954628,
      "learning_rate": 3.3241635870757887e-06,
      "loss": 0.5192,
      "step": 4817
    },
    {
      "epoch": 5.89718482252142,
      "grad_norm": 2.333996629859323,
      "learning_rate": 3.323558506801636e-06,
      "loss": 0.4911,
      "step": 4818
    },
    {
      "epoch": 5.898408812729498,
      "grad_norm": 1.2606485231361895,
      "learning_rate": 3.322953372406266e-06,
      "loss": 1.5802,
      "step": 4819
    },
    {
      "epoch": 5.899632802937576,
      "grad_norm": 0.9008448775214074,
      "learning_rate": 3.3223481839294424e-06,
      "loss": 0.326,
      "step": 4820
    },
    {
      "epoch": 5.900856793145655,
      "grad_norm": 2.543990490691916,
      "learning_rate": 3.3217429414109394e-06,
      "loss": 0.729,
      "step": 4821
    },
    {
      "epoch": 5.902080783353734,
      "grad_norm": 1.5209050747397532,
      "learning_rate": 3.3211376448905285e-06,
      "loss": 1.239,
      "step": 4822
    },
    {
      "epoch": 5.903304773561811,
      "grad_norm": 1.3616549720224305,
      "learning_rate": 3.3205322944079898e-06,
      "loss": 0.4675,
      "step": 4823
    },
    {
      "epoch": 5.90452876376989,
      "grad_norm": 1.3213013365423099,
      "learning_rate": 3.3199268900031028e-06,
      "loss": 0.5783,
      "step": 4824
    },
    {
      "epoch": 5.905752753977968,
      "grad_norm": 1.163159967187124,
      "learning_rate": 3.3193214317156535e-06,
      "loss": 0.6858,
      "step": 4825
    },
    {
      "epoch": 5.906976744186046,
      "grad_norm": 1.5966928058121546,
      "learning_rate": 3.3187159195854306e-06,
      "loss": 0.5125,
      "step": 4826
    },
    {
      "epoch": 5.908200734394125,
      "grad_norm": 1.8958917911199513,
      "learning_rate": 3.3181103536522253e-06,
      "loss": 0.4091,
      "step": 4827
    },
    {
      "epoch": 5.909424724602204,
      "grad_norm": 2.256169435761508,
      "learning_rate": 3.317504733955834e-06,
      "loss": 0.3948,
      "step": 4828
    },
    {
      "epoch": 5.910648714810281,
      "grad_norm": 1.0111287188167812,
      "learning_rate": 3.3168990605360553e-06,
      "loss": 1.2171,
      "step": 4829
    },
    {
      "epoch": 5.91187270501836,
      "grad_norm": 1.7106187783787372,
      "learning_rate": 3.3162933334326913e-06,
      "loss": 1.3457,
      "step": 4830
    },
    {
      "epoch": 5.9130966952264385,
      "grad_norm": 1.341541157659123,
      "learning_rate": 3.3156875526855483e-06,
      "loss": 1.295,
      "step": 4831
    },
    {
      "epoch": 5.914320685434516,
      "grad_norm": 2.2892187552024046,
      "learning_rate": 3.3150817183344376e-06,
      "loss": 0.5146,
      "step": 4832
    },
    {
      "epoch": 5.915544675642595,
      "grad_norm": 1.8745748454137505,
      "learning_rate": 3.3144758304191703e-06,
      "loss": 0.6372,
      "step": 4833
    },
    {
      "epoch": 5.9167686658506735,
      "grad_norm": 1.2921197420438952,
      "learning_rate": 3.313869888979564e-06,
      "loss": 1.4256,
      "step": 4834
    },
    {
      "epoch": 5.917992656058751,
      "grad_norm": 1.3197823605833092,
      "learning_rate": 3.313263894055439e-06,
      "loss": 1.2815,
      "step": 4835
    },
    {
      "epoch": 5.91921664626683,
      "grad_norm": 2.61599641710621,
      "learning_rate": 3.3126578456866192e-06,
      "loss": 0.5663,
      "step": 4836
    },
    {
      "epoch": 5.9204406364749085,
      "grad_norm": 1.8379995876734398,
      "learning_rate": 3.3120517439129317e-06,
      "loss": 0.7152,
      "step": 4837
    },
    {
      "epoch": 5.921664626682986,
      "grad_norm": 2.4503407026507724,
      "learning_rate": 3.311445588774207e-06,
      "loss": 0.5742,
      "step": 4838
    },
    {
      "epoch": 5.922888616891065,
      "grad_norm": 1.3087391985859644,
      "learning_rate": 3.310839380310279e-06,
      "loss": 0.7184,
      "step": 4839
    },
    {
      "epoch": 5.9241126070991434,
      "grad_norm": 1.9315663955820122,
      "learning_rate": 3.3102331185609867e-06,
      "loss": 0.7089,
      "step": 4840
    },
    {
      "epoch": 5.925336597307221,
      "grad_norm": 0.9673144217340133,
      "learning_rate": 3.30962680356617e-06,
      "loss": 0.5013,
      "step": 4841
    },
    {
      "epoch": 5.9265605875153,
      "grad_norm": 0.7652916638119623,
      "learning_rate": 3.309020435365675e-06,
      "loss": 0.4246,
      "step": 4842
    },
    {
      "epoch": 5.927784577723378,
      "grad_norm": 0.8775878634316255,
      "learning_rate": 3.3084140139993492e-06,
      "loss": 0.5276,
      "step": 4843
    },
    {
      "epoch": 5.929008567931456,
      "grad_norm": 1.2345580065459838,
      "learning_rate": 3.3078075395070443e-06,
      "loss": 0.7404,
      "step": 4844
    },
    {
      "epoch": 5.930232558139535,
      "grad_norm": 1.5242693477628029,
      "learning_rate": 3.3072010119286156e-06,
      "loss": 0.5551,
      "step": 4845
    },
    {
      "epoch": 5.931456548347613,
      "grad_norm": 1.7003365055637463,
      "learning_rate": 3.3065944313039227e-06,
      "loss": 0.5335,
      "step": 4846
    },
    {
      "epoch": 5.932680538555692,
      "grad_norm": 0.8641866564496203,
      "learning_rate": 3.3059877976728274e-06,
      "loss": 0.4113,
      "step": 4847
    },
    {
      "epoch": 5.93390452876377,
      "grad_norm": 1.7729617044578796,
      "learning_rate": 3.3053811110751936e-06,
      "loss": 0.5822,
      "step": 4848
    },
    {
      "epoch": 5.935128518971848,
      "grad_norm": 1.5295383567163163,
      "learning_rate": 3.304774371550894e-06,
      "loss": 1.0344,
      "step": 4849
    },
    {
      "epoch": 5.936352509179926,
      "grad_norm": 1.3516754163769535,
      "learning_rate": 3.3041675791397976e-06,
      "loss": 1.5927,
      "step": 4850
    },
    {
      "epoch": 5.937576499388005,
      "grad_norm": 2.459028388613073,
      "learning_rate": 3.303560733881783e-06,
      "loss": 1.0926,
      "step": 4851
    },
    {
      "epoch": 5.938800489596083,
      "grad_norm": 1.7193341220818141,
      "learning_rate": 3.3029538358167294e-06,
      "loss": 0.5485,
      "step": 4852
    },
    {
      "epoch": 5.940024479804162,
      "grad_norm": 0.8958037890089586,
      "learning_rate": 3.30234688498452e-06,
      "loss": 0.4882,
      "step": 4853
    },
    {
      "epoch": 5.94124847001224,
      "grad_norm": 1.309529161109099,
      "learning_rate": 3.30173988142504e-06,
      "loss": 0.5173,
      "step": 4854
    },
    {
      "epoch": 5.942472460220318,
      "grad_norm": 0.8803314992951057,
      "learning_rate": 3.3011328251781814e-06,
      "loss": 0.5988,
      "step": 4855
    },
    {
      "epoch": 5.943696450428397,
      "grad_norm": 2.810074645130276,
      "learning_rate": 3.300525716283836e-06,
      "loss": 0.5359,
      "step": 4856
    },
    {
      "epoch": 5.944920440636475,
      "grad_norm": 1.6428226581465646,
      "learning_rate": 3.2999185547819013e-06,
      "loss": 0.8818,
      "step": 4857
    },
    {
      "epoch": 5.946144430844553,
      "grad_norm": 0.9244308930211322,
      "learning_rate": 3.299311340712279e-06,
      "loss": 0.6382,
      "step": 4858
    },
    {
      "epoch": 5.947368421052632,
      "grad_norm": 1.187069018469516,
      "learning_rate": 3.298704074114871e-06,
      "loss": 0.6926,
      "step": 4859
    },
    {
      "epoch": 5.94859241126071,
      "grad_norm": 1.9975041876810142,
      "learning_rate": 3.298096755029585e-06,
      "loss": 0.6424,
      "step": 4860
    },
    {
      "epoch": 5.949816401468788,
      "grad_norm": 1.276929778727759,
      "learning_rate": 3.297489383496333e-06,
      "loss": 0.6188,
      "step": 4861
    },
    {
      "epoch": 5.951040391676867,
      "grad_norm": 1.7893598377074926,
      "learning_rate": 3.296881959555028e-06,
      "loss": 0.5154,
      "step": 4862
    },
    {
      "epoch": 5.9522643818849446,
      "grad_norm": 1.1707556100767622,
      "learning_rate": 3.2962744832455883e-06,
      "loss": 0.583,
      "step": 4863
    },
    {
      "epoch": 5.953488372093023,
      "grad_norm": 1.8480923413331225,
      "learning_rate": 3.295666954607935e-06,
      "loss": 0.554,
      "step": 4864
    },
    {
      "epoch": 5.954712362301102,
      "grad_norm": 1.4294671285819864,
      "learning_rate": 3.295059373681992e-06,
      "loss": 0.4888,
      "step": 4865
    },
    {
      "epoch": 5.95593635250918,
      "grad_norm": 2.285455436054477,
      "learning_rate": 3.294451740507688e-06,
      "loss": 0.3937,
      "step": 4866
    },
    {
      "epoch": 5.957160342717258,
      "grad_norm": 1.0299741198821855,
      "learning_rate": 3.2938440551249547e-06,
      "loss": 0.7802,
      "step": 4867
    },
    {
      "epoch": 5.958384332925337,
      "grad_norm": 1.5955510242054471,
      "learning_rate": 3.2932363175737246e-06,
      "loss": 1.0805,
      "step": 4868
    },
    {
      "epoch": 5.9596083231334145,
      "grad_norm": 1.8086153686610227,
      "learning_rate": 3.292628527893939e-06,
      "loss": 0.5216,
      "step": 4869
    },
    {
      "epoch": 5.960832313341493,
      "grad_norm": 1.8189025774809742,
      "learning_rate": 3.2920206861255378e-06,
      "loss": 0.5288,
      "step": 4870
    },
    {
      "epoch": 5.962056303549572,
      "grad_norm": 1.4376066787069752,
      "learning_rate": 3.291412792308467e-06,
      "loss": 0.6394,
      "step": 4871
    },
    {
      "epoch": 5.96328029375765,
      "grad_norm": 1.472845613801114,
      "learning_rate": 3.2908048464826736e-06,
      "loss": 0.49,
      "step": 4872
    },
    {
      "epoch": 5.964504283965728,
      "grad_norm": 0.9897971220462906,
      "learning_rate": 3.2901968486881125e-06,
      "loss": 0.5913,
      "step": 4873
    },
    {
      "epoch": 5.965728274173807,
      "grad_norm": 2.4363514916256808,
      "learning_rate": 3.289588798964736e-06,
      "loss": 0.5097,
      "step": 4874
    },
    {
      "epoch": 5.966952264381885,
      "grad_norm": 1.1798065072179091,
      "learning_rate": 3.288980697352504e-06,
      "loss": 0.6467,
      "step": 4875
    },
    {
      "epoch": 5.968176254589963,
      "grad_norm": 1.8175355720864539,
      "learning_rate": 3.2883725438913795e-06,
      "loss": 1.048,
      "step": 4876
    },
    {
      "epoch": 5.969400244798042,
      "grad_norm": 1.3088067201218474,
      "learning_rate": 3.287764338621327e-06,
      "loss": 2.3056,
      "step": 4877
    },
    {
      "epoch": 5.97062423500612,
      "grad_norm": 1.205016109987429,
      "learning_rate": 3.287156081582316e-06,
      "loss": 0.584,
      "step": 4878
    },
    {
      "epoch": 5.971848225214198,
      "grad_norm": 0.9489339373412814,
      "learning_rate": 3.286547772814319e-06,
      "loss": 0.5198,
      "step": 4879
    },
    {
      "epoch": 5.973072215422277,
      "grad_norm": 0.9223731177069728,
      "learning_rate": 3.285939412357311e-06,
      "loss": 0.5933,
      "step": 4880
    },
    {
      "epoch": 5.974296205630355,
      "grad_norm": 1.6269001679344737,
      "learning_rate": 3.285331000251272e-06,
      "loss": 0.3244,
      "step": 4881
    },
    {
      "epoch": 5.975520195838433,
      "grad_norm": 1.6156264376646563,
      "learning_rate": 3.284722536536185e-06,
      "loss": 0.4724,
      "step": 4882
    },
    {
      "epoch": 5.976744186046512,
      "grad_norm": 1.5564257299737407,
      "learning_rate": 3.284114021252035e-06,
      "loss": 0.5883,
      "step": 4883
    },
    {
      "epoch": 5.97796817625459,
      "grad_norm": 1.3290169614120093,
      "learning_rate": 3.283505454438812e-06,
      "loss": 0.5362,
      "step": 4884
    },
    {
      "epoch": 5.979192166462668,
      "grad_norm": 1.9690085907070949,
      "learning_rate": 3.2828968361365077e-06,
      "loss": 0.4594,
      "step": 4885
    },
    {
      "epoch": 5.9804161566707466,
      "grad_norm": 2.1519447675818033,
      "learning_rate": 3.2822881663851204e-06,
      "loss": 0.4781,
      "step": 4886
    },
    {
      "epoch": 5.981640146878825,
      "grad_norm": 2.4303948385563223,
      "learning_rate": 3.281679445224647e-06,
      "loss": 0.3638,
      "step": 4887
    },
    {
      "epoch": 5.982864137086903,
      "grad_norm": 1.2375575789094297,
      "learning_rate": 3.2810706726950923e-06,
      "loss": 0.5767,
      "step": 4888
    },
    {
      "epoch": 5.9840881272949815,
      "grad_norm": 1.300183353276447,
      "learning_rate": 3.2804618488364614e-06,
      "loss": 0.6508,
      "step": 4889
    },
    {
      "epoch": 5.98531211750306,
      "grad_norm": 1.1833098068489214,
      "learning_rate": 3.279852973688766e-06,
      "loss": 0.3755,
      "step": 4890
    },
    {
      "epoch": 5.986536107711139,
      "grad_norm": 1.1326039813422901,
      "learning_rate": 3.279244047292016e-06,
      "loss": 0.5112,
      "step": 4891
    },
    {
      "epoch": 5.9877600979192165,
      "grad_norm": 1.7981726625525905,
      "learning_rate": 3.27863506968623e-06,
      "loss": 1.2169,
      "step": 4892
    },
    {
      "epoch": 5.988984088127295,
      "grad_norm": 1.2042911967653704,
      "learning_rate": 3.278026040911427e-06,
      "loss": 0.7711,
      "step": 4893
    },
    {
      "epoch": 5.990208078335373,
      "grad_norm": 1.5237531444364614,
      "learning_rate": 3.277416961007631e-06,
      "loss": 0.4894,
      "step": 4894
    },
    {
      "epoch": 5.9914320685434515,
      "grad_norm": 1.6100376862468269,
      "learning_rate": 3.2768078300148675e-06,
      "loss": 0.9545,
      "step": 4895
    },
    {
      "epoch": 5.99265605875153,
      "grad_norm": 1.3857396163971971,
      "learning_rate": 3.2761986479731667e-06,
      "loss": 0.4589,
      "step": 4896
    },
    {
      "epoch": 5.993880048959609,
      "grad_norm": 1.1983890025161505,
      "learning_rate": 3.2755894149225616e-06,
      "loss": 0.6407,
      "step": 4897
    },
    {
      "epoch": 5.995104039167686,
      "grad_norm": 2.1113370217256846,
      "learning_rate": 3.274980130903088e-06,
      "loss": 0.6404,
      "step": 4898
    },
    {
      "epoch": 5.996328029375765,
      "grad_norm": 1.2567289334505396,
      "learning_rate": 3.2743707959547882e-06,
      "loss": 1.333,
      "step": 4899
    },
    {
      "epoch": 5.997552019583844,
      "grad_norm": 2.128144293218338,
      "learning_rate": 3.2737614101177027e-06,
      "loss": 0.459,
      "step": 4900
    },
    {
      "epoch": 5.998776009791921,
      "grad_norm": 2.1513753247980905,
      "learning_rate": 3.27315197343188e-06,
      "loss": 0.7201,
      "step": 4901
    },
    {
      "epoch": 6.0,
      "grad_norm": 1.7001479325391273,
      "learning_rate": 3.272542485937369e-06,
      "loss": 0.6353,
      "step": 4902
    },
    {
      "epoch": 6.001223990208079,
      "grad_norm": 1.4495315883386672,
      "learning_rate": 3.2719329476742236e-06,
      "loss": 0.5061,
      "step": 4903
    },
    {
      "epoch": 6.002447980416156,
      "grad_norm": 2.2943541500653906,
      "learning_rate": 3.2713233586824986e-06,
      "loss": 0.4978,
      "step": 4904
    },
    {
      "epoch": 6.003671970624235,
      "grad_norm": 1.2833535328103756,
      "learning_rate": 3.2707137190022575e-06,
      "loss": 0.4951,
      "step": 4905
    },
    {
      "epoch": 6.004895960832314,
      "grad_norm": 1.5240816747144867,
      "learning_rate": 3.2701040286735606e-06,
      "loss": 1.3267,
      "step": 4906
    },
    {
      "epoch": 6.006119951040391,
      "grad_norm": 1.6996284860277722,
      "learning_rate": 3.2694942877364743e-06,
      "loss": 0.4353,
      "step": 4907
    },
    {
      "epoch": 6.00734394124847,
      "grad_norm": 1.8333687270234134,
      "learning_rate": 3.2688844962310717e-06,
      "loss": 0.5468,
      "step": 4908
    },
    {
      "epoch": 6.0085679314565485,
      "grad_norm": 1.9538509524087488,
      "learning_rate": 3.268274654197422e-06,
      "loss": 0.8928,
      "step": 4909
    },
    {
      "epoch": 6.009791921664626,
      "grad_norm": 0.8095711072335581,
      "learning_rate": 3.267664761675604e-06,
      "loss": 0.5344,
      "step": 4910
    },
    {
      "epoch": 6.011015911872705,
      "grad_norm": 1.5806493481615544,
      "learning_rate": 3.267054818705698e-06,
      "loss": 0.7479,
      "step": 4911
    },
    {
      "epoch": 6.0122399020807835,
      "grad_norm": 2.074616607001054,
      "learning_rate": 3.2664448253277866e-06,
      "loss": 1.0104,
      "step": 4912
    },
    {
      "epoch": 6.013463892288861,
      "grad_norm": 1.474120598728423,
      "learning_rate": 3.265834781581955e-06,
      "loss": 0.552,
      "step": 4913
    },
    {
      "epoch": 6.01468788249694,
      "grad_norm": 0.9782131926510688,
      "learning_rate": 3.2652246875082955e-06,
      "loss": 0.5572,
      "step": 4914
    },
    {
      "epoch": 6.0159118727050185,
      "grad_norm": 1.1508011283768056,
      "learning_rate": 3.2646145431468988e-06,
      "loss": 0.5577,
      "step": 4915
    },
    {
      "epoch": 6.017135862913097,
      "grad_norm": 2.0361322162003086,
      "learning_rate": 3.2640043485378626e-06,
      "loss": 0.4324,
      "step": 4916
    },
    {
      "epoch": 6.018359853121175,
      "grad_norm": 0.8809634951167585,
      "learning_rate": 3.263394103721287e-06,
      "loss": 0.5229,
      "step": 4917
    },
    {
      "epoch": 6.0195838433292534,
      "grad_norm": 1.738804833826841,
      "learning_rate": 3.2627838087372745e-06,
      "loss": 1.225,
      "step": 4918
    },
    {
      "epoch": 6.020807833537332,
      "grad_norm": 1.3361086598061538,
      "learning_rate": 3.262173463625931e-06,
      "loss": 1.0309,
      "step": 4919
    },
    {
      "epoch": 6.02203182374541,
      "grad_norm": 2.1176828364042173,
      "learning_rate": 3.2615630684273668e-06,
      "loss": 0.6929,
      "step": 4920
    },
    {
      "epoch": 6.023255813953488,
      "grad_norm": 2.089235202496076,
      "learning_rate": 3.2609526231816945e-06,
      "loss": 0.4592,
      "step": 4921
    },
    {
      "epoch": 6.024479804161567,
      "grad_norm": 1.5026504405635486,
      "learning_rate": 3.2603421279290304e-06,
      "loss": 1.2088,
      "step": 4922
    },
    {
      "epoch": 6.025703794369645,
      "grad_norm": 2.0277189919358194,
      "learning_rate": 3.2597315827094946e-06,
      "loss": 0.6096,
      "step": 4923
    },
    {
      "epoch": 6.026927784577723,
      "grad_norm": 1.5170366602052028,
      "learning_rate": 3.2591209875632077e-06,
      "loss": 0.6619,
      "step": 4924
    },
    {
      "epoch": 6.028151774785802,
      "grad_norm": 0.7738009109717333,
      "learning_rate": 3.2585103425302984e-06,
      "loss": 0.421,
      "step": 4925
    },
    {
      "epoch": 6.02937576499388,
      "grad_norm": 2.028753672367803,
      "learning_rate": 3.257899647650895e-06,
      "loss": 1.3065,
      "step": 4926
    },
    {
      "epoch": 6.030599755201958,
      "grad_norm": 0.671258128027905,
      "learning_rate": 3.2572889029651296e-06,
      "loss": 0.3605,
      "step": 4927
    },
    {
      "epoch": 6.031823745410037,
      "grad_norm": 1.4828326507250587,
      "learning_rate": 3.256678108513139e-06,
      "loss": 0.5471,
      "step": 4928
    },
    {
      "epoch": 6.033047735618115,
      "grad_norm": 1.6933048373114348,
      "learning_rate": 3.256067264335061e-06,
      "loss": 0.8718,
      "step": 4929
    },
    {
      "epoch": 6.034271725826193,
      "grad_norm": 1.4982681246173863,
      "learning_rate": 3.2554563704710396e-06,
      "loss": 1.1907,
      "step": 4930
    },
    {
      "epoch": 6.035495716034272,
      "grad_norm": 1.7484534123447877,
      "learning_rate": 3.2548454269612193e-06,
      "loss": 0.5957,
      "step": 4931
    },
    {
      "epoch": 6.03671970624235,
      "grad_norm": 1.5418366902327179,
      "learning_rate": 3.2542344338457505e-06,
      "loss": 1.9031,
      "step": 4932
    },
    {
      "epoch": 6.037943696450428,
      "grad_norm": 2.7853430097602665,
      "learning_rate": 3.253623391164783e-06,
      "loss": 0.3078,
      "step": 4933
    },
    {
      "epoch": 6.039167686658507,
      "grad_norm": 1.6139949514992882,
      "learning_rate": 3.2530122989584745e-06,
      "loss": 0.405,
      "step": 4934
    },
    {
      "epoch": 6.0403916768665855,
      "grad_norm": 1.4379171977321241,
      "learning_rate": 3.2524011572669834e-06,
      "loss": 0.6756,
      "step": 4935
    },
    {
      "epoch": 6.041615667074663,
      "grad_norm": 1.0673365253916953,
      "learning_rate": 3.2517899661304704e-06,
      "loss": 0.7865,
      "step": 4936
    },
    {
      "epoch": 6.042839657282742,
      "grad_norm": 1.1080901689459726,
      "learning_rate": 3.2511787255891018e-06,
      "loss": 0.7534,
      "step": 4937
    },
    {
      "epoch": 6.0440636474908205,
      "grad_norm": 1.5076793890919469,
      "learning_rate": 3.2505674356830457e-06,
      "loss": 1.3819,
      "step": 4938
    },
    {
      "epoch": 6.045287637698898,
      "grad_norm": 2.616440640589218,
      "learning_rate": 3.2499560964524733e-06,
      "loss": 0.4782,
      "step": 4939
    },
    {
      "epoch": 6.046511627906977,
      "grad_norm": 1.8296444145564486,
      "learning_rate": 3.2493447079375605e-06,
      "loss": 0.6067,
      "step": 4940
    },
    {
      "epoch": 6.0477356181150554,
      "grad_norm": 2.0232005552726817,
      "learning_rate": 3.2487332701784856e-06,
      "loss": 0.5672,
      "step": 4941
    },
    {
      "epoch": 6.048959608323133,
      "grad_norm": 1.6280212747305132,
      "learning_rate": 3.24812178321543e-06,
      "loss": 0.6467,
      "step": 4942
    },
    {
      "epoch": 6.050183598531212,
      "grad_norm": 2.8111256772611104,
      "learning_rate": 3.2475102470885766e-06,
      "loss": 0.4674,
      "step": 4943
    },
    {
      "epoch": 6.05140758873929,
      "grad_norm": 0.8740828219204546,
      "learning_rate": 3.246898661838116e-06,
      "loss": 0.4091,
      "step": 4944
    },
    {
      "epoch": 6.052631578947368,
      "grad_norm": 1.4716553680920659,
      "learning_rate": 3.246287027504237e-06,
      "loss": 0.6299,
      "step": 4945
    },
    {
      "epoch": 6.053855569155447,
      "grad_norm": 2.6834229872485795,
      "learning_rate": 3.245675344127135e-06,
      "loss": 0.39,
      "step": 4946
    },
    {
      "epoch": 6.055079559363525,
      "grad_norm": 1.7413162225726788,
      "learning_rate": 3.2450636117470076e-06,
      "loss": 0.5264,
      "step": 4947
    },
    {
      "epoch": 6.056303549571603,
      "grad_norm": 1.259915324543423,
      "learning_rate": 3.244451830404055e-06,
      "loss": 1.4256,
      "step": 4948
    },
    {
      "epoch": 6.057527539779682,
      "grad_norm": 0.991520956827504,
      "learning_rate": 3.2438400001384823e-06,
      "loss": 0.9038,
      "step": 4949
    },
    {
      "epoch": 6.05875152998776,
      "grad_norm": 1.5033564267154547,
      "learning_rate": 3.2432281209904954e-06,
      "loss": 1.1085,
      "step": 4950
    },
    {
      "epoch": 6.059975520195838,
      "grad_norm": 1.2076764629437715,
      "learning_rate": 3.2426161930003056e-06,
      "loss": 0.3408,
      "step": 4951
    },
    {
      "epoch": 6.061199510403917,
      "grad_norm": 2.058934797260067,
      "learning_rate": 3.2420042162081267e-06,
      "loss": 0.6988,
      "step": 4952
    },
    {
      "epoch": 6.062423500611995,
      "grad_norm": 1.6874285590223277,
      "learning_rate": 3.2413921906541752e-06,
      "loss": 0.7552,
      "step": 4953
    },
    {
      "epoch": 6.063647490820073,
      "grad_norm": 1.3603242508113675,
      "learning_rate": 3.24078011637867e-06,
      "loss": 0.621,
      "step": 4954
    },
    {
      "epoch": 6.064871481028152,
      "grad_norm": 2.19946195533844,
      "learning_rate": 3.2401679934218374e-06,
      "loss": 0.5296,
      "step": 4955
    },
    {
      "epoch": 6.06609547123623,
      "grad_norm": 1.4069039600713957,
      "learning_rate": 3.239555821823901e-06,
      "loss": 0.5391,
      "step": 4956
    },
    {
      "epoch": 6.067319461444308,
      "grad_norm": 1.709497319656516,
      "learning_rate": 3.238943601625091e-06,
      "loss": 0.982,
      "step": 4957
    },
    {
      "epoch": 6.068543451652387,
      "grad_norm": 1.351835047675315,
      "learning_rate": 3.238331332865642e-06,
      "loss": 0.4772,
      "step": 4958
    },
    {
      "epoch": 6.069767441860465,
      "grad_norm": 1.6287000612639446,
      "learning_rate": 3.237719015585787e-06,
      "loss": 0.5716,
      "step": 4959
    },
    {
      "epoch": 6.070991432068544,
      "grad_norm": 0.7934925720420954,
      "learning_rate": 3.2371066498257675e-06,
      "loss": 0.475,
      "step": 4960
    },
    {
      "epoch": 6.072215422276622,
      "grad_norm": 1.1572582387776182,
      "learning_rate": 3.2364942356258254e-06,
      "loss": 0.4495,
      "step": 4961
    },
    {
      "epoch": 6.0734394124847,
      "grad_norm": 1.2964454007668307,
      "learning_rate": 3.2358817730262073e-06,
      "loss": 0.4238,
      "step": 4962
    },
    {
      "epoch": 6.074663402692779,
      "grad_norm": 1.0283760919459746,
      "learning_rate": 3.235269262067159e-06,
      "loss": 0.5701,
      "step": 4963
    },
    {
      "epoch": 6.0758873929008566,
      "grad_norm": 1.657870920258874,
      "learning_rate": 3.2346567027889358e-06,
      "loss": 1.4118,
      "step": 4964
    },
    {
      "epoch": 6.077111383108935,
      "grad_norm": 2.2402070919410324,
      "learning_rate": 3.234044095231791e-06,
      "loss": 0.5643,
      "step": 4965
    },
    {
      "epoch": 6.078335373317014,
      "grad_norm": 1.9195826722818026,
      "learning_rate": 3.2334314394359827e-06,
      "loss": 1.1061,
      "step": 4966
    },
    {
      "epoch": 6.0795593635250915,
      "grad_norm": 2.032270843958977,
      "learning_rate": 3.232818735441774e-06,
      "loss": 0.4939,
      "step": 4967
    },
    {
      "epoch": 6.08078335373317,
      "grad_norm": 1.202929918923349,
      "learning_rate": 3.232205983289428e-06,
      "loss": 0.6086,
      "step": 4968
    },
    {
      "epoch": 6.082007343941249,
      "grad_norm": 1.1410996729210305,
      "learning_rate": 3.2315931830192133e-06,
      "loss": 0.352,
      "step": 4969
    },
    {
      "epoch": 6.0832313341493265,
      "grad_norm": 1.2886792928236062,
      "learning_rate": 3.2309803346714003e-06,
      "loss": 0.3108,
      "step": 4970
    },
    {
      "epoch": 6.084455324357405,
      "grad_norm": 1.36977712148266,
      "learning_rate": 3.230367438286264e-06,
      "loss": 0.5209,
      "step": 4971
    },
    {
      "epoch": 6.085679314565484,
      "grad_norm": 1.8699022335926228,
      "learning_rate": 3.2297544939040797e-06,
      "loss": 0.575,
      "step": 4972
    },
    {
      "epoch": 6.0869033047735615,
      "grad_norm": 1.6695638185273625,
      "learning_rate": 3.2291415015651304e-06,
      "loss": 0.5447,
      "step": 4973
    },
    {
      "epoch": 6.08812729498164,
      "grad_norm": 0.8839407383922806,
      "learning_rate": 3.228528461309698e-06,
      "loss": 0.5216,
      "step": 4974
    },
    {
      "epoch": 6.089351285189719,
      "grad_norm": 2.220876922258344,
      "learning_rate": 3.22791537317807e-06,
      "loss": 0.3956,
      "step": 4975
    },
    {
      "epoch": 6.090575275397796,
      "grad_norm": 1.8441206543444824,
      "learning_rate": 3.2273022372105364e-06,
      "loss": 0.8481,
      "step": 4976
    },
    {
      "epoch": 6.091799265605875,
      "grad_norm": 2.363831458493239,
      "learning_rate": 3.226689053447389e-06,
      "loss": 0.4513,
      "step": 4977
    },
    {
      "epoch": 6.093023255813954,
      "grad_norm": 2.1322794472475297,
      "learning_rate": 3.2260758219289255e-06,
      "loss": 0.303,
      "step": 4978
    },
    {
      "epoch": 6.094247246022032,
      "grad_norm": 1.3042766085138409,
      "learning_rate": 3.225462542695445e-06,
      "loss": 0.7293,
      "step": 4979
    },
    {
      "epoch": 6.09547123623011,
      "grad_norm": 2.308097449716267,
      "learning_rate": 3.2248492157872484e-06,
      "loss": 0.3948,
      "step": 4980
    },
    {
      "epoch": 6.096695226438189,
      "grad_norm": 1.5837988458937426,
      "learning_rate": 3.224235841244643e-06,
      "loss": 0.3831,
      "step": 4981
    },
    {
      "epoch": 6.097919216646267,
      "grad_norm": 1.7020958565574291,
      "learning_rate": 3.2236224191079375e-06,
      "loss": 0.3368,
      "step": 4982
    },
    {
      "epoch": 6.099143206854345,
      "grad_norm": 1.5518017651156921,
      "learning_rate": 3.2230089494174416e-06,
      "loss": 0.9738,
      "step": 4983
    },
    {
      "epoch": 6.100367197062424,
      "grad_norm": 0.7805946825944557,
      "learning_rate": 3.2223954322134724e-06,
      "loss": 0.4486,
      "step": 4984
    },
    {
      "epoch": 6.101591187270502,
      "grad_norm": 1.5651929669437348,
      "learning_rate": 3.221781867536348e-06,
      "loss": 0.5452,
      "step": 4985
    },
    {
      "epoch": 6.10281517747858,
      "grad_norm": 0.9582848286462313,
      "learning_rate": 3.2211682554263882e-06,
      "loss": 0.5803,
      "step": 4986
    },
    {
      "epoch": 6.1040391676866586,
      "grad_norm": 1.3592667149010977,
      "learning_rate": 3.220554595923919e-06,
      "loss": 0.5272,
      "step": 4987
    },
    {
      "epoch": 6.105263157894737,
      "grad_norm": 3.009922420339776,
      "learning_rate": 3.2199408890692656e-06,
      "loss": 0.7927,
      "step": 4988
    },
    {
      "epoch": 6.106487148102815,
      "grad_norm": 2.9410663043049863,
      "learning_rate": 3.21932713490276e-06,
      "loss": 0.5146,
      "step": 4989
    },
    {
      "epoch": 6.1077111383108935,
      "grad_norm": 1.2589892844182529,
      "learning_rate": 3.2187133334647362e-06,
      "loss": 0.4031,
      "step": 4990
    },
    {
      "epoch": 6.108935128518972,
      "grad_norm": 1.3312796355034278,
      "learning_rate": 3.2180994847955304e-06,
      "loss": 0.5009,
      "step": 4991
    },
    {
      "epoch": 6.11015911872705,
      "grad_norm": 1.51791840909122,
      "learning_rate": 3.2174855889354826e-06,
      "loss": 0.5471,
      "step": 4992
    },
    {
      "epoch": 6.1113831089351285,
      "grad_norm": 1.2700164288612994,
      "learning_rate": 3.216871645924936e-06,
      "loss": 0.4816,
      "step": 4993
    },
    {
      "epoch": 6.112607099143207,
      "grad_norm": 1.0909084457668567,
      "learning_rate": 3.2162576558042364e-06,
      "loss": 0.7275,
      "step": 4994
    },
    {
      "epoch": 6.113831089351285,
      "grad_norm": 2.1193645569908286,
      "learning_rate": 3.2156436186137323e-06,
      "loss": 0.7039,
      "step": 4995
    },
    {
      "epoch": 6.1150550795593634,
      "grad_norm": 1.5121310364746152,
      "learning_rate": 3.215029534393777e-06,
      "loss": 0.6065,
      "step": 4996
    },
    {
      "epoch": 6.116279069767442,
      "grad_norm": 1.3540936766894813,
      "learning_rate": 3.2144154031847253e-06,
      "loss": 1.25,
      "step": 4997
    },
    {
      "epoch": 6.11750305997552,
      "grad_norm": 1.0978512810268124,
      "learning_rate": 3.213801225026935e-06,
      "loss": 0.9871,
      "step": 4998
    },
    {
      "epoch": 6.118727050183598,
      "grad_norm": 1.0908890047484585,
      "learning_rate": 3.2131869999607695e-06,
      "loss": 0.7451,
      "step": 4999
    },
    {
      "epoch": 6.119951040391677,
      "grad_norm": 1.3002445347945324,
      "learning_rate": 3.212572728026592e-06,
      "loss": 2.2988,
      "step": 5000
    },
    {
      "epoch": 6.121175030599755,
      "grad_norm": 1.8307320905392317,
      "learning_rate": 3.21195840926477e-06,
      "loss": 0.8446,
      "step": 5001
    },
    {
      "epoch": 6.122399020807833,
      "grad_norm": 1.2915360659758834,
      "learning_rate": 3.2113440437156745e-06,
      "loss": 0.6114,
      "step": 5002
    },
    {
      "epoch": 6.123623011015912,
      "grad_norm": 1.3000044621245683,
      "learning_rate": 3.21072963141968e-06,
      "loss": 1.0862,
      "step": 5003
    },
    {
      "epoch": 6.124847001223991,
      "grad_norm": 0.9578173273971511,
      "learning_rate": 3.2101151724171624e-06,
      "loss": 0.4581,
      "step": 5004
    },
    {
      "epoch": 6.126070991432068,
      "grad_norm": 1.3746566297575276,
      "learning_rate": 3.2095006667485036e-06,
      "loss": 0.5857,
      "step": 5005
    },
    {
      "epoch": 6.127294981640147,
      "grad_norm": 0.9741346731209892,
      "learning_rate": 3.208886114454084e-06,
      "loss": 0.5193,
      "step": 5006
    },
    {
      "epoch": 6.128518971848226,
      "grad_norm": 1.9796139784824802,
      "learning_rate": 3.2082715155742913e-06,
      "loss": 0.5166,
      "step": 5007
    },
    {
      "epoch": 6.129742962056303,
      "grad_norm": 0.9817650019912368,
      "learning_rate": 3.2076568701495143e-06,
      "loss": 0.5933,
      "step": 5008
    },
    {
      "epoch": 6.130966952264382,
      "grad_norm": 1.9018029548735351,
      "learning_rate": 3.207042178220145e-06,
      "loss": 0.4271,
      "step": 5009
    },
    {
      "epoch": 6.1321909424724605,
      "grad_norm": 1.8627016432403698,
      "learning_rate": 3.2064274398265792e-06,
      "loss": 1.3611,
      "step": 5010
    },
    {
      "epoch": 6.133414932680538,
      "grad_norm": 1.4922792081956022,
      "learning_rate": 3.205812655009215e-06,
      "loss": 1.6171,
      "step": 5011
    },
    {
      "epoch": 6.134638922888617,
      "grad_norm": 1.4677160719075089,
      "learning_rate": 3.2051978238084536e-06,
      "loss": 0.5994,
      "step": 5012
    },
    {
      "epoch": 6.1358629130966955,
      "grad_norm": 1.5257848740883315,
      "learning_rate": 3.2045829462646986e-06,
      "loss": 0.9325,
      "step": 5013
    },
    {
      "epoch": 6.137086903304773,
      "grad_norm": 1.5670291132870946,
      "learning_rate": 3.20396802241836e-06,
      "loss": 0.489,
      "step": 5014
    },
    {
      "epoch": 6.138310893512852,
      "grad_norm": 1.418455227717482,
      "learning_rate": 3.203353052309846e-06,
      "loss": 0.9258,
      "step": 5015
    },
    {
      "epoch": 6.1395348837209305,
      "grad_norm": 1.6156598782743565,
      "learning_rate": 3.202738035979571e-06,
      "loss": 0.4402,
      "step": 5016
    },
    {
      "epoch": 6.140758873929008,
      "grad_norm": 1.8866851472563322,
      "learning_rate": 3.2021229734679514e-06,
      "loss": 0.4104,
      "step": 5017
    },
    {
      "epoch": 6.141982864137087,
      "grad_norm": 1.5320221819207216,
      "learning_rate": 3.2015078648154076e-06,
      "loss": 0.4771,
      "step": 5018
    },
    {
      "epoch": 6.1432068543451654,
      "grad_norm": 1.2987351257435138,
      "learning_rate": 3.200892710062361e-06,
      "loss": 0.5913,
      "step": 5019
    },
    {
      "epoch": 6.144430844553243,
      "grad_norm": 2.6831334638063216,
      "learning_rate": 3.2002775092492383e-06,
      "loss": 0.5507,
      "step": 5020
    },
    {
      "epoch": 6.145654834761322,
      "grad_norm": 1.4359405972060042,
      "learning_rate": 3.1996622624164676e-06,
      "loss": 1.5403,
      "step": 5021
    },
    {
      "epoch": 6.1468788249694,
      "grad_norm": 0.7968725611581579,
      "learning_rate": 3.1990469696044806e-06,
      "loss": 0.3557,
      "step": 5022
    },
    {
      "epoch": 6.148102815177479,
      "grad_norm": 1.1878294636284004,
      "learning_rate": 3.1984316308537134e-06,
      "loss": 0.8295,
      "step": 5023
    },
    {
      "epoch": 6.149326805385557,
      "grad_norm": 2.322647729527677,
      "learning_rate": 3.197816246204601e-06,
      "loss": 1.0826,
      "step": 5024
    },
    {
      "epoch": 6.150550795593635,
      "grad_norm": 1.9805261959662368,
      "learning_rate": 3.1972008156975877e-06,
      "loss": 0.4503,
      "step": 5025
    },
    {
      "epoch": 6.151774785801714,
      "grad_norm": 0.9076356164610674,
      "learning_rate": 3.196585339373115e-06,
      "loss": 0.6925,
      "step": 5026
    },
    {
      "epoch": 6.152998776009792,
      "grad_norm": 1.2117113035791238,
      "learning_rate": 3.1959698172716296e-06,
      "loss": 0.5175,
      "step": 5027
    },
    {
      "epoch": 6.15422276621787,
      "grad_norm": 1.574836986285899,
      "learning_rate": 3.195354249433582e-06,
      "loss": 0.5781,
      "step": 5028
    },
    {
      "epoch": 6.155446756425949,
      "grad_norm": 0.9348443121813554,
      "learning_rate": 3.1947386358994263e-06,
      "loss": 0.6169,
      "step": 5029
    },
    {
      "epoch": 6.156670746634027,
      "grad_norm": 1.398795643667143,
      "learning_rate": 3.194122976709616e-06,
      "loss": 1.063,
      "step": 5030
    },
    {
      "epoch": 6.157894736842105,
      "grad_norm": 1.6884367454473714,
      "learning_rate": 3.193507271904612e-06,
      "loss": 0.9486,
      "step": 5031
    },
    {
      "epoch": 6.159118727050184,
      "grad_norm": 1.6314551664765318,
      "learning_rate": 3.1928915215248746e-06,
      "loss": 0.7668,
      "step": 5032
    },
    {
      "epoch": 6.160342717258262,
      "grad_norm": 1.3755837868993164,
      "learning_rate": 3.1922757256108694e-06,
      "loss": 0.8366,
      "step": 5033
    },
    {
      "epoch": 6.16156670746634,
      "grad_norm": 1.4307947946190678,
      "learning_rate": 3.191659884203064e-06,
      "loss": 0.5759,
      "step": 5034
    },
    {
      "epoch": 6.162790697674419,
      "grad_norm": 1.6730373100985183,
      "learning_rate": 3.1910439973419293e-06,
      "loss": 0.3415,
      "step": 5035
    },
    {
      "epoch": 6.164014687882497,
      "grad_norm": 2.0816273384449753,
      "learning_rate": 3.190428065067939e-06,
      "loss": 0.455,
      "step": 5036
    },
    {
      "epoch": 6.165238678090575,
      "grad_norm": 1.3564389628686835,
      "learning_rate": 3.18981208742157e-06,
      "loss": 0.5242,
      "step": 5037
    },
    {
      "epoch": 6.166462668298654,
      "grad_norm": 2.3430407308511656,
      "learning_rate": 3.189196064443302e-06,
      "loss": 0.991,
      "step": 5038
    },
    {
      "epoch": 6.167686658506732,
      "grad_norm": 1.6654374181301463,
      "learning_rate": 3.1885799961736173e-06,
      "loss": 0.3159,
      "step": 5039
    },
    {
      "epoch": 6.16891064871481,
      "grad_norm": 1.631860101499813,
      "learning_rate": 3.187963882653003e-06,
      "loss": 0.4632,
      "step": 5040
    },
    {
      "epoch": 6.170134638922889,
      "grad_norm": 2.2314010198724485,
      "learning_rate": 3.187347723921946e-06,
      "loss": 0.8511,
      "step": 5041
    },
    {
      "epoch": 6.1713586291309666,
      "grad_norm": 1.4974042468713582,
      "learning_rate": 3.1867315200209403e-06,
      "loss": 0.9005,
      "step": 5042
    },
    {
      "epoch": 6.172582619339045,
      "grad_norm": 1.1217089350480969,
      "learning_rate": 3.1861152709904784e-06,
      "loss": 0.6557,
      "step": 5043
    },
    {
      "epoch": 6.173806609547124,
      "grad_norm": 1.4005877669236046,
      "learning_rate": 3.1854989768710585e-06,
      "loss": 1.1179,
      "step": 5044
    },
    {
      "epoch": 6.1750305997552015,
      "grad_norm": 2.1898081613661797,
      "learning_rate": 3.1848826377031816e-06,
      "loss": 0.9773,
      "step": 5045
    },
    {
      "epoch": 6.17625458996328,
      "grad_norm": 1.2419570230519992,
      "learning_rate": 3.184266253527351e-06,
      "loss": 1.0031,
      "step": 5046
    },
    {
      "epoch": 6.177478580171359,
      "grad_norm": 1.2236150196652122,
      "learning_rate": 3.183649824384073e-06,
      "loss": 0.5666,
      "step": 5047
    },
    {
      "epoch": 6.178702570379437,
      "grad_norm": 1.5013724478247217,
      "learning_rate": 3.1830333503138565e-06,
      "loss": 0.6427,
      "step": 5048
    },
    {
      "epoch": 6.179926560587515,
      "grad_norm": 1.9394804666126197,
      "learning_rate": 3.182416831357216e-06,
      "loss": 0.5753,
      "step": 5049
    },
    {
      "epoch": 6.181150550795594,
      "grad_norm": 2.226285931797056,
      "learning_rate": 3.1818002675546643e-06,
      "loss": 0.468,
      "step": 5050
    },
    {
      "epoch": 6.182374541003672,
      "grad_norm": 0.6539060781202931,
      "learning_rate": 3.181183658946721e-06,
      "loss": 0.263,
      "step": 5051
    },
    {
      "epoch": 6.18359853121175,
      "grad_norm": 0.9438863341530361,
      "learning_rate": 3.1805670055739073e-06,
      "loss": 0.6101,
      "step": 5052
    },
    {
      "epoch": 6.184822521419829,
      "grad_norm": 1.430052836253628,
      "learning_rate": 3.1799503074767477e-06,
      "loss": 0.543,
      "step": 5053
    },
    {
      "epoch": 6.186046511627907,
      "grad_norm": 1.341516101931673,
      "learning_rate": 3.179333564695768e-06,
      "loss": 0.5105,
      "step": 5054
    },
    {
      "epoch": 6.187270501835985,
      "grad_norm": 0.8462778969209044,
      "learning_rate": 3.1787167772715e-06,
      "loss": 0.5958,
      "step": 5055
    },
    {
      "epoch": 6.188494492044064,
      "grad_norm": 1.318830512803707,
      "learning_rate": 3.178099945244475e-06,
      "loss": 1.5736,
      "step": 5056
    },
    {
      "epoch": 6.189718482252142,
      "grad_norm": 1.0599387337759185,
      "learning_rate": 3.17748306865523e-06,
      "loss": 0.789,
      "step": 5057
    },
    {
      "epoch": 6.19094247246022,
      "grad_norm": 1.0900408638308066,
      "learning_rate": 3.176866147544304e-06,
      "loss": 0.5971,
      "step": 5058
    },
    {
      "epoch": 6.192166462668299,
      "grad_norm": 1.6933337735705989,
      "learning_rate": 3.176249181952238e-06,
      "loss": 0.528,
      "step": 5059
    },
    {
      "epoch": 6.193390452876377,
      "grad_norm": 1.2528597755939546,
      "learning_rate": 3.175632171919577e-06,
      "loss": 1.1711,
      "step": 5060
    },
    {
      "epoch": 6.194614443084455,
      "grad_norm": 2.1718360054281662,
      "learning_rate": 3.1750151174868684e-06,
      "loss": 0.5186,
      "step": 5061
    },
    {
      "epoch": 6.195838433292534,
      "grad_norm": 2.1126007407629226,
      "learning_rate": 3.1743980186946634e-06,
      "loss": 0.4535,
      "step": 5062
    },
    {
      "epoch": 6.197062423500612,
      "grad_norm": 1.288316659099044,
      "learning_rate": 3.1737808755835144e-06,
      "loss": 0.9344,
      "step": 5063
    },
    {
      "epoch": 6.19828641370869,
      "grad_norm": 1.2175929083376564,
      "learning_rate": 3.1731636881939794e-06,
      "loss": 1.292,
      "step": 5064
    },
    {
      "epoch": 6.1995104039167686,
      "grad_norm": 1.778686347884478,
      "learning_rate": 3.172546456566616e-06,
      "loss": 1.2191,
      "step": 5065
    },
    {
      "epoch": 6.200734394124847,
      "grad_norm": 1.0621873240509532,
      "learning_rate": 3.1719291807419874e-06,
      "loss": 0.588,
      "step": 5066
    },
    {
      "epoch": 6.201958384332926,
      "grad_norm": 1.1461693082787718,
      "learning_rate": 3.1713118607606575e-06,
      "loss": 1.145,
      "step": 5067
    },
    {
      "epoch": 6.2031823745410035,
      "grad_norm": 1.1350699634370278,
      "learning_rate": 3.170694496663196e-06,
      "loss": 0.5032,
      "step": 5068
    },
    {
      "epoch": 6.204406364749082,
      "grad_norm": 1.461219543636155,
      "learning_rate": 3.1700770884901723e-06,
      "loss": 0.4826,
      "step": 5069
    },
    {
      "epoch": 6.205630354957161,
      "grad_norm": 1.1968034421367164,
      "learning_rate": 3.1694596362821615e-06,
      "loss": 1.0894,
      "step": 5070
    },
    {
      "epoch": 6.2068543451652385,
      "grad_norm": 1.144937641394743,
      "learning_rate": 3.1688421400797397e-06,
      "loss": 0.4813,
      "step": 5071
    },
    {
      "epoch": 6.208078335373317,
      "grad_norm": 1.5422818090704904,
      "learning_rate": 3.1682245999234853e-06,
      "loss": 1.105,
      "step": 5072
    },
    {
      "epoch": 6.209302325581396,
      "grad_norm": 1.1850659386994014,
      "learning_rate": 3.167607015853983e-06,
      "loss": 0.5625,
      "step": 5073
    },
    {
      "epoch": 6.2105263157894735,
      "grad_norm": 0.8533220512636793,
      "learning_rate": 3.1669893879118158e-06,
      "loss": 0.5228,
      "step": 5074
    },
    {
      "epoch": 6.211750305997552,
      "grad_norm": 1.862526686824757,
      "learning_rate": 3.1663717161375736e-06,
      "loss": 0.4032,
      "step": 5075
    },
    {
      "epoch": 6.212974296205631,
      "grad_norm": 1.1499090956174667,
      "learning_rate": 3.1657540005718466e-06,
      "loss": 0.4059,
      "step": 5076
    },
    {
      "epoch": 6.214198286413708,
      "grad_norm": 1.1862791728241355,
      "learning_rate": 3.1651362412552293e-06,
      "loss": 0.6401,
      "step": 5077
    },
    {
      "epoch": 6.215422276621787,
      "grad_norm": 2.1255196434374666,
      "learning_rate": 3.1645184382283183e-06,
      "loss": 0.8885,
      "step": 5078
    },
    {
      "epoch": 6.216646266829866,
      "grad_norm": 1.489064095141007,
      "learning_rate": 3.163900591531714e-06,
      "loss": 0.4685,
      "step": 5079
    },
    {
      "epoch": 6.217870257037943,
      "grad_norm": 2.0893730932803813,
      "learning_rate": 3.1632827012060176e-06,
      "loss": 0.5061,
      "step": 5080
    },
    {
      "epoch": 6.219094247246022,
      "grad_norm": 0.8719556386834397,
      "learning_rate": 3.162664767291836e-06,
      "loss": 0.6053,
      "step": 5081
    },
    {
      "epoch": 6.220318237454101,
      "grad_norm": 1.582103202157159,
      "learning_rate": 3.1620467898297774e-06,
      "loss": 1.0309,
      "step": 5082
    },
    {
      "epoch": 6.221542227662178,
      "grad_norm": 1.7475007688632729,
      "learning_rate": 3.1614287688604513e-06,
      "loss": 1.1485,
      "step": 5083
    },
    {
      "epoch": 6.222766217870257,
      "grad_norm": 1.3886314989040456,
      "learning_rate": 3.1608107044244734e-06,
      "loss": 0.7492,
      "step": 5084
    },
    {
      "epoch": 6.223990208078336,
      "grad_norm": 1.2835948556307393,
      "learning_rate": 3.1601925965624604e-06,
      "loss": 0.4931,
      "step": 5085
    },
    {
      "epoch": 6.225214198286413,
      "grad_norm": 1.2934795368254515,
      "learning_rate": 3.1595744453150316e-06,
      "loss": 0.7018,
      "step": 5086
    },
    {
      "epoch": 6.226438188494492,
      "grad_norm": 2.4593093680318145,
      "learning_rate": 3.1589562507228106e-06,
      "loss": 0.527,
      "step": 5087
    },
    {
      "epoch": 6.2276621787025706,
      "grad_norm": 1.6686127563505049,
      "learning_rate": 3.1583380128264214e-06,
      "loss": 0.5284,
      "step": 5088
    },
    {
      "epoch": 6.228886168910648,
      "grad_norm": 1.0907047036709694,
      "learning_rate": 3.1577197316664925e-06,
      "loss": 0.5283,
      "step": 5089
    },
    {
      "epoch": 6.230110159118727,
      "grad_norm": 1.4944316925675412,
      "learning_rate": 3.1571014072836564e-06,
      "loss": 1.6321,
      "step": 5090
    },
    {
      "epoch": 6.2313341493268055,
      "grad_norm": 0.9352051751066006,
      "learning_rate": 3.156483039718546e-06,
      "loss": 0.4301,
      "step": 5091
    },
    {
      "epoch": 6.232558139534884,
      "grad_norm": 0.7818813358280692,
      "learning_rate": 3.1558646290117984e-06,
      "loss": 0.3993,
      "step": 5092
    },
    {
      "epoch": 6.233782129742962,
      "grad_norm": 1.585376320549656,
      "learning_rate": 3.155246175204054e-06,
      "loss": 0.5465,
      "step": 5093
    },
    {
      "epoch": 6.2350061199510405,
      "grad_norm": 1.6912424544369857,
      "learning_rate": 3.1546276783359532e-06,
      "loss": 0.928,
      "step": 5094
    },
    {
      "epoch": 6.236230110159119,
      "grad_norm": 1.169492430758933,
      "learning_rate": 3.1540091384481436e-06,
      "loss": 0.5652,
      "step": 5095
    },
    {
      "epoch": 6.237454100367197,
      "grad_norm": 1.3709308225057861,
      "learning_rate": 3.153390555581272e-06,
      "loss": 0.7162,
      "step": 5096
    },
    {
      "epoch": 6.2386780905752754,
      "grad_norm": 0.8891838800052027,
      "learning_rate": 3.15277192977599e-06,
      "loss": 0.7231,
      "step": 5097
    },
    {
      "epoch": 6.239902080783354,
      "grad_norm": 1.1919197151629488,
      "learning_rate": 3.152153261072951e-06,
      "loss": 1.0594,
      "step": 5098
    },
    {
      "epoch": 6.241126070991432,
      "grad_norm": 1.0107906581917385,
      "learning_rate": 3.151534549512812e-06,
      "loss": 0.4474,
      "step": 5099
    },
    {
      "epoch": 6.24235006119951,
      "grad_norm": 1.632981039228178,
      "learning_rate": 3.1509157951362325e-06,
      "loss": 0.4348,
      "step": 5100
    },
    {
      "epoch": 6.243574051407589,
      "grad_norm": 1.420724425150078,
      "learning_rate": 3.1502969979838742e-06,
      "loss": 1.2868,
      "step": 5101
    },
    {
      "epoch": 6.244798041615667,
      "grad_norm": 2.457871499028654,
      "learning_rate": 3.1496781580964024e-06,
      "loss": 0.9429,
      "step": 5102
    },
    {
      "epoch": 6.246022031823745,
      "grad_norm": 1.3875144782501043,
      "learning_rate": 3.1490592755144855e-06,
      "loss": 1.5882,
      "step": 5103
    },
    {
      "epoch": 6.247246022031824,
      "grad_norm": 1.9438846992379268,
      "learning_rate": 3.148440350278793e-06,
      "loss": 0.9964,
      "step": 5104
    },
    {
      "epoch": 6.248470012239902,
      "grad_norm": 1.8481290822610643,
      "learning_rate": 3.14782138243e-06,
      "loss": 0.4048,
      "step": 5105
    },
    {
      "epoch": 6.24969400244798,
      "grad_norm": 1.1357731084986995,
      "learning_rate": 3.1472023720087814e-06,
      "loss": 0.5218,
      "step": 5106
    },
    {
      "epoch": 6.250917992656059,
      "grad_norm": 1.3402065692326268,
      "learning_rate": 3.1465833190558163e-06,
      "loss": 0.4235,
      "step": 5107
    },
    {
      "epoch": 6.252141982864137,
      "grad_norm": 1.3823666952061648,
      "learning_rate": 3.1459642236117877e-06,
      "loss": 0.5605,
      "step": 5108
    },
    {
      "epoch": 6.253365973072215,
      "grad_norm": 1.7530888874350352,
      "learning_rate": 3.1453450857173795e-06,
      "loss": 0.5969,
      "step": 5109
    },
    {
      "epoch": 6.254589963280294,
      "grad_norm": 1.6671176912499175,
      "learning_rate": 3.1447259054132796e-06,
      "loss": 0.7973,
      "step": 5110
    },
    {
      "epoch": 6.2558139534883725,
      "grad_norm": 1.193743113171425,
      "learning_rate": 3.1441066827401768e-06,
      "loss": 0.547,
      "step": 5111
    },
    {
      "epoch": 6.25703794369645,
      "grad_norm": 1.7504905587370918,
      "learning_rate": 3.1434874177387657e-06,
      "loss": 0.7895,
      "step": 5112
    },
    {
      "epoch": 6.258261933904529,
      "grad_norm": 2.063792043893467,
      "learning_rate": 3.1428681104497416e-06,
      "loss": 0.3698,
      "step": 5113
    },
    {
      "epoch": 6.2594859241126075,
      "grad_norm": 1.457827579076249,
      "learning_rate": 3.1422487609138037e-06,
      "loss": 0.5617,
      "step": 5114
    },
    {
      "epoch": 6.260709914320685,
      "grad_norm": 1.0228133107983184,
      "learning_rate": 3.1416293691716515e-06,
      "loss": 0.4843,
      "step": 5115
    },
    {
      "epoch": 6.261933904528764,
      "grad_norm": 1.6583628090496667,
      "learning_rate": 3.141009935263991e-06,
      "loss": 1.0262,
      "step": 5116
    },
    {
      "epoch": 6.2631578947368425,
      "grad_norm": 1.5620816888658107,
      "learning_rate": 3.1403904592315287e-06,
      "loss": 0.4965,
      "step": 5117
    },
    {
      "epoch": 6.26438188494492,
      "grad_norm": 1.7653846767402892,
      "learning_rate": 3.139770941114974e-06,
      "loss": 0.8557,
      "step": 5118
    },
    {
      "epoch": 6.265605875152999,
      "grad_norm": 1.7135264398400152,
      "learning_rate": 3.1391513809550387e-06,
      "loss": 0.9806,
      "step": 5119
    },
    {
      "epoch": 6.2668298653610774,
      "grad_norm": 1.8459445828426224,
      "learning_rate": 3.1385317787924404e-06,
      "loss": 0.3748,
      "step": 5120
    },
    {
      "epoch": 6.268053855569155,
      "grad_norm": 1.012213574932585,
      "learning_rate": 3.137912134667894e-06,
      "loss": 0.3959,
      "step": 5121
    },
    {
      "epoch": 6.269277845777234,
      "grad_norm": 1.3736905179138803,
      "learning_rate": 3.1372924486221213e-06,
      "loss": 0.3924,
      "step": 5122
    },
    {
      "epoch": 6.270501835985312,
      "grad_norm": 1.7890517215659805,
      "learning_rate": 3.1366727206958474e-06,
      "loss": 0.5638,
      "step": 5123
    },
    {
      "epoch": 6.27172582619339,
      "grad_norm": 1.4198794654295313,
      "learning_rate": 3.136052950929796e-06,
      "loss": 0.6805,
      "step": 5124
    },
    {
      "epoch": 6.272949816401469,
      "grad_norm": 2.5117265400490423,
      "learning_rate": 3.1354331393646974e-06,
      "loss": 0.4801,
      "step": 5125
    },
    {
      "epoch": 6.274173806609547,
      "grad_norm": 2.234094818662838,
      "learning_rate": 3.1348132860412837e-06,
      "loss": 0.972,
      "step": 5126
    },
    {
      "epoch": 6.275397796817625,
      "grad_norm": 1.5122675025102843,
      "learning_rate": 3.1341933910002887e-06,
      "loss": 0.4772,
      "step": 5127
    },
    {
      "epoch": 6.276621787025704,
      "grad_norm": 3.157732600231198,
      "learning_rate": 3.133573454282449e-06,
      "loss": 0.4188,
      "step": 5128
    },
    {
      "epoch": 6.277845777233782,
      "grad_norm": 2.0547166525053466,
      "learning_rate": 3.1329534759285067e-06,
      "loss": 0.5634,
      "step": 5129
    },
    {
      "epoch": 6.27906976744186,
      "grad_norm": 1.8119997109498913,
      "learning_rate": 3.132333455979202e-06,
      "loss": 0.9936,
      "step": 5130
    },
    {
      "epoch": 6.280293757649939,
      "grad_norm": 1.2505256539022798,
      "learning_rate": 3.131713394475282e-06,
      "loss": 0.493,
      "step": 5131
    },
    {
      "epoch": 6.281517747858017,
      "grad_norm": 1.7478300190282847,
      "learning_rate": 3.131093291457494e-06,
      "loss": 0.626,
      "step": 5132
    },
    {
      "epoch": 6.282741738066095,
      "grad_norm": 1.0821712618966253,
      "learning_rate": 3.1304731469665892e-06,
      "loss": 0.6188,
      "step": 5133
    },
    {
      "epoch": 6.283965728274174,
      "grad_norm": 1.4159576718366151,
      "learning_rate": 3.1298529610433213e-06,
      "loss": 0.4878,
      "step": 5134
    },
    {
      "epoch": 6.285189718482252,
      "grad_norm": 1.586541651319388,
      "learning_rate": 3.1292327337284463e-06,
      "loss": 0.7707,
      "step": 5135
    },
    {
      "epoch": 6.286413708690331,
      "grad_norm": 1.614459347095316,
      "learning_rate": 3.128612465062724e-06,
      "loss": 0.6243,
      "step": 5136
    },
    {
      "epoch": 6.287637698898409,
      "grad_norm": 1.666080143171431,
      "learning_rate": 3.127992155086914e-06,
      "loss": 0.3886,
      "step": 5137
    },
    {
      "epoch": 6.288861689106487,
      "grad_norm": 1.3455261655496396,
      "learning_rate": 3.1273718038417837e-06,
      "loss": 1.3145,
      "step": 5138
    },
    {
      "epoch": 6.290085679314566,
      "grad_norm": 1.1489818345039409,
      "learning_rate": 3.126751411368098e-06,
      "loss": 0.3836,
      "step": 5139
    },
    {
      "epoch": 6.291309669522644,
      "grad_norm": 1.3023961616097004,
      "learning_rate": 3.1261309777066284e-06,
      "loss": 0.6204,
      "step": 5140
    },
    {
      "epoch": 6.292533659730722,
      "grad_norm": 1.0884616396414535,
      "learning_rate": 3.125510502898147e-06,
      "loss": 0.836,
      "step": 5141
    },
    {
      "epoch": 6.293757649938801,
      "grad_norm": 1.8405492228918197,
      "learning_rate": 3.1248899869834288e-06,
      "loss": 0.4976,
      "step": 5142
    },
    {
      "epoch": 6.2949816401468786,
      "grad_norm": 1.1676190240721025,
      "learning_rate": 3.1242694300032515e-06,
      "loss": 0.3758,
      "step": 5143
    },
    {
      "epoch": 6.296205630354957,
      "grad_norm": 1.550301137259765,
      "learning_rate": 3.123648831998397e-06,
      "loss": 0.5086,
      "step": 5144
    },
    {
      "epoch": 6.297429620563036,
      "grad_norm": 1.166241352059848,
      "learning_rate": 3.1230281930096474e-06,
      "loss": 0.546,
      "step": 5145
    },
    {
      "epoch": 6.2986536107711135,
      "grad_norm": 1.944767201199805,
      "learning_rate": 3.12240751307779e-06,
      "loss": 0.5458,
      "step": 5146
    },
    {
      "epoch": 6.299877600979192,
      "grad_norm": 1.2003780009590979,
      "learning_rate": 3.1217867922436125e-06,
      "loss": 0.6627,
      "step": 5147
    },
    {
      "epoch": 6.301101591187271,
      "grad_norm": 2.1574519392974074,
      "learning_rate": 3.1211660305479063e-06,
      "loss": 1.0466,
      "step": 5148
    },
    {
      "epoch": 6.3023255813953485,
      "grad_norm": 1.9796506912149894,
      "learning_rate": 3.120545228031467e-06,
      "loss": 1.0631,
      "step": 5149
    },
    {
      "epoch": 6.303549571603427,
      "grad_norm": 0.4779134045408042,
      "learning_rate": 3.1199243847350903e-06,
      "loss": 0.1325,
      "step": 5150
    },
    {
      "epoch": 6.304773561811506,
      "grad_norm": 1.227697419748376,
      "learning_rate": 3.119303500699576e-06,
      "loss": 1.5764,
      "step": 5151
    },
    {
      "epoch": 6.3059975520195835,
      "grad_norm": 1.7204585111605968,
      "learning_rate": 3.1186825759657263e-06,
      "loss": 1.1299,
      "step": 5152
    },
    {
      "epoch": 6.307221542227662,
      "grad_norm": 1.2791661809028727,
      "learning_rate": 3.118061610574346e-06,
      "loss": 0.7256,
      "step": 5153
    },
    {
      "epoch": 6.308445532435741,
      "grad_norm": 1.0678609230562675,
      "learning_rate": 3.117440604566242e-06,
      "loss": 0.6709,
      "step": 5154
    },
    {
      "epoch": 6.309669522643819,
      "grad_norm": 0.8552403415173792,
      "learning_rate": 3.116819557982227e-06,
      "loss": 0.605,
      "step": 5155
    },
    {
      "epoch": 6.310893512851897,
      "grad_norm": 1.1351381541178502,
      "learning_rate": 3.1161984708631097e-06,
      "loss": 0.4575,
      "step": 5156
    },
    {
      "epoch": 6.312117503059976,
      "grad_norm": 1.682052578046046,
      "learning_rate": 3.11557734324971e-06,
      "loss": 0.6159,
      "step": 5157
    },
    {
      "epoch": 6.313341493268053,
      "grad_norm": 1.2294805449197728,
      "learning_rate": 3.1149561751828433e-06,
      "loss": 0.6243,
      "step": 5158
    },
    {
      "epoch": 6.314565483476132,
      "grad_norm": 0.9500196958489768,
      "learning_rate": 3.1143349667033317e-06,
      "loss": 0.7065,
      "step": 5159
    },
    {
      "epoch": 6.315789473684211,
      "grad_norm": 1.864912566819893,
      "learning_rate": 3.1137137178519983e-06,
      "loss": 0.4767,
      "step": 5160
    },
    {
      "epoch": 6.317013463892289,
      "grad_norm": 1.878214117321595,
      "learning_rate": 3.1130924286696697e-06,
      "loss": 0.6658,
      "step": 5161
    },
    {
      "epoch": 6.318237454100367,
      "grad_norm": 2.1993834806608215,
      "learning_rate": 3.1124710991971733e-06,
      "loss": 0.3772,
      "step": 5162
    },
    {
      "epoch": 6.319461444308446,
      "grad_norm": 2.0375204390524413,
      "learning_rate": 3.111849729475342e-06,
      "loss": 0.5117,
      "step": 5163
    },
    {
      "epoch": 6.320685434516524,
      "grad_norm": 1.1203286099738816,
      "learning_rate": 3.1112283195450105e-06,
      "loss": 0.6428,
      "step": 5164
    },
    {
      "epoch": 6.321909424724602,
      "grad_norm": 1.4655066593792638,
      "learning_rate": 3.1106068694470133e-06,
      "loss": 0.7106,
      "step": 5165
    },
    {
      "epoch": 6.3231334149326806,
      "grad_norm": 1.0970750291978095,
      "learning_rate": 3.109985379222192e-06,
      "loss": 1.0998,
      "step": 5166
    },
    {
      "epoch": 6.324357405140759,
      "grad_norm": 1.3555766168678587,
      "learning_rate": 3.1093638489113875e-06,
      "loss": 0.5362,
      "step": 5167
    },
    {
      "epoch": 6.325581395348837,
      "grad_norm": 0.8008787230241562,
      "learning_rate": 3.1087422785554442e-06,
      "loss": 0.4358,
      "step": 5168
    },
    {
      "epoch": 6.3268053855569155,
      "grad_norm": 1.2879381393596085,
      "learning_rate": 3.10812066819521e-06,
      "loss": 0.5318,
      "step": 5169
    },
    {
      "epoch": 6.328029375764994,
      "grad_norm": 1.369892064853444,
      "learning_rate": 3.1074990178715364e-06,
      "loss": 1.6584,
      "step": 5170
    },
    {
      "epoch": 6.329253365973072,
      "grad_norm": 1.395334417184578,
      "learning_rate": 3.106877327625273e-06,
      "loss": 1.3495,
      "step": 5171
    },
    {
      "epoch": 6.3304773561811505,
      "grad_norm": 1.6357641424075164,
      "learning_rate": 3.1062555974972765e-06,
      "loss": 0.5565,
      "step": 5172
    },
    {
      "epoch": 6.331701346389229,
      "grad_norm": 1.4183228173617424,
      "learning_rate": 3.105633827528405e-06,
      "loss": 0.4512,
      "step": 5173
    },
    {
      "epoch": 6.332925336597307,
      "grad_norm": 1.5410697356821677,
      "learning_rate": 3.1050120177595173e-06,
      "loss": 0.4814,
      "step": 5174
    },
    {
      "epoch": 6.3341493268053854,
      "grad_norm": 2.584266276358924,
      "learning_rate": 3.1043901682314782e-06,
      "loss": 0.7329,
      "step": 5175
    },
    {
      "epoch": 6.335373317013464,
      "grad_norm": 1.7167494675840742,
      "learning_rate": 3.1037682789851524e-06,
      "loss": 0.437,
      "step": 5176
    },
    {
      "epoch": 6.336597307221542,
      "grad_norm": 1.8180581243277534,
      "learning_rate": 3.103146350061409e-06,
      "loss": 0.505,
      "step": 5177
    },
    {
      "epoch": 6.33782129742962,
      "grad_norm": 1.205180711644409,
      "learning_rate": 3.102524381501118e-06,
      "loss": 0.8604,
      "step": 5178
    },
    {
      "epoch": 6.339045287637699,
      "grad_norm": 1.2432797128385729,
      "learning_rate": 3.1019023733451536e-06,
      "loss": 0.9095,
      "step": 5179
    },
    {
      "epoch": 6.340269277845778,
      "grad_norm": 2.2779805793320973,
      "learning_rate": 3.1012803256343908e-06,
      "loss": 0.5324,
      "step": 5180
    },
    {
      "epoch": 6.341493268053855,
      "grad_norm": 2.0269682157548257,
      "learning_rate": 3.1006582384097096e-06,
      "loss": 0.4462,
      "step": 5181
    },
    {
      "epoch": 6.342717258261934,
      "grad_norm": 1.9212024812437598,
      "learning_rate": 3.100036111711991e-06,
      "loss": 0.6291,
      "step": 5182
    },
    {
      "epoch": 6.343941248470013,
      "grad_norm": 1.1910032272156494,
      "learning_rate": 3.0994139455821183e-06,
      "loss": 0.8885,
      "step": 5183
    },
    {
      "epoch": 6.34516523867809,
      "grad_norm": 1.636343890719578,
      "learning_rate": 3.0987917400609775e-06,
      "loss": 0.615,
      "step": 5184
    },
    {
      "epoch": 6.346389228886169,
      "grad_norm": 1.5665048970695965,
      "learning_rate": 3.098169495189459e-06,
      "loss": 0.7179,
      "step": 5185
    },
    {
      "epoch": 6.347613219094248,
      "grad_norm": 1.7559911073947418,
      "learning_rate": 3.0975472110084536e-06,
      "loss": 1.1592,
      "step": 5186
    },
    {
      "epoch": 6.348837209302325,
      "grad_norm": 1.6628832130336493,
      "learning_rate": 3.0969248875588547e-06,
      "loss": 1.0165,
      "step": 5187
    },
    {
      "epoch": 6.350061199510404,
      "grad_norm": 1.8536656643155178,
      "learning_rate": 3.096302524881562e-06,
      "loss": 0.382,
      "step": 5188
    },
    {
      "epoch": 6.3512851897184825,
      "grad_norm": 1.4984531859967243,
      "learning_rate": 3.095680123017471e-06,
      "loss": 0.7268,
      "step": 5189
    },
    {
      "epoch": 6.35250917992656,
      "grad_norm": 1.1361964629646923,
      "learning_rate": 3.0950576820074867e-06,
      "loss": 0.536,
      "step": 5190
    },
    {
      "epoch": 6.353733170134639,
      "grad_norm": 2.0608283979117714,
      "learning_rate": 3.0944352018925123e-06,
      "loss": 1.2273,
      "step": 5191
    },
    {
      "epoch": 6.3549571603427175,
      "grad_norm": 1.1733295396474879,
      "learning_rate": 3.093812682713455e-06,
      "loss": 0.7127,
      "step": 5192
    },
    {
      "epoch": 6.356181150550795,
      "grad_norm": 2.0354595177284818,
      "learning_rate": 3.093190124511225e-06,
      "loss": 0.4023,
      "step": 5193
    },
    {
      "epoch": 6.357405140758874,
      "grad_norm": 0.862418834509687,
      "learning_rate": 3.0925675273267335e-06,
      "loss": 0.5304,
      "step": 5194
    },
    {
      "epoch": 6.3586291309669525,
      "grad_norm": 1.8980559999083995,
      "learning_rate": 3.0919448912008953e-06,
      "loss": 0.5128,
      "step": 5195
    },
    {
      "epoch": 6.35985312117503,
      "grad_norm": 2.67030720517238,
      "learning_rate": 3.091322216174629e-06,
      "loss": 0.448,
      "step": 5196
    },
    {
      "epoch": 6.361077111383109,
      "grad_norm": 0.9518121101068123,
      "learning_rate": 3.0906995022888548e-06,
      "loss": 0.6767,
      "step": 5197
    },
    {
      "epoch": 6.3623011015911874,
      "grad_norm": 1.7597818510859173,
      "learning_rate": 3.090076749584493e-06,
      "loss": 0.5717,
      "step": 5198
    },
    {
      "epoch": 6.363525091799266,
      "grad_norm": 1.384699014976833,
      "learning_rate": 3.0894539581024697e-06,
      "loss": 0.4784,
      "step": 5199
    },
    {
      "epoch": 6.364749082007344,
      "grad_norm": 0.9173856649075256,
      "learning_rate": 3.0888311278837128e-06,
      "loss": 0.5513,
      "step": 5200
    },
    {
      "epoch": 6.365973072215422,
      "grad_norm": 1.332379109725859,
      "learning_rate": 3.088208258969152e-06,
      "loss": 0.7202,
      "step": 5201
    },
    {
      "epoch": 6.3671970624235,
      "grad_norm": 1.3018639260493576,
      "learning_rate": 3.0875853513997206e-06,
      "loss": 0.5424,
      "step": 5202
    },
    {
      "epoch": 6.368421052631579,
      "grad_norm": 1.9043069127516545,
      "learning_rate": 3.0869624052163527e-06,
      "loss": 0.658,
      "step": 5203
    },
    {
      "epoch": 6.369645042839657,
      "grad_norm": 1.831940146499211,
      "learning_rate": 3.086339420459986e-06,
      "loss": 0.9778,
      "step": 5204
    },
    {
      "epoch": 6.370869033047736,
      "grad_norm": 1.9994104830299426,
      "learning_rate": 3.0857163971715627e-06,
      "loss": 0.6151,
      "step": 5205
    },
    {
      "epoch": 6.372093023255814,
      "grad_norm": 1.5002599733821465,
      "learning_rate": 3.0850933353920233e-06,
      "loss": 0.5562,
      "step": 5206
    },
    {
      "epoch": 6.373317013463892,
      "grad_norm": 1.4075090317355905,
      "learning_rate": 3.084470235162314e-06,
      "loss": 0.6611,
      "step": 5207
    },
    {
      "epoch": 6.374541003671971,
      "grad_norm": 1.226992593037473,
      "learning_rate": 3.0838470965233823e-06,
      "loss": 0.3806,
      "step": 5208
    },
    {
      "epoch": 6.375764993880049,
      "grad_norm": 0.939834204309093,
      "learning_rate": 3.08322391951618e-06,
      "loss": 0.5513,
      "step": 5209
    },
    {
      "epoch": 6.376988984088127,
      "grad_norm": 2.312702693961054,
      "learning_rate": 3.082600704181658e-06,
      "loss": 0.2388,
      "step": 5210
    },
    {
      "epoch": 6.378212974296206,
      "grad_norm": 1.3426462482089907,
      "learning_rate": 3.081977450560773e-06,
      "loss": 0.474,
      "step": 5211
    },
    {
      "epoch": 6.379436964504284,
      "grad_norm": 1.5072813796103943,
      "learning_rate": 3.081354158694483e-06,
      "loss": 1.4408,
      "step": 5212
    },
    {
      "epoch": 6.380660954712362,
      "grad_norm": 2.551152977200912,
      "learning_rate": 3.080730828623747e-06,
      "loss": 1.0565,
      "step": 5213
    },
    {
      "epoch": 6.381884944920441,
      "grad_norm": 1.3566152761019628,
      "learning_rate": 3.0801074603895296e-06,
      "loss": 0.5182,
      "step": 5214
    },
    {
      "epoch": 6.383108935128519,
      "grad_norm": 1.7062631077392574,
      "learning_rate": 3.079484054032795e-06,
      "loss": 0.6245,
      "step": 5215
    },
    {
      "epoch": 6.384332925336597,
      "grad_norm": 2.947628989028716,
      "learning_rate": 3.0788606095945117e-06,
      "loss": 0.519,
      "step": 5216
    },
    {
      "epoch": 6.385556915544676,
      "grad_norm": 1.1029716910848006,
      "learning_rate": 3.0782371271156506e-06,
      "loss": 0.9058,
      "step": 5217
    },
    {
      "epoch": 6.386780905752754,
      "grad_norm": 1.0656800446308654,
      "learning_rate": 3.0776136066371843e-06,
      "loss": 0.403,
      "step": 5218
    },
    {
      "epoch": 6.388004895960832,
      "grad_norm": 1.9789932058987998,
      "learning_rate": 3.076990048200088e-06,
      "loss": 0.4741,
      "step": 5219
    },
    {
      "epoch": 6.389228886168911,
      "grad_norm": 1.4337006865764237,
      "learning_rate": 3.076366451845341e-06,
      "loss": 0.9989,
      "step": 5220
    },
    {
      "epoch": 6.3904528763769886,
      "grad_norm": 1.259753761496176,
      "learning_rate": 3.0757428176139204e-06,
      "loss": 0.4456,
      "step": 5221
    },
    {
      "epoch": 6.391676866585067,
      "grad_norm": 2.2426976192693386,
      "learning_rate": 3.0751191455468126e-06,
      "loss": 0.843,
      "step": 5222
    },
    {
      "epoch": 6.392900856793146,
      "grad_norm": 2.048600217866285,
      "learning_rate": 3.0744954356850028e-06,
      "loss": 0.5304,
      "step": 5223
    },
    {
      "epoch": 6.394124847001224,
      "grad_norm": 1.2665923981895382,
      "learning_rate": 3.0738716880694763e-06,
      "loss": 0.3746,
      "step": 5224
    },
    {
      "epoch": 6.395348837209302,
      "grad_norm": 1.208293564212225,
      "learning_rate": 3.0732479027412254e-06,
      "loss": 0.7353,
      "step": 5225
    },
    {
      "epoch": 6.396572827417381,
      "grad_norm": 2.333393530080522,
      "learning_rate": 3.0726240797412426e-06,
      "loss": 0.3866,
      "step": 5226
    },
    {
      "epoch": 6.397796817625459,
      "grad_norm": 1.2842540126023991,
      "learning_rate": 3.0720002191105234e-06,
      "loss": 0.5805,
      "step": 5227
    },
    {
      "epoch": 6.399020807833537,
      "grad_norm": 2.3208641779037262,
      "learning_rate": 3.0713763208900656e-06,
      "loss": 0.4582,
      "step": 5228
    },
    {
      "epoch": 6.400244798041616,
      "grad_norm": 3.06443380034515,
      "learning_rate": 3.0707523851208698e-06,
      "loss": 0.398,
      "step": 5229
    },
    {
      "epoch": 6.401468788249694,
      "grad_norm": 1.783610142204136,
      "learning_rate": 3.0701284118439373e-06,
      "loss": 0.5076,
      "step": 5230
    },
    {
      "epoch": 6.402692778457772,
      "grad_norm": 1.346876347901455,
      "learning_rate": 3.069504401100275e-06,
      "loss": 0.3486,
      "step": 5231
    },
    {
      "epoch": 6.403916768665851,
      "grad_norm": 1.2568600695327496,
      "learning_rate": 3.06888035293089e-06,
      "loss": 0.636,
      "step": 5232
    },
    {
      "epoch": 6.405140758873929,
      "grad_norm": 1.2436458277631997,
      "learning_rate": 3.068256267376792e-06,
      "loss": 1.3345,
      "step": 5233
    },
    {
      "epoch": 6.406364749082007,
      "grad_norm": 0.9319070992522746,
      "learning_rate": 3.067632144478994e-06,
      "loss": 0.6092,
      "step": 5234
    },
    {
      "epoch": 6.407588739290086,
      "grad_norm": 1.710348216474738,
      "learning_rate": 3.067007984278511e-06,
      "loss": 0.7806,
      "step": 5235
    },
    {
      "epoch": 6.408812729498164,
      "grad_norm": 1.6438210061862117,
      "learning_rate": 3.066383786816361e-06,
      "loss": 0.5479,
      "step": 5236
    },
    {
      "epoch": 6.410036719706242,
      "grad_norm": 1.2297807910385883,
      "learning_rate": 3.0657595521335624e-06,
      "loss": 1.3257,
      "step": 5237
    },
    {
      "epoch": 6.411260709914321,
      "grad_norm": 2.50755360589812,
      "learning_rate": 3.0651352802711397e-06,
      "loss": 0.496,
      "step": 5238
    },
    {
      "epoch": 6.412484700122399,
      "grad_norm": 1.5089664078621159,
      "learning_rate": 3.064510971270116e-06,
      "loss": 1.577,
      "step": 5239
    },
    {
      "epoch": 6.413708690330477,
      "grad_norm": 1.0664964342359886,
      "learning_rate": 3.063886625171519e-06,
      "loss": 0.9939,
      "step": 5240
    },
    {
      "epoch": 6.414932680538556,
      "grad_norm": 1.971317859155347,
      "learning_rate": 3.0632622420163795e-06,
      "loss": 0.472,
      "step": 5241
    },
    {
      "epoch": 6.416156670746634,
      "grad_norm": 2.419454285862608,
      "learning_rate": 3.0626378218457285e-06,
      "loss": 0.4332,
      "step": 5242
    },
    {
      "epoch": 6.417380660954713,
      "grad_norm": 0.7635018570798805,
      "learning_rate": 3.0620133647006007e-06,
      "loss": 0.3582,
      "step": 5243
    },
    {
      "epoch": 6.4186046511627906,
      "grad_norm": 1.0865410691780641,
      "learning_rate": 3.0613888706220336e-06,
      "loss": 0.7329,
      "step": 5244
    },
    {
      "epoch": 6.419828641370869,
      "grad_norm": 1.9055433255030234,
      "learning_rate": 3.0607643396510654e-06,
      "loss": 0.9289,
      "step": 5245
    },
    {
      "epoch": 6.421052631578947,
      "grad_norm": 1.8061697218612756,
      "learning_rate": 3.06013977182874e-06,
      "loss": 0.5704,
      "step": 5246
    },
    {
      "epoch": 6.4222766217870255,
      "grad_norm": 1.0718263452729597,
      "learning_rate": 3.059515167196101e-06,
      "loss": 0.7646,
      "step": 5247
    },
    {
      "epoch": 6.423500611995104,
      "grad_norm": 1.6770285152662974,
      "learning_rate": 3.0588905257941936e-06,
      "loss": 0.4316,
      "step": 5248
    },
    {
      "epoch": 6.424724602203183,
      "grad_norm": 2.543981419110719,
      "learning_rate": 3.0582658476640686e-06,
      "loss": 0.4066,
      "step": 5249
    },
    {
      "epoch": 6.4259485924112605,
      "grad_norm": 0.9391602504884662,
      "learning_rate": 3.057641132846777e-06,
      "loss": 0.5594,
      "step": 5250
    },
    {
      "epoch": 6.427172582619339,
      "grad_norm": 1.76986851467784,
      "learning_rate": 3.0570163813833723e-06,
      "loss": 0.677,
      "step": 5251
    },
    {
      "epoch": 6.428396572827418,
      "grad_norm": 1.6932353011382728,
      "learning_rate": 3.056391593314912e-06,
      "loss": 0.6175,
      "step": 5252
    },
    {
      "epoch": 6.4296205630354955,
      "grad_norm": 1.8190924593358844,
      "learning_rate": 3.0557667686824538e-06,
      "loss": 0.4936,
      "step": 5253
    },
    {
      "epoch": 6.430844553243574,
      "grad_norm": 2.4533058456752337,
      "learning_rate": 3.0551419075270585e-06,
      "loss": 0.3797,
      "step": 5254
    },
    {
      "epoch": 6.432068543451653,
      "grad_norm": 1.3986935538015697,
      "learning_rate": 3.0545170098897915e-06,
      "loss": 1.1932,
      "step": 5255
    },
    {
      "epoch": 6.43329253365973,
      "grad_norm": 1.8662831289401944,
      "learning_rate": 3.053892075811716e-06,
      "loss": 0.4701,
      "step": 5256
    },
    {
      "epoch": 6.434516523867809,
      "grad_norm": 1.3789164490168464,
      "learning_rate": 3.0532671053339034e-06,
      "loss": 0.4603,
      "step": 5257
    },
    {
      "epoch": 6.435740514075888,
      "grad_norm": 1.58553808993022,
      "learning_rate": 3.0526420984974233e-06,
      "loss": 0.5039,
      "step": 5258
    },
    {
      "epoch": 6.436964504283965,
      "grad_norm": 1.5303100131541043,
      "learning_rate": 3.052017055343348e-06,
      "loss": 1.2122,
      "step": 5259
    },
    {
      "epoch": 6.438188494492044,
      "grad_norm": 0.9564738267153151,
      "learning_rate": 3.0513919759127537e-06,
      "loss": 0.547,
      "step": 5260
    },
    {
      "epoch": 6.439412484700123,
      "grad_norm": 2.329232796389088,
      "learning_rate": 3.0507668602467175e-06,
      "loss": 0.9445,
      "step": 5261
    },
    {
      "epoch": 6.4406364749082,
      "grad_norm": 2.1838685153973785,
      "learning_rate": 3.0501417083863215e-06,
      "loss": 0.541,
      "step": 5262
    },
    {
      "epoch": 6.441860465116279,
      "grad_norm": 1.6653873807847648,
      "learning_rate": 3.049516520372646e-06,
      "loss": 0.5504,
      "step": 5263
    },
    {
      "epoch": 6.443084455324358,
      "grad_norm": 0.9056731988853691,
      "learning_rate": 3.0488912962467787e-06,
      "loss": 0.6436,
      "step": 5264
    },
    {
      "epoch": 6.444308445532435,
      "grad_norm": 1.3268590707395715,
      "learning_rate": 3.0482660360498044e-06,
      "loss": 0.5142,
      "step": 5265
    },
    {
      "epoch": 6.445532435740514,
      "grad_norm": 2.0000801921448574,
      "learning_rate": 3.047640739822815e-06,
      "loss": 0.4717,
      "step": 5266
    },
    {
      "epoch": 6.4467564259485926,
      "grad_norm": 0.9103401410822293,
      "learning_rate": 3.0470154076069015e-06,
      "loss": 0.5444,
      "step": 5267
    },
    {
      "epoch": 6.447980416156671,
      "grad_norm": 1.6443557131937512,
      "learning_rate": 3.0463900394431585e-06,
      "loss": 0.5155,
      "step": 5268
    },
    {
      "epoch": 6.449204406364749,
      "grad_norm": 1.289886352517538,
      "learning_rate": 3.045764635372683e-06,
      "loss": 0.8819,
      "step": 5269
    },
    {
      "epoch": 6.4504283965728275,
      "grad_norm": 1.2803055054723436,
      "learning_rate": 3.0451391954365746e-06,
      "loss": 0.8245,
      "step": 5270
    },
    {
      "epoch": 6.451652386780906,
      "grad_norm": 1.372376821155428,
      "learning_rate": 3.0445137196759337e-06,
      "loss": 0.6877,
      "step": 5271
    },
    {
      "epoch": 6.452876376988984,
      "grad_norm": 1.9610813334768602,
      "learning_rate": 3.043888208131866e-06,
      "loss": 0.6022,
      "step": 5272
    },
    {
      "epoch": 6.4541003671970625,
      "grad_norm": 2.2279771460447964,
      "learning_rate": 3.043262660845478e-06,
      "loss": 0.4281,
      "step": 5273
    },
    {
      "epoch": 6.455324357405141,
      "grad_norm": 1.0496508251931276,
      "learning_rate": 3.042637077857876e-06,
      "loss": 1.053,
      "step": 5274
    },
    {
      "epoch": 6.456548347613219,
      "grad_norm": 0.9587009045198399,
      "learning_rate": 3.0420114592101718e-06,
      "loss": 0.6529,
      "step": 5275
    },
    {
      "epoch": 6.4577723378212974,
      "grad_norm": 1.1551059004846356,
      "learning_rate": 3.0413858049434803e-06,
      "loss": 0.5187,
      "step": 5276
    },
    {
      "epoch": 6.458996328029376,
      "grad_norm": 1.6038909151401448,
      "learning_rate": 3.0407601150989157e-06,
      "loss": 0.6111,
      "step": 5277
    },
    {
      "epoch": 6.460220318237454,
      "grad_norm": 1.1081988768511692,
      "learning_rate": 3.040134389717596e-06,
      "loss": 1.1052,
      "step": 5278
    },
    {
      "epoch": 6.461444308445532,
      "grad_norm": 1.8436172231097054,
      "learning_rate": 3.039508628840643e-06,
      "loss": 0.7039,
      "step": 5279
    },
    {
      "epoch": 6.462668298653611,
      "grad_norm": 1.2968062921750463,
      "learning_rate": 3.038882832509178e-06,
      "loss": 0.7103,
      "step": 5280
    },
    {
      "epoch": 6.463892288861689,
      "grad_norm": 1.2277560200585054,
      "learning_rate": 3.038257000764326e-06,
      "loss": 0.5758,
      "step": 5281
    },
    {
      "epoch": 6.465116279069767,
      "grad_norm": 0.9540705568723842,
      "learning_rate": 3.037631133647216e-06,
      "loss": 0.6182,
      "step": 5282
    },
    {
      "epoch": 6.466340269277846,
      "grad_norm": 1.5582370191602384,
      "learning_rate": 3.0370052311989756e-06,
      "loss": 0.9445,
      "step": 5283
    },
    {
      "epoch": 6.467564259485924,
      "grad_norm": 2.14300181581877,
      "learning_rate": 3.0363792934607372e-06,
      "loss": 0.4276,
      "step": 5284
    },
    {
      "epoch": 6.468788249694002,
      "grad_norm": 2.002123489043887,
      "learning_rate": 3.0357533204736366e-06,
      "loss": 0.497,
      "step": 5285
    },
    {
      "epoch": 6.470012239902081,
      "grad_norm": 1.1995599621362933,
      "learning_rate": 3.0351273122788093e-06,
      "loss": 0.8833,
      "step": 5286
    },
    {
      "epoch": 6.47123623011016,
      "grad_norm": 1.2764398544000135,
      "learning_rate": 3.0345012689173937e-06,
      "loss": 0.5878,
      "step": 5287
    },
    {
      "epoch": 6.472460220318237,
      "grad_norm": 1.6960604392041478,
      "learning_rate": 3.0338751904305323e-06,
      "loss": 0.3708,
      "step": 5288
    },
    {
      "epoch": 6.473684210526316,
      "grad_norm": 1.3898488478219966,
      "learning_rate": 3.0332490768593676e-06,
      "loss": 0.9983,
      "step": 5289
    },
    {
      "epoch": 6.474908200734394,
      "grad_norm": 1.5097354138570733,
      "learning_rate": 3.032622928245046e-06,
      "loss": 0.5883,
      "step": 5290
    },
    {
      "epoch": 6.476132190942472,
      "grad_norm": 1.448813662105671,
      "learning_rate": 3.0319967446287157e-06,
      "loss": 0.7709,
      "step": 5291
    },
    {
      "epoch": 6.477356181150551,
      "grad_norm": 1.8512342573519214,
      "learning_rate": 3.031370526051527e-06,
      "loss": 0.5379,
      "step": 5292
    },
    {
      "epoch": 6.4785801713586295,
      "grad_norm": 1.2724500420229987,
      "learning_rate": 3.0307442725546327e-06,
      "loss": 0.7681,
      "step": 5293
    },
    {
      "epoch": 6.479804161566707,
      "grad_norm": 1.2567625367962512,
      "learning_rate": 3.0301179841791877e-06,
      "loss": 0.9625,
      "step": 5294
    },
    {
      "epoch": 6.481028151774786,
      "grad_norm": 0.8084406221969098,
      "learning_rate": 3.0294916609663496e-06,
      "loss": 0.5048,
      "step": 5295
    },
    {
      "epoch": 6.4822521419828645,
      "grad_norm": 1.8368701515191537,
      "learning_rate": 3.0288653029572777e-06,
      "loss": 0.9878,
      "step": 5296
    },
    {
      "epoch": 6.483476132190942,
      "grad_norm": 2.137899905210895,
      "learning_rate": 3.0282389101931354e-06,
      "loss": 0.3641,
      "step": 5297
    },
    {
      "epoch": 6.484700122399021,
      "grad_norm": 1.3970213151467021,
      "learning_rate": 3.027612482715085e-06,
      "loss": 0.5713,
      "step": 5298
    },
    {
      "epoch": 6.4859241126070994,
      "grad_norm": 1.5827461302215218,
      "learning_rate": 3.026986020564294e-06,
      "loss": 1.3645,
      "step": 5299
    },
    {
      "epoch": 6.487148102815177,
      "grad_norm": 2.8455529406005473,
      "learning_rate": 3.0263595237819306e-06,
      "loss": 0.4575,
      "step": 5300
    },
    {
      "epoch": 6.488372093023256,
      "grad_norm": 2.023117784583821,
      "learning_rate": 3.025732992409166e-06,
      "loss": 0.4292,
      "step": 5301
    },
    {
      "epoch": 6.489596083231334,
      "grad_norm": 1.1445218149582992,
      "learning_rate": 3.0251064264871742e-06,
      "loss": 0.6907,
      "step": 5302
    },
    {
      "epoch": 6.490820073439412,
      "grad_norm": 1.392959995817116,
      "learning_rate": 3.02447982605713e-06,
      "loss": 1.5708,
      "step": 5303
    },
    {
      "epoch": 6.492044063647491,
      "grad_norm": 2.1209419159281726,
      "learning_rate": 3.023853191160211e-06,
      "loss": 0.6269,
      "step": 5304
    },
    {
      "epoch": 6.493268053855569,
      "grad_norm": 1.2171374034880065,
      "learning_rate": 3.0232265218375984e-06,
      "loss": 0.7547,
      "step": 5305
    },
    {
      "epoch": 6.494492044063647,
      "grad_norm": 1.819238163422663,
      "learning_rate": 3.0225998181304745e-06,
      "loss": 1.1295,
      "step": 5306
    },
    {
      "epoch": 6.495716034271726,
      "grad_norm": 1.048095177624131,
      "learning_rate": 3.021973080080023e-06,
      "loss": 0.4547,
      "step": 5307
    },
    {
      "epoch": 6.496940024479804,
      "grad_norm": 1.1315115631597263,
      "learning_rate": 3.021346307727431e-06,
      "loss": 0.5713,
      "step": 5308
    },
    {
      "epoch": 6.498164014687882,
      "grad_norm": 1.012220254352979,
      "learning_rate": 3.020719501113889e-06,
      "loss": 0.6,
      "step": 5309
    },
    {
      "epoch": 6.499388004895961,
      "grad_norm": 1.9915755567933502,
      "learning_rate": 3.020092660280586e-06,
      "loss": 0.4915,
      "step": 5310
    },
    {
      "epoch": 6.500611995104039,
      "grad_norm": 2.1145020769184795,
      "learning_rate": 3.0194657852687186e-06,
      "loss": 0.5523,
      "step": 5311
    },
    {
      "epoch": 6.501835985312118,
      "grad_norm": 1.8162380933981084,
      "learning_rate": 3.018838876119481e-06,
      "loss": 0.7393,
      "step": 5312
    },
    {
      "epoch": 6.503059975520196,
      "grad_norm": 1.7031480873185578,
      "learning_rate": 3.01821193287407e-06,
      "loss": 0.7408,
      "step": 5313
    },
    {
      "epoch": 6.504283965728274,
      "grad_norm": 1.3194847431773604,
      "learning_rate": 3.017584955573689e-06,
      "loss": 0.5861,
      "step": 5314
    },
    {
      "epoch": 6.505507955936353,
      "grad_norm": 1.4890310300084912,
      "learning_rate": 3.016957944259538e-06,
      "loss": 1.4255,
      "step": 5315
    },
    {
      "epoch": 6.506731946144431,
      "grad_norm": 2.450300828680658,
      "learning_rate": 3.016330898972824e-06,
      "loss": 0.5594,
      "step": 5316
    },
    {
      "epoch": 6.507955936352509,
      "grad_norm": 1.8329365707865262,
      "learning_rate": 3.0157038197547527e-06,
      "loss": 0.6442,
      "step": 5317
    },
    {
      "epoch": 6.509179926560588,
      "grad_norm": 2.2019053274982023,
      "learning_rate": 3.015076706646534e-06,
      "loss": 0.8209,
      "step": 5318
    },
    {
      "epoch": 6.510403916768666,
      "grad_norm": 2.2740686789385287,
      "learning_rate": 3.0144495596893787e-06,
      "loss": 0.4814,
      "step": 5319
    },
    {
      "epoch": 6.511627906976744,
      "grad_norm": 1.9957741448651014,
      "learning_rate": 3.013822378924502e-06,
      "loss": 0.5273,
      "step": 5320
    },
    {
      "epoch": 6.512851897184823,
      "grad_norm": 1.809584462562897,
      "learning_rate": 3.0131951643931178e-06,
      "loss": 1.2066,
      "step": 5321
    },
    {
      "epoch": 6.5140758873929006,
      "grad_norm": 0.9487198297794097,
      "learning_rate": 3.0125679161364467e-06,
      "loss": 0.384,
      "step": 5322
    },
    {
      "epoch": 6.515299877600979,
      "grad_norm": 2.178331614344408,
      "learning_rate": 3.011940634195708e-06,
      "loss": 0.8109,
      "step": 5323
    },
    {
      "epoch": 6.516523867809058,
      "grad_norm": 2.0595157558768777,
      "learning_rate": 3.011313318612124e-06,
      "loss": 0.428,
      "step": 5324
    },
    {
      "epoch": 6.5177478580171355,
      "grad_norm": 1.2859237099344467,
      "learning_rate": 3.01068596942692e-06,
      "loss": 0.4864,
      "step": 5325
    },
    {
      "epoch": 6.518971848225214,
      "grad_norm": 2.0920858119779036,
      "learning_rate": 3.010058586681323e-06,
      "loss": 0.348,
      "step": 5326
    },
    {
      "epoch": 6.520195838433293,
      "grad_norm": 1.3119470492294893,
      "learning_rate": 3.0094311704165625e-06,
      "loss": 1.3225,
      "step": 5327
    },
    {
      "epoch": 6.5214198286413705,
      "grad_norm": 1.2716640554776317,
      "learning_rate": 3.0088037206738683e-06,
      "loss": 1.1014,
      "step": 5328
    },
    {
      "epoch": 6.522643818849449,
      "grad_norm": 1.448519425169039,
      "learning_rate": 3.008176237494478e-06,
      "loss": 0.9268,
      "step": 5329
    },
    {
      "epoch": 6.523867809057528,
      "grad_norm": 1.6654712457980188,
      "learning_rate": 3.007548720919623e-06,
      "loss": 0.5026,
      "step": 5330
    },
    {
      "epoch": 6.525091799265606,
      "grad_norm": 1.4573903577094696,
      "learning_rate": 3.006921170990544e-06,
      "loss": 0.6987,
      "step": 5331
    },
    {
      "epoch": 6.526315789473684,
      "grad_norm": 1.8047713588873915,
      "learning_rate": 3.0062935877484807e-06,
      "loss": 1.1529,
      "step": 5332
    },
    {
      "epoch": 6.527539779681763,
      "grad_norm": 0.9497939722379454,
      "learning_rate": 3.0056659712346753e-06,
      "loss": 0.4607,
      "step": 5333
    },
    {
      "epoch": 6.52876376988984,
      "grad_norm": 1.4808702266401585,
      "learning_rate": 3.0050383214903724e-06,
      "loss": 0.9496,
      "step": 5334
    },
    {
      "epoch": 6.529987760097919,
      "grad_norm": 1.1631253939177064,
      "learning_rate": 3.0044106385568194e-06,
      "loss": 0.6246,
      "step": 5335
    },
    {
      "epoch": 6.531211750305998,
      "grad_norm": 2.614757104072889,
      "learning_rate": 3.003782922475265e-06,
      "loss": 0.4376,
      "step": 5336
    },
    {
      "epoch": 6.532435740514076,
      "grad_norm": 1.9817381634604467,
      "learning_rate": 3.0031551732869597e-06,
      "loss": 1.4363,
      "step": 5337
    },
    {
      "epoch": 6.533659730722154,
      "grad_norm": 1.4402302537811131,
      "learning_rate": 3.002527391033159e-06,
      "loss": 0.5944,
      "step": 5338
    },
    {
      "epoch": 6.534883720930233,
      "grad_norm": 2.55414225637835,
      "learning_rate": 3.0018995757551145e-06,
      "loss": 0.3378,
      "step": 5339
    },
    {
      "epoch": 6.536107711138311,
      "grad_norm": 2.0113164398195575,
      "learning_rate": 3.001271727494088e-06,
      "loss": 0.9859,
      "step": 5340
    },
    {
      "epoch": 6.537331701346389,
      "grad_norm": 1.8586821946551257,
      "learning_rate": 3.0006438462913367e-06,
      "loss": 0.5001,
      "step": 5341
    },
    {
      "epoch": 6.538555691554468,
      "grad_norm": 1.0419399346288811,
      "learning_rate": 3.0000159321881246e-06,
      "loss": 0.5146,
      "step": 5342
    },
    {
      "epoch": 6.539779681762546,
      "grad_norm": 0.8541209556563468,
      "learning_rate": 2.999387985225714e-06,
      "loss": 0.4513,
      "step": 5343
    },
    {
      "epoch": 6.541003671970624,
      "grad_norm": 1.472974372863977,
      "learning_rate": 2.9987600054453725e-06,
      "loss": 0.5924,
      "step": 5344
    },
    {
      "epoch": 6.5422276621787026,
      "grad_norm": 1.3102145201953603,
      "learning_rate": 2.998131992888368e-06,
      "loss": 1.228,
      "step": 5345
    },
    {
      "epoch": 6.543451652386781,
      "grad_norm": 1.979472870550857,
      "learning_rate": 2.9975039475959717e-06,
      "loss": 1.0113,
      "step": 5346
    },
    {
      "epoch": 6.544675642594859,
      "grad_norm": 1.4662642821153606,
      "learning_rate": 2.996875869609456e-06,
      "loss": 0.8402,
      "step": 5347
    },
    {
      "epoch": 6.5458996328029375,
      "grad_norm": 1.1574807823956437,
      "learning_rate": 2.9962477589700954e-06,
      "loss": 0.4689,
      "step": 5348
    },
    {
      "epoch": 6.547123623011016,
      "grad_norm": 1.442888307041627,
      "learning_rate": 2.9956196157191685e-06,
      "loss": 0.5196,
      "step": 5349
    },
    {
      "epoch": 6.548347613219094,
      "grad_norm": 1.675266947117676,
      "learning_rate": 2.994991439897953e-06,
      "loss": 0.4636,
      "step": 5350
    },
    {
      "epoch": 6.5495716034271725,
      "grad_norm": 1.2399756219703901,
      "learning_rate": 2.994363231547731e-06,
      "loss": 0.6642,
      "step": 5351
    },
    {
      "epoch": 6.550795593635251,
      "grad_norm": 1.5931312300036078,
      "learning_rate": 2.993734990709787e-06,
      "loss": 0.451,
      "step": 5352
    },
    {
      "epoch": 6.552019583843329,
      "grad_norm": 1.9418298761914672,
      "learning_rate": 2.993106717425404e-06,
      "loss": 0.5225,
      "step": 5353
    },
    {
      "epoch": 6.5532435740514074,
      "grad_norm": 1.2290302718264912,
      "learning_rate": 2.992478411735872e-06,
      "loss": 0.6807,
      "step": 5354
    },
    {
      "epoch": 6.554467564259486,
      "grad_norm": 1.9360700357022829,
      "learning_rate": 2.99185007368248e-06,
      "loss": 0.4497,
      "step": 5355
    },
    {
      "epoch": 6.555691554467565,
      "grad_norm": 1.4766149188297162,
      "learning_rate": 2.9912217033065212e-06,
      "loss": 1.0048,
      "step": 5356
    },
    {
      "epoch": 6.556915544675642,
      "grad_norm": 1.3618875964653336,
      "learning_rate": 2.990593300649289e-06,
      "loss": 0.51,
      "step": 5357
    },
    {
      "epoch": 6.558139534883721,
      "grad_norm": 1.5113570217936043,
      "learning_rate": 2.989964865752079e-06,
      "loss": 1.1506,
      "step": 5358
    },
    {
      "epoch": 6.5593635250918,
      "grad_norm": 1.1539356337650317,
      "learning_rate": 2.989336398656191e-06,
      "loss": 0.5519,
      "step": 5359
    },
    {
      "epoch": 6.560587515299877,
      "grad_norm": 1.1849486757467327,
      "learning_rate": 2.9887078994029232e-06,
      "loss": 0.4728,
      "step": 5360
    },
    {
      "epoch": 6.561811505507956,
      "grad_norm": 0.9549763239060722,
      "learning_rate": 2.988079368033582e-06,
      "loss": 0.593,
      "step": 5361
    },
    {
      "epoch": 6.563035495716035,
      "grad_norm": 0.9330135841833421,
      "learning_rate": 2.987450804589469e-06,
      "loss": 0.6079,
      "step": 5362
    },
    {
      "epoch": 6.564259485924112,
      "grad_norm": 0.8996704424454582,
      "learning_rate": 2.9868222091118913e-06,
      "loss": 0.5574,
      "step": 5363
    },
    {
      "epoch": 6.565483476132191,
      "grad_norm": 2.5175713480942887,
      "learning_rate": 2.9861935816421596e-06,
      "loss": 0.3545,
      "step": 5364
    },
    {
      "epoch": 6.56670746634027,
      "grad_norm": 1.4596859699325908,
      "learning_rate": 2.9855649222215837e-06,
      "loss": 1.1921,
      "step": 5365
    },
    {
      "epoch": 6.567931456548347,
      "grad_norm": 1.2311117282029895,
      "learning_rate": 2.9849362308914775e-06,
      "loss": 0.811,
      "step": 5366
    },
    {
      "epoch": 6.569155446756426,
      "grad_norm": 2.128249655309296,
      "learning_rate": 2.9843075076931556e-06,
      "loss": 0.5798,
      "step": 5367
    },
    {
      "epoch": 6.5703794369645045,
      "grad_norm": 1.6208935708575591,
      "learning_rate": 2.983678752667936e-06,
      "loss": 1.4796,
      "step": 5368
    },
    {
      "epoch": 6.571603427172582,
      "grad_norm": 1.674346233722502,
      "learning_rate": 2.9830499658571367e-06,
      "loss": 1.1967,
      "step": 5369
    },
    {
      "epoch": 6.572827417380661,
      "grad_norm": 1.708150051478475,
      "learning_rate": 2.9824211473020814e-06,
      "loss": 0.4174,
      "step": 5370
    },
    {
      "epoch": 6.5740514075887395,
      "grad_norm": 0.9698889239524006,
      "learning_rate": 2.9817922970440918e-06,
      "loss": 0.6297,
      "step": 5371
    },
    {
      "epoch": 6.575275397796817,
      "grad_norm": 1.3168192878465668,
      "learning_rate": 2.981163415124495e-06,
      "loss": 0.9485,
      "step": 5372
    },
    {
      "epoch": 6.576499388004896,
      "grad_norm": 1.766797973874093,
      "learning_rate": 2.980534501584618e-06,
      "loss": 0.691,
      "step": 5373
    },
    {
      "epoch": 6.5777233782129745,
      "grad_norm": 2.5676215612630258,
      "learning_rate": 2.9799055564657913e-06,
      "loss": 0.5803,
      "step": 5374
    },
    {
      "epoch": 6.578947368421053,
      "grad_norm": 1.799742307125125,
      "learning_rate": 2.9792765798093466e-06,
      "loss": 0.8704,
      "step": 5375
    },
    {
      "epoch": 6.580171358629131,
      "grad_norm": 1.57040681171445,
      "learning_rate": 2.978647571656618e-06,
      "loss": 1.0047,
      "step": 5376
    },
    {
      "epoch": 6.5813953488372094,
      "grad_norm": 1.1376203627100405,
      "learning_rate": 2.9780185320489403e-06,
      "loss": 0.5668,
      "step": 5377
    },
    {
      "epoch": 6.582619339045287,
      "grad_norm": 1.2464884743093998,
      "learning_rate": 2.9773894610276525e-06,
      "loss": 1.0181,
      "step": 5378
    },
    {
      "epoch": 6.583843329253366,
      "grad_norm": 1.7258870523487413,
      "learning_rate": 2.976760358634097e-06,
      "loss": 0.5147,
      "step": 5379
    },
    {
      "epoch": 6.585067319461444,
      "grad_norm": 1.5516825602924413,
      "learning_rate": 2.9761312249096125e-06,
      "loss": 0.4609,
      "step": 5380
    },
    {
      "epoch": 6.586291309669523,
      "grad_norm": 1.3933203646536483,
      "learning_rate": 2.975502059895545e-06,
      "loss": 0.507,
      "step": 5381
    },
    {
      "epoch": 6.587515299877601,
      "grad_norm": 2.1565653103212528,
      "learning_rate": 2.974872863633242e-06,
      "loss": 0.3933,
      "step": 5382
    },
    {
      "epoch": 6.588739290085679,
      "grad_norm": 1.6922789991287492,
      "learning_rate": 2.9742436361640505e-06,
      "loss": 0.755,
      "step": 5383
    },
    {
      "epoch": 6.589963280293758,
      "grad_norm": 2.1989318064180186,
      "learning_rate": 2.9736143775293204e-06,
      "loss": 0.4292,
      "step": 5384
    },
    {
      "epoch": 6.591187270501836,
      "grad_norm": 1.0230367398926992,
      "learning_rate": 2.9729850877704065e-06,
      "loss": 0.4952,
      "step": 5385
    },
    {
      "epoch": 6.592411260709914,
      "grad_norm": 1.6944706976115564,
      "learning_rate": 2.9723557669286606e-06,
      "loss": 0.7839,
      "step": 5386
    },
    {
      "epoch": 6.593635250917993,
      "grad_norm": 1.2709389015815304,
      "learning_rate": 2.9717264150454417e-06,
      "loss": 0.5421,
      "step": 5387
    },
    {
      "epoch": 6.594859241126071,
      "grad_norm": 0.962302890329117,
      "learning_rate": 2.9710970321621086e-06,
      "loss": 0.51,
      "step": 5388
    },
    {
      "epoch": 6.596083231334149,
      "grad_norm": 1.2516869582831065,
      "learning_rate": 2.9704676183200195e-06,
      "loss": 0.5824,
      "step": 5389
    },
    {
      "epoch": 6.597307221542228,
      "grad_norm": 1.3463624762072195,
      "learning_rate": 2.969838173560539e-06,
      "loss": 0.5703,
      "step": 5390
    },
    {
      "epoch": 6.598531211750306,
      "grad_norm": 1.929397164305181,
      "learning_rate": 2.969208697925032e-06,
      "loss": 0.312,
      "step": 5391
    },
    {
      "epoch": 6.599755201958384,
      "grad_norm": 1.302340256621175,
      "learning_rate": 2.9685791914548646e-06,
      "loss": 0.6277,
      "step": 5392
    },
    {
      "epoch": 6.600979192166463,
      "grad_norm": 1.561087363635551,
      "learning_rate": 2.967949654191405e-06,
      "loss": 0.5678,
      "step": 5393
    },
    {
      "epoch": 6.602203182374541,
      "grad_norm": 1.6631831684142577,
      "learning_rate": 2.9673200861760263e-06,
      "loss": 1.0699,
      "step": 5394
    },
    {
      "epoch": 6.603427172582619,
      "grad_norm": 1.0231653977924138,
      "learning_rate": 2.966690487450099e-06,
      "loss": 0.5213,
      "step": 5395
    },
    {
      "epoch": 6.604651162790698,
      "grad_norm": 1.8717213264601191,
      "learning_rate": 2.966060858055e-06,
      "loss": 0.6283,
      "step": 5396
    },
    {
      "epoch": 6.605875152998776,
      "grad_norm": 0.8614544811778199,
      "learning_rate": 2.965431198032105e-06,
      "loss": 0.4539,
      "step": 5397
    },
    {
      "epoch": 6.607099143206854,
      "grad_norm": 1.9712625975197968,
      "learning_rate": 2.9648015074227927e-06,
      "loss": 0.6961,
      "step": 5398
    },
    {
      "epoch": 6.608323133414933,
      "grad_norm": 1.388435046797862,
      "learning_rate": 2.964171786268445e-06,
      "loss": 0.4058,
      "step": 5399
    },
    {
      "epoch": 6.6095471236230114,
      "grad_norm": 0.965417792779119,
      "learning_rate": 2.963542034610444e-06,
      "loss": 0.3219,
      "step": 5400
    },
    {
      "epoch": 6.610771113831089,
      "grad_norm": 0.9941354366362283,
      "learning_rate": 2.9629122524901754e-06,
      "loss": 0.6261,
      "step": 5401
    },
    {
      "epoch": 6.611995104039168,
      "grad_norm": 2.5131805294771694,
      "learning_rate": 2.9622824399490254e-06,
      "loss": 0.479,
      "step": 5402
    },
    {
      "epoch": 6.613219094247246,
      "grad_norm": 0.979355777309658,
      "learning_rate": 2.961652597028383e-06,
      "loss": 0.6467,
      "step": 5403
    },
    {
      "epoch": 6.614443084455324,
      "grad_norm": 0.9171329090352748,
      "learning_rate": 2.96102272376964e-06,
      "loss": 0.3764,
      "step": 5404
    },
    {
      "epoch": 6.615667074663403,
      "grad_norm": 1.8806884543596591,
      "learning_rate": 2.9603928202141887e-06,
      "loss": 1.0377,
      "step": 5405
    },
    {
      "epoch": 6.616891064871481,
      "grad_norm": 1.0862109727409912,
      "learning_rate": 2.959762886403424e-06,
      "loss": 0.4916,
      "step": 5406
    },
    {
      "epoch": 6.618115055079559,
      "grad_norm": 1.475439347861243,
      "learning_rate": 2.9591329223787435e-06,
      "loss": 0.4704,
      "step": 5407
    },
    {
      "epoch": 6.619339045287638,
      "grad_norm": 1.215879909100252,
      "learning_rate": 2.9585029281815447e-06,
      "loss": 0.7927,
      "step": 5408
    },
    {
      "epoch": 6.620563035495716,
      "grad_norm": 1.3517423440951264,
      "learning_rate": 2.95787290385323e-06,
      "loss": 0.6374,
      "step": 5409
    },
    {
      "epoch": 6.621787025703794,
      "grad_norm": 1.22669424337683,
      "learning_rate": 2.9572428494352006e-06,
      "loss": 0.4945,
      "step": 5410
    },
    {
      "epoch": 6.623011015911873,
      "grad_norm": 1.5821447657905434,
      "learning_rate": 2.956612764968864e-06,
      "loss": 0.6294,
      "step": 5411
    },
    {
      "epoch": 6.624235006119951,
      "grad_norm": 0.9667091857763446,
      "learning_rate": 2.9559826504956245e-06,
      "loss": 0.4839,
      "step": 5412
    },
    {
      "epoch": 6.625458996328029,
      "grad_norm": 1.114336665359784,
      "learning_rate": 2.955352506056892e-06,
      "loss": 0.6024,
      "step": 5413
    },
    {
      "epoch": 6.626682986536108,
      "grad_norm": 1.7770633845178825,
      "learning_rate": 2.9547223316940767e-06,
      "loss": 1.0936,
      "step": 5414
    },
    {
      "epoch": 6.627906976744186,
      "grad_norm": 1.182393811289438,
      "learning_rate": 2.9540921274485913e-06,
      "loss": 0.4462,
      "step": 5415
    },
    {
      "epoch": 6.629130966952264,
      "grad_norm": 1.539929677107794,
      "learning_rate": 2.9534618933618515e-06,
      "loss": 1.0174,
      "step": 5416
    },
    {
      "epoch": 6.630354957160343,
      "grad_norm": 2.206598380218775,
      "learning_rate": 2.9528316294752733e-06,
      "loss": 0.4148,
      "step": 5417
    },
    {
      "epoch": 6.631578947368421,
      "grad_norm": 2.196446122544651,
      "learning_rate": 2.9522013358302754e-06,
      "loss": 0.4477,
      "step": 5418
    },
    {
      "epoch": 6.6328029375765,
      "grad_norm": 1.8004781787572024,
      "learning_rate": 2.9515710124682772e-06,
      "loss": 0.3281,
      "step": 5419
    },
    {
      "epoch": 6.634026927784578,
      "grad_norm": 1.0810513485652549,
      "learning_rate": 2.9509406594307037e-06,
      "loss": 0.6619,
      "step": 5420
    },
    {
      "epoch": 6.635250917992656,
      "grad_norm": 2.0830999321247186,
      "learning_rate": 2.950310276758977e-06,
      "loss": 0.5519,
      "step": 5421
    },
    {
      "epoch": 6.636474908200734,
      "grad_norm": 2.65180981563236,
      "learning_rate": 2.949679864494525e-06,
      "loss": 0.3446,
      "step": 5422
    },
    {
      "epoch": 6.6376988984088126,
      "grad_norm": 2.2037671147687354,
      "learning_rate": 2.9490494226787758e-06,
      "loss": 0.886,
      "step": 5423
    },
    {
      "epoch": 6.638922888616891,
      "grad_norm": 1.9558876856273422,
      "learning_rate": 2.948418951353159e-06,
      "loss": 0.6177,
      "step": 5424
    },
    {
      "epoch": 6.64014687882497,
      "grad_norm": 1.6503930476376285,
      "learning_rate": 2.9477884505591076e-06,
      "loss": 1.2335,
      "step": 5425
    },
    {
      "epoch": 6.6413708690330475,
      "grad_norm": 1.144128551623011,
      "learning_rate": 2.9471579203380552e-06,
      "loss": 0.5402,
      "step": 5426
    },
    {
      "epoch": 6.642594859241126,
      "grad_norm": 1.2984433590828688,
      "learning_rate": 2.9465273607314386e-06,
      "loss": 0.642,
      "step": 5427
    },
    {
      "epoch": 6.643818849449205,
      "grad_norm": 1.9452229930110212,
      "learning_rate": 2.9458967717806945e-06,
      "loss": 0.485,
      "step": 5428
    },
    {
      "epoch": 6.6450428396572825,
      "grad_norm": 1.4287439972317089,
      "learning_rate": 2.9452661535272652e-06,
      "loss": 1.2445,
      "step": 5429
    },
    {
      "epoch": 6.646266829865361,
      "grad_norm": 1.4182890500517595,
      "learning_rate": 2.9446355060125903e-06,
      "loss": 1.64,
      "step": 5430
    },
    {
      "epoch": 6.64749082007344,
      "grad_norm": 2.6400337948234105,
      "learning_rate": 2.944004829278115e-06,
      "loss": 0.6207,
      "step": 5431
    },
    {
      "epoch": 6.6487148102815175,
      "grad_norm": 2.6956186339329804,
      "learning_rate": 2.943374123365284e-06,
      "loss": 0.4149,
      "step": 5432
    },
    {
      "epoch": 6.649938800489596,
      "grad_norm": 1.7763645340940566,
      "learning_rate": 2.942743388315546e-06,
      "loss": 0.4567,
      "step": 5433
    },
    {
      "epoch": 6.651162790697675,
      "grad_norm": 1.510037479639602,
      "learning_rate": 2.9421126241703496e-06,
      "loss": 1.0287,
      "step": 5434
    },
    {
      "epoch": 6.652386780905752,
      "grad_norm": 1.6801624841069247,
      "learning_rate": 2.9414818309711475e-06,
      "loss": 1.3111,
      "step": 5435
    },
    {
      "epoch": 6.653610771113831,
      "grad_norm": 2.0088029003878773,
      "learning_rate": 2.9408510087593915e-06,
      "loss": 0.4876,
      "step": 5436
    },
    {
      "epoch": 6.65483476132191,
      "grad_norm": 2.337474247938543,
      "learning_rate": 2.940220157576539e-06,
      "loss": 0.5387,
      "step": 5437
    },
    {
      "epoch": 6.656058751529987,
      "grad_norm": 1.9002422291282646,
      "learning_rate": 2.9395892774640455e-06,
      "loss": 1.0392,
      "step": 5438
    },
    {
      "epoch": 6.657282741738066,
      "grad_norm": 1.5662421120782486,
      "learning_rate": 2.9389583684633697e-06,
      "loss": 0.3726,
      "step": 5439
    },
    {
      "epoch": 6.658506731946145,
      "grad_norm": 1.6144778880007884,
      "learning_rate": 2.9383274306159747e-06,
      "loss": 0.5079,
      "step": 5440
    },
    {
      "epoch": 6.659730722154222,
      "grad_norm": 1.538253191851325,
      "learning_rate": 2.937696463963322e-06,
      "loss": 0.7277,
      "step": 5441
    },
    {
      "epoch": 6.660954712362301,
      "grad_norm": 1.2412801331283339,
      "learning_rate": 2.9370654685468766e-06,
      "loss": 0.611,
      "step": 5442
    },
    {
      "epoch": 6.66217870257038,
      "grad_norm": 0.9867519347586725,
      "learning_rate": 2.9364344444081043e-06,
      "loss": 0.5191,
      "step": 5443
    },
    {
      "epoch": 6.663402692778458,
      "grad_norm": 2.5559085656684815,
      "learning_rate": 2.9358033915884758e-06,
      "loss": 0.5622,
      "step": 5444
    },
    {
      "epoch": 6.664626682986536,
      "grad_norm": 2.495609696901564,
      "learning_rate": 2.9351723101294593e-06,
      "loss": 0.7681,
      "step": 5445
    },
    {
      "epoch": 6.6658506731946146,
      "grad_norm": 0.998768139159904,
      "learning_rate": 2.9345412000725293e-06,
      "loss": 0.5752,
      "step": 5446
    },
    {
      "epoch": 6.667074663402692,
      "grad_norm": 1.4598716408089027,
      "learning_rate": 2.9339100614591583e-06,
      "loss": 0.5515,
      "step": 5447
    },
    {
      "epoch": 6.668298653610771,
      "grad_norm": 1.843817307650144,
      "learning_rate": 2.9332788943308234e-06,
      "loss": 0.8334,
      "step": 5448
    },
    {
      "epoch": 6.6695226438188495,
      "grad_norm": 1.3053693899756624,
      "learning_rate": 2.9326476987290016e-06,
      "loss": 0.8387,
      "step": 5449
    },
    {
      "epoch": 6.670746634026928,
      "grad_norm": 1.7916472394091607,
      "learning_rate": 2.9320164746951734e-06,
      "loss": 1.3389,
      "step": 5450
    },
    {
      "epoch": 6.671970624235006,
      "grad_norm": 1.9494926775431656,
      "learning_rate": 2.9313852222708205e-06,
      "loss": 0.3525,
      "step": 5451
    },
    {
      "epoch": 6.6731946144430845,
      "grad_norm": 1.4250091911130924,
      "learning_rate": 2.930753941497426e-06,
      "loss": 0.4398,
      "step": 5452
    },
    {
      "epoch": 6.674418604651163,
      "grad_norm": 1.8861857789345184,
      "learning_rate": 2.930122632416476e-06,
      "loss": 0.5356,
      "step": 5453
    },
    {
      "epoch": 6.675642594859241,
      "grad_norm": 1.6478732289424214,
      "learning_rate": 2.9294912950694575e-06,
      "loss": 0.7073,
      "step": 5454
    },
    {
      "epoch": 6.6768665850673194,
      "grad_norm": 1.055885576391337,
      "learning_rate": 2.92885992949786e-06,
      "loss": 0.5498,
      "step": 5455
    },
    {
      "epoch": 6.678090575275398,
      "grad_norm": 1.2651068193868458,
      "learning_rate": 2.928228535743174e-06,
      "loss": 0.6309,
      "step": 5456
    },
    {
      "epoch": 6.679314565483476,
      "grad_norm": 1.0453699558789769,
      "learning_rate": 2.927597113846892e-06,
      "loss": 0.6704,
      "step": 5457
    },
    {
      "epoch": 6.680538555691554,
      "grad_norm": 1.544603762440099,
      "learning_rate": 2.9269656638505093e-06,
      "loss": 1.2118,
      "step": 5458
    },
    {
      "epoch": 6.681762545899633,
      "grad_norm": 1.3353120729939156,
      "learning_rate": 2.9263341857955237e-06,
      "loss": 0.4868,
      "step": 5459
    },
    {
      "epoch": 6.682986536107711,
      "grad_norm": 1.0203838379611825,
      "learning_rate": 2.9257026797234304e-06,
      "loss": 0.6861,
      "step": 5460
    },
    {
      "epoch": 6.684210526315789,
      "grad_norm": 1.1820878153630456,
      "learning_rate": 2.9250711456757332e-06,
      "loss": 0.5495,
      "step": 5461
    },
    {
      "epoch": 6.685434516523868,
      "grad_norm": 1.8551744844188174,
      "learning_rate": 2.9244395836939313e-06,
      "loss": 1.0906,
      "step": 5462
    },
    {
      "epoch": 6.686658506731947,
      "grad_norm": 1.0380606894740105,
      "learning_rate": 2.9238079938195306e-06,
      "loss": 0.5433,
      "step": 5463
    },
    {
      "epoch": 6.687882496940024,
      "grad_norm": 2.800254243011946,
      "learning_rate": 2.9231763760940364e-06,
      "loss": 0.7159,
      "step": 5464
    },
    {
      "epoch": 6.689106487148103,
      "grad_norm": 1.3471435714340598,
      "learning_rate": 2.9225447305589556e-06,
      "loss": 0.4845,
      "step": 5465
    },
    {
      "epoch": 6.690330477356181,
      "grad_norm": 2.2695246259851585,
      "learning_rate": 2.9219130572557973e-06,
      "loss": 0.3888,
      "step": 5466
    },
    {
      "epoch": 6.691554467564259,
      "grad_norm": 2.32404647565442,
      "learning_rate": 2.9212813562260744e-06,
      "loss": 0.5974,
      "step": 5467
    },
    {
      "epoch": 6.692778457772338,
      "grad_norm": 2.3276412691911976,
      "learning_rate": 2.920649627511299e-06,
      "loss": 0.3857,
      "step": 5468
    },
    {
      "epoch": 6.6940024479804165,
      "grad_norm": 1.466960140415234,
      "learning_rate": 2.920017871152985e-06,
      "loss": 1.4148,
      "step": 5469
    },
    {
      "epoch": 6.695226438188494,
      "grad_norm": 2.3949770999029276,
      "learning_rate": 2.919386087192651e-06,
      "loss": 0.4816,
      "step": 5470
    },
    {
      "epoch": 6.696450428396573,
      "grad_norm": 1.0715180229634735,
      "learning_rate": 2.9187542756718133e-06,
      "loss": 0.4676,
      "step": 5471
    },
    {
      "epoch": 6.6976744186046515,
      "grad_norm": 1.362905469831542,
      "learning_rate": 2.9181224366319947e-06,
      "loss": 0.7628,
      "step": 5472
    },
    {
      "epoch": 6.698898408812729,
      "grad_norm": 1.4433289273395071,
      "learning_rate": 2.917490570114716e-06,
      "loss": 0.6354,
      "step": 5473
    },
    {
      "epoch": 6.700122399020808,
      "grad_norm": 1.5215173231764259,
      "learning_rate": 2.916858676161501e-06,
      "loss": 0.491,
      "step": 5474
    },
    {
      "epoch": 6.7013463892288865,
      "grad_norm": 1.3187886904363915,
      "learning_rate": 2.916226754813876e-06,
      "loss": 1.2755,
      "step": 5475
    },
    {
      "epoch": 6.702570379436964,
      "grad_norm": 1.0213097910209732,
      "learning_rate": 2.915594806113368e-06,
      "loss": 0.7594,
      "step": 5476
    },
    {
      "epoch": 6.703794369645043,
      "grad_norm": 2.431051679861424,
      "learning_rate": 2.914962830101506e-06,
      "loss": 0.4393,
      "step": 5477
    },
    {
      "epoch": 6.7050183598531214,
      "grad_norm": 1.3683341266244353,
      "learning_rate": 2.914330826819821e-06,
      "loss": 0.4751,
      "step": 5478
    },
    {
      "epoch": 6.706242350061199,
      "grad_norm": 1.292614744828594,
      "learning_rate": 2.9136987963098485e-06,
      "loss": 0.7983,
      "step": 5479
    },
    {
      "epoch": 6.707466340269278,
      "grad_norm": 0.9952475928764565,
      "learning_rate": 2.9130667386131197e-06,
      "loss": 0.4919,
      "step": 5480
    },
    {
      "epoch": 6.708690330477356,
      "grad_norm": 0.9581135126661421,
      "learning_rate": 2.9124346537711736e-06,
      "loss": 0.8551,
      "step": 5481
    },
    {
      "epoch": 6.709914320685434,
      "grad_norm": 1.342252398961635,
      "learning_rate": 2.9118025418255467e-06,
      "loss": 0.4827,
      "step": 5482
    },
    {
      "epoch": 6.711138310893513,
      "grad_norm": 1.2986636058898233,
      "learning_rate": 2.9111704028177813e-06,
      "loss": 0.4674,
      "step": 5483
    },
    {
      "epoch": 6.712362301101591,
      "grad_norm": 2.3285673994736853,
      "learning_rate": 2.9105382367894158e-06,
      "loss": 0.4935,
      "step": 5484
    },
    {
      "epoch": 6.713586291309669,
      "grad_norm": 1.4881997250883878,
      "learning_rate": 2.909906043781998e-06,
      "loss": 1.1776,
      "step": 5485
    },
    {
      "epoch": 6.714810281517748,
      "grad_norm": 2.01784293684291,
      "learning_rate": 2.9092738238370696e-06,
      "loss": 0.7607,
      "step": 5486
    },
    {
      "epoch": 6.716034271725826,
      "grad_norm": 1.3899049304915057,
      "learning_rate": 2.9086415769961802e-06,
      "loss": 0.489,
      "step": 5487
    },
    {
      "epoch": 6.717258261933905,
      "grad_norm": 1.7038797248660045,
      "learning_rate": 2.908009303300878e-06,
      "loss": 1.3344,
      "step": 5488
    },
    {
      "epoch": 6.718482252141983,
      "grad_norm": 1.2390325890335105,
      "learning_rate": 2.9073770027927124e-06,
      "loss": 0.6985,
      "step": 5489
    },
    {
      "epoch": 6.719706242350061,
      "grad_norm": 1.2942152571252798,
      "learning_rate": 2.906744675513237e-06,
      "loss": 0.9298,
      "step": 5490
    },
    {
      "epoch": 6.720930232558139,
      "grad_norm": 1.3993578994827387,
      "learning_rate": 2.906112321504007e-06,
      "loss": 0.4527,
      "step": 5491
    },
    {
      "epoch": 6.722154222766218,
      "grad_norm": 1.4792718585314104,
      "learning_rate": 2.905479940806577e-06,
      "loss": 1.3775,
      "step": 5492
    },
    {
      "epoch": 6.723378212974296,
      "grad_norm": 1.137744589809637,
      "learning_rate": 2.904847533462504e-06,
      "loss": 0.4232,
      "step": 5493
    },
    {
      "epoch": 6.724602203182375,
      "grad_norm": 1.282985535639114,
      "learning_rate": 2.9042150995133506e-06,
      "loss": 0.553,
      "step": 5494
    },
    {
      "epoch": 6.725826193390453,
      "grad_norm": 1.3543811744864525,
      "learning_rate": 2.9035826390006737e-06,
      "loss": 0.9521,
      "step": 5495
    },
    {
      "epoch": 6.727050183598531,
      "grad_norm": 0.9516628616008946,
      "learning_rate": 2.9029501519660403e-06,
      "loss": 0.5791,
      "step": 5496
    },
    {
      "epoch": 6.72827417380661,
      "grad_norm": 1.660470956203183,
      "learning_rate": 2.902317638451012e-06,
      "loss": 0.5148,
      "step": 5497
    },
    {
      "epoch": 6.729498164014688,
      "grad_norm": 1.7630791995749566,
      "learning_rate": 2.901685098497158e-06,
      "loss": 0.5349,
      "step": 5498
    },
    {
      "epoch": 6.730722154222766,
      "grad_norm": 1.492037638767875,
      "learning_rate": 2.9010525321460444e-06,
      "loss": 0.5163,
      "step": 5499
    },
    {
      "epoch": 6.731946144430845,
      "grad_norm": 1.1929301688068106,
      "learning_rate": 2.9004199394392414e-06,
      "loss": 0.551,
      "step": 5500
    },
    {
      "epoch": 6.7331701346389226,
      "grad_norm": 1.9769071325880443,
      "learning_rate": 2.8997873204183215e-06,
      "loss": 0.9174,
      "step": 5501
    },
    {
      "epoch": 6.734394124847001,
      "grad_norm": 2.4633897698757554,
      "learning_rate": 2.8991546751248566e-06,
      "loss": 0.8526,
      "step": 5502
    },
    {
      "epoch": 6.73561811505508,
      "grad_norm": 1.2250614516993776,
      "learning_rate": 2.898522003600424e-06,
      "loss": 0.8115,
      "step": 5503
    },
    {
      "epoch": 6.7368421052631575,
      "grad_norm": 1.113889784809873,
      "learning_rate": 2.8978893058865986e-06,
      "loss": 0.4103,
      "step": 5504
    },
    {
      "epoch": 6.738066095471236,
      "grad_norm": 1.0669494947258493,
      "learning_rate": 2.89725658202496e-06,
      "loss": 0.609,
      "step": 5505
    },
    {
      "epoch": 6.739290085679315,
      "grad_norm": 1.6488454308015879,
      "learning_rate": 2.8966238320570883e-06,
      "loss": 0.6513,
      "step": 5506
    },
    {
      "epoch": 6.740514075887393,
      "grad_norm": 1.2775021558575208,
      "learning_rate": 2.8959910560245653e-06,
      "loss": 0.2372,
      "step": 5507
    },
    {
      "epoch": 6.741738066095471,
      "grad_norm": 1.3807083664211375,
      "learning_rate": 2.895358253968975e-06,
      "loss": 0.6266,
      "step": 5508
    },
    {
      "epoch": 6.74296205630355,
      "grad_norm": 1.5308071223081277,
      "learning_rate": 2.8947254259319025e-06,
      "loss": 0.438,
      "step": 5509
    },
    {
      "epoch": 6.7441860465116275,
      "grad_norm": 1.8845323303160175,
      "learning_rate": 2.894092571954934e-06,
      "loss": 0.4424,
      "step": 5510
    },
    {
      "epoch": 6.745410036719706,
      "grad_norm": 2.1992188258494703,
      "learning_rate": 2.893459692079661e-06,
      "loss": 0.9594,
      "step": 5511
    },
    {
      "epoch": 6.746634026927785,
      "grad_norm": 1.7386594934876363,
      "learning_rate": 2.892826786347671e-06,
      "loss": 0.4589,
      "step": 5512
    },
    {
      "epoch": 6.747858017135863,
      "grad_norm": 1.4867070782167986,
      "learning_rate": 2.892193854800558e-06,
      "loss": 1.1218,
      "step": 5513
    },
    {
      "epoch": 6.749082007343941,
      "grad_norm": 1.571551289082583,
      "learning_rate": 2.891560897479916e-06,
      "loss": 0.5037,
      "step": 5514
    },
    {
      "epoch": 6.75030599755202,
      "grad_norm": 1.7238627135306095,
      "learning_rate": 2.8909279144273395e-06,
      "loss": 1.0216,
      "step": 5515
    },
    {
      "epoch": 6.751529987760098,
      "grad_norm": 1.4780157052148044,
      "learning_rate": 2.8902949056844265e-06,
      "loss": 1.558,
      "step": 5516
    },
    {
      "epoch": 6.752753977968176,
      "grad_norm": 1.0489637450628029,
      "learning_rate": 2.889661871292776e-06,
      "loss": 0.6966,
      "step": 5517
    },
    {
      "epoch": 6.753977968176255,
      "grad_norm": 2.694992085034109,
      "learning_rate": 2.889028811293989e-06,
      "loss": 0.3871,
      "step": 5518
    },
    {
      "epoch": 6.755201958384333,
      "grad_norm": 1.5168854147992572,
      "learning_rate": 2.888395725729666e-06,
      "loss": 1.4027,
      "step": 5519
    },
    {
      "epoch": 6.756425948592411,
      "grad_norm": 1.30220338816623,
      "learning_rate": 2.8877626146414147e-06,
      "loss": 0.7329,
      "step": 5520
    },
    {
      "epoch": 6.75764993880049,
      "grad_norm": 1.5936273219399018,
      "learning_rate": 2.8871294780708364e-06,
      "loss": 0.5519,
      "step": 5521
    },
    {
      "epoch": 6.758873929008568,
      "grad_norm": 1.9881903207933318,
      "learning_rate": 2.8864963160595418e-06,
      "loss": 0.395,
      "step": 5522
    },
    {
      "epoch": 6.760097919216646,
      "grad_norm": 2.3256700234874628,
      "learning_rate": 2.8858631286491388e-06,
      "loss": 0.6122,
      "step": 5523
    },
    {
      "epoch": 6.7613219094247246,
      "grad_norm": 1.5894906512465472,
      "learning_rate": 2.885229915881238e-06,
      "loss": 0.4179,
      "step": 5524
    },
    {
      "epoch": 6.762545899632803,
      "grad_norm": 1.140624848690117,
      "learning_rate": 2.8845966777974514e-06,
      "loss": 0.6463,
      "step": 5525
    },
    {
      "epoch": 6.763769889840881,
      "grad_norm": 1.292258199720588,
      "learning_rate": 2.8839634144393953e-06,
      "loss": 0.5709,
      "step": 5526
    },
    {
      "epoch": 6.7649938800489595,
      "grad_norm": 1.1726544701905492,
      "learning_rate": 2.883330125848683e-06,
      "loss": 0.6946,
      "step": 5527
    },
    {
      "epoch": 6.766217870257038,
      "grad_norm": 2.101987659938037,
      "learning_rate": 2.882696812066932e-06,
      "loss": 1.0414,
      "step": 5528
    },
    {
      "epoch": 6.767441860465116,
      "grad_norm": 1.3925568247915026,
      "learning_rate": 2.882063473135763e-06,
      "loss": 0.5688,
      "step": 5529
    },
    {
      "epoch": 6.7686658506731945,
      "grad_norm": 2.8711317697730587,
      "learning_rate": 2.881430109096795e-06,
      "loss": 0.3734,
      "step": 5530
    },
    {
      "epoch": 6.769889840881273,
      "grad_norm": 2.3068818995491998,
      "learning_rate": 2.8807967199916508e-06,
      "loss": 0.7609,
      "step": 5531
    },
    {
      "epoch": 6.771113831089352,
      "grad_norm": 2.2119229857763325,
      "learning_rate": 2.8801633058619553e-06,
      "loss": 0.4917,
      "step": 5532
    },
    {
      "epoch": 6.7723378212974294,
      "grad_norm": 1.3587437601770174,
      "learning_rate": 2.879529866749334e-06,
      "loss": 0.3765,
      "step": 5533
    },
    {
      "epoch": 6.773561811505508,
      "grad_norm": 1.6662838264223658,
      "learning_rate": 2.878896402695412e-06,
      "loss": 1.1288,
      "step": 5534
    },
    {
      "epoch": 6.774785801713586,
      "grad_norm": 1.5381589647662035,
      "learning_rate": 2.878262913741822e-06,
      "loss": 0.3143,
      "step": 5535
    },
    {
      "epoch": 6.776009791921664,
      "grad_norm": 1.2113176917642094,
      "learning_rate": 2.87762939993019e-06,
      "loss": 0.7092,
      "step": 5536
    },
    {
      "epoch": 6.777233782129743,
      "grad_norm": 1.6666538096186447,
      "learning_rate": 2.876995861302152e-06,
      "loss": 0.6003,
      "step": 5537
    },
    {
      "epoch": 6.778457772337822,
      "grad_norm": 1.9984116802527854,
      "learning_rate": 2.87636229789934e-06,
      "loss": 0.9455,
      "step": 5538
    },
    {
      "epoch": 6.779681762545899,
      "grad_norm": 2.042764625334225,
      "learning_rate": 2.8757287097633903e-06,
      "loss": 0.5823,
      "step": 5539
    },
    {
      "epoch": 6.780905752753978,
      "grad_norm": 1.5944089415693725,
      "learning_rate": 2.8750950969359394e-06,
      "loss": 0.5314,
      "step": 5540
    },
    {
      "epoch": 6.782129742962057,
      "grad_norm": 0.909134969934406,
      "learning_rate": 2.874461459458626e-06,
      "loss": 0.5253,
      "step": 5541
    },
    {
      "epoch": 6.783353733170134,
      "grad_norm": 3.104198609857638,
      "learning_rate": 2.8738277973730903e-06,
      "loss": 0.4548,
      "step": 5542
    },
    {
      "epoch": 6.784577723378213,
      "grad_norm": 1.3730505978422332,
      "learning_rate": 2.8731941107209744e-06,
      "loss": 0.597,
      "step": 5543
    },
    {
      "epoch": 6.785801713586292,
      "grad_norm": 1.6106426423431808,
      "learning_rate": 2.872560399543922e-06,
      "loss": 1.5078,
      "step": 5544
    },
    {
      "epoch": 6.787025703794369,
      "grad_norm": 2.2294096730536332,
      "learning_rate": 2.8719266638835784e-06,
      "loss": 0.4211,
      "step": 5545
    },
    {
      "epoch": 6.788249694002448,
      "grad_norm": 2.5661905201238944,
      "learning_rate": 2.87129290378159e-06,
      "loss": 0.4292,
      "step": 5546
    },
    {
      "epoch": 6.7894736842105265,
      "grad_norm": 1.5694390956300333,
      "learning_rate": 2.870659119279605e-06,
      "loss": 0.9499,
      "step": 5547
    },
    {
      "epoch": 6.790697674418604,
      "grad_norm": 1.762397592582257,
      "learning_rate": 2.8700253104192743e-06,
      "loss": 0.6753,
      "step": 5548
    },
    {
      "epoch": 6.791921664626683,
      "grad_norm": 1.2192227787029608,
      "learning_rate": 2.8693914772422484e-06,
      "loss": 0.4167,
      "step": 5549
    },
    {
      "epoch": 6.7931456548347615,
      "grad_norm": 2.0348021446065467,
      "learning_rate": 2.868757619790181e-06,
      "loss": 0.4782,
      "step": 5550
    },
    {
      "epoch": 6.79436964504284,
      "grad_norm": 1.766855908897202,
      "learning_rate": 2.8681237381047268e-06,
      "loss": 0.5736,
      "step": 5551
    },
    {
      "epoch": 6.795593635250918,
      "grad_norm": 1.8174082994667973,
      "learning_rate": 2.8674898322275415e-06,
      "loss": 0.5579,
      "step": 5552
    },
    {
      "epoch": 6.7968176254589965,
      "grad_norm": 1.473994682999887,
      "learning_rate": 2.8668559022002847e-06,
      "loss": 1.176,
      "step": 5553
    },
    {
      "epoch": 6.798041615667074,
      "grad_norm": 1.2405617184968123,
      "learning_rate": 2.866221948064614e-06,
      "loss": 1.132,
      "step": 5554
    },
    {
      "epoch": 6.799265605875153,
      "grad_norm": 1.8038109729808949,
      "learning_rate": 2.865587969862192e-06,
      "loss": 0.4156,
      "step": 5555
    },
    {
      "epoch": 6.8004895960832314,
      "grad_norm": 1.2807488800389606,
      "learning_rate": 2.8649539676346806e-06,
      "loss": 1.0399,
      "step": 5556
    },
    {
      "epoch": 6.80171358629131,
      "grad_norm": 2.157830705854145,
      "learning_rate": 2.8643199414237444e-06,
      "loss": 0.5333,
      "step": 5557
    },
    {
      "epoch": 6.802937576499388,
      "grad_norm": 1.5374685861606177,
      "learning_rate": 2.8636858912710492e-06,
      "loss": 0.4798,
      "step": 5558
    },
    {
      "epoch": 6.804161566707466,
      "grad_norm": 1.7072757834044476,
      "learning_rate": 2.8630518172182624e-06,
      "loss": 0.5616,
      "step": 5559
    },
    {
      "epoch": 6.805385556915545,
      "grad_norm": 1.763419050909636,
      "learning_rate": 2.8624177193070527e-06,
      "loss": 0.5572,
      "step": 5560
    },
    {
      "epoch": 6.806609547123623,
      "grad_norm": 2.2086367623723335,
      "learning_rate": 2.8617835975790914e-06,
      "loss": 0.4924,
      "step": 5561
    },
    {
      "epoch": 6.807833537331701,
      "grad_norm": 1.6581662187082307,
      "learning_rate": 2.8611494520760503e-06,
      "loss": 0.5681,
      "step": 5562
    },
    {
      "epoch": 6.80905752753978,
      "grad_norm": 1.2912542498895163,
      "learning_rate": 2.8605152828396025e-06,
      "loss": 0.6669,
      "step": 5563
    },
    {
      "epoch": 6.810281517747858,
      "grad_norm": 1.2773360113086716,
      "learning_rate": 2.8598810899114242e-06,
      "loss": 1.2456,
      "step": 5564
    },
    {
      "epoch": 6.811505507955936,
      "grad_norm": 1.7239189412695697,
      "learning_rate": 2.859246873333192e-06,
      "loss": 1.0573,
      "step": 5565
    },
    {
      "epoch": 6.812729498164015,
      "grad_norm": 1.409111801451644,
      "learning_rate": 2.8586126331465837e-06,
      "loss": 1.1864,
      "step": 5566
    },
    {
      "epoch": 6.813953488372093,
      "grad_norm": 2.4140424828853235,
      "learning_rate": 2.8579783693932793e-06,
      "loss": 0.4393,
      "step": 5567
    },
    {
      "epoch": 6.815177478580171,
      "grad_norm": 1.7893217408104554,
      "learning_rate": 2.857344082114961e-06,
      "loss": 1.1748,
      "step": 5568
    },
    {
      "epoch": 6.81640146878825,
      "grad_norm": 1.4942723152045594,
      "learning_rate": 2.856709771353311e-06,
      "loss": 0.5438,
      "step": 5569
    },
    {
      "epoch": 6.817625458996328,
      "grad_norm": 1.6136583237999798,
      "learning_rate": 2.8560754371500144e-06,
      "loss": 1.0412,
      "step": 5570
    },
    {
      "epoch": 6.818849449204406,
      "grad_norm": 1.0855419495397352,
      "learning_rate": 2.855441079546757e-06,
      "loss": 0.533,
      "step": 5571
    },
    {
      "epoch": 6.820073439412485,
      "grad_norm": 1.1226668682009042,
      "learning_rate": 2.854806698585226e-06,
      "loss": 0.775,
      "step": 5572
    },
    {
      "epoch": 6.821297429620563,
      "grad_norm": 1.7615917203309266,
      "learning_rate": 2.854172294307112e-06,
      "loss": 0.6564,
      "step": 5573
    },
    {
      "epoch": 6.822521419828641,
      "grad_norm": 1.9358701893859358,
      "learning_rate": 2.8535378667541048e-06,
      "loss": 0.6999,
      "step": 5574
    },
    {
      "epoch": 6.82374541003672,
      "grad_norm": 0.9561991426415091,
      "learning_rate": 2.8529034159678957e-06,
      "loss": 0.5384,
      "step": 5575
    },
    {
      "epoch": 6.8249694002447985,
      "grad_norm": 1.222447857645459,
      "learning_rate": 2.8522689419901808e-06,
      "loss": 0.6322,
      "step": 5576
    },
    {
      "epoch": 6.826193390452876,
      "grad_norm": 1.3817302873636137,
      "learning_rate": 2.851634444862653e-06,
      "loss": 0.5453,
      "step": 5577
    },
    {
      "epoch": 6.827417380660955,
      "grad_norm": 1.5789549227460191,
      "learning_rate": 2.85099992462701e-06,
      "loss": 0.3586,
      "step": 5578
    },
    {
      "epoch": 6.828641370869033,
      "grad_norm": 2.151668502742587,
      "learning_rate": 2.850365381324951e-06,
      "loss": 0.2601,
      "step": 5579
    },
    {
      "epoch": 6.829865361077111,
      "grad_norm": 1.1444473203719696,
      "learning_rate": 2.8497308149981743e-06,
      "loss": 0.3236,
      "step": 5580
    },
    {
      "epoch": 6.83108935128519,
      "grad_norm": 1.411328254441265,
      "learning_rate": 2.8490962256883825e-06,
      "loss": 0.3274,
      "step": 5581
    },
    {
      "epoch": 6.832313341493268,
      "grad_norm": 1.7457198940619485,
      "learning_rate": 2.8484616134372777e-06,
      "loss": 0.4341,
      "step": 5582
    },
    {
      "epoch": 6.833537331701346,
      "grad_norm": 1.5333463904461255,
      "learning_rate": 2.8478269782865647e-06,
      "loss": 1.8214,
      "step": 5583
    },
    {
      "epoch": 6.834761321909425,
      "grad_norm": 1.4720810898924692,
      "learning_rate": 2.8471923202779485e-06,
      "loss": 1.4728,
      "step": 5584
    },
    {
      "epoch": 6.835985312117503,
      "grad_norm": 2.592203672423223,
      "learning_rate": 2.8465576394531387e-06,
      "loss": 0.6075,
      "step": 5585
    },
    {
      "epoch": 6.837209302325581,
      "grad_norm": 2.215516301459949,
      "learning_rate": 2.845922935853841e-06,
      "loss": 0.6339,
      "step": 5586
    },
    {
      "epoch": 6.83843329253366,
      "grad_norm": 2.4041657301943355,
      "learning_rate": 2.845288209521768e-06,
      "loss": 0.6214,
      "step": 5587
    },
    {
      "epoch": 6.839657282741738,
      "grad_norm": 1.3649307736723348,
      "learning_rate": 2.8446534604986316e-06,
      "loss": 1.4189,
      "step": 5588
    },
    {
      "epoch": 6.840881272949816,
      "grad_norm": 0.9574641350600663,
      "learning_rate": 2.844018688826144e-06,
      "loss": 0.654,
      "step": 5589
    },
    {
      "epoch": 6.842105263157895,
      "grad_norm": 1.5451458449232431,
      "learning_rate": 2.8433838945460207e-06,
      "loss": 0.3156,
      "step": 5590
    },
    {
      "epoch": 6.843329253365973,
      "grad_norm": 1.7067177890896719,
      "learning_rate": 2.8427490776999782e-06,
      "loss": 0.9798,
      "step": 5591
    },
    {
      "epoch": 6.844553243574051,
      "grad_norm": 1.4948734450041181,
      "learning_rate": 2.842114238329734e-06,
      "loss": 0.537,
      "step": 5592
    },
    {
      "epoch": 6.84577723378213,
      "grad_norm": 2.129204498136481,
      "learning_rate": 2.8414793764770066e-06,
      "loss": 0.4402,
      "step": 5593
    },
    {
      "epoch": 6.847001223990208,
      "grad_norm": 1.3930568137526673,
      "learning_rate": 2.8408444921835186e-06,
      "loss": 1.2634,
      "step": 5594
    },
    {
      "epoch": 6.848225214198287,
      "grad_norm": 1.9202199638406743,
      "learning_rate": 2.840209585490991e-06,
      "loss": 0.5557,
      "step": 5595
    },
    {
      "epoch": 6.849449204406365,
      "grad_norm": 2.0247105830522703,
      "learning_rate": 2.8395746564411473e-06,
      "loss": 0.5018,
      "step": 5596
    },
    {
      "epoch": 6.850673194614443,
      "grad_norm": 1.2873235442825794,
      "learning_rate": 2.838939705075714e-06,
      "loss": 0.3714,
      "step": 5597
    },
    {
      "epoch": 6.851897184822521,
      "grad_norm": 1.2695055235056605,
      "learning_rate": 2.838304731436417e-06,
      "loss": 0.2878,
      "step": 5598
    },
    {
      "epoch": 6.8531211750306,
      "grad_norm": 1.0802077898535527,
      "learning_rate": 2.837669735564984e-06,
      "loss": 0.857,
      "step": 5599
    },
    {
      "epoch": 6.854345165238678,
      "grad_norm": 1.563819332121256,
      "learning_rate": 2.837034717503145e-06,
      "loss": 0.4073,
      "step": 5600
    },
    {
      "epoch": 6.855569155446757,
      "grad_norm": 2.5685221140215315,
      "learning_rate": 2.83639967729263e-06,
      "loss": 0.7211,
      "step": 5601
    },
    {
      "epoch": 6.8567931456548346,
      "grad_norm": 1.2740984068263674,
      "learning_rate": 2.835764614975174e-06,
      "loss": 0.5242,
      "step": 5602
    },
    {
      "epoch": 6.858017135862913,
      "grad_norm": 1.6622183366915677,
      "learning_rate": 2.835129530592509e-06,
      "loss": 0.4811,
      "step": 5603
    },
    {
      "epoch": 6.859241126070992,
      "grad_norm": 2.0154729641193305,
      "learning_rate": 2.83449442418637e-06,
      "loss": 0.4418,
      "step": 5604
    },
    {
      "epoch": 6.8604651162790695,
      "grad_norm": 1.401774333585441,
      "learning_rate": 2.833859295798495e-06,
      "loss": 0.7093,
      "step": 5605
    },
    {
      "epoch": 6.861689106487148,
      "grad_norm": 1.4757713740131093,
      "learning_rate": 2.833224145470622e-06,
      "loss": 0.6489,
      "step": 5606
    },
    {
      "epoch": 6.862913096695227,
      "grad_norm": 1.462399271457787,
      "learning_rate": 2.8325889732444913e-06,
      "loss": 0.944,
      "step": 5607
    },
    {
      "epoch": 6.8641370869033045,
      "grad_norm": 1.1416131305379702,
      "learning_rate": 2.831953779161843e-06,
      "loss": 0.5225,
      "step": 5608
    },
    {
      "epoch": 6.865361077111383,
      "grad_norm": 2.542894341869147,
      "learning_rate": 2.8313185632644204e-06,
      "loss": 0.3781,
      "step": 5609
    },
    {
      "epoch": 6.866585067319462,
      "grad_norm": 1.103869734737693,
      "learning_rate": 2.830683325593966e-06,
      "loss": 0.4179,
      "step": 5610
    },
    {
      "epoch": 6.8678090575275395,
      "grad_norm": 1.8955038725499673,
      "learning_rate": 2.8300480661922274e-06,
      "loss": 0.3973,
      "step": 5611
    },
    {
      "epoch": 6.869033047735618,
      "grad_norm": 2.1629217074522002,
      "learning_rate": 2.829412785100951e-06,
      "loss": 0.5431,
      "step": 5612
    },
    {
      "epoch": 6.870257037943697,
      "grad_norm": 1.806640460690669,
      "learning_rate": 2.828777482361885e-06,
      "loss": 0.5942,
      "step": 5613
    },
    {
      "epoch": 6.871481028151774,
      "grad_norm": 1.5896452251647475,
      "learning_rate": 2.8281421580167784e-06,
      "loss": 0.7486,
      "step": 5614
    },
    {
      "epoch": 6.872705018359853,
      "grad_norm": 1.1884174332178827,
      "learning_rate": 2.8275068121073828e-06,
      "loss": 0.4513,
      "step": 5615
    },
    {
      "epoch": 6.873929008567932,
      "grad_norm": 1.5484639936027278,
      "learning_rate": 2.8268714446754513e-06,
      "loss": 0.4439,
      "step": 5616
    },
    {
      "epoch": 6.875152998776009,
      "grad_norm": 1.2693783182580014,
      "learning_rate": 2.8262360557627373e-06,
      "loss": 0.685,
      "step": 5617
    },
    {
      "epoch": 6.876376988984088,
      "grad_norm": 1.385174129542064,
      "learning_rate": 2.8256006454109963e-06,
      "loss": 0.6424,
      "step": 5618
    },
    {
      "epoch": 6.877600979192167,
      "grad_norm": 1.503480466250667,
      "learning_rate": 2.824965213661985e-06,
      "loss": 0.4484,
      "step": 5619
    },
    {
      "epoch": 6.878824969400245,
      "grad_norm": 1.4160362091416847,
      "learning_rate": 2.824329760557462e-06,
      "loss": 0.6639,
      "step": 5620
    },
    {
      "epoch": 6.880048959608323,
      "grad_norm": 1.1593963561084202,
      "learning_rate": 2.8236942861391866e-06,
      "loss": 0.5689,
      "step": 5621
    },
    {
      "epoch": 6.881272949816402,
      "grad_norm": 1.5986474443815986,
      "learning_rate": 2.8230587904489198e-06,
      "loss": 0.7542,
      "step": 5622
    },
    {
      "epoch": 6.882496940024479,
      "grad_norm": 1.0606291595186266,
      "learning_rate": 2.8224232735284252e-06,
      "loss": 0.8528,
      "step": 5623
    },
    {
      "epoch": 6.883720930232558,
      "grad_norm": 2.0414041826212985,
      "learning_rate": 2.8217877354194646e-06,
      "loss": 0.6062,
      "step": 5624
    },
    {
      "epoch": 6.8849449204406366,
      "grad_norm": 1.5628322833357626,
      "learning_rate": 2.821152176163804e-06,
      "loss": 0.5227,
      "step": 5625
    },
    {
      "epoch": 6.886168910648715,
      "grad_norm": 2.033286739920621,
      "learning_rate": 2.8205165958032114e-06,
      "loss": 0.4747,
      "step": 5626
    },
    {
      "epoch": 6.887392900856793,
      "grad_norm": 1.709131282192089,
      "learning_rate": 2.8198809943794536e-06,
      "loss": 0.4376,
      "step": 5627
    },
    {
      "epoch": 6.8886168910648715,
      "grad_norm": 1.2018914774896028,
      "learning_rate": 2.819245371934299e-06,
      "loss": 0.3833,
      "step": 5628
    },
    {
      "epoch": 6.88984088127295,
      "grad_norm": 0.7315140730800727,
      "learning_rate": 2.8186097285095205e-06,
      "loss": 0.4407,
      "step": 5629
    },
    {
      "epoch": 6.891064871481028,
      "grad_norm": 1.8206093292659014,
      "learning_rate": 2.8179740641468877e-06,
      "loss": 0.5082,
      "step": 5630
    },
    {
      "epoch": 6.8922888616891065,
      "grad_norm": 1.4726296290721987,
      "learning_rate": 2.8173383788881765e-06,
      "loss": 1.5164,
      "step": 5631
    },
    {
      "epoch": 6.893512851897185,
      "grad_norm": 2.2236524484736786,
      "learning_rate": 2.81670267277516e-06,
      "loss": 0.5094,
      "step": 5632
    },
    {
      "epoch": 6.894736842105263,
      "grad_norm": 2.5426804695466525,
      "learning_rate": 2.816066945849616e-06,
      "loss": 0.5464,
      "step": 5633
    },
    {
      "epoch": 6.8959608323133414,
      "grad_norm": 1.3481794610529918,
      "learning_rate": 2.815431198153321e-06,
      "loss": 0.9403,
      "step": 5634
    },
    {
      "epoch": 6.89718482252142,
      "grad_norm": 1.5207084979225358,
      "learning_rate": 2.814795429728055e-06,
      "loss": 0.4273,
      "step": 5635
    },
    {
      "epoch": 6.898408812729498,
      "grad_norm": 2.961267542614461,
      "learning_rate": 2.814159640615597e-06,
      "loss": 0.3902,
      "step": 5636
    },
    {
      "epoch": 6.899632802937576,
      "grad_norm": 1.321469085314527,
      "learning_rate": 2.8135238308577294e-06,
      "loss": 0.983,
      "step": 5637
    },
    {
      "epoch": 6.900856793145655,
      "grad_norm": 1.411216602371665,
      "learning_rate": 2.8128880004962354e-06,
      "loss": 1.189,
      "step": 5638
    },
    {
      "epoch": 6.902080783353734,
      "grad_norm": 1.5209734737620877,
      "learning_rate": 2.8122521495728993e-06,
      "loss": 0.5049,
      "step": 5639
    },
    {
      "epoch": 6.903304773561811,
      "grad_norm": 1.6266607480507114,
      "learning_rate": 2.8116162781295074e-06,
      "loss": 0.5653,
      "step": 5640
    },
    {
      "epoch": 6.90452876376989,
      "grad_norm": 2.074697253979892,
      "learning_rate": 2.8109803862078463e-06,
      "loss": 0.6308,
      "step": 5641
    },
    {
      "epoch": 6.905752753977968,
      "grad_norm": 1.4722037906119299,
      "learning_rate": 2.810344473849704e-06,
      "loss": 0.7524,
      "step": 5642
    },
    {
      "epoch": 6.906976744186046,
      "grad_norm": 0.9497172256863463,
      "learning_rate": 2.80970854109687e-06,
      "loss": 0.5778,
      "step": 5643
    },
    {
      "epoch": 6.908200734394125,
      "grad_norm": 2.144475825860658,
      "learning_rate": 2.8090725879911375e-06,
      "loss": 0.6466,
      "step": 5644
    },
    {
      "epoch": 6.909424724602204,
      "grad_norm": 1.628375054206427,
      "learning_rate": 2.808436614574297e-06,
      "loss": 0.606,
      "step": 5645
    },
    {
      "epoch": 6.910648714810281,
      "grad_norm": 1.5280337361002931,
      "learning_rate": 2.807800620888143e-06,
      "loss": 0.4826,
      "step": 5646
    },
    {
      "epoch": 6.91187270501836,
      "grad_norm": 1.1099564867713991,
      "learning_rate": 2.8071646069744713e-06,
      "loss": 0.6397,
      "step": 5647
    },
    {
      "epoch": 6.9130966952264385,
      "grad_norm": 1.1603910321236541,
      "learning_rate": 2.8065285728750774e-06,
      "loss": 0.7302,
      "step": 5648
    },
    {
      "epoch": 6.914320685434516,
      "grad_norm": 1.1903171311391147,
      "learning_rate": 2.8058925186317585e-06,
      "loss": 0.6879,
      "step": 5649
    },
    {
      "epoch": 6.915544675642595,
      "grad_norm": 0.9832356004428477,
      "learning_rate": 2.8052564442863162e-06,
      "loss": 0.5862,
      "step": 5650
    },
    {
      "epoch": 6.9167686658506735,
      "grad_norm": 1.423661062851523,
      "learning_rate": 2.8046203498805486e-06,
      "loss": 0.7372,
      "step": 5651
    },
    {
      "epoch": 6.917992656058751,
      "grad_norm": 1.2716321302258267,
      "learning_rate": 2.8039842354562584e-06,
      "loss": 0.5774,
      "step": 5652
    },
    {
      "epoch": 6.91921664626683,
      "grad_norm": 1.247396849369937,
      "learning_rate": 2.8033481010552492e-06,
      "loss": 1.1678,
      "step": 5653
    },
    {
      "epoch": 6.9204406364749085,
      "grad_norm": 1.4306493095549107,
      "learning_rate": 2.8027119467193243e-06,
      "loss": 0.8759,
      "step": 5654
    },
    {
      "epoch": 6.921664626682986,
      "grad_norm": 1.994299187587238,
      "learning_rate": 2.8020757724902902e-06,
      "loss": 1.0572,
      "step": 5655
    },
    {
      "epoch": 6.922888616891065,
      "grad_norm": 1.5424570527700001,
      "learning_rate": 2.8014395784099535e-06,
      "loss": 0.65,
      "step": 5656
    },
    {
      "epoch": 6.9241126070991434,
      "grad_norm": 2.2896733074433167,
      "learning_rate": 2.8008033645201226e-06,
      "loss": 0.6201,
      "step": 5657
    },
    {
      "epoch": 6.925336597307221,
      "grad_norm": 2.3345054837354673,
      "learning_rate": 2.8001671308626072e-06,
      "loss": 0.9247,
      "step": 5658
    },
    {
      "epoch": 6.9265605875153,
      "grad_norm": 1.2501600181123813,
      "learning_rate": 2.799530877479218e-06,
      "loss": 0.4548,
      "step": 5659
    },
    {
      "epoch": 6.927784577723378,
      "grad_norm": 1.1776687445336698,
      "learning_rate": 2.798894604411767e-06,
      "loss": 0.9079,
      "step": 5660
    },
    {
      "epoch": 6.929008567931456,
      "grad_norm": 1.6765667558287263,
      "learning_rate": 2.798258311702069e-06,
      "loss": 0.4492,
      "step": 5661
    },
    {
      "epoch": 6.930232558139535,
      "grad_norm": 1.6844512402481335,
      "learning_rate": 2.797621999391938e-06,
      "loss": 0.4179,
      "step": 5662
    },
    {
      "epoch": 6.931456548347613,
      "grad_norm": 0.9554788384749119,
      "learning_rate": 2.7969856675231892e-06,
      "loss": 0.5953,
      "step": 5663
    },
    {
      "epoch": 6.932680538555692,
      "grad_norm": 1.0187225692161708,
      "learning_rate": 2.7963493161376416e-06,
      "loss": 0.4889,
      "step": 5664
    },
    {
      "epoch": 6.93390452876377,
      "grad_norm": 1.504079583591393,
      "learning_rate": 2.7957129452771124e-06,
      "loss": 0.3289,
      "step": 5665
    },
    {
      "epoch": 6.935128518971848,
      "grad_norm": 0.705651972375692,
      "learning_rate": 2.795076554983423e-06,
      "loss": 0.2258,
      "step": 5666
    },
    {
      "epoch": 6.936352509179926,
      "grad_norm": 1.2209188390346009,
      "learning_rate": 2.7944401452983933e-06,
      "loss": 1.5845,
      "step": 5667
    },
    {
      "epoch": 6.937576499388005,
      "grad_norm": 1.039276224584157,
      "learning_rate": 2.7938037162638467e-06,
      "loss": 0.5855,
      "step": 5668
    },
    {
      "epoch": 6.938800489596083,
      "grad_norm": 1.7470168816925766,
      "learning_rate": 2.7931672679216055e-06,
      "loss": 0.6362,
      "step": 5669
    },
    {
      "epoch": 6.940024479804162,
      "grad_norm": 1.6136024828947029,
      "learning_rate": 2.7925308003134965e-06,
      "loss": 0.6726,
      "step": 5670
    },
    {
      "epoch": 6.94124847001224,
      "grad_norm": 2.8119470035877248,
      "learning_rate": 2.791894313481346e-06,
      "loss": 0.3785,
      "step": 5671
    },
    {
      "epoch": 6.942472460220318,
      "grad_norm": 1.4586942044300744,
      "learning_rate": 2.79125780746698e-06,
      "loss": 0.8718,
      "step": 5672
    },
    {
      "epoch": 6.943696450428397,
      "grad_norm": 2.311431681790971,
      "learning_rate": 2.790621282312228e-06,
      "loss": 1.2055,
      "step": 5673
    },
    {
      "epoch": 6.944920440636475,
      "grad_norm": 0.9864373083228465,
      "learning_rate": 2.789984738058921e-06,
      "loss": 0.4061,
      "step": 5674
    },
    {
      "epoch": 6.946144430844553,
      "grad_norm": 1.8878327347618347,
      "learning_rate": 2.7893481747488887e-06,
      "loss": 0.5095,
      "step": 5675
    },
    {
      "epoch": 6.947368421052632,
      "grad_norm": 2.4931241462147025,
      "learning_rate": 2.788711592423966e-06,
      "loss": 0.8309,
      "step": 5676
    },
    {
      "epoch": 6.94859241126071,
      "grad_norm": 1.08635165012773,
      "learning_rate": 2.7880749911259835e-06,
      "loss": 0.5723,
      "step": 5677
    },
    {
      "epoch": 6.949816401468788,
      "grad_norm": 1.165966432302402,
      "learning_rate": 2.7874383708967795e-06,
      "loss": 0.7778,
      "step": 5678
    },
    {
      "epoch": 6.951040391676867,
      "grad_norm": 1.077425618290188,
      "learning_rate": 2.7868017317781893e-06,
      "loss": 0.5074,
      "step": 5679
    },
    {
      "epoch": 6.9522643818849446,
      "grad_norm": 1.322945283083331,
      "learning_rate": 2.7861650738120483e-06,
      "loss": 1.1806,
      "step": 5680
    },
    {
      "epoch": 6.953488372093023,
      "grad_norm": 1.1650117298399576,
      "learning_rate": 2.785528397040198e-06,
      "loss": 1.3465,
      "step": 5681
    },
    {
      "epoch": 6.954712362301102,
      "grad_norm": 2.092081053163567,
      "learning_rate": 2.784891701504478e-06,
      "loss": 0.7386,
      "step": 5682
    },
    {
      "epoch": 6.95593635250918,
      "grad_norm": 1.6265146638502619,
      "learning_rate": 2.784254987246729e-06,
      "loss": 0.3867,
      "step": 5683
    },
    {
      "epoch": 6.957160342717258,
      "grad_norm": 0.9563106239619373,
      "learning_rate": 2.7836182543087925e-06,
      "loss": 0.3989,
      "step": 5684
    },
    {
      "epoch": 6.958384332925337,
      "grad_norm": 1.3121335740374447,
      "learning_rate": 2.782981502732515e-06,
      "loss": 1.133,
      "step": 5685
    },
    {
      "epoch": 6.9596083231334145,
      "grad_norm": 1.853527352489472,
      "learning_rate": 2.7823447325597385e-06,
      "loss": 0.5775,
      "step": 5686
    },
    {
      "epoch": 6.960832313341493,
      "grad_norm": 1.3518298250920942,
      "learning_rate": 2.7817079438323108e-06,
      "loss": 0.6364,
      "step": 5687
    },
    {
      "epoch": 6.962056303549572,
      "grad_norm": 1.9995846681058083,
      "learning_rate": 2.7810711365920797e-06,
      "loss": 0.4477,
      "step": 5688
    },
    {
      "epoch": 6.96328029375765,
      "grad_norm": 1.6077324491211626,
      "learning_rate": 2.7804343108808925e-06,
      "loss": 0.4776,
      "step": 5689
    },
    {
      "epoch": 6.964504283965728,
      "grad_norm": 1.8115933211891238,
      "learning_rate": 2.779797466740599e-06,
      "loss": 0.631,
      "step": 5690
    },
    {
      "epoch": 6.965728274173807,
      "grad_norm": 1.1595684306073137,
      "learning_rate": 2.7791606042130527e-06,
      "loss": 0.5911,
      "step": 5691
    },
    {
      "epoch": 6.966952264381885,
      "grad_norm": 1.3859485479142015,
      "learning_rate": 2.778523723340103e-06,
      "loss": 1.4986,
      "step": 5692
    },
    {
      "epoch": 6.968176254589963,
      "grad_norm": 2.0279516233768367,
      "learning_rate": 2.7778868241636037e-06,
      "loss": 0.5236,
      "step": 5693
    },
    {
      "epoch": 6.969400244798042,
      "grad_norm": 1.2921969119899792,
      "learning_rate": 2.7772499067254116e-06,
      "loss": 0.5807,
      "step": 5694
    },
    {
      "epoch": 6.97062423500612,
      "grad_norm": 1.5267217629225196,
      "learning_rate": 2.7766129710673793e-06,
      "loss": 0.4332,
      "step": 5695
    },
    {
      "epoch": 6.971848225214198,
      "grad_norm": 1.1166284550278673,
      "learning_rate": 2.775976017231367e-06,
      "loss": 1.2112,
      "step": 5696
    },
    {
      "epoch": 6.973072215422277,
      "grad_norm": 1.8305482993659905,
      "learning_rate": 2.7753390452592315e-06,
      "loss": 1.1402,
      "step": 5697
    },
    {
      "epoch": 6.974296205630355,
      "grad_norm": 1.3097586202556044,
      "learning_rate": 2.7747020551928324e-06,
      "loss": 0.5906,
      "step": 5698
    },
    {
      "epoch": 6.975520195838433,
      "grad_norm": 2.081456878253612,
      "learning_rate": 2.774065047074029e-06,
      "loss": 0.5131,
      "step": 5699
    },
    {
      "epoch": 6.976744186046512,
      "grad_norm": 1.722744498888617,
      "learning_rate": 2.773428020944687e-06,
      "loss": 0.5964,
      "step": 5700
    },
    {
      "epoch": 6.97796817625459,
      "grad_norm": 0.9771996886752283,
      "learning_rate": 2.7727909768466643e-06,
      "loss": 0.6322,
      "step": 5701
    },
    {
      "epoch": 6.979192166462668,
      "grad_norm": 1.4519669875846333,
      "learning_rate": 2.7721539148218284e-06,
      "loss": 1.1009,
      "step": 5702
    },
    {
      "epoch": 6.9804161566707466,
      "grad_norm": 1.893936512748418,
      "learning_rate": 2.7715168349120448e-06,
      "loss": 0.4188,
      "step": 5703
    },
    {
      "epoch": 6.981640146878825,
      "grad_norm": 1.0380596268232443,
      "learning_rate": 2.770879737159178e-06,
      "loss": 0.5522,
      "step": 5704
    },
    {
      "epoch": 6.982864137086903,
      "grad_norm": 1.5654329586133502,
      "learning_rate": 2.770242621605097e-06,
      "loss": 0.4964,
      "step": 5705
    },
    {
      "epoch": 6.9840881272949815,
      "grad_norm": 1.0099856866704504,
      "learning_rate": 2.769605488291671e-06,
      "loss": 0.5704,
      "step": 5706
    },
    {
      "epoch": 6.98531211750306,
      "grad_norm": 1.4013700428713258,
      "learning_rate": 2.7689683372607683e-06,
      "loss": 0.6112,
      "step": 5707
    },
    {
      "epoch": 6.986536107711139,
      "grad_norm": 2.0375718517328227,
      "learning_rate": 2.7683311685542626e-06,
      "loss": 0.5716,
      "step": 5708
    },
    {
      "epoch": 6.9877600979192165,
      "grad_norm": 2.441580915229278,
      "learning_rate": 2.7676939822140242e-06,
      "loss": 0.3858,
      "step": 5709
    },
    {
      "epoch": 6.988984088127295,
      "grad_norm": 1.7775613873638167,
      "learning_rate": 2.7670567782819276e-06,
      "loss": 0.4033,
      "step": 5710
    },
    {
      "epoch": 6.990208078335373,
      "grad_norm": 1.4286067491090981,
      "learning_rate": 2.766419556799847e-06,
      "loss": 0.5067,
      "step": 5711
    },
    {
      "epoch": 6.9914320685434515,
      "grad_norm": 2.3175541776858264,
      "learning_rate": 2.7657823178096593e-06,
      "loss": 0.9759,
      "step": 5712
    },
    {
      "epoch": 6.99265605875153,
      "grad_norm": 1.3174463191761763,
      "learning_rate": 2.7651450613532406e-06,
      "loss": 0.8898,
      "step": 5713
    },
    {
      "epoch": 6.993880048959609,
      "grad_norm": 2.6627553236610284,
      "learning_rate": 2.7645077874724686e-06,
      "loss": 0.4901,
      "step": 5714
    },
    {
      "epoch": 6.995104039167686,
      "grad_norm": 1.307970970516204,
      "learning_rate": 2.7638704962092235e-06,
      "loss": 1.2953,
      "step": 5715
    },
    {
      "epoch": 6.996328029375765,
      "grad_norm": 1.0771487535088442,
      "learning_rate": 2.7632331876053853e-06,
      "loss": 0.6143,
      "step": 5716
    },
    {
      "epoch": 6.997552019583844,
      "grad_norm": 1.7717526762143052,
      "learning_rate": 2.762595861702836e-06,
      "loss": 0.7555,
      "step": 5717
    },
    {
      "epoch": 6.998776009791921,
      "grad_norm": 1.6528459547647463,
      "learning_rate": 2.7619585185434577e-06,
      "loss": 0.6436,
      "step": 5718
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.8407809597322985,
      "learning_rate": 2.761321158169134e-06,
      "loss": 0.4557,
      "step": 5719
    },
    {
      "epoch": 7.001223990208079,
      "grad_norm": 1.3936232359981666,
      "learning_rate": 2.7606837806217507e-06,
      "loss": 0.5045,
      "step": 5720
    },
    {
      "epoch": 7.002447980416156,
      "grad_norm": 1.9081785522443069,
      "learning_rate": 2.7600463859431935e-06,
      "loss": 0.696,
      "step": 5721
    },
    {
      "epoch": 7.003671970624235,
      "grad_norm": 1.86631256931455,
      "learning_rate": 2.7594089741753497e-06,
      "loss": 0.8365,
      "step": 5722
    },
    {
      "epoch": 7.004895960832314,
      "grad_norm": 1.2755530683852594,
      "learning_rate": 2.758771545360108e-06,
      "loss": 0.8842,
      "step": 5723
    },
    {
      "epoch": 7.006119951040391,
      "grad_norm": 1.5822070272052304,
      "learning_rate": 2.7581340995393576e-06,
      "loss": 1.1843,
      "step": 5724
    },
    {
      "epoch": 7.00734394124847,
      "grad_norm": 1.3173955583038957,
      "learning_rate": 2.7574966367549875e-06,
      "loss": 0.2338,
      "step": 5725
    },
    {
      "epoch": 7.0085679314565485,
      "grad_norm": 1.4004833582546488,
      "learning_rate": 2.7568591570488922e-06,
      "loss": 0.6541,
      "step": 5726
    },
    {
      "epoch": 7.009791921664626,
      "grad_norm": 1.447144346013208,
      "learning_rate": 2.7562216604629626e-06,
      "loss": 0.7426,
      "step": 5727
    },
    {
      "epoch": 7.011015911872705,
      "grad_norm": 1.8803523907042863,
      "learning_rate": 2.7555841470390937e-06,
      "loss": 0.462,
      "step": 5728
    },
    {
      "epoch": 7.0122399020807835,
      "grad_norm": 1.4348414073562572,
      "learning_rate": 2.7549466168191797e-06,
      "loss": 1.6506,
      "step": 5729
    },
    {
      "epoch": 7.013463892288861,
      "grad_norm": 2.2653560519163563,
      "learning_rate": 2.7543090698451168e-06,
      "loss": 0.684,
      "step": 5730
    },
    {
      "epoch": 7.01468788249694,
      "grad_norm": 1.3654100184497586,
      "learning_rate": 2.7536715061588028e-06,
      "loss": 0.492,
      "step": 5731
    },
    {
      "epoch": 7.0159118727050185,
      "grad_norm": 1.3452738442029217,
      "learning_rate": 2.7530339258021354e-06,
      "loss": 2.2912,
      "step": 5732
    },
    {
      "epoch": 7.017135862913097,
      "grad_norm": 1.1645464442942937,
      "learning_rate": 2.752396328817015e-06,
      "loss": 0.6895,
      "step": 5733
    },
    {
      "epoch": 7.018359853121175,
      "grad_norm": 1.6113214323656389,
      "learning_rate": 2.75175871524534e-06,
      "loss": 1.0966,
      "step": 5734
    },
    {
      "epoch": 7.0195838433292534,
      "grad_norm": 1.1610481549853344,
      "learning_rate": 2.751121085129015e-06,
      "loss": 0.7724,
      "step": 5735
    },
    {
      "epoch": 7.020807833537332,
      "grad_norm": 2.093481195984658,
      "learning_rate": 2.75048343850994e-06,
      "loss": 0.5418,
      "step": 5736
    },
    {
      "epoch": 7.02203182374541,
      "grad_norm": 2.4098117403833563,
      "learning_rate": 2.74984577543002e-06,
      "loss": 1.0422,
      "step": 5737
    },
    {
      "epoch": 7.023255813953488,
      "grad_norm": 0.9765361027707729,
      "learning_rate": 2.74920809593116e-06,
      "loss": 0.4784,
      "step": 5738
    },
    {
      "epoch": 7.024479804161567,
      "grad_norm": 1.1526812328428344,
      "learning_rate": 2.748570400055266e-06,
      "loss": 0.3773,
      "step": 5739
    },
    {
      "epoch": 7.025703794369645,
      "grad_norm": 1.1743774687069874,
      "learning_rate": 2.747932687844244e-06,
      "loss": 0.6403,
      "step": 5740
    },
    {
      "epoch": 7.026927784577723,
      "grad_norm": 1.5008926975731758,
      "learning_rate": 2.747294959340004e-06,
      "loss": 0.4337,
      "step": 5741
    },
    {
      "epoch": 7.028151774785802,
      "grad_norm": 1.1481467747500385,
      "learning_rate": 2.7466572145844526e-06,
      "loss": 0.6808,
      "step": 5742
    },
    {
      "epoch": 7.02937576499388,
      "grad_norm": 1.7680299683236922,
      "learning_rate": 2.746019453619502e-06,
      "loss": 0.9775,
      "step": 5743
    },
    {
      "epoch": 7.030599755201958,
      "grad_norm": 2.1993521413858903,
      "learning_rate": 2.7453816764870635e-06,
      "loss": 0.5715,
      "step": 5744
    },
    {
      "epoch": 7.031823745410037,
      "grad_norm": 1.7408233380372644,
      "learning_rate": 2.7447438832290472e-06,
      "loss": 0.4052,
      "step": 5745
    },
    {
      "epoch": 7.033047735618115,
      "grad_norm": 1.1310739777274708,
      "learning_rate": 2.744106073887369e-06,
      "loss": 0.7272,
      "step": 5746
    },
    {
      "epoch": 7.034271725826193,
      "grad_norm": 1.3316798823595026,
      "learning_rate": 2.743468248503943e-06,
      "loss": 0.501,
      "step": 5747
    },
    {
      "epoch": 7.035495716034272,
      "grad_norm": 1.8943219444810213,
      "learning_rate": 2.742830407120684e-06,
      "loss": 0.966,
      "step": 5748
    },
    {
      "epoch": 7.03671970624235,
      "grad_norm": 2.4968384597846263,
      "learning_rate": 2.742192549779508e-06,
      "loss": 0.707,
      "step": 5749
    },
    {
      "epoch": 7.037943696450428,
      "grad_norm": 1.508006557868597,
      "learning_rate": 2.7415546765223348e-06,
      "loss": 0.6423,
      "step": 5750
    },
    {
      "epoch": 7.039167686658507,
      "grad_norm": 1.1023914614340453,
      "learning_rate": 2.740916787391081e-06,
      "loss": 0.6931,
      "step": 5751
    },
    {
      "epoch": 7.0403916768665855,
      "grad_norm": 1.7197382424757552,
      "learning_rate": 2.7402788824276666e-06,
      "loss": 0.4326,
      "step": 5752
    },
    {
      "epoch": 7.041615667074663,
      "grad_norm": 1.6195769915097413,
      "learning_rate": 2.739640961674013e-06,
      "loss": 1.5709,
      "step": 5753
    },
    {
      "epoch": 7.042839657282742,
      "grad_norm": 1.8645274344379392,
      "learning_rate": 2.7390030251720424e-06,
      "loss": 0.6702,
      "step": 5754
    },
    {
      "epoch": 7.0440636474908205,
      "grad_norm": 2.0852137070919396,
      "learning_rate": 2.738365072963677e-06,
      "loss": 0.52,
      "step": 5755
    },
    {
      "epoch": 7.045287637698898,
      "grad_norm": 1.7589598655329923,
      "learning_rate": 2.73772710509084e-06,
      "loss": 0.4233,
      "step": 5756
    },
    {
      "epoch": 7.046511627906977,
      "grad_norm": 1.1172507436350694,
      "learning_rate": 2.7370891215954572e-06,
      "loss": 0.4873,
      "step": 5757
    },
    {
      "epoch": 7.0477356181150554,
      "grad_norm": 1.4946805867168533,
      "learning_rate": 2.736451122519454e-06,
      "loss": 0.4886,
      "step": 5758
    },
    {
      "epoch": 7.048959608323133,
      "grad_norm": 2.030585292545271,
      "learning_rate": 2.7358131079047586e-06,
      "loss": 0.5588,
      "step": 5759
    },
    {
      "epoch": 7.050183598531212,
      "grad_norm": 1.6193525253161094,
      "learning_rate": 2.735175077793296e-06,
      "loss": 0.4692,
      "step": 5760
    },
    {
      "epoch": 7.05140758873929,
      "grad_norm": 1.1418665943312767,
      "learning_rate": 2.734537032226999e-06,
      "loss": 0.5283,
      "step": 5761
    },
    {
      "epoch": 7.052631578947368,
      "grad_norm": 1.9771389968736977,
      "learning_rate": 2.733898971247795e-06,
      "loss": 0.3952,
      "step": 5762
    },
    {
      "epoch": 7.053855569155447,
      "grad_norm": 1.6324033415189536,
      "learning_rate": 2.7332608948976156e-06,
      "loss": 0.9971,
      "step": 5763
    },
    {
      "epoch": 7.055079559363525,
      "grad_norm": 2.7914129226149016,
      "learning_rate": 2.732622803218393e-06,
      "loss": 0.613,
      "step": 5764
    },
    {
      "epoch": 7.056303549571603,
      "grad_norm": 2.9666117003008337,
      "learning_rate": 2.7319846962520607e-06,
      "loss": 0.3603,
      "step": 5765
    },
    {
      "epoch": 7.057527539779682,
      "grad_norm": 1.2779417711179746,
      "learning_rate": 2.7313465740405515e-06,
      "loss": 0.8239,
      "step": 5766
    },
    {
      "epoch": 7.05875152998776,
      "grad_norm": 1.2079209765318863,
      "learning_rate": 2.730708436625802e-06,
      "loss": 0.5858,
      "step": 5767
    },
    {
      "epoch": 7.059975520195838,
      "grad_norm": 2.5068398972809938,
      "learning_rate": 2.7300702840497468e-06,
      "loss": 0.8193,
      "step": 5768
    },
    {
      "epoch": 7.061199510403917,
      "grad_norm": 2.3953955602104866,
      "learning_rate": 2.729432116354323e-06,
      "loss": 0.3867,
      "step": 5769
    },
    {
      "epoch": 7.062423500611995,
      "grad_norm": 1.8815610494355783,
      "learning_rate": 2.72879393358147e-06,
      "loss": 1.2113,
      "step": 5770
    },
    {
      "epoch": 7.063647490820073,
      "grad_norm": 1.2963416540082595,
      "learning_rate": 2.728155735773126e-06,
      "loss": 1.2845,
      "step": 5771
    },
    {
      "epoch": 7.064871481028152,
      "grad_norm": 1.4069806454930105,
      "learning_rate": 2.7275175229712304e-06,
      "loss": 0.6298,
      "step": 5772
    },
    {
      "epoch": 7.06609547123623,
      "grad_norm": 1.211515031542482,
      "learning_rate": 2.726879295217725e-06,
      "loss": 0.6859,
      "step": 5773
    },
    {
      "epoch": 7.067319461444308,
      "grad_norm": 1.21644057716479,
      "learning_rate": 2.7262410525545513e-06,
      "loss": 1.0965,
      "step": 5774
    },
    {
      "epoch": 7.068543451652387,
      "grad_norm": 1.0192296458061059,
      "learning_rate": 2.725602795023652e-06,
      "loss": 0.5154,
      "step": 5775
    },
    {
      "epoch": 7.069767441860465,
      "grad_norm": 1.351963391175913,
      "learning_rate": 2.724964522666973e-06,
      "loss": 0.5391,
      "step": 5776
    },
    {
      "epoch": 7.070991432068544,
      "grad_norm": 1.9329311037111725,
      "learning_rate": 2.724326235526456e-06,
      "loss": 0.4105,
      "step": 5777
    },
    {
      "epoch": 7.072215422276622,
      "grad_norm": 2.1154178307669653,
      "learning_rate": 2.723687933644049e-06,
      "loss": 0.4184,
      "step": 5778
    },
    {
      "epoch": 7.0734394124847,
      "grad_norm": 1.2904874668197714,
      "learning_rate": 2.7230496170616986e-06,
      "loss": 1.315,
      "step": 5779
    },
    {
      "epoch": 7.074663402692779,
      "grad_norm": 1.402154418545157,
      "learning_rate": 2.7224112858213523e-06,
      "loss": 1.2562,
      "step": 5780
    },
    {
      "epoch": 7.0758873929008566,
      "grad_norm": 1.3615501179704175,
      "learning_rate": 2.721772939964959e-06,
      "loss": 0.4416,
      "step": 5781
    },
    {
      "epoch": 7.077111383108935,
      "grad_norm": 1.3710532484153344,
      "learning_rate": 2.7211345795344683e-06,
      "loss": 0.5076,
      "step": 5782
    },
    {
      "epoch": 7.078335373317014,
      "grad_norm": 2.303215137439167,
      "learning_rate": 2.7204962045718313e-06,
      "loss": 0.4201,
      "step": 5783
    },
    {
      "epoch": 7.0795593635250915,
      "grad_norm": 2.5233256678333165,
      "learning_rate": 2.7198578151189988e-06,
      "loss": 0.4376,
      "step": 5784
    },
    {
      "epoch": 7.08078335373317,
      "grad_norm": 2.288407472573978,
      "learning_rate": 2.719219411217924e-06,
      "loss": 0.4178,
      "step": 5785
    },
    {
      "epoch": 7.082007343941249,
      "grad_norm": 1.1597742968340323,
      "learning_rate": 2.71858099291056e-06,
      "loss": 0.9023,
      "step": 5786
    },
    {
      "epoch": 7.0832313341493265,
      "grad_norm": 2.179264712309661,
      "learning_rate": 2.717942560238862e-06,
      "loss": 0.5424,
      "step": 5787
    },
    {
      "epoch": 7.084455324357405,
      "grad_norm": 1.4521444669365413,
      "learning_rate": 2.717304113244785e-06,
      "loss": 0.5228,
      "step": 5788
    },
    {
      "epoch": 7.085679314565484,
      "grad_norm": 2.0958335284910823,
      "learning_rate": 2.716665651970286e-06,
      "loss": 0.49,
      "step": 5789
    },
    {
      "epoch": 7.0869033047735615,
      "grad_norm": 1.028511948278021,
      "learning_rate": 2.7160271764573205e-06,
      "loss": 0.3915,
      "step": 5790
    },
    {
      "epoch": 7.08812729498164,
      "grad_norm": 1.2489147081014447,
      "learning_rate": 2.7153886867478497e-06,
      "loss": 0.5455,
      "step": 5791
    },
    {
      "epoch": 7.089351285189719,
      "grad_norm": 1.1290295322071788,
      "learning_rate": 2.7147501828838295e-06,
      "loss": 0.3979,
      "step": 5792
    },
    {
      "epoch": 7.090575275397796,
      "grad_norm": 1.895202978000537,
      "learning_rate": 2.7141116649072218e-06,
      "loss": 0.6168,
      "step": 5793
    },
    {
      "epoch": 7.091799265605875,
      "grad_norm": 1.57876117672094,
      "learning_rate": 2.713473132859988e-06,
      "loss": 0.9971,
      "step": 5794
    },
    {
      "epoch": 7.093023255813954,
      "grad_norm": 1.3512490624807578,
      "learning_rate": 2.7128345867840895e-06,
      "loss": 0.3961,
      "step": 5795
    },
    {
      "epoch": 7.094247246022032,
      "grad_norm": 1.2153718824515682,
      "learning_rate": 2.7121960267214894e-06,
      "loss": 0.9036,
      "step": 5796
    },
    {
      "epoch": 7.09547123623011,
      "grad_norm": 1.8156048887736884,
      "learning_rate": 2.7115574527141502e-06,
      "loss": 1.4047,
      "step": 5797
    },
    {
      "epoch": 7.096695226438189,
      "grad_norm": 0.966839330274678,
      "learning_rate": 2.7109188648040384e-06,
      "loss": 0.5913,
      "step": 5798
    },
    {
      "epoch": 7.097919216646267,
      "grad_norm": 2.715517175384114,
      "learning_rate": 2.7102802630331188e-06,
      "loss": 0.4344,
      "step": 5799
    },
    {
      "epoch": 7.099143206854345,
      "grad_norm": 1.1708328711818636,
      "learning_rate": 2.7096416474433585e-06,
      "loss": 0.5583,
      "step": 5800
    },
    {
      "epoch": 7.100367197062424,
      "grad_norm": 1.6515946316676944,
      "learning_rate": 2.709003018076724e-06,
      "loss": 1.0335,
      "step": 5801
    },
    {
      "epoch": 7.101591187270502,
      "grad_norm": 1.1642748315058615,
      "learning_rate": 2.7083643749751843e-06,
      "loss": 0.6499,
      "step": 5802
    },
    {
      "epoch": 7.10281517747858,
      "grad_norm": 1.41584899377622,
      "learning_rate": 2.707725718180709e-06,
      "loss": 0.5613,
      "step": 5803
    },
    {
      "epoch": 7.1040391676866586,
      "grad_norm": 1.7248928239434664,
      "learning_rate": 2.7070870477352674e-06,
      "loss": 0.7623,
      "step": 5804
    },
    {
      "epoch": 7.105263157894737,
      "grad_norm": 1.465756853129191,
      "learning_rate": 2.7064483636808314e-06,
      "loss": 0.5543,
      "step": 5805
    },
    {
      "epoch": 7.106487148102815,
      "grad_norm": 1.6159171993392503,
      "learning_rate": 2.705809666059372e-06,
      "loss": 0.7101,
      "step": 5806
    },
    {
      "epoch": 7.1077111383108935,
      "grad_norm": 1.5614535475125977,
      "learning_rate": 2.705170954912863e-06,
      "loss": 0.5435,
      "step": 5807
    },
    {
      "epoch": 7.108935128518972,
      "grad_norm": 2.6224495865992252,
      "learning_rate": 2.704532230283277e-06,
      "loss": 0.3255,
      "step": 5808
    },
    {
      "epoch": 7.11015911872705,
      "grad_norm": 0.9372957961476086,
      "learning_rate": 2.70389349221259e-06,
      "loss": 0.5726,
      "step": 5809
    },
    {
      "epoch": 7.1113831089351285,
      "grad_norm": 2.0172986779017057,
      "learning_rate": 2.703254740742777e-06,
      "loss": 0.6097,
      "step": 5810
    },
    {
      "epoch": 7.112607099143207,
      "grad_norm": 1.317932336170072,
      "learning_rate": 2.702615975915814e-06,
      "loss": 0.5641,
      "step": 5811
    },
    {
      "epoch": 7.113831089351285,
      "grad_norm": 2.0018421366614683,
      "learning_rate": 2.701977197773678e-06,
      "loss": 0.3822,
      "step": 5812
    },
    {
      "epoch": 7.1150550795593634,
      "grad_norm": 2.111066790359107,
      "learning_rate": 2.701338406358348e-06,
      "loss": 0.3948,
      "step": 5813
    },
    {
      "epoch": 7.116279069767442,
      "grad_norm": 1.0414407840743822,
      "learning_rate": 2.7006996017118033e-06,
      "loss": 0.5466,
      "step": 5814
    },
    {
      "epoch": 7.11750305997552,
      "grad_norm": 1.4741199032552008,
      "learning_rate": 2.700060783876022e-06,
      "loss": 0.9177,
      "step": 5815
    },
    {
      "epoch": 7.118727050183598,
      "grad_norm": 1.2079149455759826,
      "learning_rate": 2.6994219528929866e-06,
      "loss": 0.3707,
      "step": 5816
    },
    {
      "epoch": 7.119951040391677,
      "grad_norm": 2.407055732054247,
      "learning_rate": 2.698783108804678e-06,
      "loss": 0.5865,
      "step": 5817
    },
    {
      "epoch": 7.121175030599755,
      "grad_norm": 0.9583433603472471,
      "learning_rate": 2.698144251653079e-06,
      "loss": 0.5449,
      "step": 5818
    },
    {
      "epoch": 7.122399020807833,
      "grad_norm": 2.7918799769077456,
      "learning_rate": 2.6975053814801715e-06,
      "loss": 0.3722,
      "step": 5819
    },
    {
      "epoch": 7.123623011015912,
      "grad_norm": 1.0866525244794647,
      "learning_rate": 2.6968664983279417e-06,
      "loss": 0.4602,
      "step": 5820
    },
    {
      "epoch": 7.124847001223991,
      "grad_norm": 3.165613390544089,
      "learning_rate": 2.6962276022383742e-06,
      "loss": 0.3035,
      "step": 5821
    },
    {
      "epoch": 7.126070991432068,
      "grad_norm": 2.2064632569062894,
      "learning_rate": 2.6955886932534543e-06,
      "loss": 0.4203,
      "step": 5822
    },
    {
      "epoch": 7.127294981640147,
      "grad_norm": 1.5878167289346539,
      "learning_rate": 2.6949497714151686e-06,
      "loss": 1.395,
      "step": 5823
    },
    {
      "epoch": 7.128518971848226,
      "grad_norm": 2.0011138110996676,
      "learning_rate": 2.6943108367655053e-06,
      "loss": 0.6551,
      "step": 5824
    },
    {
      "epoch": 7.129742962056303,
      "grad_norm": 2.7029290724635495,
      "learning_rate": 2.6936718893464514e-06,
      "loss": 0.4773,
      "step": 5825
    },
    {
      "epoch": 7.130966952264382,
      "grad_norm": 1.3243529047019005,
      "learning_rate": 2.693032929199999e-06,
      "loss": 0.9968,
      "step": 5826
    },
    {
      "epoch": 7.1321909424724605,
      "grad_norm": 1.7906347674586385,
      "learning_rate": 2.692393956368135e-06,
      "loss": 0.7506,
      "step": 5827
    },
    {
      "epoch": 7.133414932680538,
      "grad_norm": 1.6451484260988838,
      "learning_rate": 2.6917549708928525e-06,
      "loss": 1.3737,
      "step": 5828
    },
    {
      "epoch": 7.134638922888617,
      "grad_norm": 2.6803967381959968,
      "learning_rate": 2.6911159728161424e-06,
      "loss": 0.4264,
      "step": 5829
    },
    {
      "epoch": 7.1358629130966955,
      "grad_norm": 2.218286829247927,
      "learning_rate": 2.6904769621799974e-06,
      "loss": 1.0412,
      "step": 5830
    },
    {
      "epoch": 7.137086903304773,
      "grad_norm": 1.8631819927982056,
      "learning_rate": 2.689837939026411e-06,
      "loss": 0.5298,
      "step": 5831
    },
    {
      "epoch": 7.138310893512852,
      "grad_norm": 1.61503109370927,
      "learning_rate": 2.6891989033973783e-06,
      "loss": 1.3184,
      "step": 5832
    },
    {
      "epoch": 7.1395348837209305,
      "grad_norm": 1.2004967683126637,
      "learning_rate": 2.6885598553348924e-06,
      "loss": 0.6179,
      "step": 5833
    },
    {
      "epoch": 7.140758873929008,
      "grad_norm": 2.579940059803031,
      "learning_rate": 2.68792079488095e-06,
      "loss": 0.5485,
      "step": 5834
    },
    {
      "epoch": 7.141982864137087,
      "grad_norm": 1.5079298477541492,
      "learning_rate": 2.6872817220775493e-06,
      "loss": 0.6302,
      "step": 5835
    },
    {
      "epoch": 7.1432068543451654,
      "grad_norm": 1.2315752382671101,
      "learning_rate": 2.6866426369666856e-06,
      "loss": 0.4112,
      "step": 5836
    },
    {
      "epoch": 7.144430844553243,
      "grad_norm": 1.2808646563410333,
      "learning_rate": 2.6860035395903583e-06,
      "loss": 0.7328,
      "step": 5837
    },
    {
      "epoch": 7.145654834761322,
      "grad_norm": 1.5640654567690786,
      "learning_rate": 2.6853644299905663e-06,
      "loss": 0.8384,
      "step": 5838
    },
    {
      "epoch": 7.1468788249694,
      "grad_norm": 1.7972076491829738,
      "learning_rate": 2.6847253082093093e-06,
      "loss": 0.6267,
      "step": 5839
    },
    {
      "epoch": 7.148102815177479,
      "grad_norm": 1.2860188435452964,
      "learning_rate": 2.6840861742885887e-06,
      "loss": 0.5705,
      "step": 5840
    },
    {
      "epoch": 7.149326805385557,
      "grad_norm": 1.4526572457315046,
      "learning_rate": 2.6834470282704063e-06,
      "loss": 0.7109,
      "step": 5841
    },
    {
      "epoch": 7.150550795593635,
      "grad_norm": 1.7306703202028026,
      "learning_rate": 2.682807870196762e-06,
      "loss": 1.0638,
      "step": 5842
    },
    {
      "epoch": 7.151774785801714,
      "grad_norm": 2.7733239265768876,
      "learning_rate": 2.682168700109662e-06,
      "loss": 0.3336,
      "step": 5843
    },
    {
      "epoch": 7.152998776009792,
      "grad_norm": 2.1403124186898097,
      "learning_rate": 2.6815295180511093e-06,
      "loss": 0.8832,
      "step": 5844
    },
    {
      "epoch": 7.15422276621787,
      "grad_norm": 1.607808075817778,
      "learning_rate": 2.680890324063106e-06,
      "loss": 0.323,
      "step": 5845
    },
    {
      "epoch": 7.155446756425949,
      "grad_norm": 2.509185596461943,
      "learning_rate": 2.6802511181876613e-06,
      "loss": 0.3759,
      "step": 5846
    },
    {
      "epoch": 7.156670746634027,
      "grad_norm": 1.6144407534646787,
      "learning_rate": 2.6796119004667796e-06,
      "loss": 0.4731,
      "step": 5847
    },
    {
      "epoch": 7.157894736842105,
      "grad_norm": 1.5713802882615826,
      "learning_rate": 2.678972670942468e-06,
      "loss": 0.5806,
      "step": 5848
    },
    {
      "epoch": 7.159118727050184,
      "grad_norm": 1.984274542863448,
      "learning_rate": 2.678333429656734e-06,
      "loss": 0.586,
      "step": 5849
    },
    {
      "epoch": 7.160342717258262,
      "grad_norm": 1.5198611656988819,
      "learning_rate": 2.6776941766515876e-06,
      "loss": 0.763,
      "step": 5850
    },
    {
      "epoch": 7.16156670746634,
      "grad_norm": 1.709463100978918,
      "learning_rate": 2.677054911969036e-06,
      "loss": 1.0194,
      "step": 5851
    },
    {
      "epoch": 7.162790697674419,
      "grad_norm": 1.9640060704857298,
      "learning_rate": 2.6764156356510913e-06,
      "loss": 0.6903,
      "step": 5852
    },
    {
      "epoch": 7.164014687882497,
      "grad_norm": 0.8293969431768294,
      "learning_rate": 2.6757763477397636e-06,
      "loss": 0.4327,
      "step": 5853
    },
    {
      "epoch": 7.165238678090575,
      "grad_norm": 1.6254361874901109,
      "learning_rate": 2.6751370482770638e-06,
      "loss": 1.5005,
      "step": 5854
    },
    {
      "epoch": 7.166462668298654,
      "grad_norm": 1.672238312705694,
      "learning_rate": 2.6744977373050057e-06,
      "loss": 0.6036,
      "step": 5855
    },
    {
      "epoch": 7.167686658506732,
      "grad_norm": 1.4939509545259069,
      "learning_rate": 2.6738584148656012e-06,
      "loss": 0.5766,
      "step": 5856
    },
    {
      "epoch": 7.16891064871481,
      "grad_norm": 1.5699765549997182,
      "learning_rate": 2.673219081000865e-06,
      "loss": 0.4812,
      "step": 5857
    },
    {
      "epoch": 7.170134638922889,
      "grad_norm": 1.6963738264336345,
      "learning_rate": 2.672579735752811e-06,
      "loss": 1.1211,
      "step": 5858
    },
    {
      "epoch": 7.1713586291309666,
      "grad_norm": 1.2084049652055546,
      "learning_rate": 2.6719403791634564e-06,
      "loss": 0.529,
      "step": 5859
    },
    {
      "epoch": 7.172582619339045,
      "grad_norm": 1.5779265910682159,
      "learning_rate": 2.6713010112748145e-06,
      "loss": 0.3068,
      "step": 5860
    },
    {
      "epoch": 7.173806609547124,
      "grad_norm": 1.745005180667403,
      "learning_rate": 2.6706616321289043e-06,
      "loss": 1.3273,
      "step": 5861
    },
    {
      "epoch": 7.1750305997552015,
      "grad_norm": 1.4749477798653052,
      "learning_rate": 2.670022241767743e-06,
      "loss": 0.5125,
      "step": 5862
    },
    {
      "epoch": 7.17625458996328,
      "grad_norm": 1.8456439993950582,
      "learning_rate": 2.669382840233349e-06,
      "loss": 1.1315,
      "step": 5863
    },
    {
      "epoch": 7.177478580171359,
      "grad_norm": 2.268773411982329,
      "learning_rate": 2.668743427567741e-06,
      "loss": 0.4819,
      "step": 5864
    },
    {
      "epoch": 7.178702570379437,
      "grad_norm": 1.3854112947657111,
      "learning_rate": 2.668104003812939e-06,
      "loss": 0.6203,
      "step": 5865
    },
    {
      "epoch": 7.179926560587515,
      "grad_norm": 2.2591000717236915,
      "learning_rate": 2.6674645690109636e-06,
      "loss": 0.802,
      "step": 5866
    },
    {
      "epoch": 7.181150550795594,
      "grad_norm": 1.41360072156089,
      "learning_rate": 2.666825123203836e-06,
      "loss": 0.5322,
      "step": 5867
    },
    {
      "epoch": 7.182374541003672,
      "grad_norm": 1.7975226052932693,
      "learning_rate": 2.6661856664335795e-06,
      "loss": 0.7294,
      "step": 5868
    },
    {
      "epoch": 7.18359853121175,
      "grad_norm": 1.8318051464720373,
      "learning_rate": 2.665546198742214e-06,
      "loss": 0.933,
      "step": 5869
    },
    {
      "epoch": 7.184822521419829,
      "grad_norm": 1.5092581838754413,
      "learning_rate": 2.6649067201717655e-06,
      "loss": 1.1137,
      "step": 5870
    },
    {
      "epoch": 7.186046511627907,
      "grad_norm": 1.2820211146030684,
      "learning_rate": 2.6642672307642575e-06,
      "loss": 0.282,
      "step": 5871
    },
    {
      "epoch": 7.187270501835985,
      "grad_norm": 1.302893622153774,
      "learning_rate": 2.6636277305617144e-06,
      "loss": 0.8562,
      "step": 5872
    },
    {
      "epoch": 7.188494492044064,
      "grad_norm": 1.232705315086875,
      "learning_rate": 2.6629882196061625e-06,
      "loss": 0.4457,
      "step": 5873
    },
    {
      "epoch": 7.189718482252142,
      "grad_norm": 1.332025558056674,
      "learning_rate": 2.6623486979396266e-06,
      "loss": 1.5668,
      "step": 5874
    },
    {
      "epoch": 7.19094247246022,
      "grad_norm": 1.094990909472066,
      "learning_rate": 2.6617091656041356e-06,
      "loss": 0.4791,
      "step": 5875
    },
    {
      "epoch": 7.192166462668299,
      "grad_norm": 1.5973759854999787,
      "learning_rate": 2.6610696226417163e-06,
      "loss": 0.4623,
      "step": 5876
    },
    {
      "epoch": 7.193390452876377,
      "grad_norm": 1.7111798491239965,
      "learning_rate": 2.660430069094397e-06,
      "loss": 1.1996,
      "step": 5877
    },
    {
      "epoch": 7.194614443084455,
      "grad_norm": 1.0396945210202908,
      "learning_rate": 2.6597905050042065e-06,
      "loss": 0.6421,
      "step": 5878
    },
    {
      "epoch": 7.195838433292534,
      "grad_norm": 2.013406270396875,
      "learning_rate": 2.659150930413176e-06,
      "loss": 0.5062,
      "step": 5879
    },
    {
      "epoch": 7.197062423500612,
      "grad_norm": 1.3246048634039544,
      "learning_rate": 2.6585113453633344e-06,
      "loss": 0.7236,
      "step": 5880
    },
    {
      "epoch": 7.19828641370869,
      "grad_norm": 0.9990101921829487,
      "learning_rate": 2.657871749896714e-06,
      "loss": 0.6484,
      "step": 5881
    },
    {
      "epoch": 7.1995104039167686,
      "grad_norm": 1.288076751192695,
      "learning_rate": 2.6572321440553454e-06,
      "loss": 0.513,
      "step": 5882
    },
    {
      "epoch": 7.200734394124847,
      "grad_norm": 2.263192158343417,
      "learning_rate": 2.656592527881262e-06,
      "loss": 0.3586,
      "step": 5883
    },
    {
      "epoch": 7.201958384332926,
      "grad_norm": 1.6352870729896452,
      "learning_rate": 2.6559529014164965e-06,
      "loss": 0.4557,
      "step": 5884
    },
    {
      "epoch": 7.2031823745410035,
      "grad_norm": 0.9476113974041005,
      "learning_rate": 2.6553132647030844e-06,
      "loss": 0.4031,
      "step": 5885
    },
    {
      "epoch": 7.204406364749082,
      "grad_norm": 0.9195531689459152,
      "learning_rate": 2.654673617783058e-06,
      "loss": 0.5928,
      "step": 5886
    },
    {
      "epoch": 7.205630354957161,
      "grad_norm": 2.227331124798192,
      "learning_rate": 2.654033960698454e-06,
      "loss": 0.8769,
      "step": 5887
    },
    {
      "epoch": 7.2068543451652385,
      "grad_norm": 1.4416628815693862,
      "learning_rate": 2.6533942934913077e-06,
      "loss": 0.9927,
      "step": 5888
    },
    {
      "epoch": 7.208078335373317,
      "grad_norm": 1.0928965417837828,
      "learning_rate": 2.6527546162036555e-06,
      "loss": 0.5947,
      "step": 5889
    },
    {
      "epoch": 7.209302325581396,
      "grad_norm": 2.2981471921257857,
      "learning_rate": 2.652114928877535e-06,
      "loss": 0.2964,
      "step": 5890
    },
    {
      "epoch": 7.2105263157894735,
      "grad_norm": 1.3102980647399354,
      "learning_rate": 2.651475231554985e-06,
      "loss": 0.9764,
      "step": 5891
    },
    {
      "epoch": 7.211750305997552,
      "grad_norm": 2.3313128639464744,
      "learning_rate": 2.6508355242780413e-06,
      "loss": 0.9511,
      "step": 5892
    },
    {
      "epoch": 7.212974296205631,
      "grad_norm": 1.361908127702271,
      "learning_rate": 2.6501958070887455e-06,
      "loss": 0.8778,
      "step": 5893
    },
    {
      "epoch": 7.214198286413708,
      "grad_norm": 0.6675559554317547,
      "learning_rate": 2.6495560800291383e-06,
      "loss": 0.261,
      "step": 5894
    },
    {
      "epoch": 7.215422276621787,
      "grad_norm": 1.0811648712669348,
      "learning_rate": 2.648916343141257e-06,
      "loss": 0.582,
      "step": 5895
    },
    {
      "epoch": 7.216646266829866,
      "grad_norm": 1.3773367077176948,
      "learning_rate": 2.648276596467145e-06,
      "loss": 0.8325,
      "step": 5896
    },
    {
      "epoch": 7.217870257037943,
      "grad_norm": 1.0346416586004556,
      "learning_rate": 2.6476368400488434e-06,
      "loss": 0.3171,
      "step": 5897
    },
    {
      "epoch": 7.219094247246022,
      "grad_norm": 1.877392196829962,
      "learning_rate": 2.6469970739283944e-06,
      "loss": 0.625,
      "step": 5898
    },
    {
      "epoch": 7.220318237454101,
      "grad_norm": 1.756247564291727,
      "learning_rate": 2.6463572981478412e-06,
      "loss": 0.3334,
      "step": 5899
    },
    {
      "epoch": 7.221542227662178,
      "grad_norm": 1.884875022901362,
      "learning_rate": 2.645717512749229e-06,
      "loss": 1.2184,
      "step": 5900
    },
    {
      "epoch": 7.222766217870257,
      "grad_norm": 1.0973380772363157,
      "learning_rate": 2.6450777177745996e-06,
      "loss": 0.5023,
      "step": 5901
    },
    {
      "epoch": 7.223990208078336,
      "grad_norm": 2.209907119930594,
      "learning_rate": 2.644437913266e-06,
      "loss": 0.4843,
      "step": 5902
    },
    {
      "epoch": 7.225214198286413,
      "grad_norm": 1.2687934738354933,
      "learning_rate": 2.6437980992654745e-06,
      "loss": 0.4682,
      "step": 5903
    },
    {
      "epoch": 7.226438188494492,
      "grad_norm": 2.1804890435223254,
      "learning_rate": 2.6431582758150705e-06,
      "loss": 0.5208,
      "step": 5904
    },
    {
      "epoch": 7.2276621787025706,
      "grad_norm": 1.5865355633507352,
      "learning_rate": 2.6425184429568333e-06,
      "loss": 0.7021,
      "step": 5905
    },
    {
      "epoch": 7.228886168910648,
      "grad_norm": 1.4808872647046532,
      "learning_rate": 2.6418786007328113e-06,
      "loss": 1.2383,
      "step": 5906
    },
    {
      "epoch": 7.230110159118727,
      "grad_norm": 1.2806652204508921,
      "learning_rate": 2.6412387491850526e-06,
      "loss": 1.291,
      "step": 5907
    },
    {
      "epoch": 7.2313341493268055,
      "grad_norm": 2.071364629070407,
      "learning_rate": 2.6405988883556056e-06,
      "loss": 0.6058,
      "step": 5908
    },
    {
      "epoch": 7.232558139534884,
      "grad_norm": 1.0192633611003123,
      "learning_rate": 2.6399590182865202e-06,
      "loss": 0.6255,
      "step": 5909
    },
    {
      "epoch": 7.233782129742962,
      "grad_norm": 1.2026378932120672,
      "learning_rate": 2.6393191390198447e-06,
      "loss": 0.5126,
      "step": 5910
    },
    {
      "epoch": 7.2350061199510405,
      "grad_norm": 1.0104356844195284,
      "learning_rate": 2.6386792505976314e-06,
      "loss": 0.6137,
      "step": 5911
    },
    {
      "epoch": 7.236230110159119,
      "grad_norm": 1.640903824144805,
      "learning_rate": 2.6380393530619307e-06,
      "loss": 0.6647,
      "step": 5912
    },
    {
      "epoch": 7.237454100367197,
      "grad_norm": 1.5307607419372122,
      "learning_rate": 2.6373994464547947e-06,
      "loss": 0.673,
      "step": 5913
    },
    {
      "epoch": 7.2386780905752754,
      "grad_norm": 1.2720899772676348,
      "learning_rate": 2.6367595308182747e-06,
      "loss": 0.6244,
      "step": 5914
    },
    {
      "epoch": 7.239902080783354,
      "grad_norm": 1.467685696574331,
      "learning_rate": 2.6361196061944244e-06,
      "loss": 1.1098,
      "step": 5915
    },
    {
      "epoch": 7.241126070991432,
      "grad_norm": 1.2181300036311464,
      "learning_rate": 2.635479672625296e-06,
      "loss": 0.5552,
      "step": 5916
    },
    {
      "epoch": 7.24235006119951,
      "grad_norm": 1.9832348584423176,
      "learning_rate": 2.634839730152946e-06,
      "loss": 0.5662,
      "step": 5917
    },
    {
      "epoch": 7.243574051407589,
      "grad_norm": 1.5891830929701114,
      "learning_rate": 2.6341997788194272e-06,
      "loss": 0.5983,
      "step": 5918
    },
    {
      "epoch": 7.244798041615667,
      "grad_norm": 1.787364039028482,
      "learning_rate": 2.6335598186667956e-06,
      "loss": 0.8687,
      "step": 5919
    },
    {
      "epoch": 7.246022031823745,
      "grad_norm": 0.9189615488026313,
      "learning_rate": 2.632919849737107e-06,
      "loss": 0.5209,
      "step": 5920
    },
    {
      "epoch": 7.247246022031824,
      "grad_norm": 1.8269469117948611,
      "learning_rate": 2.6322798720724167e-06,
      "loss": 0.5502,
      "step": 5921
    },
    {
      "epoch": 7.248470012239902,
      "grad_norm": 1.7829480355878096,
      "learning_rate": 2.631639885714783e-06,
      "loss": 0.5483,
      "step": 5922
    },
    {
      "epoch": 7.24969400244798,
      "grad_norm": 1.7055412705309752,
      "learning_rate": 2.630999890706262e-06,
      "loss": 0.5085,
      "step": 5923
    },
    {
      "epoch": 7.250917992656059,
      "grad_norm": 1.2682953854879155,
      "learning_rate": 2.630359887088914e-06,
      "loss": 1.0128,
      "step": 5924
    },
    {
      "epoch": 7.252141982864137,
      "grad_norm": 1.5465753148061108,
      "learning_rate": 2.6297198749047954e-06,
      "loss": 0.3064,
      "step": 5925
    },
    {
      "epoch": 7.253365973072215,
      "grad_norm": 1.1236670885975288,
      "learning_rate": 2.6290798541959665e-06,
      "loss": 0.7801,
      "step": 5926
    },
    {
      "epoch": 7.254589963280294,
      "grad_norm": 1.091761526240801,
      "learning_rate": 2.628439825004487e-06,
      "loss": 1.0497,
      "step": 5927
    },
    {
      "epoch": 7.2558139534883725,
      "grad_norm": 1.2859221097143547,
      "learning_rate": 2.627799787372418e-06,
      "loss": 0.8181,
      "step": 5928
    },
    {
      "epoch": 7.25703794369645,
      "grad_norm": 2.0907452929278243,
      "learning_rate": 2.6271597413418197e-06,
      "loss": 0.5141,
      "step": 5929
    },
    {
      "epoch": 7.258261933904529,
      "grad_norm": 1.3245765287036302,
      "learning_rate": 2.6265196869547533e-06,
      "loss": 0.6607,
      "step": 5930
    },
    {
      "epoch": 7.2594859241126075,
      "grad_norm": 0.9908278544764718,
      "learning_rate": 2.62587962425328e-06,
      "loss": 0.5812,
      "step": 5931
    },
    {
      "epoch": 7.260709914320685,
      "grad_norm": 1.0146535435761823,
      "learning_rate": 2.625239553279465e-06,
      "loss": 0.6024,
      "step": 5932
    },
    {
      "epoch": 7.261933904528764,
      "grad_norm": 1.4003264418018755,
      "learning_rate": 2.624599474075369e-06,
      "loss": 0.3057,
      "step": 5933
    },
    {
      "epoch": 7.2631578947368425,
      "grad_norm": 1.5714193863852155,
      "learning_rate": 2.623959386683056e-06,
      "loss": 1.1716,
      "step": 5934
    },
    {
      "epoch": 7.26438188494492,
      "grad_norm": 0.755908209444401,
      "learning_rate": 2.6233192911445916e-06,
      "loss": 0.4384,
      "step": 5935
    },
    {
      "epoch": 7.265605875152999,
      "grad_norm": 1.2231739750509312,
      "learning_rate": 2.6226791875020384e-06,
      "loss": 1.5779,
      "step": 5936
    },
    {
      "epoch": 7.2668298653610774,
      "grad_norm": 2.6746240235651935,
      "learning_rate": 2.6220390757974633e-06,
      "loss": 0.344,
      "step": 5937
    },
    {
      "epoch": 7.268053855569155,
      "grad_norm": 1.324689355851732,
      "learning_rate": 2.621398956072932e-06,
      "loss": 0.6377,
      "step": 5938
    },
    {
      "epoch": 7.269277845777234,
      "grad_norm": 1.56170755267866,
      "learning_rate": 2.6207588283705097e-06,
      "loss": 1.4327,
      "step": 5939
    },
    {
      "epoch": 7.270501835985312,
      "grad_norm": 1.5984482975105423,
      "learning_rate": 2.6201186927322635e-06,
      "loss": 1.4181,
      "step": 5940
    },
    {
      "epoch": 7.27172582619339,
      "grad_norm": 0.9832418775241379,
      "learning_rate": 2.6194785492002624e-06,
      "loss": 0.7188,
      "step": 5941
    },
    {
      "epoch": 7.272949816401469,
      "grad_norm": 1.6637590657795454,
      "learning_rate": 2.6188383978165716e-06,
      "loss": 0.7225,
      "step": 5942
    },
    {
      "epoch": 7.274173806609547,
      "grad_norm": 2.433531763712246,
      "learning_rate": 2.618198238623261e-06,
      "loss": 0.5257,
      "step": 5943
    },
    {
      "epoch": 7.275397796817625,
      "grad_norm": 2.031816040792014,
      "learning_rate": 2.6175580716624003e-06,
      "loss": 0.5394,
      "step": 5944
    },
    {
      "epoch": 7.276621787025704,
      "grad_norm": 2.536938681651498,
      "learning_rate": 2.6169178969760572e-06,
      "loss": 0.8441,
      "step": 5945
    },
    {
      "epoch": 7.277845777233782,
      "grad_norm": 1.2830184813922128,
      "learning_rate": 2.616277714606303e-06,
      "loss": 0.5582,
      "step": 5946
    },
    {
      "epoch": 7.27906976744186,
      "grad_norm": 2.72375927826823,
      "learning_rate": 2.615637524595207e-06,
      "loss": 0.4874,
      "step": 5947
    },
    {
      "epoch": 7.280293757649939,
      "grad_norm": 2.025424088420344,
      "learning_rate": 2.6149973269848408e-06,
      "loss": 0.6501,
      "step": 5948
    },
    {
      "epoch": 7.281517747858017,
      "grad_norm": 1.7579887450282297,
      "learning_rate": 2.6143571218172747e-06,
      "loss": 0.607,
      "step": 5949
    },
    {
      "epoch": 7.282741738066095,
      "grad_norm": 1.296260958604775,
      "learning_rate": 2.6137169091345833e-06,
      "loss": 0.6268,
      "step": 5950
    },
    {
      "epoch": 7.283965728274174,
      "grad_norm": 2.1291471912702944,
      "learning_rate": 2.6130766889788357e-06,
      "loss": 0.5655,
      "step": 5951
    },
    {
      "epoch": 7.285189718482252,
      "grad_norm": 2.5245829321509876,
      "learning_rate": 2.6124364613921066e-06,
      "loss": 0.4693,
      "step": 5952
    },
    {
      "epoch": 7.286413708690331,
      "grad_norm": 1.717234010234117,
      "learning_rate": 2.61179622641647e-06,
      "loss": 0.5375,
      "step": 5953
    },
    {
      "epoch": 7.287637698898409,
      "grad_norm": 1.6623661177813271,
      "learning_rate": 2.6111559840939983e-06,
      "loss": 0.489,
      "step": 5954
    },
    {
      "epoch": 7.288861689106487,
      "grad_norm": 1.2960466308787184,
      "learning_rate": 2.610515734466766e-06,
      "loss": 1.1256,
      "step": 5955
    },
    {
      "epoch": 7.290085679314566,
      "grad_norm": 1.2281747926141695,
      "learning_rate": 2.6098754775768494e-06,
      "loss": 0.516,
      "step": 5956
    },
    {
      "epoch": 7.291309669522644,
      "grad_norm": 1.4678302051065844,
      "learning_rate": 2.6092352134663222e-06,
      "loss": 0.681,
      "step": 5957
    },
    {
      "epoch": 7.292533659730722,
      "grad_norm": 2.002304125943287,
      "learning_rate": 2.6085949421772607e-06,
      "loss": 0.5953,
      "step": 5958
    },
    {
      "epoch": 7.293757649938801,
      "grad_norm": 2.1409577108970095,
      "learning_rate": 2.6079546637517416e-06,
      "loss": 0.6231,
      "step": 5959
    },
    {
      "epoch": 7.2949816401468786,
      "grad_norm": 1.4855107135850432,
      "learning_rate": 2.6073143782318412e-06,
      "loss": 0.8634,
      "step": 5960
    },
    {
      "epoch": 7.296205630354957,
      "grad_norm": 1.7539476588827991,
      "learning_rate": 2.606674085659637e-06,
      "loss": 0.6078,
      "step": 5961
    },
    {
      "epoch": 7.297429620563036,
      "grad_norm": 1.2106428384136332,
      "learning_rate": 2.6060337860772054e-06,
      "loss": 0.3185,
      "step": 5962
    },
    {
      "epoch": 7.2986536107711135,
      "grad_norm": 0.981218988353405,
      "learning_rate": 2.605393479526627e-06,
      "loss": 0.6046,
      "step": 5963
    },
    {
      "epoch": 7.299877600979192,
      "grad_norm": 1.6016474203887976,
      "learning_rate": 2.604753166049978e-06,
      "loss": 0.8943,
      "step": 5964
    },
    {
      "epoch": 7.301101591187271,
      "grad_norm": 1.5208217593969724,
      "learning_rate": 2.604112845689339e-06,
      "loss": 0.5869,
      "step": 5965
    },
    {
      "epoch": 7.3023255813953485,
      "grad_norm": 1.0932391912893304,
      "learning_rate": 2.6034725184867883e-06,
      "loss": 0.4406,
      "step": 5966
    },
    {
      "epoch": 7.303549571603427,
      "grad_norm": 2.6431264067426596,
      "learning_rate": 2.602832184484407e-06,
      "loss": 0.596,
      "step": 5967
    },
    {
      "epoch": 7.304773561811506,
      "grad_norm": 1.9138816723419527,
      "learning_rate": 2.602191843724274e-06,
      "loss": 1.1225,
      "step": 5968
    },
    {
      "epoch": 7.3059975520195835,
      "grad_norm": 0.6847921245352586,
      "learning_rate": 2.6015514962484716e-06,
      "loss": 0.2228,
      "step": 5969
    },
    {
      "epoch": 7.307221542227662,
      "grad_norm": 1.4688405815860193,
      "learning_rate": 2.6009111420990808e-06,
      "loss": 0.4803,
      "step": 5970
    },
    {
      "epoch": 7.308445532435741,
      "grad_norm": 0.9951747235405431,
      "learning_rate": 2.6002707813181825e-06,
      "loss": 0.6129,
      "step": 5971
    },
    {
      "epoch": 7.309669522643819,
      "grad_norm": 1.014353845365715,
      "learning_rate": 2.5996304139478597e-06,
      "loss": 0.5061,
      "step": 5972
    },
    {
      "epoch": 7.310893512851897,
      "grad_norm": 1.9414572846582285,
      "learning_rate": 2.5989900400301945e-06,
      "loss": 0.6188,
      "step": 5973
    },
    {
      "epoch": 7.312117503059976,
      "grad_norm": 1.4018957202375015,
      "learning_rate": 2.59834965960727e-06,
      "loss": 0.7583,
      "step": 5974
    },
    {
      "epoch": 7.313341493268053,
      "grad_norm": 2.3622873183132382,
      "learning_rate": 2.597709272721169e-06,
      "loss": 0.8148,
      "step": 5975
    },
    {
      "epoch": 7.314565483476132,
      "grad_norm": 1.2311652248125766,
      "learning_rate": 2.5970688794139763e-06,
      "loss": 0.4426,
      "step": 5976
    },
    {
      "epoch": 7.315789473684211,
      "grad_norm": 2.3853226213476018,
      "learning_rate": 2.596428479727776e-06,
      "loss": 0.973,
      "step": 5977
    },
    {
      "epoch": 7.317013463892289,
      "grad_norm": 3.0120655541927737,
      "learning_rate": 2.595788073704653e-06,
      "loss": 0.3755,
      "step": 5978
    },
    {
      "epoch": 7.318237454100367,
      "grad_norm": 2.1868471704892785,
      "learning_rate": 2.595147661386692e-06,
      "loss": 0.4996,
      "step": 5979
    },
    {
      "epoch": 7.319461444308446,
      "grad_norm": 2.004182219973011,
      "learning_rate": 2.5945072428159783e-06,
      "loss": 0.6209,
      "step": 5980
    },
    {
      "epoch": 7.320685434516524,
      "grad_norm": 0.9648632114561964,
      "learning_rate": 2.5938668180345983e-06,
      "loss": 0.6395,
      "step": 5981
    },
    {
      "epoch": 7.321909424724602,
      "grad_norm": 1.4389759625970304,
      "learning_rate": 2.5932263870846386e-06,
      "loss": 0.5362,
      "step": 5982
    },
    {
      "epoch": 7.3231334149326806,
      "grad_norm": 1.0769278770404218,
      "learning_rate": 2.5925859500081858e-06,
      "loss": 0.4523,
      "step": 5983
    },
    {
      "epoch": 7.324357405140759,
      "grad_norm": 2.46873006709113,
      "learning_rate": 2.5919455068473254e-06,
      "loss": 0.4443,
      "step": 5984
    },
    {
      "epoch": 7.325581395348837,
      "grad_norm": 1.6602813380514658,
      "learning_rate": 2.591305057644148e-06,
      "loss": 0.5221,
      "step": 5985
    },
    {
      "epoch": 7.3268053855569155,
      "grad_norm": 0.8371027727180844,
      "learning_rate": 2.5906646024407385e-06,
      "loss": 0.5304,
      "step": 5986
    },
    {
      "epoch": 7.328029375764994,
      "grad_norm": 1.7377561833099442,
      "learning_rate": 2.5900241412791876e-06,
      "loss": 0.4555,
      "step": 5987
    },
    {
      "epoch": 7.329253365973072,
      "grad_norm": 1.6330951714974256,
      "learning_rate": 2.589383674201583e-06,
      "loss": 0.6249,
      "step": 5988
    },
    {
      "epoch": 7.3304773561811505,
      "grad_norm": 1.7607216820367415,
      "learning_rate": 2.5887432012500134e-06,
      "loss": 0.4961,
      "step": 5989
    },
    {
      "epoch": 7.331701346389229,
      "grad_norm": 1.9510708580996678,
      "learning_rate": 2.588102722466569e-06,
      "loss": 0.827,
      "step": 5990
    },
    {
      "epoch": 7.332925336597307,
      "grad_norm": 1.313933161242551,
      "learning_rate": 2.58746223789334e-06,
      "loss": 0.8067,
      "step": 5991
    },
    {
      "epoch": 7.3341493268053854,
      "grad_norm": 1.2796146010636669,
      "learning_rate": 2.5868217475724157e-06,
      "loss": 0.6088,
      "step": 5992
    },
    {
      "epoch": 7.335373317013464,
      "grad_norm": 1.4422793519326782,
      "learning_rate": 2.586181251545888e-06,
      "loss": 1.172,
      "step": 5993
    },
    {
      "epoch": 7.336597307221542,
      "grad_norm": 0.8816985204942966,
      "learning_rate": 2.5855407498558467e-06,
      "loss": 0.4529,
      "step": 5994
    },
    {
      "epoch": 7.33782129742962,
      "grad_norm": 1.581866011162923,
      "learning_rate": 2.5849002425443837e-06,
      "loss": 0.4768,
      "step": 5995
    },
    {
      "epoch": 7.339045287637699,
      "grad_norm": 1.212705674231151,
      "learning_rate": 2.584259729653591e-06,
      "loss": 0.5177,
      "step": 5996
    },
    {
      "epoch": 7.340269277845778,
      "grad_norm": 0.8489667315689776,
      "learning_rate": 2.5836192112255603e-06,
      "loss": 0.4455,
      "step": 5997
    },
    {
      "epoch": 7.341493268053855,
      "grad_norm": 2.301183042340691,
      "learning_rate": 2.5829786873023844e-06,
      "loss": 0.4969,
      "step": 5998
    },
    {
      "epoch": 7.342717258261934,
      "grad_norm": 1.008677662090527,
      "learning_rate": 2.5823381579261556e-06,
      "loss": 0.6714,
      "step": 5999
    },
    {
      "epoch": 7.343941248470013,
      "grad_norm": 1.6874019047827935,
      "learning_rate": 2.581697623138969e-06,
      "loss": 0.6237,
      "step": 6000
    },
    {
      "epoch": 7.34516523867809,
      "grad_norm": 2.215407184418501,
      "learning_rate": 2.581057082982915e-06,
      "loss": 0.731,
      "step": 6001
    },
    {
      "epoch": 7.346389228886169,
      "grad_norm": 2.4253753179489266,
      "learning_rate": 2.5804165375000904e-06,
      "loss": 0.4602,
      "step": 6002
    },
    {
      "epoch": 7.347613219094248,
      "grad_norm": 1.6370392655039696,
      "learning_rate": 2.5797759867325884e-06,
      "loss": 0.9677,
      "step": 6003
    },
    {
      "epoch": 7.348837209302325,
      "grad_norm": 1.5754950385118511,
      "learning_rate": 2.5791354307225034e-06,
      "loss": 1.4106,
      "step": 6004
    },
    {
      "epoch": 7.350061199510404,
      "grad_norm": 1.833278887728968,
      "learning_rate": 2.5784948695119305e-06,
      "loss": 0.8468,
      "step": 6005
    },
    {
      "epoch": 7.3512851897184825,
      "grad_norm": 1.340711200342193,
      "learning_rate": 2.5778543031429655e-06,
      "loss": 0.6134,
      "step": 6006
    },
    {
      "epoch": 7.35250917992656,
      "grad_norm": 1.3162518796254616,
      "learning_rate": 2.577213731657703e-06,
      "loss": 0.3344,
      "step": 6007
    },
    {
      "epoch": 7.353733170134639,
      "grad_norm": 1.73884636872668,
      "learning_rate": 2.5765731550982403e-06,
      "loss": 0.3819,
      "step": 6008
    },
    {
      "epoch": 7.3549571603427175,
      "grad_norm": 1.5559656779269873,
      "learning_rate": 2.575932573506673e-06,
      "loss": 0.4208,
      "step": 6009
    },
    {
      "epoch": 7.356181150550795,
      "grad_norm": 1.6185953193398317,
      "learning_rate": 2.5752919869250977e-06,
      "loss": 1.6089,
      "step": 6010
    },
    {
      "epoch": 7.357405140758874,
      "grad_norm": 1.4542937513659768,
      "learning_rate": 2.5746513953956115e-06,
      "loss": 0.4844,
      "step": 6011
    },
    {
      "epoch": 7.3586291309669525,
      "grad_norm": 1.2174560620722166,
      "learning_rate": 2.5740107989603116e-06,
      "loss": 0.7816,
      "step": 6012
    },
    {
      "epoch": 7.35985312117503,
      "grad_norm": 1.6340708255199425,
      "learning_rate": 2.5733701976612956e-06,
      "loss": 1.471,
      "step": 6013
    },
    {
      "epoch": 7.361077111383109,
      "grad_norm": 1.7273296411214234,
      "learning_rate": 2.5727295915406614e-06,
      "loss": 0.5708,
      "step": 6014
    },
    {
      "epoch": 7.3623011015911874,
      "grad_norm": 2.6164929763134586,
      "learning_rate": 2.5720889806405087e-06,
      "loss": 0.3668,
      "step": 6015
    },
    {
      "epoch": 7.363525091799266,
      "grad_norm": 0.9640342274610135,
      "learning_rate": 2.5714483650029333e-06,
      "loss": 0.5473,
      "step": 6016
    },
    {
      "epoch": 7.364749082007344,
      "grad_norm": 1.3860637429164129,
      "learning_rate": 2.5708077446700363e-06,
      "loss": 0.9244,
      "step": 6017
    },
    {
      "epoch": 7.365973072215422,
      "grad_norm": 1.4351166462162042,
      "learning_rate": 2.5701671196839162e-06,
      "loss": 0.5123,
      "step": 6018
    },
    {
      "epoch": 7.3671970624235,
      "grad_norm": 3.1295697945608487,
      "learning_rate": 2.569526490086673e-06,
      "loss": 0.4442,
      "step": 6019
    },
    {
      "epoch": 7.368421052631579,
      "grad_norm": 1.7943172655387614,
      "learning_rate": 2.5688858559204056e-06,
      "loss": 0.7769,
      "step": 6020
    },
    {
      "epoch": 7.369645042839657,
      "grad_norm": 2.317596419975166,
      "learning_rate": 2.5682452172272143e-06,
      "loss": 0.8776,
      "step": 6021
    },
    {
      "epoch": 7.370869033047736,
      "grad_norm": 1.3949180133138774,
      "learning_rate": 2.5676045740492e-06,
      "loss": 0.8289,
      "step": 6022
    },
    {
      "epoch": 7.372093023255814,
      "grad_norm": 1.9283980290116884,
      "learning_rate": 2.5669639264284635e-06,
      "loss": 0.5892,
      "step": 6023
    },
    {
      "epoch": 7.373317013463892,
      "grad_norm": 1.1589183346938958,
      "learning_rate": 2.566323274407105e-06,
      "loss": 0.5916,
      "step": 6024
    },
    {
      "epoch": 7.374541003671971,
      "grad_norm": 0.7931431294781681,
      "learning_rate": 2.5656826180272254e-06,
      "loss": 0.3552,
      "step": 6025
    },
    {
      "epoch": 7.375764993880049,
      "grad_norm": 1.5628744207451057,
      "learning_rate": 2.565041957330928e-06,
      "loss": 0.4626,
      "step": 6026
    },
    {
      "epoch": 7.376988984088127,
      "grad_norm": 2.702241728521057,
      "learning_rate": 2.564401292360314e-06,
      "loss": 0.4195,
      "step": 6027
    },
    {
      "epoch": 7.378212974296206,
      "grad_norm": 1.2085926337562602,
      "learning_rate": 2.5637606231574854e-06,
      "loss": 0.4061,
      "step": 6028
    },
    {
      "epoch": 7.379436964504284,
      "grad_norm": 1.7875288677743737,
      "learning_rate": 2.5631199497645437e-06,
      "loss": 0.5626,
      "step": 6029
    },
    {
      "epoch": 7.380660954712362,
      "grad_norm": 1.778969673265318,
      "learning_rate": 2.5624792722235924e-06,
      "loss": 0.778,
      "step": 6030
    },
    {
      "epoch": 7.381884944920441,
      "grad_norm": 1.4537225247494032,
      "learning_rate": 2.5618385905767348e-06,
      "loss": 0.8705,
      "step": 6031
    },
    {
      "epoch": 7.383108935128519,
      "grad_norm": 2.2894793684285157,
      "learning_rate": 2.561197904866074e-06,
      "loss": 0.3559,
      "step": 6032
    },
    {
      "epoch": 7.384332925336597,
      "grad_norm": 1.959485171071054,
      "learning_rate": 2.560557215133713e-06,
      "loss": 0.5003,
      "step": 6033
    },
    {
      "epoch": 7.385556915544676,
      "grad_norm": 2.563648503324109,
      "learning_rate": 2.5599165214217554e-06,
      "loss": 0.4882,
      "step": 6034
    },
    {
      "epoch": 7.386780905752754,
      "grad_norm": 2.031621338900395,
      "learning_rate": 2.559275823772306e-06,
      "loss": 0.342,
      "step": 6035
    },
    {
      "epoch": 7.388004895960832,
      "grad_norm": 1.327101003354006,
      "learning_rate": 2.5586351222274687e-06,
      "loss": 0.6587,
      "step": 6036
    },
    {
      "epoch": 7.389228886168911,
      "grad_norm": 1.5013341551652672,
      "learning_rate": 2.557994416829348e-06,
      "loss": 0.5072,
      "step": 6037
    },
    {
      "epoch": 7.3904528763769886,
      "grad_norm": 1.576727615781655,
      "learning_rate": 2.557353707620049e-06,
      "loss": 0.593,
      "step": 6038
    },
    {
      "epoch": 7.391676866585067,
      "grad_norm": 1.4368151303979397,
      "learning_rate": 2.5567129946416765e-06,
      "loss": 0.7033,
      "step": 6039
    },
    {
      "epoch": 7.392900856793146,
      "grad_norm": 2.0002799647481324,
      "learning_rate": 2.5560722779363345e-06,
      "loss": 0.4908,
      "step": 6040
    },
    {
      "epoch": 7.394124847001224,
      "grad_norm": 2.472582335810784,
      "learning_rate": 2.5554315575461314e-06,
      "loss": 0.4673,
      "step": 6041
    },
    {
      "epoch": 7.395348837209302,
      "grad_norm": 1.2782960917000092,
      "learning_rate": 2.5547908335131704e-06,
      "loss": 0.5448,
      "step": 6042
    },
    {
      "epoch": 7.396572827417381,
      "grad_norm": 2.350259984611304,
      "learning_rate": 2.554150105879559e-06,
      "loss": 0.3804,
      "step": 6043
    },
    {
      "epoch": 7.397796817625459,
      "grad_norm": 2.0428847989715573,
      "learning_rate": 2.5535093746874025e-06,
      "loss": 0.484,
      "step": 6044
    },
    {
      "epoch": 7.399020807833537,
      "grad_norm": 2.563410262197895,
      "learning_rate": 2.5528686399788084e-06,
      "loss": 0.5497,
      "step": 6045
    },
    {
      "epoch": 7.400244798041616,
      "grad_norm": 2.064452724969936,
      "learning_rate": 2.5522279017958825e-06,
      "loss": 1.0837,
      "step": 6046
    },
    {
      "epoch": 7.401468788249694,
      "grad_norm": 1.3582553212127324,
      "learning_rate": 2.5515871601807317e-06,
      "loss": 0.4853,
      "step": 6047
    },
    {
      "epoch": 7.402692778457772,
      "grad_norm": 1.4551110282720963,
      "learning_rate": 2.550946415175464e-06,
      "loss": 0.6064,
      "step": 6048
    },
    {
      "epoch": 7.403916768665851,
      "grad_norm": 1.307933988581647,
      "learning_rate": 2.550305666822186e-06,
      "loss": 0.3652,
      "step": 6049
    },
    {
      "epoch": 7.405140758873929,
      "grad_norm": 2.3156079333811928,
      "learning_rate": 2.5496649151630066e-06,
      "loss": 0.5322,
      "step": 6050
    },
    {
      "epoch": 7.406364749082007,
      "grad_norm": 2.157254253031142,
      "learning_rate": 2.5490241602400314e-06,
      "loss": 0.4637,
      "step": 6051
    },
    {
      "epoch": 7.407588739290086,
      "grad_norm": 1.3861457953854517,
      "learning_rate": 2.54838340209537e-06,
      "loss": 0.6231,
      "step": 6052
    },
    {
      "epoch": 7.408812729498164,
      "grad_norm": 1.8153677261853844,
      "learning_rate": 2.547742640771131e-06,
      "loss": 0.7423,
      "step": 6053
    },
    {
      "epoch": 7.410036719706242,
      "grad_norm": 2.2002649238526257,
      "learning_rate": 2.5471018763094217e-06,
      "loss": 0.5958,
      "step": 6054
    },
    {
      "epoch": 7.411260709914321,
      "grad_norm": 1.4210700281835575,
      "learning_rate": 2.5464611087523512e-06,
      "loss": 1.0351,
      "step": 6055
    },
    {
      "epoch": 7.412484700122399,
      "grad_norm": 1.3243872910376806,
      "learning_rate": 2.5458203381420293e-06,
      "loss": 0.5714,
      "step": 6056
    },
    {
      "epoch": 7.413708690330477,
      "grad_norm": 1.1507425252721053,
      "learning_rate": 2.545179564520563e-06,
      "loss": 0.6652,
      "step": 6057
    },
    {
      "epoch": 7.414932680538556,
      "grad_norm": 2.1944067274760153,
      "learning_rate": 2.5445387879300635e-06,
      "loss": 0.441,
      "step": 6058
    },
    {
      "epoch": 7.416156670746634,
      "grad_norm": 1.8559542235995465,
      "learning_rate": 2.54389800841264e-06,
      "loss": 1.084,
      "step": 6059
    },
    {
      "epoch": 7.417380660954713,
      "grad_norm": 1.7419001103062157,
      "learning_rate": 2.5432572260104015e-06,
      "loss": 0.6576,
      "step": 6060
    },
    {
      "epoch": 7.4186046511627906,
      "grad_norm": 2.177037347828862,
      "learning_rate": 2.542616440765458e-06,
      "loss": 0.5749,
      "step": 6061
    },
    {
      "epoch": 7.419828641370869,
      "grad_norm": 1.773442561004639,
      "learning_rate": 2.5419756527199197e-06,
      "loss": 0.5626,
      "step": 6062
    },
    {
      "epoch": 7.421052631578947,
      "grad_norm": 2.261743758341832,
      "learning_rate": 2.5413348619158966e-06,
      "loss": 1.2993,
      "step": 6063
    },
    {
      "epoch": 7.4222766217870255,
      "grad_norm": 1.6215971394101416,
      "learning_rate": 2.540694068395499e-06,
      "loss": 0.4993,
      "step": 6064
    },
    {
      "epoch": 7.423500611995104,
      "grad_norm": 1.661564268430868,
      "learning_rate": 2.5400532722008387e-06,
      "loss": 0.3755,
      "step": 6065
    },
    {
      "epoch": 7.424724602203183,
      "grad_norm": 2.0565498046155204,
      "learning_rate": 2.5394124733740244e-06,
      "loss": 1.4262,
      "step": 6066
    },
    {
      "epoch": 7.4259485924112605,
      "grad_norm": 1.1584912495717106,
      "learning_rate": 2.538771671957169e-06,
      "loss": 0.6035,
      "step": 6067
    },
    {
      "epoch": 7.427172582619339,
      "grad_norm": 1.114414684605462,
      "learning_rate": 2.5381308679923823e-06,
      "loss": 0.9898,
      "step": 6068
    },
    {
      "epoch": 7.428396572827418,
      "grad_norm": 1.115430515591259,
      "learning_rate": 2.537490061521776e-06,
      "loss": 0.5815,
      "step": 6069
    },
    {
      "epoch": 7.4296205630354955,
      "grad_norm": 2.056839365082865,
      "learning_rate": 2.5368492525874618e-06,
      "loss": 1.0949,
      "step": 6070
    },
    {
      "epoch": 7.430844553243574,
      "grad_norm": 1.3166015936150672,
      "learning_rate": 2.53620844123155e-06,
      "loss": 1.0839,
      "step": 6071
    },
    {
      "epoch": 7.432068543451653,
      "grad_norm": 1.4857091172164065,
      "learning_rate": 2.5355676274961545e-06,
      "loss": 1.1826,
      "step": 6072
    },
    {
      "epoch": 7.43329253365973,
      "grad_norm": 3.3417370795016774,
      "learning_rate": 2.534926811423386e-06,
      "loss": 0.3828,
      "step": 6073
    },
    {
      "epoch": 7.434516523867809,
      "grad_norm": 2.6423860827186,
      "learning_rate": 2.5342859930553565e-06,
      "loss": 0.3655,
      "step": 6074
    },
    {
      "epoch": 7.435740514075888,
      "grad_norm": 1.0231370277035152,
      "learning_rate": 2.533645172434178e-06,
      "loss": 0.7014,
      "step": 6075
    },
    {
      "epoch": 7.436964504283965,
      "grad_norm": 0.8758688470822585,
      "learning_rate": 2.5330043496019627e-06,
      "loss": 0.4196,
      "step": 6076
    },
    {
      "epoch": 7.438188494492044,
      "grad_norm": 1.3175164223023497,
      "learning_rate": 2.532363524600825e-06,
      "loss": 0.6184,
      "step": 6077
    },
    {
      "epoch": 7.439412484700123,
      "grad_norm": 1.5641720443453775,
      "learning_rate": 2.5317226974728755e-06,
      "loss": 0.6691,
      "step": 6078
    },
    {
      "epoch": 7.4406364749082,
      "grad_norm": 2.473677380793289,
      "learning_rate": 2.5310818682602272e-06,
      "loss": 0.3745,
      "step": 6079
    },
    {
      "epoch": 7.441860465116279,
      "grad_norm": 1.0078888435839604,
      "learning_rate": 2.530441037004994e-06,
      "loss": 0.5138,
      "step": 6080
    },
    {
      "epoch": 7.443084455324358,
      "grad_norm": 1.2697819413615676,
      "learning_rate": 2.529800203749288e-06,
      "loss": 0.6325,
      "step": 6081
    },
    {
      "epoch": 7.444308445532435,
      "grad_norm": 1.261019962161274,
      "learning_rate": 2.529159368535224e-06,
      "loss": 0.8768,
      "step": 6082
    },
    {
      "epoch": 7.445532435740514,
      "grad_norm": 1.4605885455653669,
      "learning_rate": 2.528518531404913e-06,
      "loss": 0.5653,
      "step": 6083
    },
    {
      "epoch": 7.4467564259485926,
      "grad_norm": 1.3866405830366417,
      "learning_rate": 2.52787769240047e-06,
      "loss": 0.3686,
      "step": 6084
    },
    {
      "epoch": 7.447980416156671,
      "grad_norm": 2.1603253230839345,
      "learning_rate": 2.5272368515640084e-06,
      "loss": 0.4335,
      "step": 6085
    },
    {
      "epoch": 7.449204406364749,
      "grad_norm": 2.0814573678520008,
      "learning_rate": 2.5265960089376413e-06,
      "loss": 0.9118,
      "step": 6086
    },
    {
      "epoch": 7.4504283965728275,
      "grad_norm": 1.428007629846331,
      "learning_rate": 2.5259551645634828e-06,
      "loss": 0.5262,
      "step": 6087
    },
    {
      "epoch": 7.451652386780906,
      "grad_norm": 1.7400594089519827,
      "learning_rate": 2.525314318483647e-06,
      "loss": 0.553,
      "step": 6088
    },
    {
      "epoch": 7.452876376988984,
      "grad_norm": 2.359503683843971,
      "learning_rate": 2.524673470740248e-06,
      "loss": 0.9138,
      "step": 6089
    },
    {
      "epoch": 7.4541003671970625,
      "grad_norm": 2.253000641667843,
      "learning_rate": 2.5240326213753995e-06,
      "loss": 0.6265,
      "step": 6090
    },
    {
      "epoch": 7.455324357405141,
      "grad_norm": 0.9975225721173564,
      "learning_rate": 2.5233917704312167e-06,
      "loss": 0.6489,
      "step": 6091
    },
    {
      "epoch": 7.456548347613219,
      "grad_norm": 1.135679768286183,
      "learning_rate": 2.5227509179498123e-06,
      "loss": 0.8544,
      "step": 6092
    },
    {
      "epoch": 7.4577723378212974,
      "grad_norm": 1.5668430875013584,
      "learning_rate": 2.522110063973302e-06,
      "loss": 0.5319,
      "step": 6093
    },
    {
      "epoch": 7.458996328029376,
      "grad_norm": 1.5782835417774252,
      "learning_rate": 2.5214692085438003e-06,
      "loss": 0.5376,
      "step": 6094
    },
    {
      "epoch": 7.460220318237454,
      "grad_norm": 1.4155577174887026,
      "learning_rate": 2.5208283517034216e-06,
      "loss": 1.5809,
      "step": 6095
    },
    {
      "epoch": 7.461444308445532,
      "grad_norm": 0.9646864837251788,
      "learning_rate": 2.5201874934942804e-06,
      "loss": 0.5343,
      "step": 6096
    },
    {
      "epoch": 7.462668298653611,
      "grad_norm": 1.3259814230881448,
      "learning_rate": 2.519546633958492e-06,
      "loss": 1.3215,
      "step": 6097
    },
    {
      "epoch": 7.463892288861689,
      "grad_norm": 1.343208098366222,
      "learning_rate": 2.5189057731381716e-06,
      "loss": 1.2401,
      "step": 6098
    },
    {
      "epoch": 7.465116279069767,
      "grad_norm": 1.7666065002358464,
      "learning_rate": 2.5182649110754325e-06,
      "loss": 0.9722,
      "step": 6099
    },
    {
      "epoch": 7.466340269277846,
      "grad_norm": 1.4353883403584344,
      "learning_rate": 2.5176240478123926e-06,
      "loss": 0.5292,
      "step": 6100
    },
    {
      "epoch": 7.467564259485924,
      "grad_norm": 2.6125114949282318,
      "learning_rate": 2.516983183391164e-06,
      "loss": 0.9789,
      "step": 6101
    },
    {
      "epoch": 7.468788249694002,
      "grad_norm": 1.1786854395072885,
      "learning_rate": 2.5163423178538643e-06,
      "loss": 0.5218,
      "step": 6102
    },
    {
      "epoch": 7.470012239902081,
      "grad_norm": 1.5921586691449594,
      "learning_rate": 2.5157014512426074e-06,
      "loss": 0.6438,
      "step": 6103
    },
    {
      "epoch": 7.47123623011016,
      "grad_norm": 1.4335768291245108,
      "learning_rate": 2.5150605835995097e-06,
      "loss": 1.181,
      "step": 6104
    },
    {
      "epoch": 7.472460220318237,
      "grad_norm": 1.600357431311894,
      "learning_rate": 2.514419714966685e-06,
      "loss": 0.9433,
      "step": 6105
    },
    {
      "epoch": 7.473684210526316,
      "grad_norm": 2.634090099506393,
      "learning_rate": 2.5137788453862515e-06,
      "loss": 0.4718,
      "step": 6106
    },
    {
      "epoch": 7.474908200734394,
      "grad_norm": 1.1243857294647757,
      "learning_rate": 2.513137974900322e-06,
      "loss": 0.6561,
      "step": 6107
    },
    {
      "epoch": 7.476132190942472,
      "grad_norm": 1.0107949875225157,
      "learning_rate": 2.5124971035510143e-06,
      "loss": 0.5745,
      "step": 6108
    },
    {
      "epoch": 7.477356181150551,
      "grad_norm": 1.682807901670551,
      "learning_rate": 2.5118562313804433e-06,
      "loss": 0.9745,
      "step": 6109
    },
    {
      "epoch": 7.4785801713586295,
      "grad_norm": 1.8956441938366775,
      "learning_rate": 2.5112153584307237e-06,
      "loss": 0.5204,
      "step": 6110
    },
    {
      "epoch": 7.479804161566707,
      "grad_norm": 1.7273283634810426,
      "learning_rate": 2.5105744847439727e-06,
      "loss": 0.5925,
      "step": 6111
    },
    {
      "epoch": 7.481028151774786,
      "grad_norm": 2.519069616280341,
      "learning_rate": 2.5099336103623057e-06,
      "loss": 0.4316,
      "step": 6112
    },
    {
      "epoch": 7.4822521419828645,
      "grad_norm": 1.0888019478896982,
      "learning_rate": 2.5092927353278385e-06,
      "loss": 0.5887,
      "step": 6113
    },
    {
      "epoch": 7.483476132190942,
      "grad_norm": 1.307803378318719,
      "learning_rate": 2.508651859682687e-06,
      "loss": 0.5175,
      "step": 6114
    },
    {
      "epoch": 7.484700122399021,
      "grad_norm": 1.2957058222450901,
      "learning_rate": 2.508010983468968e-06,
      "loss": 0.5343,
      "step": 6115
    },
    {
      "epoch": 7.4859241126070994,
      "grad_norm": 2.473302128036359,
      "learning_rate": 2.5073701067287957e-06,
      "loss": 0.6085,
      "step": 6116
    },
    {
      "epoch": 7.487148102815177,
      "grad_norm": 1.9825347141532652,
      "learning_rate": 2.506729229504288e-06,
      "loss": 1.3518,
      "step": 6117
    },
    {
      "epoch": 7.488372093023256,
      "grad_norm": 2.4278534696682805,
      "learning_rate": 2.50608835183756e-06,
      "loss": 0.452,
      "step": 6118
    },
    {
      "epoch": 7.489596083231334,
      "grad_norm": 1.4042621705679088,
      "learning_rate": 2.505447473770728e-06,
      "loss": 0.4876,
      "step": 6119
    },
    {
      "epoch": 7.490820073439412,
      "grad_norm": 2.853619906164199,
      "learning_rate": 2.5048065953459084e-06,
      "loss": 0.4658,
      "step": 6120
    },
    {
      "epoch": 7.492044063647491,
      "grad_norm": 1.0801733655163022,
      "learning_rate": 2.5041657166052167e-06,
      "loss": 0.4502,
      "step": 6121
    },
    {
      "epoch": 7.493268053855569,
      "grad_norm": 1.301191635370929,
      "learning_rate": 2.50352483759077e-06,
      "loss": 0.9583,
      "step": 6122
    },
    {
      "epoch": 7.494492044063647,
      "grad_norm": 1.3723130538358295,
      "learning_rate": 2.502883958344683e-06,
      "loss": 0.4184,
      "step": 6123
    },
    {
      "epoch": 7.495716034271726,
      "grad_norm": 1.5822303801143285,
      "learning_rate": 2.5022430789090744e-06,
      "loss": 0.5624,
      "step": 6124
    },
    {
      "epoch": 7.496940024479804,
      "grad_norm": 1.7427664277732544,
      "learning_rate": 2.5016021993260576e-06,
      "loss": 0.5342,
      "step": 6125
    },
    {
      "epoch": 7.498164014687882,
      "grad_norm": 1.749077395392347,
      "learning_rate": 2.500961319637751e-06,
      "loss": 0.5482,
      "step": 6126
    },
    {
      "epoch": 7.499388004895961,
      "grad_norm": 1.4061645417990274,
      "learning_rate": 2.50032043988627e-06,
      "loss": 0.4169,
      "step": 6127
    },
    {
      "epoch": 7.500611995104039,
      "grad_norm": 1.2150993147614775,
      "learning_rate": 2.4996795601137306e-06,
      "loss": 0.747,
      "step": 6128
    },
    {
      "epoch": 7.501835985312118,
      "grad_norm": 2.3880122818659246,
      "learning_rate": 2.4990386803622496e-06,
      "loss": 0.936,
      "step": 6129
    },
    {
      "epoch": 7.503059975520196,
      "grad_norm": 1.494976701947248,
      "learning_rate": 2.4983978006739423e-06,
      "loss": 0.9921,
      "step": 6130
    },
    {
      "epoch": 7.504283965728274,
      "grad_norm": 1.8449890083394016,
      "learning_rate": 2.4977569210909264e-06,
      "loss": 0.6514,
      "step": 6131
    },
    {
      "epoch": 7.505507955936353,
      "grad_norm": 2.324051551638898,
      "learning_rate": 2.4971160416553175e-06,
      "loss": 0.4027,
      "step": 6132
    },
    {
      "epoch": 7.506731946144431,
      "grad_norm": 1.3527543558257142,
      "learning_rate": 2.4964751624092313e-06,
      "loss": 0.5866,
      "step": 6133
    },
    {
      "epoch": 7.507955936352509,
      "grad_norm": 1.4325695097896,
      "learning_rate": 2.4958342833947845e-06,
      "loss": 1.2231,
      "step": 6134
    },
    {
      "epoch": 7.509179926560588,
      "grad_norm": 1.137352236656296,
      "learning_rate": 2.4951934046540925e-06,
      "loss": 0.5679,
      "step": 6135
    },
    {
      "epoch": 7.510403916768666,
      "grad_norm": 1.8264961678546414,
      "learning_rate": 2.494552526229273e-06,
      "loss": 1.0493,
      "step": 6136
    },
    {
      "epoch": 7.511627906976744,
      "grad_norm": 1.7644000961128532,
      "learning_rate": 2.493911648162441e-06,
      "loss": 0.3761,
      "step": 6137
    },
    {
      "epoch": 7.512851897184823,
      "grad_norm": 1.381992645957713,
      "learning_rate": 2.493270770495713e-06,
      "loss": 1.2692,
      "step": 6138
    },
    {
      "epoch": 7.5140758873929006,
      "grad_norm": 1.4655481617006292,
      "learning_rate": 2.4926298932712047e-06,
      "loss": 0.4009,
      "step": 6139
    },
    {
      "epoch": 7.515299877600979,
      "grad_norm": 1.1307880941080095,
      "learning_rate": 2.491989016531033e-06,
      "loss": 0.7596,
      "step": 6140
    },
    {
      "epoch": 7.516523867809058,
      "grad_norm": 1.9805973774014445,
      "learning_rate": 2.491348140317314e-06,
      "loss": 0.5981,
      "step": 6141
    },
    {
      "epoch": 7.5177478580171355,
      "grad_norm": 1.4961849676718475,
      "learning_rate": 2.4907072646721624e-06,
      "loss": 0.5663,
      "step": 6142
    },
    {
      "epoch": 7.518971848225214,
      "grad_norm": 1.8049682202330257,
      "learning_rate": 2.4900663896376943e-06,
      "loss": 1.1664,
      "step": 6143
    },
    {
      "epoch": 7.520195838433293,
      "grad_norm": 1.585377021525601,
      "learning_rate": 2.4894255152560277e-06,
      "loss": 0.7413,
      "step": 6144
    },
    {
      "epoch": 7.5214198286413705,
      "grad_norm": 1.0410506622265685,
      "learning_rate": 2.488784641569277e-06,
      "loss": 1.2065,
      "step": 6145
    },
    {
      "epoch": 7.522643818849449,
      "grad_norm": 0.8798615939374048,
      "learning_rate": 2.488143768619558e-06,
      "loss": 0.4493,
      "step": 6146
    },
    {
      "epoch": 7.523867809057528,
      "grad_norm": 2.04603809562867,
      "learning_rate": 2.4875028964489866e-06,
      "loss": 0.9204,
      "step": 6147
    },
    {
      "epoch": 7.525091799265606,
      "grad_norm": 1.0347449185868562,
      "learning_rate": 2.486862025099678e-06,
      "loss": 0.754,
      "step": 6148
    },
    {
      "epoch": 7.526315789473684,
      "grad_norm": 2.0068491129132755,
      "learning_rate": 2.486221154613749e-06,
      "loss": 0.9766,
      "step": 6149
    },
    {
      "epoch": 7.527539779681763,
      "grad_norm": 1.7001312515044822,
      "learning_rate": 2.4855802850333154e-06,
      "loss": 0.4098,
      "step": 6150
    },
    {
      "epoch": 7.52876376988984,
      "grad_norm": 1.2976965266991234,
      "learning_rate": 2.4849394164004915e-06,
      "loss": 0.3456,
      "step": 6151
    },
    {
      "epoch": 7.529987760097919,
      "grad_norm": 1.3653949192717019,
      "learning_rate": 2.484298548757393e-06,
      "loss": 0.9337,
      "step": 6152
    },
    {
      "epoch": 7.531211750305998,
      "grad_norm": 0.9637523287868172,
      "learning_rate": 2.4836576821461365e-06,
      "loss": 0.5394,
      "step": 6153
    },
    {
      "epoch": 7.532435740514076,
      "grad_norm": 1.3216007338905937,
      "learning_rate": 2.483016816608837e-06,
      "loss": 0.3456,
      "step": 6154
    },
    {
      "epoch": 7.533659730722154,
      "grad_norm": 2.8816887578333383,
      "learning_rate": 2.4823759521876086e-06,
      "loss": 0.5446,
      "step": 6155
    },
    {
      "epoch": 7.534883720930233,
      "grad_norm": 1.270419355149507,
      "learning_rate": 2.4817350889245675e-06,
      "loss": 0.6048,
      "step": 6156
    },
    {
      "epoch": 7.536107711138311,
      "grad_norm": 1.1831225778341916,
      "learning_rate": 2.481094226861829e-06,
      "loss": 1.0994,
      "step": 6157
    },
    {
      "epoch": 7.537331701346389,
      "grad_norm": 1.6050683035933266,
      "learning_rate": 2.4804533660415083e-06,
      "loss": 0.5377,
      "step": 6158
    },
    {
      "epoch": 7.538555691554468,
      "grad_norm": 1.5722216752246725,
      "learning_rate": 2.4798125065057204e-06,
      "loss": 0.5369,
      "step": 6159
    },
    {
      "epoch": 7.539779681762546,
      "grad_norm": 1.3144109460143343,
      "learning_rate": 2.4791716482965797e-06,
      "loss": 0.6304,
      "step": 6160
    },
    {
      "epoch": 7.541003671970624,
      "grad_norm": 1.3440486915253003,
      "learning_rate": 2.4785307914562006e-06,
      "loss": 0.5402,
      "step": 6161
    },
    {
      "epoch": 7.5422276621787026,
      "grad_norm": 1.855158171545453,
      "learning_rate": 2.477889936026699e-06,
      "loss": 0.328,
      "step": 6162
    },
    {
      "epoch": 7.543451652386781,
      "grad_norm": 2.555566079549113,
      "learning_rate": 2.4772490820501885e-06,
      "loss": 0.4215,
      "step": 6163
    },
    {
      "epoch": 7.544675642594859,
      "grad_norm": 2.225302554369664,
      "learning_rate": 2.4766082295687846e-06,
      "loss": 0.9965,
      "step": 6164
    },
    {
      "epoch": 7.5458996328029375,
      "grad_norm": 1.782038690005208,
      "learning_rate": 2.475967378624601e-06,
      "loss": 0.5003,
      "step": 6165
    },
    {
      "epoch": 7.547123623011016,
      "grad_norm": 1.5245038204727228,
      "learning_rate": 2.4753265292597526e-06,
      "loss": 0.6906,
      "step": 6166
    },
    {
      "epoch": 7.548347613219094,
      "grad_norm": 0.9250030276374989,
      "learning_rate": 2.4746856815163532e-06,
      "loss": 0.5258,
      "step": 6167
    },
    {
      "epoch": 7.5495716034271725,
      "grad_norm": 2.6412871717958306,
      "learning_rate": 2.474044835436518e-06,
      "loss": 0.7593,
      "step": 6168
    },
    {
      "epoch": 7.550795593635251,
      "grad_norm": 2.1788147743753123,
      "learning_rate": 2.473403991062359e-06,
      "loss": 0.4403,
      "step": 6169
    },
    {
      "epoch": 7.552019583843329,
      "grad_norm": 1.0332792672357254,
      "learning_rate": 2.4727631484359924e-06,
      "loss": 0.5177,
      "step": 6170
    },
    {
      "epoch": 7.5532435740514074,
      "grad_norm": 1.0163362117626535,
      "learning_rate": 2.4721223075995305e-06,
      "loss": 0.5431,
      "step": 6171
    },
    {
      "epoch": 7.554467564259486,
      "grad_norm": 1.7209597917710562,
      "learning_rate": 2.4714814685950877e-06,
      "loss": 0.5419,
      "step": 6172
    },
    {
      "epoch": 7.555691554467565,
      "grad_norm": 1.4928979692641338,
      "learning_rate": 2.4708406314647774e-06,
      "loss": 0.4453,
      "step": 6173
    },
    {
      "epoch": 7.556915544675642,
      "grad_norm": 1.937759142300473,
      "learning_rate": 2.4701997962507118e-06,
      "loss": 1.0288,
      "step": 6174
    },
    {
      "epoch": 7.558139534883721,
      "grad_norm": 2.487098659903279,
      "learning_rate": 2.4695589629950067e-06,
      "loss": 0.5094,
      "step": 6175
    },
    {
      "epoch": 7.5593635250918,
      "grad_norm": 1.9549178695520475,
      "learning_rate": 2.468918131739773e-06,
      "loss": 0.467,
      "step": 6176
    },
    {
      "epoch": 7.560587515299877,
      "grad_norm": 1.1030077766603401,
      "learning_rate": 2.4682773025271257e-06,
      "loss": 0.5383,
      "step": 6177
    },
    {
      "epoch": 7.561811505507956,
      "grad_norm": 2.3391975890814485,
      "learning_rate": 2.4676364753991755e-06,
      "loss": 0.3654,
      "step": 6178
    },
    {
      "epoch": 7.563035495716035,
      "grad_norm": 1.1007951263944946,
      "learning_rate": 2.4669956503980377e-06,
      "loss": 0.5637,
      "step": 6179
    },
    {
      "epoch": 7.564259485924112,
      "grad_norm": 1.2349550291088451,
      "learning_rate": 2.466354827565823e-06,
      "loss": 0.5676,
      "step": 6180
    },
    {
      "epoch": 7.565483476132191,
      "grad_norm": 1.5544509845760528,
      "learning_rate": 2.465714006944645e-06,
      "loss": 1.3462,
      "step": 6181
    },
    {
      "epoch": 7.56670746634027,
      "grad_norm": 2.2877357376139735,
      "learning_rate": 2.4650731885766148e-06,
      "loss": 0.6382,
      "step": 6182
    },
    {
      "epoch": 7.567931456548347,
      "grad_norm": 1.6477121465072988,
      "learning_rate": 2.464432372503846e-06,
      "loss": 0.4418,
      "step": 6183
    },
    {
      "epoch": 7.569155446756426,
      "grad_norm": 1.582173657706063,
      "learning_rate": 2.4637915587684504e-06,
      "loss": 1.1874,
      "step": 6184
    },
    {
      "epoch": 7.5703794369645045,
      "grad_norm": 2.1312024366042808,
      "learning_rate": 2.4631507474125395e-06,
      "loss": 1.215,
      "step": 6185
    },
    {
      "epoch": 7.571603427172582,
      "grad_norm": 2.258728231016943,
      "learning_rate": 2.4625099384782255e-06,
      "loss": 0.6891,
      "step": 6186
    },
    {
      "epoch": 7.572827417380661,
      "grad_norm": 1.3687724820410727,
      "learning_rate": 2.461869132007618e-06,
      "loss": 0.5797,
      "step": 6187
    },
    {
      "epoch": 7.5740514075887395,
      "grad_norm": 1.6136110482120973,
      "learning_rate": 2.461228328042832e-06,
      "loss": 0.3648,
      "step": 6188
    },
    {
      "epoch": 7.575275397796817,
      "grad_norm": 1.9060356335013666,
      "learning_rate": 2.4605875266259765e-06,
      "loss": 0.3199,
      "step": 6189
    },
    {
      "epoch": 7.576499388004896,
      "grad_norm": 1.5012683925296622,
      "learning_rate": 2.4599467277991626e-06,
      "loss": 1.187,
      "step": 6190
    },
    {
      "epoch": 7.5777233782129745,
      "grad_norm": 0.8857766948880677,
      "learning_rate": 2.459305931604501e-06,
      "loss": 0.4477,
      "step": 6191
    },
    {
      "epoch": 7.578947368421053,
      "grad_norm": 1.4412875126489144,
      "learning_rate": 2.458665138084104e-06,
      "loss": 0.7029,
      "step": 6192
    },
    {
      "epoch": 7.580171358629131,
      "grad_norm": 1.4201882135030912,
      "learning_rate": 2.458024347280081e-06,
      "loss": 0.4768,
      "step": 6193
    },
    {
      "epoch": 7.5813953488372094,
      "grad_norm": 1.4089622610472727,
      "learning_rate": 2.4573835592345427e-06,
      "loss": 0.5801,
      "step": 6194
    },
    {
      "epoch": 7.582619339045287,
      "grad_norm": 1.4999634611024162,
      "learning_rate": 2.456742773989599e-06,
      "loss": 1.2802,
      "step": 6195
    },
    {
      "epoch": 7.583843329253366,
      "grad_norm": 1.8881913110797863,
      "learning_rate": 2.4561019915873604e-06,
      "loss": 0.7345,
      "step": 6196
    },
    {
      "epoch": 7.585067319461444,
      "grad_norm": 1.0666398175570457,
      "learning_rate": 2.455461212069937e-06,
      "loss": 0.4911,
      "step": 6197
    },
    {
      "epoch": 7.586291309669523,
      "grad_norm": 1.0347178879115895,
      "learning_rate": 2.4548204354794375e-06,
      "loss": 0.5655,
      "step": 6198
    },
    {
      "epoch": 7.587515299877601,
      "grad_norm": 2.313442434996858,
      "learning_rate": 2.4541796618579723e-06,
      "loss": 0.4504,
      "step": 6199
    },
    {
      "epoch": 7.588739290085679,
      "grad_norm": 1.4687889938849121,
      "learning_rate": 2.453538891247649e-06,
      "loss": 1.3075,
      "step": 6200
    },
    {
      "epoch": 7.589963280293758,
      "grad_norm": 1.5248883463126752,
      "learning_rate": 2.452898123690579e-06,
      "loss": 0.5861,
      "step": 6201
    },
    {
      "epoch": 7.591187270501836,
      "grad_norm": 1.5927991781122914,
      "learning_rate": 2.45225735922887e-06,
      "loss": 1.5123,
      "step": 6202
    },
    {
      "epoch": 7.592411260709914,
      "grad_norm": 1.8443181155383792,
      "learning_rate": 2.4516165979046307e-06,
      "loss": 1.3071,
      "step": 6203
    },
    {
      "epoch": 7.593635250917993,
      "grad_norm": 2.06521145936607,
      "learning_rate": 2.450975839759969e-06,
      "loss": 1.0036,
      "step": 6204
    },
    {
      "epoch": 7.594859241126071,
      "grad_norm": 2.204816751222035,
      "learning_rate": 2.4503350848369946e-06,
      "loss": 0.5562,
      "step": 6205
    },
    {
      "epoch": 7.596083231334149,
      "grad_norm": 1.6438652922680626,
      "learning_rate": 2.4496943331778148e-06,
      "loss": 0.5493,
      "step": 6206
    },
    {
      "epoch": 7.597307221542228,
      "grad_norm": 1.2916760846874298,
      "learning_rate": 2.449053584824537e-06,
      "loss": 0.4622,
      "step": 6207
    },
    {
      "epoch": 7.598531211750306,
      "grad_norm": 2.742208568838488,
      "learning_rate": 2.4484128398192687e-06,
      "loss": 0.3963,
      "step": 6208
    },
    {
      "epoch": 7.599755201958384,
      "grad_norm": 0.973799891074145,
      "learning_rate": 2.4477720982041183e-06,
      "loss": 0.5163,
      "step": 6209
    },
    {
      "epoch": 7.600979192166463,
      "grad_norm": 1.6912694135163884,
      "learning_rate": 2.447131360021193e-06,
      "loss": 0.9395,
      "step": 6210
    },
    {
      "epoch": 7.602203182374541,
      "grad_norm": 1.9911398351479648,
      "learning_rate": 2.446490625312598e-06,
      "loss": 0.5565,
      "step": 6211
    },
    {
      "epoch": 7.603427172582619,
      "grad_norm": 1.51081141073198,
      "learning_rate": 2.4458498941204425e-06,
      "loss": 1.4627,
      "step": 6212
    },
    {
      "epoch": 7.604651162790698,
      "grad_norm": 1.7309570874065212,
      "learning_rate": 2.44520916648683e-06,
      "loss": 1.1202,
      "step": 6213
    },
    {
      "epoch": 7.605875152998776,
      "grad_norm": 1.6369259654068378,
      "learning_rate": 2.4445684424538694e-06,
      "loss": 0.4763,
      "step": 6214
    },
    {
      "epoch": 7.607099143206854,
      "grad_norm": 2.0860915338917656,
      "learning_rate": 2.4439277220636664e-06,
      "loss": 0.395,
      "step": 6215
    },
    {
      "epoch": 7.608323133414933,
      "grad_norm": 1.4879094317381925,
      "learning_rate": 2.443287005358325e-06,
      "loss": 0.48,
      "step": 6216
    },
    {
      "epoch": 7.6095471236230114,
      "grad_norm": 2.8472398931786556,
      "learning_rate": 2.442646292379952e-06,
      "loss": 0.7206,
      "step": 6217
    },
    {
      "epoch": 7.610771113831089,
      "grad_norm": 1.471902825244244,
      "learning_rate": 2.4420055831706527e-06,
      "loss": 1.056,
      "step": 6218
    },
    {
      "epoch": 7.611995104039168,
      "grad_norm": 1.237349400651744,
      "learning_rate": 2.441364877772532e-06,
      "loss": 0.5482,
      "step": 6219
    },
    {
      "epoch": 7.613219094247246,
      "grad_norm": 0.5161203670723004,
      "learning_rate": 2.440724176227695e-06,
      "loss": 0.1315,
      "step": 6220
    },
    {
      "epoch": 7.614443084455324,
      "grad_norm": 1.1971232670851863,
      "learning_rate": 2.440083478578245e-06,
      "loss": 0.4746,
      "step": 6221
    },
    {
      "epoch": 7.615667074663403,
      "grad_norm": 0.9967736846013807,
      "learning_rate": 2.4394427848662878e-06,
      "loss": 0.5871,
      "step": 6222
    },
    {
      "epoch": 7.616891064871481,
      "grad_norm": 1.8068241737006503,
      "learning_rate": 2.438802095133927e-06,
      "loss": 0.5218,
      "step": 6223
    },
    {
      "epoch": 7.618115055079559,
      "grad_norm": 1.4531866880299504,
      "learning_rate": 2.438161409423266e-06,
      "loss": 0.4691,
      "step": 6224
    },
    {
      "epoch": 7.619339045287638,
      "grad_norm": 1.5649251337875758,
      "learning_rate": 2.4375207277764085e-06,
      "loss": 1.2042,
      "step": 6225
    },
    {
      "epoch": 7.620563035495716,
      "grad_norm": 1.7562804351514105,
      "learning_rate": 2.436880050235457e-06,
      "loss": 0.4066,
      "step": 6226
    },
    {
      "epoch": 7.621787025703794,
      "grad_norm": 1.8638113156111307,
      "learning_rate": 2.436239376842516e-06,
      "loss": 0.3976,
      "step": 6227
    },
    {
      "epoch": 7.623011015911873,
      "grad_norm": 1.1807536612498002,
      "learning_rate": 2.4355987076396866e-06,
      "loss": 0.7388,
      "step": 6228
    },
    {
      "epoch": 7.624235006119951,
      "grad_norm": 0.9468665454483438,
      "learning_rate": 2.4349580426690726e-06,
      "loss": 0.5534,
      "step": 6229
    },
    {
      "epoch": 7.625458996328029,
      "grad_norm": 1.572080801639012,
      "learning_rate": 2.4343173819727745e-06,
      "loss": 0.9236,
      "step": 6230
    },
    {
      "epoch": 7.626682986536108,
      "grad_norm": 1.6746264866290577,
      "learning_rate": 2.433676725592896e-06,
      "loss": 0.9248,
      "step": 6231
    },
    {
      "epoch": 7.627906976744186,
      "grad_norm": 1.585817840031627,
      "learning_rate": 2.4330360735715373e-06,
      "loss": 1.1442,
      "step": 6232
    },
    {
      "epoch": 7.629130966952264,
      "grad_norm": 1.6831160490355674,
      "learning_rate": 2.4323954259508013e-06,
      "loss": 1.224,
      "step": 6233
    },
    {
      "epoch": 7.630354957160343,
      "grad_norm": 1.8084710852761299,
      "learning_rate": 2.4317547827727857e-06,
      "loss": 0.6087,
      "step": 6234
    },
    {
      "epoch": 7.631578947368421,
      "grad_norm": 1.5035544444211026,
      "learning_rate": 2.4311141440795956e-06,
      "loss": 1.6324,
      "step": 6235
    },
    {
      "epoch": 7.6328029375765,
      "grad_norm": 1.8984633157375777,
      "learning_rate": 2.430473509913328e-06,
      "loss": 0.5269,
      "step": 6236
    },
    {
      "epoch": 7.634026927784578,
      "grad_norm": 1.7178517287730044,
      "learning_rate": 2.4298328803160846e-06,
      "loss": 0.4117,
      "step": 6237
    },
    {
      "epoch": 7.635250917992656,
      "grad_norm": 2.2175263216684113,
      "learning_rate": 2.429192255329964e-06,
      "loss": 0.4293,
      "step": 6238
    },
    {
      "epoch": 7.636474908200734,
      "grad_norm": 1.7116906762761694,
      "learning_rate": 2.428551634997067e-06,
      "loss": 0.5445,
      "step": 6239
    },
    {
      "epoch": 7.6376988984088126,
      "grad_norm": 1.4045885904958415,
      "learning_rate": 2.427911019359492e-06,
      "loss": 0.8066,
      "step": 6240
    },
    {
      "epoch": 7.638922888616891,
      "grad_norm": 1.4135724080715015,
      "learning_rate": 2.4272704084593395e-06,
      "loss": 0.5844,
      "step": 6241
    },
    {
      "epoch": 7.64014687882497,
      "grad_norm": 1.3644228705805737,
      "learning_rate": 2.4266298023387057e-06,
      "loss": 1.1678,
      "step": 6242
    },
    {
      "epoch": 7.6413708690330475,
      "grad_norm": 1.4871828035988977,
      "learning_rate": 2.425989201039689e-06,
      "loss": 1.5654,
      "step": 6243
    },
    {
      "epoch": 7.642594859241126,
      "grad_norm": 1.892404838136676,
      "learning_rate": 2.4253486046043894e-06,
      "loss": 0.6167,
      "step": 6244
    },
    {
      "epoch": 7.643818849449205,
      "grad_norm": 1.619577195135675,
      "learning_rate": 2.424708013074903e-06,
      "loss": 1.8137,
      "step": 6245
    },
    {
      "epoch": 7.6450428396572825,
      "grad_norm": 1.8095615753452048,
      "learning_rate": 2.424067426493328e-06,
      "loss": 0.7472,
      "step": 6246
    },
    {
      "epoch": 7.646266829865361,
      "grad_norm": 1.912249938441006,
      "learning_rate": 2.42342684490176e-06,
      "loss": 0.4095,
      "step": 6247
    },
    {
      "epoch": 7.64749082007344,
      "grad_norm": 2.3371232499700407,
      "learning_rate": 2.422786268342297e-06,
      "loss": 0.4365,
      "step": 6248
    },
    {
      "epoch": 7.6487148102815175,
      "grad_norm": 1.4694554086745333,
      "learning_rate": 2.422145696857035e-06,
      "loss": 0.4678,
      "step": 6249
    },
    {
      "epoch": 7.649938800489596,
      "grad_norm": 0.9788789431326116,
      "learning_rate": 2.4215051304880703e-06,
      "loss": 0.3716,
      "step": 6250
    },
    {
      "epoch": 7.651162790697675,
      "grad_norm": 2.6841692152994026,
      "learning_rate": 2.420864569277498e-06,
      "loss": 0.4684,
      "step": 6251
    },
    {
      "epoch": 7.652386780905752,
      "grad_norm": 1.503854197678877,
      "learning_rate": 2.420224013267412e-06,
      "loss": 0.3198,
      "step": 6252
    },
    {
      "epoch": 7.653610771113831,
      "grad_norm": 1.685191763320888,
      "learning_rate": 2.41958346249991e-06,
      "loss": 0.5617,
      "step": 6253
    },
    {
      "epoch": 7.65483476132191,
      "grad_norm": 2.1161571476178858,
      "learning_rate": 2.4189429170170855e-06,
      "loss": 0.4662,
      "step": 6254
    },
    {
      "epoch": 7.656058751529987,
      "grad_norm": 2.007661624166391,
      "learning_rate": 2.4183023768610326e-06,
      "loss": 0.3695,
      "step": 6255
    },
    {
      "epoch": 7.657282741738066,
      "grad_norm": 1.47509801785704,
      "learning_rate": 2.4176618420738448e-06,
      "loss": 0.3845,
      "step": 6256
    },
    {
      "epoch": 7.658506731946145,
      "grad_norm": 1.5606134156352238,
      "learning_rate": 2.417021312697616e-06,
      "loss": 0.514,
      "step": 6257
    },
    {
      "epoch": 7.659730722154222,
      "grad_norm": 1.085935129413594,
      "learning_rate": 2.41638078877444e-06,
      "loss": 0.5511,
      "step": 6258
    },
    {
      "epoch": 7.660954712362301,
      "grad_norm": 1.460596803726325,
      "learning_rate": 2.4157402703464104e-06,
      "loss": 0.9451,
      "step": 6259
    },
    {
      "epoch": 7.66217870257038,
      "grad_norm": 2.178821744866951,
      "learning_rate": 2.4150997574556163e-06,
      "loss": 0.4788,
      "step": 6260
    },
    {
      "epoch": 7.663402692778458,
      "grad_norm": 1.4466052506333684,
      "learning_rate": 2.4144592501441537e-06,
      "loss": 0.9296,
      "step": 6261
    },
    {
      "epoch": 7.664626682986536,
      "grad_norm": 3.25146181825888,
      "learning_rate": 2.4138187484541133e-06,
      "loss": 0.4451,
      "step": 6262
    },
    {
      "epoch": 7.6658506731946146,
      "grad_norm": 1.5887728459523804,
      "learning_rate": 2.413178252427585e-06,
      "loss": 1.3736,
      "step": 6263
    },
    {
      "epoch": 7.667074663402692,
      "grad_norm": 1.7056958099439616,
      "learning_rate": 2.412537762106661e-06,
      "loss": 0.4412,
      "step": 6264
    },
    {
      "epoch": 7.668298653610771,
      "grad_norm": 1.707456420735935,
      "learning_rate": 2.411897277533431e-06,
      "loss": 0.4722,
      "step": 6265
    },
    {
      "epoch": 7.6695226438188495,
      "grad_norm": 1.4126627764745834,
      "learning_rate": 2.411256798749987e-06,
      "loss": 0.3693,
      "step": 6266
    },
    {
      "epoch": 7.670746634026928,
      "grad_norm": 1.2354146276656335,
      "learning_rate": 2.410616325798418e-06,
      "loss": 0.5632,
      "step": 6267
    },
    {
      "epoch": 7.671970624235006,
      "grad_norm": 1.35179943647802,
      "learning_rate": 2.4099758587208137e-06,
      "loss": 0.6805,
      "step": 6268
    },
    {
      "epoch": 7.6731946144430845,
      "grad_norm": 2.406035828537886,
      "learning_rate": 2.4093353975592615e-06,
      "loss": 1.197,
      "step": 6269
    },
    {
      "epoch": 7.674418604651163,
      "grad_norm": 1.9472459813692993,
      "learning_rate": 2.408694942355853e-06,
      "loss": 1.189,
      "step": 6270
    },
    {
      "epoch": 7.675642594859241,
      "grad_norm": 1.557370363693338,
      "learning_rate": 2.4080544931526754e-06,
      "loss": 0.4453,
      "step": 6271
    },
    {
      "epoch": 7.6768665850673194,
      "grad_norm": 1.5985013741129968,
      "learning_rate": 2.407414049991816e-06,
      "loss": 0.5127,
      "step": 6272
    },
    {
      "epoch": 7.678090575275398,
      "grad_norm": 1.2860299535465822,
      "learning_rate": 2.406773612915362e-06,
      "loss": 0.6761,
      "step": 6273
    },
    {
      "epoch": 7.679314565483476,
      "grad_norm": 1.7220747341400386,
      "learning_rate": 2.4061331819654017e-06,
      "loss": 0.5378,
      "step": 6274
    },
    {
      "epoch": 7.680538555691554,
      "grad_norm": 2.563087509405035,
      "learning_rate": 2.405492757184022e-06,
      "loss": 0.515,
      "step": 6275
    },
    {
      "epoch": 7.681762545899633,
      "grad_norm": 0.9585274519546628,
      "learning_rate": 2.4048523386133083e-06,
      "loss": 0.5008,
      "step": 6276
    },
    {
      "epoch": 7.682986536107711,
      "grad_norm": 1.284346851400765,
      "learning_rate": 2.404211926295348e-06,
      "loss": 1.1623,
      "step": 6277
    },
    {
      "epoch": 7.684210526315789,
      "grad_norm": 1.4836959481927843,
      "learning_rate": 2.403571520272224e-06,
      "loss": 0.7308,
      "step": 6278
    },
    {
      "epoch": 7.685434516523868,
      "grad_norm": 1.674298073222216,
      "learning_rate": 2.402931120586024e-06,
      "loss": 0.3522,
      "step": 6279
    },
    {
      "epoch": 7.686658506731947,
      "grad_norm": 1.160416681341403,
      "learning_rate": 2.4022907272788316e-06,
      "loss": 0.7253,
      "step": 6280
    },
    {
      "epoch": 7.687882496940024,
      "grad_norm": 1.9402700990365052,
      "learning_rate": 2.401650340392731e-06,
      "loss": 0.4848,
      "step": 6281
    },
    {
      "epoch": 7.689106487148103,
      "grad_norm": 1.701940655457374,
      "learning_rate": 2.4010099599698063e-06,
      "loss": 0.4956,
      "step": 6282
    },
    {
      "epoch": 7.690330477356181,
      "grad_norm": 1.0851092405916003,
      "learning_rate": 2.4003695860521407e-06,
      "loss": 0.483,
      "step": 6283
    },
    {
      "epoch": 7.691554467564259,
      "grad_norm": 1.469383587352406,
      "learning_rate": 2.399729218681818e-06,
      "loss": 1.0257,
      "step": 6284
    },
    {
      "epoch": 7.692778457772338,
      "grad_norm": 1.620170030492761,
      "learning_rate": 2.39908885790092e-06,
      "loss": 0.5549,
      "step": 6285
    },
    {
      "epoch": 7.6940024479804165,
      "grad_norm": 1.0742852397369316,
      "learning_rate": 2.3984485037515296e-06,
      "loss": 0.6346,
      "step": 6286
    },
    {
      "epoch": 7.695226438188494,
      "grad_norm": 1.6728796504819454,
      "learning_rate": 2.3978081562757263e-06,
      "loss": 0.4313,
      "step": 6287
    },
    {
      "epoch": 7.696450428396573,
      "grad_norm": 1.8087834824117477,
      "learning_rate": 2.3971678155155944e-06,
      "loss": 0.7892,
      "step": 6288
    },
    {
      "epoch": 7.6976744186046515,
      "grad_norm": 1.6663826636637065,
      "learning_rate": 2.3965274815132125e-06,
      "loss": 0.4694,
      "step": 6289
    },
    {
      "epoch": 7.698898408812729,
      "grad_norm": 1.810430028534234,
      "learning_rate": 2.3958871543106625e-06,
      "loss": 0.5074,
      "step": 6290
    },
    {
      "epoch": 7.700122399020808,
      "grad_norm": 1.9761849178085087,
      "learning_rate": 2.3952468339500223e-06,
      "loss": 0.5615,
      "step": 6291
    },
    {
      "epoch": 7.7013463892288865,
      "grad_norm": 1.3203164613433664,
      "learning_rate": 2.394606520473374e-06,
      "loss": 0.6015,
      "step": 6292
    },
    {
      "epoch": 7.702570379436964,
      "grad_norm": 1.9003992897669886,
      "learning_rate": 2.393966213922795e-06,
      "loss": 0.4477,
      "step": 6293
    },
    {
      "epoch": 7.703794369645043,
      "grad_norm": 1.5205187979013979,
      "learning_rate": 2.3933259143403643e-06,
      "loss": 1.0942,
      "step": 6294
    },
    {
      "epoch": 7.7050183598531214,
      "grad_norm": 2.075422300090956,
      "learning_rate": 2.392685621768159e-06,
      "loss": 0.4797,
      "step": 6295
    },
    {
      "epoch": 7.706242350061199,
      "grad_norm": 2.1080163922110353,
      "learning_rate": 2.392045336248259e-06,
      "loss": 1.054,
      "step": 6296
    },
    {
      "epoch": 7.707466340269278,
      "grad_norm": 2.521287934213974,
      "learning_rate": 2.39140505782274e-06,
      "loss": 0.482,
      "step": 6297
    },
    {
      "epoch": 7.708690330477356,
      "grad_norm": 1.1255339854096258,
      "learning_rate": 2.390764786533679e-06,
      "loss": 0.6667,
      "step": 6298
    },
    {
      "epoch": 7.709914320685434,
      "grad_norm": 2.0873958870050355,
      "learning_rate": 2.390124522423152e-06,
      "loss": 0.5135,
      "step": 6299
    },
    {
      "epoch": 7.711138310893513,
      "grad_norm": 1.7285187535577733,
      "learning_rate": 2.3894842655332345e-06,
      "loss": 0.6161,
      "step": 6300
    },
    {
      "epoch": 7.712362301101591,
      "grad_norm": 2.141739276923362,
      "learning_rate": 2.388844015906003e-06,
      "loss": 0.6883,
      "step": 6301
    },
    {
      "epoch": 7.713586291309669,
      "grad_norm": 1.8117702710525805,
      "learning_rate": 2.388203773583531e-06,
      "loss": 0.4267,
      "step": 6302
    },
    {
      "epoch": 7.714810281517748,
      "grad_norm": 2.054150420700014,
      "learning_rate": 2.387563538607894e-06,
      "loss": 0.5478,
      "step": 6303
    },
    {
      "epoch": 7.716034271725826,
      "grad_norm": 1.7056453358179677,
      "learning_rate": 2.3869233110211647e-06,
      "loss": 1.0227,
      "step": 6304
    },
    {
      "epoch": 7.717258261933905,
      "grad_norm": 1.5546274238279476,
      "learning_rate": 2.3862830908654176e-06,
      "loss": 0.5198,
      "step": 6305
    },
    {
      "epoch": 7.718482252141983,
      "grad_norm": 1.3456353053041012,
      "learning_rate": 2.3856428781827257e-06,
      "loss": 0.5727,
      "step": 6306
    },
    {
      "epoch": 7.719706242350061,
      "grad_norm": 1.3914494861276754,
      "learning_rate": 2.3850026730151605e-06,
      "loss": 0.472,
      "step": 6307
    },
    {
      "epoch": 7.720930232558139,
      "grad_norm": 2.399434701292347,
      "learning_rate": 2.3843624754047936e-06,
      "loss": 0.9649,
      "step": 6308
    },
    {
      "epoch": 7.722154222766218,
      "grad_norm": 2.4004347447445302,
      "learning_rate": 2.383722285393698e-06,
      "loss": 0.5537,
      "step": 6309
    },
    {
      "epoch": 7.723378212974296,
      "grad_norm": 1.2993857470447243,
      "learning_rate": 2.3830821030239436e-06,
      "loss": 0.5429,
      "step": 6310
    },
    {
      "epoch": 7.724602203182375,
      "grad_norm": 1.3721975438360952,
      "learning_rate": 2.3824419283376005e-06,
      "loss": 1.3298,
      "step": 6311
    },
    {
      "epoch": 7.725826193390453,
      "grad_norm": 3.27607214112602,
      "learning_rate": 2.3818017613767392e-06,
      "loss": 0.788,
      "step": 6312
    },
    {
      "epoch": 7.727050183598531,
      "grad_norm": 1.5472215007488848,
      "learning_rate": 2.381161602183429e-06,
      "loss": 1.2436,
      "step": 6313
    },
    {
      "epoch": 7.72827417380661,
      "grad_norm": 2.128367492821983,
      "learning_rate": 2.3805214507997384e-06,
      "loss": 0.4874,
      "step": 6314
    },
    {
      "epoch": 7.729498164014688,
      "grad_norm": 1.2876238287066035,
      "learning_rate": 2.3798813072677373e-06,
      "loss": 1.3427,
      "step": 6315
    },
    {
      "epoch": 7.730722154222766,
      "grad_norm": 1.4484059469227706,
      "learning_rate": 2.3792411716294916e-06,
      "loss": 0.48,
      "step": 6316
    },
    {
      "epoch": 7.731946144430845,
      "grad_norm": 2.3858442345723665,
      "learning_rate": 2.378601043927069e-06,
      "loss": 0.8368,
      "step": 6317
    },
    {
      "epoch": 7.7331701346389226,
      "grad_norm": 1.0493042938955108,
      "learning_rate": 2.377960924202537e-06,
      "loss": 0.509,
      "step": 6318
    },
    {
      "epoch": 7.734394124847001,
      "grad_norm": 1.8323422053817044,
      "learning_rate": 2.3773208124979624e-06,
      "loss": 0.5863,
      "step": 6319
    },
    {
      "epoch": 7.73561811505508,
      "grad_norm": 2.260552549434383,
      "learning_rate": 2.3766807088554096e-06,
      "loss": 0.5954,
      "step": 6320
    },
    {
      "epoch": 7.7368421052631575,
      "grad_norm": 2.069251599982254,
      "learning_rate": 2.376040613316944e-06,
      "loss": 1.0334,
      "step": 6321
    },
    {
      "epoch": 7.738066095471236,
      "grad_norm": 1.4603185437822357,
      "learning_rate": 2.375400525924632e-06,
      "loss": 0.614,
      "step": 6322
    },
    {
      "epoch": 7.739290085679315,
      "grad_norm": 1.5334338214125567,
      "learning_rate": 2.3747604467205358e-06,
      "loss": 0.4542,
      "step": 6323
    },
    {
      "epoch": 7.740514075887393,
      "grad_norm": 2.117592068512626,
      "learning_rate": 2.3741203757467207e-06,
      "loss": 0.5541,
      "step": 6324
    },
    {
      "epoch": 7.741738066095471,
      "grad_norm": 1.098187444815803,
      "learning_rate": 2.3734803130452484e-06,
      "loss": 0.8942,
      "step": 6325
    },
    {
      "epoch": 7.74296205630355,
      "grad_norm": 1.498316405164169,
      "learning_rate": 2.372840258658181e-06,
      "loss": 0.7155,
      "step": 6326
    },
    {
      "epoch": 7.7441860465116275,
      "grad_norm": 1.482855421167908,
      "learning_rate": 2.3722002126275826e-06,
      "loss": 0.4709,
      "step": 6327
    },
    {
      "epoch": 7.745410036719706,
      "grad_norm": 1.3195068943399302,
      "learning_rate": 2.371560174995513e-06,
      "loss": 0.3725,
      "step": 6328
    },
    {
      "epoch": 7.746634026927785,
      "grad_norm": 2.5282448420952837,
      "learning_rate": 2.370920145804034e-06,
      "loss": 0.4291,
      "step": 6329
    },
    {
      "epoch": 7.747858017135863,
      "grad_norm": 2.004223365493421,
      "learning_rate": 2.370280125095205e-06,
      "loss": 0.5266,
      "step": 6330
    },
    {
      "epoch": 7.749082007343941,
      "grad_norm": 1.7539124504367005,
      "learning_rate": 2.369640112911087e-06,
      "loss": 0.744,
      "step": 6331
    },
    {
      "epoch": 7.75030599755202,
      "grad_norm": 1.7978011052772684,
      "learning_rate": 2.3690001092937383e-06,
      "loss": 0.5888,
      "step": 6332
    },
    {
      "epoch": 7.751529987760098,
      "grad_norm": 2.042500572979575,
      "learning_rate": 2.3683601142852184e-06,
      "loss": 0.6346,
      "step": 6333
    },
    {
      "epoch": 7.752753977968176,
      "grad_norm": 1.1373810389311845,
      "learning_rate": 2.3677201279275837e-06,
      "loss": 0.8285,
      "step": 6334
    },
    {
      "epoch": 7.753977968176255,
      "grad_norm": 1.8786890086638723,
      "learning_rate": 2.367080150262894e-06,
      "loss": 0.6426,
      "step": 6335
    },
    {
      "epoch": 7.755201958384333,
      "grad_norm": 1.0178942015443644,
      "learning_rate": 2.366440181333205e-06,
      "loss": 0.3776,
      "step": 6336
    },
    {
      "epoch": 7.756425948592411,
      "grad_norm": 2.251537826628075,
      "learning_rate": 2.365800221180573e-06,
      "loss": 0.2485,
      "step": 6337
    },
    {
      "epoch": 7.75764993880049,
      "grad_norm": 1.2350812837365237,
      "learning_rate": 2.3651602698470546e-06,
      "loss": 0.9825,
      "step": 6338
    },
    {
      "epoch": 7.758873929008568,
      "grad_norm": 2.056608369937828,
      "learning_rate": 2.3645203273747038e-06,
      "loss": 0.4328,
      "step": 6339
    },
    {
      "epoch": 7.760097919216646,
      "grad_norm": 2.1123474810372285,
      "learning_rate": 2.3638803938055765e-06,
      "loss": 0.3032,
      "step": 6340
    },
    {
      "epoch": 7.7613219094247246,
      "grad_norm": 1.4472691014130052,
      "learning_rate": 2.363240469181726e-06,
      "loss": 0.4991,
      "step": 6341
    },
    {
      "epoch": 7.762545899632803,
      "grad_norm": 1.9241302467442303,
      "learning_rate": 2.362600553545207e-06,
      "loss": 1.0187,
      "step": 6342
    },
    {
      "epoch": 7.763769889840881,
      "grad_norm": 1.6489050147899753,
      "learning_rate": 2.3619606469380693e-06,
      "loss": 0.72,
      "step": 6343
    },
    {
      "epoch": 7.7649938800489595,
      "grad_norm": 1.0742292818660073,
      "learning_rate": 2.361320749402369e-06,
      "loss": 0.4866,
      "step": 6344
    },
    {
      "epoch": 7.766217870257038,
      "grad_norm": 2.465369289846065,
      "learning_rate": 2.3606808609801558e-06,
      "loss": 0.9686,
      "step": 6345
    },
    {
      "epoch": 7.767441860465116,
      "grad_norm": 1.3968199881787542,
      "learning_rate": 2.360040981713481e-06,
      "loss": 1.0953,
      "step": 6346
    },
    {
      "epoch": 7.7686658506731945,
      "grad_norm": 1.760733944969053,
      "learning_rate": 2.3594011116443944e-06,
      "loss": 0.5066,
      "step": 6347
    },
    {
      "epoch": 7.769889840881273,
      "grad_norm": 1.3996303004579171,
      "learning_rate": 2.3587612508149478e-06,
      "loss": 1.413,
      "step": 6348
    },
    {
      "epoch": 7.771113831089352,
      "grad_norm": 1.6047709436251225,
      "learning_rate": 2.358121399267189e-06,
      "loss": 0.4265,
      "step": 6349
    },
    {
      "epoch": 7.7723378212974294,
      "grad_norm": 1.3355744914131111,
      "learning_rate": 2.3574815570431675e-06,
      "loss": 0.4883,
      "step": 6350
    },
    {
      "epoch": 7.773561811505508,
      "grad_norm": 1.33032966528709,
      "learning_rate": 2.356841724184931e-06,
      "loss": 0.4623,
      "step": 6351
    },
    {
      "epoch": 7.774785801713586,
      "grad_norm": 1.4077646611104828,
      "learning_rate": 2.3562019007345255e-06,
      "loss": 0.5741,
      "step": 6352
    },
    {
      "epoch": 7.776009791921664,
      "grad_norm": 1.7243617817751655,
      "learning_rate": 2.355562086734001e-06,
      "loss": 0.5009,
      "step": 6353
    },
    {
      "epoch": 7.777233782129743,
      "grad_norm": 1.6802018562370642,
      "learning_rate": 2.354922282225401e-06,
      "loss": 1.3587,
      "step": 6354
    },
    {
      "epoch": 7.778457772337822,
      "grad_norm": 1.3457874947541661,
      "learning_rate": 2.3542824872507723e-06,
      "loss": 0.7931,
      "step": 6355
    },
    {
      "epoch": 7.779681762545899,
      "grad_norm": 1.7305667743062343,
      "learning_rate": 2.3536427018521588e-06,
      "loss": 0.4725,
      "step": 6356
    },
    {
      "epoch": 7.780905752753978,
      "grad_norm": 2.024355493387664,
      "learning_rate": 2.353002926071606e-06,
      "loss": 0.4963,
      "step": 6357
    },
    {
      "epoch": 7.782129742962057,
      "grad_norm": 1.885957084810247,
      "learning_rate": 2.3523631599511574e-06,
      "loss": 1.2003,
      "step": 6358
    },
    {
      "epoch": 7.783353733170134,
      "grad_norm": 1.6099951447344143,
      "learning_rate": 2.3517234035328563e-06,
      "loss": 0.7466,
      "step": 6359
    },
    {
      "epoch": 7.784577723378213,
      "grad_norm": 2.7128212155577365,
      "learning_rate": 2.3510836568587434e-06,
      "loss": 0.567,
      "step": 6360
    },
    {
      "epoch": 7.785801713586292,
      "grad_norm": 1.6942467173483322,
      "learning_rate": 2.3504439199708625e-06,
      "loss": 0.4741,
      "step": 6361
    },
    {
      "epoch": 7.787025703794369,
      "grad_norm": 1.7649952586292341,
      "learning_rate": 2.349804192911255e-06,
      "loss": 0.4415,
      "step": 6362
    },
    {
      "epoch": 7.788249694002448,
      "grad_norm": 1.6012010501675753,
      "learning_rate": 2.3491644757219595e-06,
      "loss": 1.2034,
      "step": 6363
    },
    {
      "epoch": 7.7894736842105265,
      "grad_norm": 2.236564495282154,
      "learning_rate": 2.3485247684450164e-06,
      "loss": 0.3381,
      "step": 6364
    },
    {
      "epoch": 7.790697674418604,
      "grad_norm": 1.7695946863961953,
      "learning_rate": 2.3478850711224655e-06,
      "loss": 1.1516,
      "step": 6365
    },
    {
      "epoch": 7.791921664626683,
      "grad_norm": 1.945488839822804,
      "learning_rate": 2.3472453837963454e-06,
      "loss": 0.569,
      "step": 6366
    },
    {
      "epoch": 7.7931456548347615,
      "grad_norm": 1.2522616716007797,
      "learning_rate": 2.3466057065086935e-06,
      "loss": 0.4403,
      "step": 6367
    },
    {
      "epoch": 7.79436964504284,
      "grad_norm": 2.367155316204829,
      "learning_rate": 2.3459660393015478e-06,
      "loss": 0.7491,
      "step": 6368
    },
    {
      "epoch": 7.795593635250918,
      "grad_norm": 1.006276573669132,
      "learning_rate": 2.345326382216942e-06,
      "loss": 0.5452,
      "step": 6369
    },
    {
      "epoch": 7.7968176254589965,
      "grad_norm": 2.1771923151439396,
      "learning_rate": 2.3446867352969164e-06,
      "loss": 0.4818,
      "step": 6370
    },
    {
      "epoch": 7.798041615667074,
      "grad_norm": 1.4827532711194809,
      "learning_rate": 2.344047098583504e-06,
      "loss": 0.6535,
      "step": 6371
    },
    {
      "epoch": 7.799265605875153,
      "grad_norm": 1.6928400523280505,
      "learning_rate": 2.343407472118739e-06,
      "loss": 1.8947,
      "step": 6372
    },
    {
      "epoch": 7.8004895960832314,
      "grad_norm": 2.999869818653983,
      "learning_rate": 2.3427678559446554e-06,
      "loss": 0.4963,
      "step": 6373
    },
    {
      "epoch": 7.80171358629131,
      "grad_norm": 1.2997550884305609,
      "learning_rate": 2.342128250103287e-06,
      "loss": 0.598,
      "step": 6374
    },
    {
      "epoch": 7.802937576499388,
      "grad_norm": 1.2982665403350968,
      "learning_rate": 2.3414886546366665e-06,
      "loss": 0.8818,
      "step": 6375
    },
    {
      "epoch": 7.804161566707466,
      "grad_norm": 1.7790068173929274,
      "learning_rate": 2.340849069586825e-06,
      "loss": 0.5406,
      "step": 6376
    },
    {
      "epoch": 7.805385556915545,
      "grad_norm": 1.7934018327783172,
      "learning_rate": 2.3402094949957948e-06,
      "loss": 0.3088,
      "step": 6377
    },
    {
      "epoch": 7.806609547123623,
      "grad_norm": 2.884363091439924,
      "learning_rate": 2.3395699309056036e-06,
      "loss": 0.4004,
      "step": 6378
    },
    {
      "epoch": 7.807833537331701,
      "grad_norm": 1.7104944041457413,
      "learning_rate": 2.338930377358284e-06,
      "loss": 0.5988,
      "step": 6379
    },
    {
      "epoch": 7.80905752753978,
      "grad_norm": 1.828553207410003,
      "learning_rate": 2.3382908343958656e-06,
      "loss": 0.4237,
      "step": 6380
    },
    {
      "epoch": 7.810281517747858,
      "grad_norm": 0.8560473536461888,
      "learning_rate": 2.3376513020603742e-06,
      "loss": 0.3522,
      "step": 6381
    },
    {
      "epoch": 7.811505507955936,
      "grad_norm": 2.187164878712133,
      "learning_rate": 2.3370117803938388e-06,
      "loss": 0.5092,
      "step": 6382
    },
    {
      "epoch": 7.812729498164015,
      "grad_norm": 0.8581788747608755,
      "learning_rate": 2.336372269438286e-06,
      "loss": 0.3962,
      "step": 6383
    },
    {
      "epoch": 7.813953488372093,
      "grad_norm": 2.2238904289202077,
      "learning_rate": 2.3357327692357434e-06,
      "loss": 1.0314,
      "step": 6384
    },
    {
      "epoch": 7.815177478580171,
      "grad_norm": 1.811620978599382,
      "learning_rate": 2.335093279828235e-06,
      "loss": 1.009,
      "step": 6385
    },
    {
      "epoch": 7.81640146878825,
      "grad_norm": 2.5321778542782707,
      "learning_rate": 2.3344538012577863e-06,
      "loss": 0.2292,
      "step": 6386
    },
    {
      "epoch": 7.817625458996328,
      "grad_norm": 1.9701060897032372,
      "learning_rate": 2.3338143335664213e-06,
      "loss": 0.9867,
      "step": 6387
    },
    {
      "epoch": 7.818849449204406,
      "grad_norm": 2.7290725194464724,
      "learning_rate": 2.3331748767961647e-06,
      "loss": 0.3862,
      "step": 6388
    },
    {
      "epoch": 7.820073439412485,
      "grad_norm": 1.3232573521713817,
      "learning_rate": 2.3325354309890373e-06,
      "loss": 0.7225,
      "step": 6389
    },
    {
      "epoch": 7.821297429620563,
      "grad_norm": 2.0045064400462578,
      "learning_rate": 2.331895996187062e-06,
      "loss": 0.3664,
      "step": 6390
    },
    {
      "epoch": 7.822521419828641,
      "grad_norm": 2.1190042295328255,
      "learning_rate": 2.3312565724322596e-06,
      "loss": 0.436,
      "step": 6391
    },
    {
      "epoch": 7.82374541003672,
      "grad_norm": 2.9902093146959623,
      "learning_rate": 2.330617159766652e-06,
      "loss": 0.3684,
      "step": 6392
    },
    {
      "epoch": 7.8249694002447985,
      "grad_norm": 1.4003514046150987,
      "learning_rate": 2.329977758232258e-06,
      "loss": 1.4202,
      "step": 6393
    },
    {
      "epoch": 7.826193390452876,
      "grad_norm": 1.4795763990798554,
      "learning_rate": 2.3293383678710965e-06,
      "loss": 0.6359,
      "step": 6394
    },
    {
      "epoch": 7.827417380660955,
      "grad_norm": 2.4029286591475456,
      "learning_rate": 2.328698988725186e-06,
      "loss": 0.6086,
      "step": 6395
    },
    {
      "epoch": 7.828641370869033,
      "grad_norm": 2.568768644899412,
      "learning_rate": 2.328059620836545e-06,
      "loss": 0.3747,
      "step": 6396
    },
    {
      "epoch": 7.829865361077111,
      "grad_norm": 1.2690905295348887,
      "learning_rate": 2.32742026424719e-06,
      "loss": 1.0527,
      "step": 6397
    },
    {
      "epoch": 7.83108935128519,
      "grad_norm": 1.4997677769604212,
      "learning_rate": 2.3267809189991363e-06,
      "loss": 0.4673,
      "step": 6398
    },
    {
      "epoch": 7.832313341493268,
      "grad_norm": 1.43383328323072,
      "learning_rate": 2.3261415851343987e-06,
      "loss": 0.5766,
      "step": 6399
    },
    {
      "epoch": 7.833537331701346,
      "grad_norm": 2.402075423468278,
      "learning_rate": 2.325502262694995e-06,
      "loss": 0.4113,
      "step": 6400
    },
    {
      "epoch": 7.834761321909425,
      "grad_norm": 2.002035304717365,
      "learning_rate": 2.3248629517229366e-06,
      "loss": 0.389,
      "step": 6401
    },
    {
      "epoch": 7.835985312117503,
      "grad_norm": 2.15502089538727,
      "learning_rate": 2.3242236522602373e-06,
      "loss": 0.9879,
      "step": 6402
    },
    {
      "epoch": 7.837209302325581,
      "grad_norm": 0.8787532889013411,
      "learning_rate": 2.3235843643489095e-06,
      "loss": 0.4706,
      "step": 6403
    },
    {
      "epoch": 7.83843329253366,
      "grad_norm": 1.9807158728249974,
      "learning_rate": 2.322945088030964e-06,
      "loss": 1.1447,
      "step": 6404
    },
    {
      "epoch": 7.839657282741738,
      "grad_norm": 1.9719233390286839,
      "learning_rate": 2.322305823348413e-06,
      "loss": 0.7824,
      "step": 6405
    },
    {
      "epoch": 7.840881272949816,
      "grad_norm": 1.0417988660377782,
      "learning_rate": 2.3216665703432667e-06,
      "loss": 0.6053,
      "step": 6406
    },
    {
      "epoch": 7.842105263157895,
      "grad_norm": 1.4186704140327342,
      "learning_rate": 2.3210273290575333e-06,
      "loss": 0.7206,
      "step": 6407
    },
    {
      "epoch": 7.843329253365973,
      "grad_norm": 0.9068147912213598,
      "learning_rate": 2.3203880995332208e-06,
      "loss": 0.5186,
      "step": 6408
    },
    {
      "epoch": 7.844553243574051,
      "grad_norm": 1.2460194225261694,
      "learning_rate": 2.319748881812339e-06,
      "loss": 0.4177,
      "step": 6409
    },
    {
      "epoch": 7.84577723378213,
      "grad_norm": 1.2848295013715836,
      "learning_rate": 2.3191096759368943e-06,
      "loss": 0.4538,
      "step": 6410
    },
    {
      "epoch": 7.847001223990208,
      "grad_norm": 1.0454490295934582,
      "learning_rate": 2.3184704819488923e-06,
      "loss": 0.3956,
      "step": 6411
    },
    {
      "epoch": 7.848225214198287,
      "grad_norm": 1.2479954063672936,
      "learning_rate": 2.317831299890338e-06,
      "loss": 0.7033,
      "step": 6412
    },
    {
      "epoch": 7.849449204406365,
      "grad_norm": 1.717865191901609,
      "learning_rate": 2.317192129803238e-06,
      "loss": 0.4988,
      "step": 6413
    },
    {
      "epoch": 7.850673194614443,
      "grad_norm": 1.38676001828163,
      "learning_rate": 2.316552971729595e-06,
      "loss": 0.4838,
      "step": 6414
    },
    {
      "epoch": 7.851897184822521,
      "grad_norm": 1.5680380507891953,
      "learning_rate": 2.315913825711412e-06,
      "loss": 1.5325,
      "step": 6415
    },
    {
      "epoch": 7.8531211750306,
      "grad_norm": 1.4821264761333008,
      "learning_rate": 2.3152746917906916e-06,
      "loss": 0.5067,
      "step": 6416
    },
    {
      "epoch": 7.854345165238678,
      "grad_norm": 1.8098155642268712,
      "learning_rate": 2.3146355700094346e-06,
      "loss": 0.4316,
      "step": 6417
    },
    {
      "epoch": 7.855569155446757,
      "grad_norm": 1.2846579422771365,
      "learning_rate": 2.3139964604096426e-06,
      "loss": 1.1405,
      "step": 6418
    },
    {
      "epoch": 7.8567931456548346,
      "grad_norm": 1.0021234331769409,
      "learning_rate": 2.3133573630333157e-06,
      "loss": 0.5545,
      "step": 6419
    },
    {
      "epoch": 7.858017135862913,
      "grad_norm": 1.1620423428888107,
      "learning_rate": 2.312718277922452e-06,
      "loss": 0.6131,
      "step": 6420
    },
    {
      "epoch": 7.859241126070992,
      "grad_norm": 1.2451838058765947,
      "learning_rate": 2.3120792051190498e-06,
      "loss": 0.3992,
      "step": 6421
    },
    {
      "epoch": 7.8604651162790695,
      "grad_norm": 2.124742232014125,
      "learning_rate": 2.3114401446651084e-06,
      "loss": 0.537,
      "step": 6422
    },
    {
      "epoch": 7.861689106487148,
      "grad_norm": 1.2268336856535829,
      "learning_rate": 2.3108010966026226e-06,
      "loss": 0.6358,
      "step": 6423
    },
    {
      "epoch": 7.862913096695227,
      "grad_norm": 1.3706022485440579,
      "learning_rate": 2.3101620609735898e-06,
      "loss": 1.1281,
      "step": 6424
    },
    {
      "epoch": 7.8641370869033045,
      "grad_norm": 1.3991198972302028,
      "learning_rate": 2.309523037820002e-06,
      "loss": 0.5473,
      "step": 6425
    },
    {
      "epoch": 7.865361077111383,
      "grad_norm": 1.0517474575405679,
      "learning_rate": 2.308884027183858e-06,
      "loss": 0.8532,
      "step": 6426
    },
    {
      "epoch": 7.866585067319462,
      "grad_norm": 2.5357071214159426,
      "learning_rate": 2.3082450291071484e-06,
      "loss": 1.0745,
      "step": 6427
    },
    {
      "epoch": 7.8678090575275395,
      "grad_norm": 2.1568818226425774,
      "learning_rate": 2.3076060436318656e-06,
      "loss": 0.5034,
      "step": 6428
    },
    {
      "epoch": 7.869033047735618,
      "grad_norm": 2.2941299182510395,
      "learning_rate": 2.3069670708000026e-06,
      "loss": 0.621,
      "step": 6429
    },
    {
      "epoch": 7.870257037943697,
      "grad_norm": 2.0064956137754004,
      "learning_rate": 2.3063281106535486e-06,
      "loss": 0.5618,
      "step": 6430
    },
    {
      "epoch": 7.871481028151774,
      "grad_norm": 1.4646369976533984,
      "learning_rate": 2.305689163234496e-06,
      "loss": 1.553,
      "step": 6431
    },
    {
      "epoch": 7.872705018359853,
      "grad_norm": 1.5949644017463636,
      "learning_rate": 2.3050502285848322e-06,
      "loss": 0.6333,
      "step": 6432
    },
    {
      "epoch": 7.873929008567932,
      "grad_norm": 0.9464520176511044,
      "learning_rate": 2.304411306746547e-06,
      "loss": 0.6017,
      "step": 6433
    },
    {
      "epoch": 7.875152998776009,
      "grad_norm": 2.6321413989284608,
      "learning_rate": 2.303772397761626e-06,
      "loss": 0.5349,
      "step": 6434
    },
    {
      "epoch": 7.876376988984088,
      "grad_norm": 1.4101368969377561,
      "learning_rate": 2.3031335016720587e-06,
      "loss": 0.7616,
      "step": 6435
    },
    {
      "epoch": 7.877600979192167,
      "grad_norm": 1.005176405975124,
      "learning_rate": 2.302494618519829e-06,
      "loss": 0.4248,
      "step": 6436
    },
    {
      "epoch": 7.878824969400245,
      "grad_norm": 2.1429830850601372,
      "learning_rate": 2.301855748346922e-06,
      "loss": 0.4186,
      "step": 6437
    },
    {
      "epoch": 7.880048959608323,
      "grad_norm": 1.3705738582832643,
      "learning_rate": 2.301216891195323e-06,
      "loss": 1.174,
      "step": 6438
    },
    {
      "epoch": 7.881272949816402,
      "grad_norm": 1.9023268429407107,
      "learning_rate": 2.300578047107014e-06,
      "loss": 1.1379,
      "step": 6439
    },
    {
      "epoch": 7.882496940024479,
      "grad_norm": 2.0008588840539696,
      "learning_rate": 2.2999392161239783e-06,
      "loss": 0.8351,
      "step": 6440
    },
    {
      "epoch": 7.883720930232558,
      "grad_norm": 1.6679836005669844,
      "learning_rate": 2.2993003982881976e-06,
      "loss": 0.7588,
      "step": 6441
    },
    {
      "epoch": 7.8849449204406366,
      "grad_norm": 2.834046041133698,
      "learning_rate": 2.298661593641653e-06,
      "loss": 0.7057,
      "step": 6442
    },
    {
      "epoch": 7.886168910648715,
      "grad_norm": 1.8748266833418534,
      "learning_rate": 2.298022802226322e-06,
      "loss": 0.4014,
      "step": 6443
    },
    {
      "epoch": 7.887392900856793,
      "grad_norm": 1.6324551054584329,
      "learning_rate": 2.297384024084187e-06,
      "loss": 0.4017,
      "step": 6444
    },
    {
      "epoch": 7.8886168910648715,
      "grad_norm": 1.0721594644763368,
      "learning_rate": 2.296745259257224e-06,
      "loss": 0.5127,
      "step": 6445
    },
    {
      "epoch": 7.88984088127295,
      "grad_norm": 2.104718850953799,
      "learning_rate": 2.2961065077874106e-06,
      "loss": 0.4634,
      "step": 6446
    },
    {
      "epoch": 7.891064871481028,
      "grad_norm": 2.0213950996699226,
      "learning_rate": 2.2954677697167234e-06,
      "loss": 1.0477,
      "step": 6447
    },
    {
      "epoch": 7.8922888616891065,
      "grad_norm": 1.6426395057854148,
      "learning_rate": 2.294829045087138e-06,
      "loss": 0.4854,
      "step": 6448
    },
    {
      "epoch": 7.893512851897185,
      "grad_norm": 1.163177707520498,
      "learning_rate": 2.2941903339406284e-06,
      "loss": 0.7707,
      "step": 6449
    },
    {
      "epoch": 7.894736842105263,
      "grad_norm": 1.0519899776675787,
      "learning_rate": 2.2935516363191695e-06,
      "loss": 0.5757,
      "step": 6450
    },
    {
      "epoch": 7.8959608323133414,
      "grad_norm": 1.3041922177912244,
      "learning_rate": 2.292912952264734e-06,
      "loss": 0.695,
      "step": 6451
    },
    {
      "epoch": 7.89718482252142,
      "grad_norm": 2.1438570895912794,
      "learning_rate": 2.2922742818192916e-06,
      "loss": 0.7514,
      "step": 6452
    },
    {
      "epoch": 7.898408812729498,
      "grad_norm": 1.4938356877684331,
      "learning_rate": 2.291635625024816e-06,
      "loss": 0.9376,
      "step": 6453
    },
    {
      "epoch": 7.899632802937576,
      "grad_norm": 1.5404968556870977,
      "learning_rate": 2.290996981923277e-06,
      "loss": 0.5708,
      "step": 6454
    },
    {
      "epoch": 7.900856793145655,
      "grad_norm": 1.6330119023981398,
      "learning_rate": 2.2903583525566423e-06,
      "loss": 0.5375,
      "step": 6455
    },
    {
      "epoch": 7.902080783353734,
      "grad_norm": 0.9900299997270264,
      "learning_rate": 2.2897197369668812e-06,
      "loss": 0.5993,
      "step": 6456
    },
    {
      "epoch": 7.903304773561811,
      "grad_norm": 2.1715794479895796,
      "learning_rate": 2.289081135195962e-06,
      "loss": 0.471,
      "step": 6457
    },
    {
      "epoch": 7.90452876376989,
      "grad_norm": 1.5522241457021897,
      "learning_rate": 2.28844254728585e-06,
      "loss": 0.9422,
      "step": 6458
    },
    {
      "epoch": 7.905752753977968,
      "grad_norm": 3.4193560319395075,
      "learning_rate": 2.287803973278512e-06,
      "loss": 0.4074,
      "step": 6459
    },
    {
      "epoch": 7.906976744186046,
      "grad_norm": 2.431470915588443,
      "learning_rate": 2.287165413215911e-06,
      "loss": 0.5222,
      "step": 6460
    },
    {
      "epoch": 7.908200734394125,
      "grad_norm": 2.999078908870587,
      "learning_rate": 2.286526867140012e-06,
      "loss": 0.3796,
      "step": 6461
    },
    {
      "epoch": 7.909424724602204,
      "grad_norm": 2.6513877585482364,
      "learning_rate": 2.2858883350927786e-06,
      "loss": 0.4295,
      "step": 6462
    },
    {
      "epoch": 7.910648714810281,
      "grad_norm": 1.8947110048370153,
      "learning_rate": 2.2852498171161713e-06,
      "loss": 0.8614,
      "step": 6463
    },
    {
      "epoch": 7.91187270501836,
      "grad_norm": 2.3669011247685763,
      "learning_rate": 2.284611313252152e-06,
      "loss": 0.3775,
      "step": 6464
    },
    {
      "epoch": 7.9130966952264385,
      "grad_norm": 1.8013226038777947,
      "learning_rate": 2.2839728235426795e-06,
      "loss": 0.6667,
      "step": 6465
    },
    {
      "epoch": 7.914320685434516,
      "grad_norm": 1.8964834194622402,
      "learning_rate": 2.283334348029715e-06,
      "loss": 1.3314,
      "step": 6466
    },
    {
      "epoch": 7.915544675642595,
      "grad_norm": 1.2948188057470875,
      "learning_rate": 2.2826958867552153e-06,
      "loss": 0.3766,
      "step": 6467
    },
    {
      "epoch": 7.9167686658506735,
      "grad_norm": 1.0286402988524281,
      "learning_rate": 2.282057439761139e-06,
      "loss": 0.6279,
      "step": 6468
    },
    {
      "epoch": 7.917992656058751,
      "grad_norm": 1.6925426449765872,
      "learning_rate": 2.2814190070894402e-06,
      "loss": 1.1022,
      "step": 6469
    },
    {
      "epoch": 7.91921664626683,
      "grad_norm": 1.8427883041555264,
      "learning_rate": 2.2807805887820762e-06,
      "loss": 0.5196,
      "step": 6470
    },
    {
      "epoch": 7.9204406364749085,
      "grad_norm": 1.2956385665239043,
      "learning_rate": 2.2801421848810025e-06,
      "loss": 0.7867,
      "step": 6471
    },
    {
      "epoch": 7.921664626682986,
      "grad_norm": 2.0664780524073256,
      "learning_rate": 2.2795037954281703e-06,
      "loss": 0.4626,
      "step": 6472
    },
    {
      "epoch": 7.922888616891065,
      "grad_norm": 1.4566234344919349,
      "learning_rate": 2.278865420465532e-06,
      "loss": 0.9413,
      "step": 6473
    },
    {
      "epoch": 7.9241126070991434,
      "grad_norm": 2.311016848382608,
      "learning_rate": 2.2782270600350413e-06,
      "loss": 0.4418,
      "step": 6474
    },
    {
      "epoch": 7.925336597307221,
      "grad_norm": 1.5712197457368773,
      "learning_rate": 2.277588714178648e-06,
      "loss": 0.4979,
      "step": 6475
    },
    {
      "epoch": 7.9265605875153,
      "grad_norm": 1.3068030445040366,
      "learning_rate": 2.276950382938302e-06,
      "loss": 1.5698,
      "step": 6476
    },
    {
      "epoch": 7.927784577723378,
      "grad_norm": 1.82240820721155,
      "learning_rate": 2.276312066355952e-06,
      "loss": 0.9399,
      "step": 6477
    },
    {
      "epoch": 7.929008567931456,
      "grad_norm": 1.079701556070988,
      "learning_rate": 2.2756737644735446e-06,
      "loss": 0.571,
      "step": 6478
    },
    {
      "epoch": 7.930232558139535,
      "grad_norm": 1.6582408780794728,
      "learning_rate": 2.275035477333028e-06,
      "loss": 0.7474,
      "step": 6479
    },
    {
      "epoch": 7.931456548347613,
      "grad_norm": 1.4723081966904492,
      "learning_rate": 2.2743972049763484e-06,
      "loss": 0.7276,
      "step": 6480
    },
    {
      "epoch": 7.932680538555692,
      "grad_norm": 0.7479043622663435,
      "learning_rate": 2.27375894744545e-06,
      "loss": 0.3566,
      "step": 6481
    },
    {
      "epoch": 7.93390452876377,
      "grad_norm": 1.7851883474000476,
      "learning_rate": 2.2731207047822756e-06,
      "loss": 0.9204,
      "step": 6482
    },
    {
      "epoch": 7.935128518971848,
      "grad_norm": 2.3491979260505866,
      "learning_rate": 2.2724824770287704e-06,
      "loss": 0.527,
      "step": 6483
    },
    {
      "epoch": 7.936352509179926,
      "grad_norm": 1.2274876281734133,
      "learning_rate": 2.2718442642268753e-06,
      "loss": 0.5635,
      "step": 6484
    },
    {
      "epoch": 7.937576499388005,
      "grad_norm": 1.2271815659565355,
      "learning_rate": 2.2712060664185306e-06,
      "loss": 0.4973,
      "step": 6485
    },
    {
      "epoch": 7.938800489596083,
      "grad_norm": 1.573782003874472,
      "learning_rate": 2.2705678836456767e-06,
      "loss": 0.5436,
      "step": 6486
    },
    {
      "epoch": 7.940024479804162,
      "grad_norm": 1.5731067546142778,
      "learning_rate": 2.2699297159502536e-06,
      "loss": 1.0202,
      "step": 6487
    },
    {
      "epoch": 7.94124847001224,
      "grad_norm": 1.2063900980605007,
      "learning_rate": 2.269291563374199e-06,
      "loss": 0.5372,
      "step": 6488
    },
    {
      "epoch": 7.942472460220318,
      "grad_norm": 3.1352360942448128,
      "learning_rate": 2.2686534259594494e-06,
      "loss": 0.4528,
      "step": 6489
    },
    {
      "epoch": 7.943696450428397,
      "grad_norm": 1.844702157764708,
      "learning_rate": 2.2680153037479405e-06,
      "loss": 0.4569,
      "step": 6490
    },
    {
      "epoch": 7.944920440636475,
      "grad_norm": 2.5384215567415787,
      "learning_rate": 2.2673771967816077e-06,
      "loss": 0.5199,
      "step": 6491
    },
    {
      "epoch": 7.946144430844553,
      "grad_norm": 1.7335183536862504,
      "learning_rate": 2.2667391051023852e-06,
      "loss": 0.6368,
      "step": 6492
    },
    {
      "epoch": 7.947368421052632,
      "grad_norm": 1.7198996127612873,
      "learning_rate": 2.266101028752206e-06,
      "loss": 0.6977,
      "step": 6493
    },
    {
      "epoch": 7.94859241126071,
      "grad_norm": 1.055801014759983,
      "learning_rate": 2.265462967773002e-06,
      "loss": 0.6194,
      "step": 6494
    },
    {
      "epoch": 7.949816401468788,
      "grad_norm": 1.2478402803516269,
      "learning_rate": 2.264824922206704e-06,
      "loss": 0.6557,
      "step": 6495
    },
    {
      "epoch": 7.951040391676867,
      "grad_norm": 1.2738984262595374,
      "learning_rate": 2.2641868920952427e-06,
      "loss": 0.4489,
      "step": 6496
    },
    {
      "epoch": 7.9522643818849446,
      "grad_norm": 1.4174841342947835,
      "learning_rate": 2.263548877480547e-06,
      "loss": 0.6936,
      "step": 6497
    },
    {
      "epoch": 7.953488372093023,
      "grad_norm": 1.2633803266280825,
      "learning_rate": 2.262910878404544e-06,
      "loss": 0.7056,
      "step": 6498
    },
    {
      "epoch": 7.954712362301102,
      "grad_norm": 1.392550027616769,
      "learning_rate": 2.2622728949091603e-06,
      "loss": 0.9022,
      "step": 6499
    },
    {
      "epoch": 7.95593635250918,
      "grad_norm": 1.8769648888263015,
      "learning_rate": 2.261634927036324e-06,
      "loss": 0.4263,
      "step": 6500
    },
    {
      "epoch": 7.957160342717258,
      "grad_norm": 1.0889524895514484,
      "learning_rate": 2.260996974827958e-06,
      "loss": 0.6873,
      "step": 6501
    },
    {
      "epoch": 7.958384332925337,
      "grad_norm": 1.759003368501378,
      "learning_rate": 2.2603590383259873e-06,
      "loss": 0.6407,
      "step": 6502
    },
    {
      "epoch": 7.9596083231334145,
      "grad_norm": 3.213795418342855,
      "learning_rate": 2.259721117572334e-06,
      "loss": 0.5033,
      "step": 6503
    },
    {
      "epoch": 7.960832313341493,
      "grad_norm": 1.6570475961779154,
      "learning_rate": 2.2590832126089195e-06,
      "loss": 0.5572,
      "step": 6504
    },
    {
      "epoch": 7.962056303549572,
      "grad_norm": 2.401486414392633,
      "learning_rate": 2.2584453234776656e-06,
      "loss": 0.5347,
      "step": 6505
    },
    {
      "epoch": 7.96328029375765,
      "grad_norm": 2.3915733876965044,
      "learning_rate": 2.2578074502204923e-06,
      "loss": 0.4216,
      "step": 6506
    },
    {
      "epoch": 7.964504283965728,
      "grad_norm": 1.8723587409096716,
      "learning_rate": 2.257169592879317e-06,
      "loss": 0.4498,
      "step": 6507
    },
    {
      "epoch": 7.965728274173807,
      "grad_norm": 1.4795229286384264,
      "learning_rate": 2.256531751496057e-06,
      "loss": 1.4942,
      "step": 6508
    },
    {
      "epoch": 7.966952264381885,
      "grad_norm": 1.408643815080826,
      "learning_rate": 2.2558939261126313e-06,
      "loss": 0.6326,
      "step": 6509
    },
    {
      "epoch": 7.968176254589963,
      "grad_norm": 1.8632149437330823,
      "learning_rate": 2.255256116770953e-06,
      "loss": 0.364,
      "step": 6510
    },
    {
      "epoch": 7.969400244798042,
      "grad_norm": 1.1677212037084943,
      "learning_rate": 2.254618323512938e-06,
      "loss": 0.8503,
      "step": 6511
    },
    {
      "epoch": 7.97062423500612,
      "grad_norm": 2.74362637601098,
      "learning_rate": 2.253980546380498e-06,
      "loss": 0.9329,
      "step": 6512
    },
    {
      "epoch": 7.971848225214198,
      "grad_norm": 1.2138541003201544,
      "learning_rate": 2.2533427854155478e-06,
      "loss": 0.5336,
      "step": 6513
    },
    {
      "epoch": 7.973072215422277,
      "grad_norm": 1.9490635302675117,
      "learning_rate": 2.2527050406599968e-06,
      "loss": 0.4981,
      "step": 6514
    },
    {
      "epoch": 7.974296205630355,
      "grad_norm": 2.076823666080513,
      "learning_rate": 2.2520673121557567e-06,
      "loss": 0.4441,
      "step": 6515
    },
    {
      "epoch": 7.975520195838433,
      "grad_norm": 2.3446952090547604,
      "learning_rate": 2.2514295999447353e-06,
      "loss": 0.4177,
      "step": 6516
    },
    {
      "epoch": 7.976744186046512,
      "grad_norm": 1.4463576397344817,
      "learning_rate": 2.25079190406884e-06,
      "loss": 0.6029,
      "step": 6517
    },
    {
      "epoch": 7.97796817625459,
      "grad_norm": 0.9614984293437456,
      "learning_rate": 2.250154224569981e-06,
      "loss": 0.4051,
      "step": 6518
    },
    {
      "epoch": 7.979192166462668,
      "grad_norm": 2.439891115854889,
      "learning_rate": 2.249516561490061e-06,
      "loss": 0.6975,
      "step": 6519
    },
    {
      "epoch": 7.9804161566707466,
      "grad_norm": 2.366961466184464,
      "learning_rate": 2.2488789148709863e-06,
      "loss": 0.495,
      "step": 6520
    },
    {
      "epoch": 7.981640146878825,
      "grad_norm": 1.4314400639263924,
      "learning_rate": 2.2482412847546604e-06,
      "loss": 1.0787,
      "step": 6521
    },
    {
      "epoch": 7.982864137086903,
      "grad_norm": 1.7201384868054848,
      "learning_rate": 2.247603671182986e-06,
      "loss": 0.4982,
      "step": 6522
    },
    {
      "epoch": 7.9840881272949815,
      "grad_norm": 2.4314344159525048,
      "learning_rate": 2.2469660741978654e-06,
      "loss": 0.599,
      "step": 6523
    },
    {
      "epoch": 7.98531211750306,
      "grad_norm": 1.0378853743489331,
      "learning_rate": 2.2463284938411985e-06,
      "loss": 0.4567,
      "step": 6524
    },
    {
      "epoch": 7.986536107711139,
      "grad_norm": 1.4300654604795238,
      "learning_rate": 2.2456909301548832e-06,
      "loss": 0.59,
      "step": 6525
    },
    {
      "epoch": 7.9877600979192165,
      "grad_norm": 1.901056556159614,
      "learning_rate": 2.2450533831808215e-06,
      "loss": 0.6842,
      "step": 6526
    },
    {
      "epoch": 7.988984088127295,
      "grad_norm": 2.3694972643588987,
      "learning_rate": 2.244415852960907e-06,
      "loss": 0.4838,
      "step": 6527
    },
    {
      "epoch": 7.990208078335373,
      "grad_norm": 2.6010002579570903,
      "learning_rate": 2.243778339537038e-06,
      "loss": 0.8433,
      "step": 6528
    },
    {
      "epoch": 7.9914320685434515,
      "grad_norm": 1.8891821354845706,
      "learning_rate": 2.2431408429511086e-06,
      "loss": 0.9764,
      "step": 6529
    },
    {
      "epoch": 7.99265605875153,
      "grad_norm": 1.6584500545661083,
      "learning_rate": 2.2425033632450125e-06,
      "loss": 0.4366,
      "step": 6530
    },
    {
      "epoch": 7.993880048959609,
      "grad_norm": 1.6116162532409417,
      "learning_rate": 2.2418659004606437e-06,
      "loss": 1.0089,
      "step": 6531
    },
    {
      "epoch": 7.995104039167686,
      "grad_norm": 1.4242967264481996,
      "learning_rate": 2.2412284546398926e-06,
      "loss": 0.48,
      "step": 6532
    },
    {
      "epoch": 7.996328029375765,
      "grad_norm": 1.2222910631498214,
      "learning_rate": 2.2405910258246515e-06,
      "loss": 0.7472,
      "step": 6533
    },
    {
      "epoch": 7.997552019583844,
      "grad_norm": 1.175125652308538,
      "learning_rate": 2.2399536140568065e-06,
      "loss": 0.4111,
      "step": 6534
    },
    {
      "epoch": 7.998776009791921,
      "grad_norm": 1.607805746201495,
      "learning_rate": 2.23931621937825e-06,
      "loss": 1.6259,
      "step": 6535
    },
    {
      "epoch": 8.0,
      "grad_norm": 1.0664715413083938,
      "learning_rate": 2.238678841830867e-06,
      "loss": 0.6796,
      "step": 6536
    },
    {
      "epoch": 8.001223990208079,
      "grad_norm": 1.852407278952052,
      "learning_rate": 2.2380414814565436e-06,
      "loss": 0.4172,
      "step": 6537
    },
    {
      "epoch": 8.002447980416157,
      "grad_norm": 1.2034229308038695,
      "learning_rate": 2.237404138297165e-06,
      "loss": 1.0923,
      "step": 6538
    },
    {
      "epoch": 8.003671970624236,
      "grad_norm": 2.234676285346996,
      "learning_rate": 2.236766812394615e-06,
      "loss": 0.4209,
      "step": 6539
    },
    {
      "epoch": 8.004895960832313,
      "grad_norm": 2.595893648366043,
      "learning_rate": 2.236129503790777e-06,
      "loss": 0.5418,
      "step": 6540
    },
    {
      "epoch": 8.006119951040391,
      "grad_norm": 1.9606420224909622,
      "learning_rate": 2.2354922125275323e-06,
      "loss": 0.4593,
      "step": 6541
    },
    {
      "epoch": 8.00734394124847,
      "grad_norm": 2.5706821317399595,
      "learning_rate": 2.234854938646761e-06,
      "loss": 0.8358,
      "step": 6542
    },
    {
      "epoch": 8.008567931456549,
      "grad_norm": 1.6788531132023272,
      "learning_rate": 2.234217682190341e-06,
      "loss": 0.9327,
      "step": 6543
    },
    {
      "epoch": 8.009791921664627,
      "grad_norm": 1.831913180628728,
      "learning_rate": 2.2335804432001534e-06,
      "loss": 0.3546,
      "step": 6544
    },
    {
      "epoch": 8.011015911872706,
      "grad_norm": 1.6276527680994484,
      "learning_rate": 2.2329432217180737e-06,
      "loss": 0.4239,
      "step": 6545
    },
    {
      "epoch": 8.012239902080783,
      "grad_norm": 2.952098102199096,
      "learning_rate": 2.232306017785977e-06,
      "loss": 0.3968,
      "step": 6546
    },
    {
      "epoch": 8.013463892288861,
      "grad_norm": 1.9907334436472974,
      "learning_rate": 2.2316688314457382e-06,
      "loss": 1.2117,
      "step": 6547
    },
    {
      "epoch": 8.01468788249694,
      "grad_norm": 2.7088756035862187,
      "learning_rate": 2.231031662739232e-06,
      "loss": 0.4099,
      "step": 6548
    },
    {
      "epoch": 8.015911872705018,
      "grad_norm": 1.6520868462835037,
      "learning_rate": 2.2303945117083304e-06,
      "loss": 1.1773,
      "step": 6549
    },
    {
      "epoch": 8.017135862913097,
      "grad_norm": 1.1601104717803261,
      "learning_rate": 2.229757378394904e-06,
      "loss": 0.4985,
      "step": 6550
    },
    {
      "epoch": 8.018359853121176,
      "grad_norm": 1.3819156123124103,
      "learning_rate": 2.2291202628408225e-06,
      "loss": 0.611,
      "step": 6551
    },
    {
      "epoch": 8.019583843329253,
      "grad_norm": 1.8373062072800515,
      "learning_rate": 2.228483165087956e-06,
      "loss": 0.5109,
      "step": 6552
    },
    {
      "epoch": 8.020807833537331,
      "grad_norm": 1.7391937407116769,
      "learning_rate": 2.2278460851781724e-06,
      "loss": 0.4007,
      "step": 6553
    },
    {
      "epoch": 8.02203182374541,
      "grad_norm": 1.8163978202841773,
      "learning_rate": 2.2272090231533365e-06,
      "loss": 0.4412,
      "step": 6554
    },
    {
      "epoch": 8.023255813953488,
      "grad_norm": 1.0365210928806816,
      "learning_rate": 2.2265719790553147e-06,
      "loss": 0.6446,
      "step": 6555
    },
    {
      "epoch": 8.024479804161567,
      "grad_norm": 1.5101181448802954,
      "learning_rate": 2.225934952925971e-06,
      "loss": 0.6972,
      "step": 6556
    },
    {
      "epoch": 8.025703794369646,
      "grad_norm": 1.282521961338584,
      "learning_rate": 2.2252979448071685e-06,
      "loss": 0.4626,
      "step": 6557
    },
    {
      "epoch": 8.026927784577722,
      "grad_norm": 1.496127508815973,
      "learning_rate": 2.224660954740769e-06,
      "loss": 0.498,
      "step": 6558
    },
    {
      "epoch": 8.028151774785801,
      "grad_norm": 1.281316904699095,
      "learning_rate": 2.2240239827686335e-06,
      "loss": 0.7813,
      "step": 6559
    },
    {
      "epoch": 8.02937576499388,
      "grad_norm": 1.0207430624149103,
      "learning_rate": 2.2233870289326203e-06,
      "loss": 0.6158,
      "step": 6560
    },
    {
      "epoch": 8.030599755201958,
      "grad_norm": 0.8493580114493111,
      "learning_rate": 2.2227500932745893e-06,
      "loss": 0.5268,
      "step": 6561
    },
    {
      "epoch": 8.031823745410037,
      "grad_norm": 2.151843795840132,
      "learning_rate": 2.222113175836397e-06,
      "loss": 1.4202,
      "step": 6562
    },
    {
      "epoch": 8.033047735618116,
      "grad_norm": 1.2724303590932984,
      "learning_rate": 2.2214762766598985e-06,
      "loss": 0.4424,
      "step": 6563
    },
    {
      "epoch": 8.034271725826194,
      "grad_norm": 1.1824886137481352,
      "learning_rate": 2.220839395786948e-06,
      "loss": 0.7238,
      "step": 6564
    },
    {
      "epoch": 8.035495716034271,
      "grad_norm": 1.5981847336758548,
      "learning_rate": 2.220202533259401e-06,
      "loss": 0.5605,
      "step": 6565
    },
    {
      "epoch": 8.03671970624235,
      "grad_norm": 1.834556106330547,
      "learning_rate": 2.2195656891191083e-06,
      "loss": 0.9681,
      "step": 6566
    },
    {
      "epoch": 8.037943696450428,
      "grad_norm": 1.6821814050332506,
      "learning_rate": 2.218928863407921e-06,
      "loss": 0.5547,
      "step": 6567
    },
    {
      "epoch": 8.039167686658507,
      "grad_norm": 1.6306047756115023,
      "learning_rate": 2.2182920561676896e-06,
      "loss": 1.3683,
      "step": 6568
    },
    {
      "epoch": 8.040391676866586,
      "grad_norm": 2.2780416034381323,
      "learning_rate": 2.217655267440262e-06,
      "loss": 0.5337,
      "step": 6569
    },
    {
      "epoch": 8.041615667074664,
      "grad_norm": 1.556982369483097,
      "learning_rate": 2.2170184972674856e-06,
      "loss": 0.4385,
      "step": 6570
    },
    {
      "epoch": 8.042839657282741,
      "grad_norm": 2.3618280438293655,
      "learning_rate": 2.2163817456912083e-06,
      "loss": 0.3458,
      "step": 6571
    },
    {
      "epoch": 8.04406364749082,
      "grad_norm": 1.636695555283927,
      "learning_rate": 2.215745012753273e-06,
      "loss": 0.5754,
      "step": 6572
    },
    {
      "epoch": 8.045287637698898,
      "grad_norm": 1.4238920586087913,
      "learning_rate": 2.2151082984955228e-06,
      "loss": 0.435,
      "step": 6573
    },
    {
      "epoch": 8.046511627906977,
      "grad_norm": 2.3307723102488693,
      "learning_rate": 2.2144716029598025e-06,
      "loss": 0.6126,
      "step": 6574
    },
    {
      "epoch": 8.047735618115055,
      "grad_norm": 3.2714291160695126,
      "learning_rate": 2.213834926187952e-06,
      "loss": 0.4289,
      "step": 6575
    },
    {
      "epoch": 8.048959608323134,
      "grad_norm": 1.2835388956626008,
      "learning_rate": 2.2131982682218124e-06,
      "loss": 0.3714,
      "step": 6576
    },
    {
      "epoch": 8.050183598531211,
      "grad_norm": 1.1486199663859957,
      "learning_rate": 2.212561629103221e-06,
      "loss": 0.8461,
      "step": 6577
    },
    {
      "epoch": 8.05140758873929,
      "grad_norm": 2.0539881862262774,
      "learning_rate": 2.2119250088740165e-06,
      "loss": 1.0233,
      "step": 6578
    },
    {
      "epoch": 8.052631578947368,
      "grad_norm": 1.850919262021815,
      "learning_rate": 2.211288407576035e-06,
      "loss": 0.4154,
      "step": 6579
    },
    {
      "epoch": 8.053855569155447,
      "grad_norm": 1.739758323024896,
      "learning_rate": 2.210651825251112e-06,
      "loss": 1.5668,
      "step": 6580
    },
    {
      "epoch": 8.055079559363525,
      "grad_norm": 2.2217845303409,
      "learning_rate": 2.2100152619410805e-06,
      "loss": 0.913,
      "step": 6581
    },
    {
      "epoch": 8.056303549571604,
      "grad_norm": 1.9126609029246495,
      "learning_rate": 2.2093787176877723e-06,
      "loss": 1.1169,
      "step": 6582
    },
    {
      "epoch": 8.057527539779683,
      "grad_norm": 1.3409938918644844,
      "learning_rate": 2.2087421925330208e-06,
      "loss": 0.9533,
      "step": 6583
    },
    {
      "epoch": 8.05875152998776,
      "grad_norm": 1.3082520502415809,
      "learning_rate": 2.208105686518655e-06,
      "loss": 0.5391,
      "step": 6584
    },
    {
      "epoch": 8.059975520195838,
      "grad_norm": 1.4857668539596944,
      "learning_rate": 2.2074691996865043e-06,
      "loss": 0.6487,
      "step": 6585
    },
    {
      "epoch": 8.061199510403917,
      "grad_norm": 1.4408452366081115,
      "learning_rate": 2.2068327320783945e-06,
      "loss": 1.235,
      "step": 6586
    },
    {
      "epoch": 8.062423500611995,
      "grad_norm": 1.954758532434582,
      "learning_rate": 2.206196283736154e-06,
      "loss": 0.8205,
      "step": 6587
    },
    {
      "epoch": 8.063647490820074,
      "grad_norm": 2.6520138235005217,
      "learning_rate": 2.2055598547016075e-06,
      "loss": 0.3566,
      "step": 6588
    },
    {
      "epoch": 8.064871481028153,
      "grad_norm": 2.3329818206342017,
      "learning_rate": 2.204923445016578e-06,
      "loss": 0.4829,
      "step": 6589
    },
    {
      "epoch": 8.06609547123623,
      "grad_norm": 1.638126250963845,
      "learning_rate": 2.204287054722887e-06,
      "loss": 1.1961,
      "step": 6590
    },
    {
      "epoch": 8.067319461444308,
      "grad_norm": 1.6622609886613309,
      "learning_rate": 2.2036506838623593e-06,
      "loss": 1.8884,
      "step": 6591
    },
    {
      "epoch": 8.068543451652387,
      "grad_norm": 2.6595975837950703,
      "learning_rate": 2.203014332476811e-06,
      "loss": 0.4776,
      "step": 6592
    },
    {
      "epoch": 8.069767441860465,
      "grad_norm": 1.03603033962083,
      "learning_rate": 2.202378000608063e-06,
      "loss": 0.4519,
      "step": 6593
    },
    {
      "epoch": 8.070991432068544,
      "grad_norm": 2.684968126513587,
      "learning_rate": 2.2017416882979316e-06,
      "loss": 0.5044,
      "step": 6594
    },
    {
      "epoch": 8.072215422276622,
      "grad_norm": 1.0340528404807305,
      "learning_rate": 2.2011053955882332e-06,
      "loss": 0.5022,
      "step": 6595
    },
    {
      "epoch": 8.0734394124847,
      "grad_norm": 1.8083352965813093,
      "learning_rate": 2.2004691225207825e-06,
      "loss": 0.5253,
      "step": 6596
    },
    {
      "epoch": 8.074663402692778,
      "grad_norm": 2.0069766418215487,
      "learning_rate": 2.1998328691373936e-06,
      "loss": 0.456,
      "step": 6597
    },
    {
      "epoch": 8.075887392900857,
      "grad_norm": 1.6744848094951486,
      "learning_rate": 2.1991966354798787e-06,
      "loss": 0.651,
      "step": 6598
    },
    {
      "epoch": 8.077111383108935,
      "grad_norm": 0.8180308965548361,
      "learning_rate": 2.198560421590047e-06,
      "loss": 0.3518,
      "step": 6599
    },
    {
      "epoch": 8.078335373317014,
      "grad_norm": 1.6823419391440262,
      "learning_rate": 2.1979242275097106e-06,
      "loss": 0.4815,
      "step": 6600
    },
    {
      "epoch": 8.079559363525092,
      "grad_norm": 1.0552436767092936,
      "learning_rate": 2.1972880532806765e-06,
      "loss": 0.4815,
      "step": 6601
    },
    {
      "epoch": 8.080783353733171,
      "grad_norm": 1.7345020217486629,
      "learning_rate": 2.1966518989447516e-06,
      "loss": 1.1959,
      "step": 6602
    },
    {
      "epoch": 8.082007343941248,
      "grad_norm": 1.5237870269831193,
      "learning_rate": 2.196015764543742e-06,
      "loss": 0.7239,
      "step": 6603
    },
    {
      "epoch": 8.083231334149326,
      "grad_norm": 1.8228339648224645,
      "learning_rate": 2.1953796501194514e-06,
      "loss": 0.3744,
      "step": 6604
    },
    {
      "epoch": 8.084455324357405,
      "grad_norm": 2.755178929686862,
      "learning_rate": 2.194743555713684e-06,
      "loss": 0.4591,
      "step": 6605
    },
    {
      "epoch": 8.085679314565484,
      "grad_norm": 0.8518916534997869,
      "learning_rate": 2.194107481368242e-06,
      "loss": 0.4681,
      "step": 6606
    },
    {
      "epoch": 8.086903304773562,
      "grad_norm": 1.2180052140000721,
      "learning_rate": 2.1934714271249243e-06,
      "loss": 0.608,
      "step": 6607
    },
    {
      "epoch": 8.088127294981641,
      "grad_norm": 1.1778763596836528,
      "learning_rate": 2.192835393025529e-06,
      "loss": 0.4067,
      "step": 6608
    },
    {
      "epoch": 8.089351285189718,
      "grad_norm": 2.5546624246829968,
      "learning_rate": 2.1921993791118574e-06,
      "loss": 0.2179,
      "step": 6609
    },
    {
      "epoch": 8.090575275397796,
      "grad_norm": 2.9113226101020553,
      "learning_rate": 2.191563385425704e-06,
      "loss": 0.3246,
      "step": 6610
    },
    {
      "epoch": 8.091799265605875,
      "grad_norm": 2.153578585260385,
      "learning_rate": 2.1909274120088638e-06,
      "loss": 0.4268,
      "step": 6611
    },
    {
      "epoch": 8.093023255813954,
      "grad_norm": 2.2384837814527563,
      "learning_rate": 2.19029145890313e-06,
      "loss": 0.5483,
      "step": 6612
    },
    {
      "epoch": 8.094247246022032,
      "grad_norm": 1.6083340656080023,
      "learning_rate": 2.189655526150297e-06,
      "loss": 1.6205,
      "step": 6613
    },
    {
      "epoch": 8.095471236230111,
      "grad_norm": 1.558093894148372,
      "learning_rate": 2.189019613792155e-06,
      "loss": 0.38,
      "step": 6614
    },
    {
      "epoch": 8.096695226438188,
      "grad_norm": 2.739626162532667,
      "learning_rate": 2.188383721870494e-06,
      "loss": 0.3574,
      "step": 6615
    },
    {
      "epoch": 8.097919216646266,
      "grad_norm": 1.579293419500051,
      "learning_rate": 2.1877478504271015e-06,
      "loss": 0.4278,
      "step": 6616
    },
    {
      "epoch": 8.099143206854345,
      "grad_norm": 1.7546863226970089,
      "learning_rate": 2.1871119995037646e-06,
      "loss": 0.5293,
      "step": 6617
    },
    {
      "epoch": 8.100367197062424,
      "grad_norm": 0.9651305705126904,
      "learning_rate": 2.1864761691422715e-06,
      "loss": 0.4008,
      "step": 6618
    },
    {
      "epoch": 8.101591187270502,
      "grad_norm": 1.261085419908414,
      "learning_rate": 2.185840359384404e-06,
      "loss": 0.395,
      "step": 6619
    },
    {
      "epoch": 8.10281517747858,
      "grad_norm": 0.8989441457127049,
      "learning_rate": 2.1852045702719464e-06,
      "loss": 0.3937,
      "step": 6620
    },
    {
      "epoch": 8.104039167686658,
      "grad_norm": 1.7656444366257238,
      "learning_rate": 2.1845688018466793e-06,
      "loss": 0.9946,
      "step": 6621
    },
    {
      "epoch": 8.105263157894736,
      "grad_norm": 2.38707802991172,
      "learning_rate": 2.1839330541503846e-06,
      "loss": 0.4113,
      "step": 6622
    },
    {
      "epoch": 8.106487148102815,
      "grad_norm": 1.6094169226634656,
      "learning_rate": 2.1832973272248403e-06,
      "loss": 0.4574,
      "step": 6623
    },
    {
      "epoch": 8.107711138310894,
      "grad_norm": 1.848057597046354,
      "learning_rate": 2.182661621111825e-06,
      "loss": 0.5808,
      "step": 6624
    },
    {
      "epoch": 8.108935128518972,
      "grad_norm": 1.3617645402363812,
      "learning_rate": 2.1820259358531127e-06,
      "loss": 0.545,
      "step": 6625
    },
    {
      "epoch": 8.11015911872705,
      "grad_norm": 1.1044490361287267,
      "learning_rate": 2.1813902714904807e-06,
      "loss": 0.3881,
      "step": 6626
    },
    {
      "epoch": 8.11138310893513,
      "grad_norm": 1.7041014492511182,
      "learning_rate": 2.1807546280657023e-06,
      "loss": 1.0949,
      "step": 6627
    },
    {
      "epoch": 8.112607099143206,
      "grad_norm": 1.3435808071205826,
      "learning_rate": 2.180119005620548e-06,
      "loss": 1.1226,
      "step": 6628
    },
    {
      "epoch": 8.113831089351285,
      "grad_norm": 2.9039464362574616,
      "learning_rate": 2.1794834041967895e-06,
      "loss": 0.7097,
      "step": 6629
    },
    {
      "epoch": 8.115055079559363,
      "grad_norm": 1.8387532100957535,
      "learning_rate": 2.178847823836196e-06,
      "loss": 1.3199,
      "step": 6630
    },
    {
      "epoch": 8.116279069767442,
      "grad_norm": 2.195197198773073,
      "learning_rate": 2.178212264580536e-06,
      "loss": 0.6144,
      "step": 6631
    },
    {
      "epoch": 8.11750305997552,
      "grad_norm": 1.2453756904687139,
      "learning_rate": 2.1775767264715756e-06,
      "loss": 0.6351,
      "step": 6632
    },
    {
      "epoch": 8.1187270501836,
      "grad_norm": 1.4057463204214742,
      "learning_rate": 2.176941209551081e-06,
      "loss": 0.8791,
      "step": 6633
    },
    {
      "epoch": 8.119951040391676,
      "grad_norm": 2.0494387136333647,
      "learning_rate": 2.176305713860814e-06,
      "loss": 0.4886,
      "step": 6634
    },
    {
      "epoch": 8.121175030599755,
      "grad_norm": 3.2447267396527124,
      "learning_rate": 2.1756702394425386e-06,
      "loss": 0.7742,
      "step": 6635
    },
    {
      "epoch": 8.122399020807833,
      "grad_norm": 1.585722094621414,
      "learning_rate": 2.1750347863380162e-06,
      "loss": 0.5811,
      "step": 6636
    },
    {
      "epoch": 8.123623011015912,
      "grad_norm": 2.5140095035140586,
      "learning_rate": 2.1743993545890054e-06,
      "loss": 0.9073,
      "step": 6637
    },
    {
      "epoch": 8.12484700122399,
      "grad_norm": 1.5246166388034237,
      "learning_rate": 2.1737639442372636e-06,
      "loss": 0.7068,
      "step": 6638
    },
    {
      "epoch": 8.12607099143207,
      "grad_norm": 1.039561841201002,
      "learning_rate": 2.1731285553245495e-06,
      "loss": 0.4207,
      "step": 6639
    },
    {
      "epoch": 8.127294981640146,
      "grad_norm": 2.4200675142431303,
      "learning_rate": 2.1724931878926176e-06,
      "loss": 0.5107,
      "step": 6640
    },
    {
      "epoch": 8.128518971848225,
      "grad_norm": 1.9694600450717241,
      "learning_rate": 2.1718578419832225e-06,
      "loss": 1.1118,
      "step": 6641
    },
    {
      "epoch": 8.129742962056303,
      "grad_norm": 2.679731976729549,
      "learning_rate": 2.1712225176381164e-06,
      "loss": 0.4174,
      "step": 6642
    },
    {
      "epoch": 8.130966952264382,
      "grad_norm": 3.0519648641709716,
      "learning_rate": 2.170587214899049e-06,
      "loss": 0.5244,
      "step": 6643
    },
    {
      "epoch": 8.13219094247246,
      "grad_norm": 1.0215783000774044,
      "learning_rate": 2.169951933807773e-06,
      "loss": 0.5428,
      "step": 6644
    },
    {
      "epoch": 8.13341493268054,
      "grad_norm": 2.14200161193924,
      "learning_rate": 2.1693166744060344e-06,
      "loss": 0.4949,
      "step": 6645
    },
    {
      "epoch": 8.134638922888616,
      "grad_norm": 1.4375016099115383,
      "learning_rate": 2.1686814367355813e-06,
      "loss": 0.4682,
      "step": 6646
    },
    {
      "epoch": 8.135862913096695,
      "grad_norm": 1.6534159060911062,
      "learning_rate": 2.168046220838158e-06,
      "loss": 0.5319,
      "step": 6647
    },
    {
      "epoch": 8.137086903304773,
      "grad_norm": 0.9916513924430601,
      "learning_rate": 2.1674110267555095e-06,
      "loss": 0.5308,
      "step": 6648
    },
    {
      "epoch": 8.138310893512852,
      "grad_norm": 2.1021229821731717,
      "learning_rate": 2.1667758545293783e-06,
      "loss": 0.5569,
      "step": 6649
    },
    {
      "epoch": 8.13953488372093,
      "grad_norm": 1.3537435303763818,
      "learning_rate": 2.1661407042015055e-06,
      "loss": 1.3237,
      "step": 6650
    },
    {
      "epoch": 8.140758873929009,
      "grad_norm": 1.2355430849580833,
      "learning_rate": 2.16550557581363e-06,
      "loss": 0.4008,
      "step": 6651
    },
    {
      "epoch": 8.141982864137088,
      "grad_norm": 1.7402211961817422,
      "learning_rate": 2.1648704694074916e-06,
      "loss": 1.494,
      "step": 6652
    },
    {
      "epoch": 8.143206854345165,
      "grad_norm": 1.8623640443822076,
      "learning_rate": 2.1642353850248273e-06,
      "loss": 0.6018,
      "step": 6653
    },
    {
      "epoch": 8.144430844553243,
      "grad_norm": 1.5327530630083277,
      "learning_rate": 2.1636003227073704e-06,
      "loss": 1.5603,
      "step": 6654
    },
    {
      "epoch": 8.145654834761322,
      "grad_norm": 2.125362372051552,
      "learning_rate": 2.1629652824968563e-06,
      "loss": 0.456,
      "step": 6655
    },
    {
      "epoch": 8.1468788249694,
      "grad_norm": 1.3044534533739336,
      "learning_rate": 2.1623302644350168e-06,
      "loss": 0.8784,
      "step": 6656
    },
    {
      "epoch": 8.148102815177479,
      "grad_norm": 2.3539004988269667,
      "learning_rate": 2.1616952685635837e-06,
      "loss": 0.562,
      "step": 6657
    },
    {
      "epoch": 8.149326805385558,
      "grad_norm": 1.2825701965030385,
      "learning_rate": 2.1610602949242863e-06,
      "loss": 0.5321,
      "step": 6658
    },
    {
      "epoch": 8.150550795593634,
      "grad_norm": 1.3544175396790887,
      "learning_rate": 2.160425343558853e-06,
      "loss": 0.6234,
      "step": 6659
    },
    {
      "epoch": 8.151774785801713,
      "grad_norm": 1.299887202012589,
      "learning_rate": 2.1597904145090094e-06,
      "loss": 1.5635,
      "step": 6660
    },
    {
      "epoch": 8.152998776009792,
      "grad_norm": 2.333933363151094,
      "learning_rate": 2.159155507816482e-06,
      "loss": 0.4757,
      "step": 6661
    },
    {
      "epoch": 8.15422276621787,
      "grad_norm": 1.4684572616599354,
      "learning_rate": 2.158520623522994e-06,
      "loss": 1.0486,
      "step": 6662
    },
    {
      "epoch": 8.155446756425949,
      "grad_norm": 1.4752317033059412,
      "learning_rate": 2.1578857616702674e-06,
      "loss": 1.0724,
      "step": 6663
    },
    {
      "epoch": 8.156670746634028,
      "grad_norm": 1.64374938038092,
      "learning_rate": 2.157250922300022e-06,
      "loss": 1.4275,
      "step": 6664
    },
    {
      "epoch": 8.157894736842104,
      "grad_norm": 2.013130125002096,
      "learning_rate": 2.1566161054539797e-06,
      "loss": 0.4911,
      "step": 6665
    },
    {
      "epoch": 8.159118727050183,
      "grad_norm": 1.487332898870085,
      "learning_rate": 2.1559813111738567e-06,
      "loss": 0.7539,
      "step": 6666
    },
    {
      "epoch": 8.160342717258262,
      "grad_norm": 2.0264269394419285,
      "learning_rate": 2.1553465395013693e-06,
      "loss": 0.4774,
      "step": 6667
    },
    {
      "epoch": 8.16156670746634,
      "grad_norm": 1.772637673248803,
      "learning_rate": 2.1547117904782324e-06,
      "loss": 1.0138,
      "step": 6668
    },
    {
      "epoch": 8.162790697674419,
      "grad_norm": 1.2854701511936402,
      "learning_rate": 2.154077064146159e-06,
      "loss": 0.614,
      "step": 6669
    },
    {
      "epoch": 8.164014687882498,
      "grad_norm": 2.13341912020497,
      "learning_rate": 2.1534423605468626e-06,
      "loss": 0.2966,
      "step": 6670
    },
    {
      "epoch": 8.165238678090576,
      "grad_norm": 1.7009529264148802,
      "learning_rate": 2.152807679722052e-06,
      "loss": 1.6012,
      "step": 6671
    },
    {
      "epoch": 8.166462668298653,
      "grad_norm": 1.961206018779308,
      "learning_rate": 2.152173021713437e-06,
      "loss": 0.6095,
      "step": 6672
    },
    {
      "epoch": 8.167686658506732,
      "grad_norm": 1.787273137692857,
      "learning_rate": 2.151538386562723e-06,
      "loss": 0.5964,
      "step": 6673
    },
    {
      "epoch": 8.16891064871481,
      "grad_norm": 0.8946737230648709,
      "learning_rate": 2.1509037743116183e-06,
      "loss": 0.4176,
      "step": 6674
    },
    {
      "epoch": 8.170134638922889,
      "grad_norm": 2.2226810664160697,
      "learning_rate": 2.1502691850018265e-06,
      "loss": 0.4573,
      "step": 6675
    },
    {
      "epoch": 8.171358629130967,
      "grad_norm": 2.28428571044889,
      "learning_rate": 2.14963461867505e-06,
      "loss": 0.4298,
      "step": 6676
    },
    {
      "epoch": 8.172582619339046,
      "grad_norm": 2.997896380223036,
      "learning_rate": 2.14900007537299e-06,
      "loss": 0.3779,
      "step": 6677
    },
    {
      "epoch": 8.173806609547123,
      "grad_norm": 2.5589920230247296,
      "learning_rate": 2.1483655551373474e-06,
      "loss": 1.0753,
      "step": 6678
    },
    {
      "epoch": 8.175030599755202,
      "grad_norm": 1.570323946009387,
      "learning_rate": 2.1477310580098197e-06,
      "loss": 0.314,
      "step": 6679
    },
    {
      "epoch": 8.17625458996328,
      "grad_norm": 1.9497693534675653,
      "learning_rate": 2.147096584032105e-06,
      "loss": 0.4015,
      "step": 6680
    },
    {
      "epoch": 8.177478580171359,
      "grad_norm": 1.6538646220109985,
      "learning_rate": 2.1464621332458965e-06,
      "loss": 0.3021,
      "step": 6681
    },
    {
      "epoch": 8.178702570379437,
      "grad_norm": 1.8859867396907404,
      "learning_rate": 2.1458277056928885e-06,
      "loss": 0.5354,
      "step": 6682
    },
    {
      "epoch": 8.179926560587516,
      "grad_norm": 2.0462923770334,
      "learning_rate": 2.145193301414774e-06,
      "loss": 0.4839,
      "step": 6683
    },
    {
      "epoch": 8.181150550795593,
      "grad_norm": 3.1764638328194432,
      "learning_rate": 2.144558920453244e-06,
      "loss": 0.4366,
      "step": 6684
    },
    {
      "epoch": 8.182374541003671,
      "grad_norm": 1.4940195153223896,
      "learning_rate": 2.1439245628499864e-06,
      "loss": 0.8251,
      "step": 6685
    },
    {
      "epoch": 8.18359853121175,
      "grad_norm": 0.8803815968032952,
      "learning_rate": 2.1432902286466893e-06,
      "loss": 0.4974,
      "step": 6686
    },
    {
      "epoch": 8.184822521419829,
      "grad_norm": 2.3973895988894656,
      "learning_rate": 2.1426559178850396e-06,
      "loss": 0.5138,
      "step": 6687
    },
    {
      "epoch": 8.186046511627907,
      "grad_norm": 2.399922407032271,
      "learning_rate": 2.142021630606721e-06,
      "loss": 0.3593,
      "step": 6688
    },
    {
      "epoch": 8.187270501835986,
      "grad_norm": 2.458582555614869,
      "learning_rate": 2.1413873668534175e-06,
      "loss": 0.9569,
      "step": 6689
    },
    {
      "epoch": 8.188494492044065,
      "grad_norm": 1.3219841045953808,
      "learning_rate": 2.1407531266668084e-06,
      "loss": 0.5078,
      "step": 6690
    },
    {
      "epoch": 8.189718482252141,
      "grad_norm": 1.786886327529939,
      "learning_rate": 2.140118910088576e-06,
      "loss": 0.5379,
      "step": 6691
    },
    {
      "epoch": 8.19094247246022,
      "grad_norm": 1.7820163825414106,
      "learning_rate": 2.139484717160398e-06,
      "loss": 0.6181,
      "step": 6692
    },
    {
      "epoch": 8.192166462668299,
      "grad_norm": 1.82121968401776,
      "learning_rate": 2.1388505479239506e-06,
      "loss": 0.6339,
      "step": 6693
    },
    {
      "epoch": 8.193390452876377,
      "grad_norm": 1.0360683861721705,
      "learning_rate": 2.1382164024209094e-06,
      "loss": 0.57,
      "step": 6694
    },
    {
      "epoch": 8.194614443084456,
      "grad_norm": 2.5714272393481226,
      "learning_rate": 2.1375822806929477e-06,
      "loss": 0.8338,
      "step": 6695
    },
    {
      "epoch": 8.195838433292534,
      "grad_norm": 1.5600166080691629,
      "learning_rate": 2.136948182781738e-06,
      "loss": 0.5656,
      "step": 6696
    },
    {
      "epoch": 8.197062423500611,
      "grad_norm": 1.6523175232907512,
      "learning_rate": 2.136314108728951e-06,
      "loss": 0.6632,
      "step": 6697
    },
    {
      "epoch": 8.19828641370869,
      "grad_norm": 0.9734671190130147,
      "learning_rate": 2.1356800585762564e-06,
      "loss": 0.5981,
      "step": 6698
    },
    {
      "epoch": 8.199510403916769,
      "grad_norm": 1.5994324357962955,
      "learning_rate": 2.1350460323653194e-06,
      "loss": 0.5245,
      "step": 6699
    },
    {
      "epoch": 8.200734394124847,
      "grad_norm": 1.682272062621435,
      "learning_rate": 2.134412030137809e-06,
      "loss": 0.5384,
      "step": 6700
    },
    {
      "epoch": 8.201958384332926,
      "grad_norm": 1.7850944820183734,
      "learning_rate": 2.133778051935387e-06,
      "loss": 0.4931,
      "step": 6701
    },
    {
      "epoch": 8.203182374541004,
      "grad_norm": 1.0821734833681462,
      "learning_rate": 2.1331440977997166e-06,
      "loss": 0.7508,
      "step": 6702
    },
    {
      "epoch": 8.204406364749081,
      "grad_norm": 2.5916685060296683,
      "learning_rate": 2.1325101677724585e-06,
      "loss": 0.5467,
      "step": 6703
    },
    {
      "epoch": 8.20563035495716,
      "grad_norm": 1.8478674357515412,
      "learning_rate": 2.131876261895274e-06,
      "loss": 0.9338,
      "step": 6704
    },
    {
      "epoch": 8.206854345165238,
      "grad_norm": 1.8155312794217595,
      "learning_rate": 2.13124238020982e-06,
      "loss": 0.7548,
      "step": 6705
    },
    {
      "epoch": 8.208078335373317,
      "grad_norm": 1.5256660612015858,
      "learning_rate": 2.130608522757752e-06,
      "loss": 0.4622,
      "step": 6706
    },
    {
      "epoch": 8.209302325581396,
      "grad_norm": 1.3451092969048304,
      "learning_rate": 2.129974689580727e-06,
      "loss": 0.7154,
      "step": 6707
    },
    {
      "epoch": 8.210526315789474,
      "grad_norm": 1.3987444071945088,
      "learning_rate": 2.129340880720395e-06,
      "loss": 0.6329,
      "step": 6708
    },
    {
      "epoch": 8.211750305997551,
      "grad_norm": 1.1621294693742426,
      "learning_rate": 2.128707096218411e-06,
      "loss": 0.754,
      "step": 6709
    },
    {
      "epoch": 8.21297429620563,
      "grad_norm": 1.1845142642246196,
      "learning_rate": 2.1280733361164224e-06,
      "loss": 0.5784,
      "step": 6710
    },
    {
      "epoch": 8.214198286413708,
      "grad_norm": 2.189692872493779,
      "learning_rate": 2.1274396004560788e-06,
      "loss": 0.4255,
      "step": 6711
    },
    {
      "epoch": 8.215422276621787,
      "grad_norm": 1.7069615360247032,
      "learning_rate": 2.126805889279026e-06,
      "loss": 0.4953,
      "step": 6712
    },
    {
      "epoch": 8.216646266829866,
      "grad_norm": 1.8455025414397845,
      "learning_rate": 2.1261722026269105e-06,
      "loss": 0.5336,
      "step": 6713
    },
    {
      "epoch": 8.217870257037944,
      "grad_norm": 2.215870657127642,
      "learning_rate": 2.1255385405413746e-06,
      "loss": 0.5535,
      "step": 6714
    },
    {
      "epoch": 8.219094247246023,
      "grad_norm": 1.8008465592282343,
      "learning_rate": 2.1249049030640614e-06,
      "loss": 0.7374,
      "step": 6715
    },
    {
      "epoch": 8.2203182374541,
      "grad_norm": 2.2221660810178148,
      "learning_rate": 2.12427129023661e-06,
      "loss": 0.6016,
      "step": 6716
    },
    {
      "epoch": 8.221542227662178,
      "grad_norm": 1.514802575350776,
      "learning_rate": 2.1236377021006598e-06,
      "loss": 0.6302,
      "step": 6717
    },
    {
      "epoch": 8.222766217870257,
      "grad_norm": 1.3686064572897325,
      "learning_rate": 2.123004138697849e-06,
      "loss": 0.5531,
      "step": 6718
    },
    {
      "epoch": 8.223990208078336,
      "grad_norm": 1.5506624409823542,
      "learning_rate": 2.1223706000698106e-06,
      "loss": 0.559,
      "step": 6719
    },
    {
      "epoch": 8.225214198286414,
      "grad_norm": 1.2619792515703245,
      "learning_rate": 2.12173708625818e-06,
      "loss": 0.6791,
      "step": 6720
    },
    {
      "epoch": 8.226438188494493,
      "grad_norm": 2.699893192131862,
      "learning_rate": 2.121103597304588e-06,
      "loss": 0.7501,
      "step": 6721
    },
    {
      "epoch": 8.22766217870257,
      "grad_norm": 1.8301710128733761,
      "learning_rate": 2.1204701332506674e-06,
      "loss": 0.4253,
      "step": 6722
    },
    {
      "epoch": 8.228886168910648,
      "grad_norm": 2.3546894013600017,
      "learning_rate": 2.119836694138045e-06,
      "loss": 0.4906,
      "step": 6723
    },
    {
      "epoch": 8.230110159118727,
      "grad_norm": 1.4132850068347453,
      "learning_rate": 2.11920328000835e-06,
      "loss": 0.6256,
      "step": 6724
    },
    {
      "epoch": 8.231334149326806,
      "grad_norm": 3.1660562586079974,
      "learning_rate": 2.1185698909032054e-06,
      "loss": 0.3519,
      "step": 6725
    },
    {
      "epoch": 8.232558139534884,
      "grad_norm": 2.030107706445244,
      "learning_rate": 2.1179365268642377e-06,
      "loss": 1.196,
      "step": 6726
    },
    {
      "epoch": 8.233782129742963,
      "grad_norm": 1.608322012543041,
      "learning_rate": 2.117303187933069e-06,
      "loss": 1.1044,
      "step": 6727
    },
    {
      "epoch": 8.23500611995104,
      "grad_norm": 1.4258471824206793,
      "learning_rate": 2.1166698741513184e-06,
      "loss": 0.5584,
      "step": 6728
    },
    {
      "epoch": 8.236230110159118,
      "grad_norm": 1.7014759508851782,
      "learning_rate": 2.116036585560605e-06,
      "loss": 1.3916,
      "step": 6729
    },
    {
      "epoch": 8.237454100367197,
      "grad_norm": 1.6110456457897848,
      "learning_rate": 2.115403322202548e-06,
      "loss": 0.6262,
      "step": 6730
    },
    {
      "epoch": 8.238678090575275,
      "grad_norm": 1.369057315777079,
      "learning_rate": 2.114770084118763e-06,
      "loss": 0.3673,
      "step": 6731
    },
    {
      "epoch": 8.239902080783354,
      "grad_norm": 1.5856641013252073,
      "learning_rate": 2.114136871350862e-06,
      "loss": 0.449,
      "step": 6732
    },
    {
      "epoch": 8.241126070991433,
      "grad_norm": 1.5896961609791502,
      "learning_rate": 2.1135036839404595e-06,
      "loss": 0.9894,
      "step": 6733
    },
    {
      "epoch": 8.24235006119951,
      "grad_norm": 1.1095919475049956,
      "learning_rate": 2.1128705219291636e-06,
      "loss": 0.487,
      "step": 6734
    },
    {
      "epoch": 8.243574051407588,
      "grad_norm": 1.771171627510569,
      "learning_rate": 2.1122373853585866e-06,
      "loss": 1.1139,
      "step": 6735
    },
    {
      "epoch": 8.244798041615667,
      "grad_norm": 2.384446093686923,
      "learning_rate": 2.1116042742703342e-06,
      "loss": 0.3326,
      "step": 6736
    },
    {
      "epoch": 8.246022031823745,
      "grad_norm": 1.4817463362080683,
      "learning_rate": 2.1109711887060124e-06,
      "loss": 0.5037,
      "step": 6737
    },
    {
      "epoch": 8.247246022031824,
      "grad_norm": 1.4327859517819457,
      "learning_rate": 2.1103381287072244e-06,
      "loss": 0.72,
      "step": 6738
    },
    {
      "epoch": 8.248470012239903,
      "grad_norm": 1.3495614381404395,
      "learning_rate": 2.1097050943155744e-06,
      "loss": 0.9724,
      "step": 6739
    },
    {
      "epoch": 8.249694002447981,
      "grad_norm": 1.2732064228089517,
      "learning_rate": 2.1090720855726613e-06,
      "loss": 0.7176,
      "step": 6740
    },
    {
      "epoch": 8.250917992656058,
      "grad_norm": 1.9495001343135345,
      "learning_rate": 2.108439102520085e-06,
      "loss": 0.678,
      "step": 6741
    },
    {
      "epoch": 8.252141982864137,
      "grad_norm": 2.245210237420504,
      "learning_rate": 2.107806145199442e-06,
      "loss": 0.4962,
      "step": 6742
    },
    {
      "epoch": 8.253365973072215,
      "grad_norm": 1.6548151967739417,
      "learning_rate": 2.1071732136523293e-06,
      "loss": 1.0057,
      "step": 6743
    },
    {
      "epoch": 8.254589963280294,
      "grad_norm": 2.193276884181038,
      "learning_rate": 2.1065403079203397e-06,
      "loss": 0.5579,
      "step": 6744
    },
    {
      "epoch": 8.255813953488373,
      "grad_norm": 2.348411526038012,
      "learning_rate": 2.1059074280450663e-06,
      "loss": 0.5225,
      "step": 6745
    },
    {
      "epoch": 8.257037943696451,
      "grad_norm": 2.5345396484067813,
      "learning_rate": 2.1052745740680988e-06,
      "loss": 0.5894,
      "step": 6746
    },
    {
      "epoch": 8.258261933904528,
      "grad_norm": 1.5737495522063873,
      "learning_rate": 2.1046417460310257e-06,
      "loss": 0.4743,
      "step": 6747
    },
    {
      "epoch": 8.259485924112607,
      "grad_norm": 1.0999047407078313,
      "learning_rate": 2.104008943975435e-06,
      "loss": 0.8891,
      "step": 6748
    },
    {
      "epoch": 8.260709914320685,
      "grad_norm": 2.3640012164651627,
      "learning_rate": 2.1033761679429125e-06,
      "loss": 0.6317,
      "step": 6749
    },
    {
      "epoch": 8.261933904528764,
      "grad_norm": 1.7448125991572676,
      "learning_rate": 2.1027434179750404e-06,
      "loss": 1.4651,
      "step": 6750
    },
    {
      "epoch": 8.263157894736842,
      "grad_norm": 2.082843579032236,
      "learning_rate": 2.1021106941134014e-06,
      "loss": 0.5306,
      "step": 6751
    },
    {
      "epoch": 8.264381884944921,
      "grad_norm": 1.285677467447035,
      "learning_rate": 2.1014779963995767e-06,
      "loss": 0.56,
      "step": 6752
    },
    {
      "epoch": 8.265605875152998,
      "grad_norm": 2.1964523325078935,
      "learning_rate": 2.1008453248751442e-06,
      "loss": 0.4775,
      "step": 6753
    },
    {
      "epoch": 8.266829865361077,
      "grad_norm": 1.6362339127780214,
      "learning_rate": 2.10021267958168e-06,
      "loss": 0.3189,
      "step": 6754
    },
    {
      "epoch": 8.268053855569155,
      "grad_norm": 1.0116216538068679,
      "learning_rate": 2.09958006056076e-06,
      "loss": 0.367,
      "step": 6755
    },
    {
      "epoch": 8.269277845777234,
      "grad_norm": 1.4266712203175176,
      "learning_rate": 2.098947467853957e-06,
      "loss": 0.498,
      "step": 6756
    },
    {
      "epoch": 8.270501835985312,
      "grad_norm": 1.3497383671407535,
      "learning_rate": 2.0983149015028433e-06,
      "loss": 1.1624,
      "step": 6757
    },
    {
      "epoch": 8.271725826193391,
      "grad_norm": 1.3987575352186479,
      "learning_rate": 2.0976823615489884e-06,
      "loss": 0.6565,
      "step": 6758
    },
    {
      "epoch": 8.27294981640147,
      "grad_norm": 1.4749288224159403,
      "learning_rate": 2.097049848033961e-06,
      "loss": 1.2508,
      "step": 6759
    },
    {
      "epoch": 8.274173806609546,
      "grad_norm": 2.0940035104170978,
      "learning_rate": 2.096417360999326e-06,
      "loss": 0.5891,
      "step": 6760
    },
    {
      "epoch": 8.275397796817625,
      "grad_norm": 1.6991679231192904,
      "learning_rate": 2.0957849004866503e-06,
      "loss": 0.4715,
      "step": 6761
    },
    {
      "epoch": 8.276621787025704,
      "grad_norm": 1.7989150696472913,
      "learning_rate": 2.0951524665374966e-06,
      "loss": 0.4668,
      "step": 6762
    },
    {
      "epoch": 8.277845777233782,
      "grad_norm": 1.431294075021027,
      "learning_rate": 2.0945200591934247e-06,
      "loss": 0.5813,
      "step": 6763
    },
    {
      "epoch": 8.279069767441861,
      "grad_norm": 1.4361816478408258,
      "learning_rate": 2.0938876784959932e-06,
      "loss": 0.6765,
      "step": 6764
    },
    {
      "epoch": 8.28029375764994,
      "grad_norm": 1.0359744260475334,
      "learning_rate": 2.0932553244867632e-06,
      "loss": 0.3745,
      "step": 6765
    },
    {
      "epoch": 8.281517747858016,
      "grad_norm": 1.8166112840470832,
      "learning_rate": 2.0926229972072884e-06,
      "loss": 0.3704,
      "step": 6766
    },
    {
      "epoch": 8.282741738066095,
      "grad_norm": 1.5874678889682423,
      "learning_rate": 2.0919906966991234e-06,
      "loss": 0.6657,
      "step": 6767
    },
    {
      "epoch": 8.283965728274174,
      "grad_norm": 1.3002318697371347,
      "learning_rate": 2.091358423003821e-06,
      "loss": 0.8721,
      "step": 6768
    },
    {
      "epoch": 8.285189718482252,
      "grad_norm": 2.072009868604084,
      "learning_rate": 2.0907261761629304e-06,
      "loss": 0.6825,
      "step": 6769
    },
    {
      "epoch": 8.286413708690331,
      "grad_norm": 1.951644430293242,
      "learning_rate": 2.0900939562180028e-06,
      "loss": 0.6618,
      "step": 6770
    },
    {
      "epoch": 8.28763769889841,
      "grad_norm": 1.7413664251402752,
      "learning_rate": 2.0894617632105846e-06,
      "loss": 0.6916,
      "step": 6771
    },
    {
      "epoch": 8.288861689106486,
      "grad_norm": 1.053326936700101,
      "learning_rate": 2.0888295971822204e-06,
      "loss": 0.62,
      "step": 6772
    },
    {
      "epoch": 8.290085679314565,
      "grad_norm": 1.914578599713639,
      "learning_rate": 2.088197458174453e-06,
      "loss": 1.1472,
      "step": 6773
    },
    {
      "epoch": 8.291309669522644,
      "grad_norm": 1.59381467019456,
      "learning_rate": 2.0875653462288272e-06,
      "loss": 0.6854,
      "step": 6774
    },
    {
      "epoch": 8.292533659730722,
      "grad_norm": 2.992613059046619,
      "learning_rate": 2.0869332613868807e-06,
      "loss": 0.4548,
      "step": 6775
    },
    {
      "epoch": 8.2937576499388,
      "grad_norm": 2.0344607930791074,
      "learning_rate": 2.0863012036901528e-06,
      "loss": 1.0239,
      "step": 6776
    },
    {
      "epoch": 8.29498164014688,
      "grad_norm": 2.3698672687306375,
      "learning_rate": 2.085669173180179e-06,
      "loss": 0.4126,
      "step": 6777
    },
    {
      "epoch": 8.296205630354958,
      "grad_norm": 1.4317860778082927,
      "learning_rate": 2.085037169898495e-06,
      "loss": 1.0913,
      "step": 6778
    },
    {
      "epoch": 8.297429620563035,
      "grad_norm": 2.112292106529208,
      "learning_rate": 2.0844051938866334e-06,
      "loss": 0.3344,
      "step": 6779
    },
    {
      "epoch": 8.298653610771114,
      "grad_norm": 2.0735705521766588,
      "learning_rate": 2.0837732451861254e-06,
      "loss": 0.6421,
      "step": 6780
    },
    {
      "epoch": 8.299877600979192,
      "grad_norm": 1.246949879564877,
      "learning_rate": 2.0831413238385005e-06,
      "loss": 0.7354,
      "step": 6781
    },
    {
      "epoch": 8.30110159118727,
      "grad_norm": 2.720861952136872,
      "learning_rate": 2.082509429885285e-06,
      "loss": 0.9224,
      "step": 6782
    },
    {
      "epoch": 8.30232558139535,
      "grad_norm": 1.7113489685599366,
      "learning_rate": 2.0818775633680057e-06,
      "loss": 0.3012,
      "step": 6783
    },
    {
      "epoch": 8.303549571603428,
      "grad_norm": 1.5961815147557212,
      "learning_rate": 2.081245724328187e-06,
      "loss": 0.5085,
      "step": 6784
    },
    {
      "epoch": 8.304773561811505,
      "grad_norm": 2.4260714758487714,
      "learning_rate": 2.0806139128073504e-06,
      "loss": 0.4103,
      "step": 6785
    },
    {
      "epoch": 8.305997552019583,
      "grad_norm": 2.7867483053975266,
      "learning_rate": 2.0799821288470155e-06,
      "loss": 0.3339,
      "step": 6786
    },
    {
      "epoch": 8.307221542227662,
      "grad_norm": 1.1644117937877905,
      "learning_rate": 2.079350372488702e-06,
      "loss": 1.0455,
      "step": 6787
    },
    {
      "epoch": 8.30844553243574,
      "grad_norm": 2.4071487147458277,
      "learning_rate": 2.078718643773926e-06,
      "loss": 0.4442,
      "step": 6788
    },
    {
      "epoch": 8.30966952264382,
      "grad_norm": 1.3280722816880743,
      "learning_rate": 2.0780869427442035e-06,
      "loss": 1.1564,
      "step": 6789
    },
    {
      "epoch": 8.310893512851898,
      "grad_norm": 1.2839188514873148,
      "learning_rate": 2.077455269441045e-06,
      "loss": 0.3658,
      "step": 6790
    },
    {
      "epoch": 8.312117503059975,
      "grad_norm": 2.0866053189124747,
      "learning_rate": 2.076823623905965e-06,
      "loss": 1.3442,
      "step": 6791
    },
    {
      "epoch": 8.313341493268053,
      "grad_norm": 1.6545847559824294,
      "learning_rate": 2.07619200618047e-06,
      "loss": 0.8894,
      "step": 6792
    },
    {
      "epoch": 8.314565483476132,
      "grad_norm": 1.4727434613282415,
      "learning_rate": 2.075560416306069e-06,
      "loss": 1.4901,
      "step": 6793
    },
    {
      "epoch": 8.31578947368421,
      "grad_norm": 2.605610783118542,
      "learning_rate": 2.074928854324268e-06,
      "loss": 0.4296,
      "step": 6794
    },
    {
      "epoch": 8.31701346389229,
      "grad_norm": 0.8857399561794008,
      "learning_rate": 2.0742973202765696e-06,
      "loss": 0.3487,
      "step": 6795
    },
    {
      "epoch": 8.318237454100368,
      "grad_norm": 2.101797883365381,
      "learning_rate": 2.0736658142044776e-06,
      "loss": 0.9794,
      "step": 6796
    },
    {
      "epoch": 8.319461444308445,
      "grad_norm": 1.0652121702932194,
      "learning_rate": 2.073034336149491e-06,
      "loss": 0.6028,
      "step": 6797
    },
    {
      "epoch": 8.320685434516523,
      "grad_norm": 1.3839643983213603,
      "learning_rate": 2.072402886153109e-06,
      "loss": 0.5129,
      "step": 6798
    },
    {
      "epoch": 8.321909424724602,
      "grad_norm": 1.2438367460226005,
      "learning_rate": 2.0717714642568264e-06,
      "loss": 0.6627,
      "step": 6799
    },
    {
      "epoch": 8.32313341493268,
      "grad_norm": 1.3737488812909973,
      "learning_rate": 2.0711400705021413e-06,
      "loss": 0.5324,
      "step": 6800
    },
    {
      "epoch": 8.32435740514076,
      "grad_norm": 1.774709631141625,
      "learning_rate": 2.0705087049305433e-06,
      "loss": 0.4884,
      "step": 6801
    },
    {
      "epoch": 8.325581395348838,
      "grad_norm": 1.0732136486811348,
      "learning_rate": 2.069877367583525e-06,
      "loss": 0.4756,
      "step": 6802
    },
    {
      "epoch": 8.326805385556916,
      "grad_norm": 1.0719627827164726,
      "learning_rate": 2.0692460585025743e-06,
      "loss": 0.6092,
      "step": 6803
    },
    {
      "epoch": 8.328029375764993,
      "grad_norm": 2.4598849394385995,
      "learning_rate": 2.0686147777291803e-06,
      "loss": 0.5021,
      "step": 6804
    },
    {
      "epoch": 8.329253365973072,
      "grad_norm": 1.425740124878504,
      "learning_rate": 2.0679835253048275e-06,
      "loss": 0.8715,
      "step": 6805
    },
    {
      "epoch": 8.33047735618115,
      "grad_norm": 1.6707733517804741,
      "learning_rate": 2.0673523012709997e-06,
      "loss": 1.1657,
      "step": 6806
    },
    {
      "epoch": 8.331701346389229,
      "grad_norm": 1.7249642200112596,
      "learning_rate": 2.0667211056691783e-06,
      "loss": 0.5078,
      "step": 6807
    },
    {
      "epoch": 8.332925336597308,
      "grad_norm": 1.453933640824531,
      "learning_rate": 2.066089938540842e-06,
      "loss": 0.3626,
      "step": 6808
    },
    {
      "epoch": 8.334149326805386,
      "grad_norm": 1.3321100873739509,
      "learning_rate": 2.0654587999274715e-06,
      "loss": 0.4366,
      "step": 6809
    },
    {
      "epoch": 8.335373317013463,
      "grad_norm": 1.1667372925937674,
      "learning_rate": 2.064827689870541e-06,
      "loss": 0.392,
      "step": 6810
    },
    {
      "epoch": 8.336597307221542,
      "grad_norm": 3.1394919175102993,
      "learning_rate": 2.064196608411525e-06,
      "loss": 0.4861,
      "step": 6811
    },
    {
      "epoch": 8.33782129742962,
      "grad_norm": 1.1134408590171994,
      "learning_rate": 2.0635655555918957e-06,
      "loss": 0.5081,
      "step": 6812
    },
    {
      "epoch": 8.339045287637699,
      "grad_norm": 1.8774082004774948,
      "learning_rate": 2.062934531453124e-06,
      "loss": 0.3019,
      "step": 6813
    },
    {
      "epoch": 8.340269277845778,
      "grad_norm": 1.5663781361641809,
      "learning_rate": 2.062303536036679e-06,
      "loss": 0.7362,
      "step": 6814
    },
    {
      "epoch": 8.341493268053856,
      "grad_norm": 1.9858781973250021,
      "learning_rate": 2.0616725693840257e-06,
      "loss": 0.5444,
      "step": 6815
    },
    {
      "epoch": 8.342717258261933,
      "grad_norm": 1.1553833772943796,
      "learning_rate": 2.06104163153663e-06,
      "loss": 0.5603,
      "step": 6816
    },
    {
      "epoch": 8.343941248470012,
      "grad_norm": 1.1608865636745855,
      "learning_rate": 2.060410722535955e-06,
      "loss": 0.6063,
      "step": 6817
    },
    {
      "epoch": 8.34516523867809,
      "grad_norm": 1.3641510293732337,
      "learning_rate": 2.0597798424234624e-06,
      "loss": 0.5947,
      "step": 6818
    },
    {
      "epoch": 8.346389228886169,
      "grad_norm": 1.6032515499589617,
      "learning_rate": 2.059148991240609e-06,
      "loss": 0.4921,
      "step": 6819
    },
    {
      "epoch": 8.347613219094248,
      "grad_norm": 2.144048809414431,
      "learning_rate": 2.0585181690288538e-06,
      "loss": 0.5399,
      "step": 6820
    },
    {
      "epoch": 8.348837209302326,
      "grad_norm": 1.9850893023138876,
      "learning_rate": 2.057887375829651e-06,
      "loss": 0.5216,
      "step": 6821
    },
    {
      "epoch": 8.350061199510403,
      "grad_norm": 2.227170451793553,
      "learning_rate": 2.0572566116844547e-06,
      "loss": 0.8738,
      "step": 6822
    },
    {
      "epoch": 8.351285189718482,
      "grad_norm": 2.0383547852124706,
      "learning_rate": 2.0566258766347167e-06,
      "loss": 1.1282,
      "step": 6823
    },
    {
      "epoch": 8.35250917992656,
      "grad_norm": 2.6014354088055485,
      "learning_rate": 2.0559951707218863e-06,
      "loss": 0.3638,
      "step": 6824
    },
    {
      "epoch": 8.353733170134639,
      "grad_norm": 1.2688177141995514,
      "learning_rate": 2.05536449398741e-06,
      "loss": 0.5223,
      "step": 6825
    },
    {
      "epoch": 8.354957160342718,
      "grad_norm": 1.682948828933325,
      "learning_rate": 2.054733846472735e-06,
      "loss": 0.9381,
      "step": 6826
    },
    {
      "epoch": 8.356181150550796,
      "grad_norm": 2.3556846243576572,
      "learning_rate": 2.054103228219306e-06,
      "loss": 0.9946,
      "step": 6827
    },
    {
      "epoch": 8.357405140758875,
      "grad_norm": 1.666616377118988,
      "learning_rate": 2.0534726392685627e-06,
      "loss": 0.6282,
      "step": 6828
    },
    {
      "epoch": 8.358629130966952,
      "grad_norm": 1.3707270378830119,
      "learning_rate": 2.052842079661945e-06,
      "loss": 1.2796,
      "step": 6829
    },
    {
      "epoch": 8.35985312117503,
      "grad_norm": 1.2083573403036485,
      "learning_rate": 2.0522115494408932e-06,
      "loss": 0.7671,
      "step": 6830
    },
    {
      "epoch": 8.361077111383109,
      "grad_norm": 1.1174026788841374,
      "learning_rate": 2.0515810486468416e-06,
      "loss": 0.5437,
      "step": 6831
    },
    {
      "epoch": 8.362301101591187,
      "grad_norm": 1.546058902351172,
      "learning_rate": 2.050950577321225e-06,
      "loss": 1.6447,
      "step": 6832
    },
    {
      "epoch": 8.363525091799266,
      "grad_norm": 1.252668060576506,
      "learning_rate": 2.050320135505476e-06,
      "loss": 0.6309,
      "step": 6833
    },
    {
      "epoch": 8.364749082007345,
      "grad_norm": 1.8654232756845972,
      "learning_rate": 2.049689723241023e-06,
      "loss": 0.3997,
      "step": 6834
    },
    {
      "epoch": 8.365973072215422,
      "grad_norm": 1.6231936388585335,
      "learning_rate": 2.0490593405692967e-06,
      "loss": 0.7407,
      "step": 6835
    },
    {
      "epoch": 8.3671970624235,
      "grad_norm": 1.3457571095214387,
      "learning_rate": 2.0484289875317236e-06,
      "loss": 1.0304,
      "step": 6836
    },
    {
      "epoch": 8.368421052631579,
      "grad_norm": 1.14068004119968,
      "learning_rate": 2.0477986641697263e-06,
      "loss": 0.5336,
      "step": 6837
    },
    {
      "epoch": 8.369645042839657,
      "grad_norm": 1.2752132172781216,
      "learning_rate": 2.0471683705247275e-06,
      "loss": 0.6511,
      "step": 6838
    },
    {
      "epoch": 8.370869033047736,
      "grad_norm": 2.1348581090097833,
      "learning_rate": 2.046538106638149e-06,
      "loss": 0.4558,
      "step": 6839
    },
    {
      "epoch": 8.372093023255815,
      "grad_norm": 1.4098613119270778,
      "learning_rate": 2.045907872551409e-06,
      "loss": 0.8978,
      "step": 6840
    },
    {
      "epoch": 8.373317013463891,
      "grad_norm": 2.4792318228813985,
      "learning_rate": 2.0452776683059246e-06,
      "loss": 0.4757,
      "step": 6841
    },
    {
      "epoch": 8.37454100367197,
      "grad_norm": 1.7922595657013023,
      "learning_rate": 2.0446474939431086e-06,
      "loss": 0.5988,
      "step": 6842
    },
    {
      "epoch": 8.375764993880049,
      "grad_norm": 3.2039573023020425,
      "learning_rate": 2.044017349504376e-06,
      "loss": 0.2865,
      "step": 6843
    },
    {
      "epoch": 8.376988984088127,
      "grad_norm": 1.656442882211272,
      "learning_rate": 2.0433872350311367e-06,
      "loss": 0.4233,
      "step": 6844
    },
    {
      "epoch": 8.378212974296206,
      "grad_norm": 1.7115662539335517,
      "learning_rate": 2.0427571505648e-06,
      "loss": 0.4689,
      "step": 6845
    },
    {
      "epoch": 8.379436964504285,
      "grad_norm": 1.597843038260663,
      "learning_rate": 2.0421270961467713e-06,
      "loss": 0.5481,
      "step": 6846
    },
    {
      "epoch": 8.380660954712361,
      "grad_norm": 1.8851945680330533,
      "learning_rate": 2.0414970718184557e-06,
      "loss": 0.4514,
      "step": 6847
    },
    {
      "epoch": 8.38188494492044,
      "grad_norm": 1.3415422344928152,
      "learning_rate": 2.0408670776212573e-06,
      "loss": 0.4068,
      "step": 6848
    },
    {
      "epoch": 8.383108935128519,
      "grad_norm": 2.4239336931726476,
      "learning_rate": 2.0402371135965765e-06,
      "loss": 0.68,
      "step": 6849
    },
    {
      "epoch": 8.384332925336597,
      "grad_norm": 1.0978519134438605,
      "learning_rate": 2.039607179785812e-06,
      "loss": 0.5674,
      "step": 6850
    },
    {
      "epoch": 8.385556915544676,
      "grad_norm": 1.1190642771781085,
      "learning_rate": 2.0389772762303604e-06,
      "loss": 0.6769,
      "step": 6851
    },
    {
      "epoch": 8.386780905752754,
      "grad_norm": 1.8389414534254926,
      "learning_rate": 2.0383474029716173e-06,
      "loss": 1.0561,
      "step": 6852
    },
    {
      "epoch": 8.388004895960833,
      "grad_norm": 1.9656203623550008,
      "learning_rate": 2.037717560050975e-06,
      "loss": 0.496,
      "step": 6853
    },
    {
      "epoch": 8.38922888616891,
      "grad_norm": 1.327016169674342,
      "learning_rate": 2.0370877475098262e-06,
      "loss": 0.5376,
      "step": 6854
    },
    {
      "epoch": 8.390452876376989,
      "grad_norm": 1.8536413053123195,
      "learning_rate": 2.036457965389556e-06,
      "loss": 0.3275,
      "step": 6855
    },
    {
      "epoch": 8.391676866585067,
      "grad_norm": 1.0021245001200034,
      "learning_rate": 2.035828213731556e-06,
      "loss": 0.5228,
      "step": 6856
    },
    {
      "epoch": 8.392900856793146,
      "grad_norm": 1.8241781320089239,
      "learning_rate": 2.035198492577208e-06,
      "loss": 0.6322,
      "step": 6857
    },
    {
      "epoch": 8.394124847001224,
      "grad_norm": 2.8210332212182134,
      "learning_rate": 2.0345688019678963e-06,
      "loss": 0.58,
      "step": 6858
    },
    {
      "epoch": 8.395348837209303,
      "grad_norm": 2.6151300744916948,
      "learning_rate": 2.033939141945001e-06,
      "loss": 0.5957,
      "step": 6859
    },
    {
      "epoch": 8.39657282741738,
      "grad_norm": 2.165004402826451,
      "learning_rate": 2.033309512549901e-06,
      "loss": 1.0409,
      "step": 6860
    },
    {
      "epoch": 8.397796817625458,
      "grad_norm": 1.7571171236797694,
      "learning_rate": 2.032679913823974e-06,
      "loss": 0.4679,
      "step": 6861
    },
    {
      "epoch": 8.399020807833537,
      "grad_norm": 1.9053920841864824,
      "learning_rate": 2.0320503458085956e-06,
      "loss": 0.5575,
      "step": 6862
    },
    {
      "epoch": 8.400244798041616,
      "grad_norm": 1.9773040311635894,
      "learning_rate": 2.031420808545137e-06,
      "loss": 0.5818,
      "step": 6863
    },
    {
      "epoch": 8.401468788249694,
      "grad_norm": 3.0110220403143764,
      "learning_rate": 2.0307913020749685e-06,
      "loss": 0.3926,
      "step": 6864
    },
    {
      "epoch": 8.402692778457773,
      "grad_norm": 1.0568991420469478,
      "learning_rate": 2.0301618264394617e-06,
      "loss": 0.3997,
      "step": 6865
    },
    {
      "epoch": 8.403916768665852,
      "grad_norm": 1.7182677194320992,
      "learning_rate": 2.0295323816799814e-06,
      "loss": 1.3688,
      "step": 6866
    },
    {
      "epoch": 8.405140758873928,
      "grad_norm": 1.4008665143296746,
      "learning_rate": 2.028902967837893e-06,
      "loss": 0.6216,
      "step": 6867
    },
    {
      "epoch": 8.406364749082007,
      "grad_norm": 2.4883738778729336,
      "learning_rate": 2.0282735849545583e-06,
      "loss": 1.1887,
      "step": 6868
    },
    {
      "epoch": 8.407588739290086,
      "grad_norm": 1.3358477596711156,
      "learning_rate": 2.0276442330713394e-06,
      "loss": 0.8498,
      "step": 6869
    },
    {
      "epoch": 8.408812729498164,
      "grad_norm": 1.602101616691074,
      "learning_rate": 2.0270149122295943e-06,
      "loss": 0.5024,
      "step": 6870
    },
    {
      "epoch": 8.410036719706243,
      "grad_norm": 1.7691185932609315,
      "learning_rate": 2.0263856224706805e-06,
      "loss": 0.3458,
      "step": 6871
    },
    {
      "epoch": 8.411260709914322,
      "grad_norm": 1.1145475603032757,
      "learning_rate": 2.0257563638359507e-06,
      "loss": 1.2042,
      "step": 6872
    },
    {
      "epoch": 8.412484700122398,
      "grad_norm": 1.18557478379905,
      "learning_rate": 2.025127136366758e-06,
      "loss": 0.7738,
      "step": 6873
    },
    {
      "epoch": 8.413708690330477,
      "grad_norm": 1.2968808701250418,
      "learning_rate": 2.0244979401044553e-06,
      "loss": 0.7448,
      "step": 6874
    },
    {
      "epoch": 8.414932680538556,
      "grad_norm": 1.0343370882932446,
      "learning_rate": 2.0238687750903883e-06,
      "loss": 0.5108,
      "step": 6875
    },
    {
      "epoch": 8.416156670746634,
      "grad_norm": 2.296990632500212,
      "learning_rate": 2.0232396413659043e-06,
      "loss": 0.5308,
      "step": 6876
    },
    {
      "epoch": 8.417380660954713,
      "grad_norm": 2.2444572686856623,
      "learning_rate": 2.022610538972347e-06,
      "loss": 0.4093,
      "step": 6877
    },
    {
      "epoch": 8.418604651162791,
      "grad_norm": 2.1598296098055196,
      "learning_rate": 2.02198146795106e-06,
      "loss": 0.5268,
      "step": 6878
    },
    {
      "epoch": 8.419828641370868,
      "grad_norm": 1.6325921715529566,
      "learning_rate": 2.0213524283433834e-06,
      "loss": 0.4158,
      "step": 6879
    },
    {
      "epoch": 8.421052631578947,
      "grad_norm": 1.8089362406027387,
      "learning_rate": 2.0207234201906546e-06,
      "loss": 1.3975,
      "step": 6880
    },
    {
      "epoch": 8.422276621787026,
      "grad_norm": 1.8612527890644999,
      "learning_rate": 2.0200944435342086e-06,
      "loss": 0.5012,
      "step": 6881
    },
    {
      "epoch": 8.423500611995104,
      "grad_norm": 1.9504900453712164,
      "learning_rate": 2.019465498415382e-06,
      "loss": 0.6228,
      "step": 6882
    },
    {
      "epoch": 8.424724602203183,
      "grad_norm": 2.6515030430435997,
      "learning_rate": 2.0188365848755057e-06,
      "loss": 0.4604,
      "step": 6883
    },
    {
      "epoch": 8.425948592411261,
      "grad_norm": 1.0143521450619137,
      "learning_rate": 2.0182077029559086e-06,
      "loss": 0.3916,
      "step": 6884
    },
    {
      "epoch": 8.427172582619338,
      "grad_norm": 1.7655041446453574,
      "learning_rate": 2.01757885269792e-06,
      "loss": 1.2224,
      "step": 6885
    },
    {
      "epoch": 8.428396572827417,
      "grad_norm": 0.9272933985207185,
      "learning_rate": 2.0169500341428637e-06,
      "loss": 0.4503,
      "step": 6886
    },
    {
      "epoch": 8.429620563035495,
      "grad_norm": 2.757552124711763,
      "learning_rate": 2.016321247332065e-06,
      "loss": 0.4304,
      "step": 6887
    },
    {
      "epoch": 8.430844553243574,
      "grad_norm": 1.8491804811230579,
      "learning_rate": 2.015692492306845e-06,
      "loss": 0.9155,
      "step": 6888
    },
    {
      "epoch": 8.432068543451653,
      "grad_norm": 2.0338262293146614,
      "learning_rate": 2.015063769108524e-06,
      "loss": 0.9725,
      "step": 6889
    },
    {
      "epoch": 8.433292533659731,
      "grad_norm": 1.324387009278543,
      "learning_rate": 2.0144350777784163e-06,
      "loss": 1.1346,
      "step": 6890
    },
    {
      "epoch": 8.43451652386781,
      "grad_norm": 1.0707878985759098,
      "learning_rate": 2.0138064183578412e-06,
      "loss": 0.6247,
      "step": 6891
    },
    {
      "epoch": 8.435740514075887,
      "grad_norm": 2.2611799007059346,
      "learning_rate": 2.0131777908881096e-06,
      "loss": 0.4719,
      "step": 6892
    },
    {
      "epoch": 8.436964504283965,
      "grad_norm": 1.6975677601065493,
      "learning_rate": 2.0125491954105324e-06,
      "loss": 1.0275,
      "step": 6893
    },
    {
      "epoch": 8.438188494492044,
      "grad_norm": 2.0861918018359993,
      "learning_rate": 2.011920631966419e-06,
      "loss": 0.5208,
      "step": 6894
    },
    {
      "epoch": 8.439412484700123,
      "grad_norm": 1.754974238545466,
      "learning_rate": 2.0112921005970768e-06,
      "loss": 0.6605,
      "step": 6895
    },
    {
      "epoch": 8.440636474908201,
      "grad_norm": 2.1497824673124355,
      "learning_rate": 2.0106636013438103e-06,
      "loss": 0.6481,
      "step": 6896
    },
    {
      "epoch": 8.44186046511628,
      "grad_norm": 2.4560190232931527,
      "learning_rate": 2.010035134247922e-06,
      "loss": 0.6869,
      "step": 6897
    },
    {
      "epoch": 8.443084455324357,
      "grad_norm": 1.367054637174487,
      "learning_rate": 2.0094066993507128e-06,
      "loss": 1.0083,
      "step": 6898
    },
    {
      "epoch": 8.444308445532435,
      "grad_norm": 2.667424874141127,
      "learning_rate": 2.0087782966934788e-06,
      "loss": 0.435,
      "step": 6899
    },
    {
      "epoch": 8.445532435740514,
      "grad_norm": 1.4306981438521915,
      "learning_rate": 2.0081499263175204e-06,
      "loss": 1.3167,
      "step": 6900
    },
    {
      "epoch": 8.446756425948593,
      "grad_norm": 1.4035655127370477,
      "learning_rate": 2.007521588264129e-06,
      "loss": 0.8143,
      "step": 6901
    },
    {
      "epoch": 8.447980416156671,
      "grad_norm": 1.0646452283768477,
      "learning_rate": 2.0068932825745973e-06,
      "loss": 0.5515,
      "step": 6902
    },
    {
      "epoch": 8.44920440636475,
      "grad_norm": 1.0594162101167974,
      "learning_rate": 2.006265009290214e-06,
      "loss": 0.5885,
      "step": 6903
    },
    {
      "epoch": 8.450428396572827,
      "grad_norm": 1.3454459498404339,
      "learning_rate": 2.0056367684522693e-06,
      "loss": 0.7009,
      "step": 6904
    },
    {
      "epoch": 8.451652386780905,
      "grad_norm": 1.731456978484616,
      "learning_rate": 2.0050085601020476e-06,
      "loss": 0.5149,
      "step": 6905
    },
    {
      "epoch": 8.452876376988984,
      "grad_norm": 2.1124859246293677,
      "learning_rate": 2.0043803842808323e-06,
      "loss": 0.3843,
      "step": 6906
    },
    {
      "epoch": 8.454100367197062,
      "grad_norm": 1.756268694537351,
      "learning_rate": 2.003752241029906e-06,
      "loss": 0.4338,
      "step": 6907
    },
    {
      "epoch": 8.455324357405141,
      "grad_norm": 1.5286467358429099,
      "learning_rate": 2.0031241303905445e-06,
      "loss": 0.942,
      "step": 6908
    },
    {
      "epoch": 8.45654834761322,
      "grad_norm": 1.3663857157089,
      "learning_rate": 2.002496052404029e-06,
      "loss": 0.5985,
      "step": 6909
    },
    {
      "epoch": 8.457772337821297,
      "grad_norm": 1.4107113093896,
      "learning_rate": 2.001868007111633e-06,
      "loss": 0.6929,
      "step": 6910
    },
    {
      "epoch": 8.458996328029375,
      "grad_norm": 1.6320947535621373,
      "learning_rate": 2.0012399945546283e-06,
      "loss": 1.8083,
      "step": 6911
    },
    {
      "epoch": 8.460220318237454,
      "grad_norm": 1.4744483172046594,
      "learning_rate": 2.0006120147742865e-06,
      "loss": 0.7232,
      "step": 6912
    },
    {
      "epoch": 8.461444308445532,
      "grad_norm": 2.00352122411292,
      "learning_rate": 1.9999840678118767e-06,
      "loss": 0.9811,
      "step": 6913
    },
    {
      "epoch": 8.462668298653611,
      "grad_norm": 1.7565456244570796,
      "learning_rate": 1.9993561537086637e-06,
      "loss": 1.0189,
      "step": 6914
    },
    {
      "epoch": 8.46389228886169,
      "grad_norm": 1.2860435004003699,
      "learning_rate": 1.998728272505913e-06,
      "loss": 0.5951,
      "step": 6915
    },
    {
      "epoch": 8.465116279069768,
      "grad_norm": 1.4516205440411636,
      "learning_rate": 1.9981004242448855e-06,
      "loss": 0.9199,
      "step": 6916
    },
    {
      "epoch": 8.466340269277845,
      "grad_norm": 1.0543644906493497,
      "learning_rate": 1.997472608966842e-06,
      "loss": 0.6671,
      "step": 6917
    },
    {
      "epoch": 8.467564259485924,
      "grad_norm": 2.362233999394828,
      "learning_rate": 1.996844826713041e-06,
      "loss": 0.5139,
      "step": 6918
    },
    {
      "epoch": 8.468788249694002,
      "grad_norm": 1.6159726855292025,
      "learning_rate": 1.9962170775247365e-06,
      "loss": 0.4396,
      "step": 6919
    },
    {
      "epoch": 8.470012239902081,
      "grad_norm": 2.017775189588905,
      "learning_rate": 1.995589361443182e-06,
      "loss": 0.3846,
      "step": 6920
    },
    {
      "epoch": 8.47123623011016,
      "grad_norm": 2.1827549794974397,
      "learning_rate": 1.994961678509628e-06,
      "loss": 0.4734,
      "step": 6921
    },
    {
      "epoch": 8.472460220318238,
      "grad_norm": 1.3248508693903187,
      "learning_rate": 1.9943340287653255e-06,
      "loss": 0.4416,
      "step": 6922
    },
    {
      "epoch": 8.473684210526315,
      "grad_norm": 1.9989058329621847,
      "learning_rate": 1.9937064122515205e-06,
      "loss": 0.8537,
      "step": 6923
    },
    {
      "epoch": 8.474908200734394,
      "grad_norm": 1.6317226144446266,
      "learning_rate": 1.993078829009457e-06,
      "loss": 0.8315,
      "step": 6924
    },
    {
      "epoch": 8.476132190942472,
      "grad_norm": 2.4504369021737,
      "learning_rate": 1.9924512790803775e-06,
      "loss": 0.7389,
      "step": 6925
    },
    {
      "epoch": 8.477356181150551,
      "grad_norm": 0.7875174469814209,
      "learning_rate": 1.991823762505523e-06,
      "loss": 0.4365,
      "step": 6926
    },
    {
      "epoch": 8.47858017135863,
      "grad_norm": 2.1042315666532097,
      "learning_rate": 1.991196279326132e-06,
      "loss": 0.5648,
      "step": 6927
    },
    {
      "epoch": 8.479804161566708,
      "grad_norm": 1.8001978364625342,
      "learning_rate": 1.990568829583439e-06,
      "loss": 0.4665,
      "step": 6928
    },
    {
      "epoch": 8.481028151774785,
      "grad_norm": 1.9621950415326832,
      "learning_rate": 1.989941413318678e-06,
      "loss": 1.2034,
      "step": 6929
    },
    {
      "epoch": 8.482252141982864,
      "grad_norm": 1.1252543620823954,
      "learning_rate": 1.9893140305730805e-06,
      "loss": 0.6317,
      "step": 6930
    },
    {
      "epoch": 8.483476132190942,
      "grad_norm": 1.082171300932542,
      "learning_rate": 1.9886866813878767e-06,
      "loss": 0.563,
      "step": 6931
    },
    {
      "epoch": 8.48470012239902,
      "grad_norm": 2.2937667187947244,
      "learning_rate": 1.988059365804293e-06,
      "loss": 0.5672,
      "step": 6932
    },
    {
      "epoch": 8.4859241126071,
      "grad_norm": 1.9160811262149635,
      "learning_rate": 1.9874320838635537e-06,
      "loss": 1.0121,
      "step": 6933
    },
    {
      "epoch": 8.487148102815178,
      "grad_norm": 1.270493485851037,
      "learning_rate": 1.9868048356068818e-06,
      "loss": 0.5519,
      "step": 6934
    },
    {
      "epoch": 8.488372093023255,
      "grad_norm": 2.1302149793885845,
      "learning_rate": 1.9861776210754988e-06,
      "loss": 0.8275,
      "step": 6935
    },
    {
      "epoch": 8.489596083231334,
      "grad_norm": 1.4841212997947781,
      "learning_rate": 1.985550440310622e-06,
      "loss": 1.1688,
      "step": 6936
    },
    {
      "epoch": 8.490820073439412,
      "grad_norm": 2.491020025593202,
      "learning_rate": 1.9849232933534672e-06,
      "loss": 0.6713,
      "step": 6937
    },
    {
      "epoch": 8.49204406364749,
      "grad_norm": 1.5403835970613065,
      "learning_rate": 1.9842961802452477e-06,
      "loss": 0.4644,
      "step": 6938
    },
    {
      "epoch": 8.49326805385557,
      "grad_norm": 1.5383925625316852,
      "learning_rate": 1.983669101027177e-06,
      "loss": 0.4793,
      "step": 6939
    },
    {
      "epoch": 8.494492044063648,
      "grad_norm": 1.4579316438762524,
      "learning_rate": 1.9830420557404623e-06,
      "loss": 2.2863,
      "step": 6940
    },
    {
      "epoch": 8.495716034271727,
      "grad_norm": 1.4960720306087048,
      "learning_rate": 1.982415044426312e-06,
      "loss": 0.483,
      "step": 6941
    },
    {
      "epoch": 8.496940024479803,
      "grad_norm": 1.4837610998494806,
      "learning_rate": 1.98178806712593e-06,
      "loss": 0.4806,
      "step": 6942
    },
    {
      "epoch": 8.498164014687882,
      "grad_norm": 1.7611504411036183,
      "learning_rate": 1.9811611238805203e-06,
      "loss": 0.743,
      "step": 6943
    },
    {
      "epoch": 8.49938800489596,
      "grad_norm": 1.524582252481321,
      "learning_rate": 1.980534214731282e-06,
      "loss": 0.4711,
      "step": 6944
    },
    {
      "epoch": 8.50061199510404,
      "grad_norm": 1.215590223477337,
      "learning_rate": 1.9799073397194142e-06,
      "loss": 0.9796,
      "step": 6945
    },
    {
      "epoch": 8.501835985312118,
      "grad_norm": 1.4353806036948056,
      "learning_rate": 1.979280498886112e-06,
      "loss": 0.4796,
      "step": 6946
    },
    {
      "epoch": 8.503059975520197,
      "grad_norm": 1.6490080261176043,
      "learning_rate": 1.9786536922725692e-06,
      "loss": 0.6383,
      "step": 6947
    },
    {
      "epoch": 8.504283965728273,
      "grad_norm": 1.6226611269231428,
      "learning_rate": 1.9780269199199776e-06,
      "loss": 0.531,
      "step": 6948
    },
    {
      "epoch": 8.505507955936352,
      "grad_norm": 1.5774658530954,
      "learning_rate": 1.9774001818695264e-06,
      "loss": 1.3392,
      "step": 6949
    },
    {
      "epoch": 8.50673194614443,
      "grad_norm": 2.510875266684185,
      "learning_rate": 1.976773478162402e-06,
      "loss": 0.6027,
      "step": 6950
    },
    {
      "epoch": 8.50795593635251,
      "grad_norm": 1.7567774012251876,
      "learning_rate": 1.976146808839789e-06,
      "loss": 0.4946,
      "step": 6951
    },
    {
      "epoch": 8.509179926560588,
      "grad_norm": 2.7038831977134774,
      "learning_rate": 1.9755201739428707e-06,
      "loss": 0.4625,
      "step": 6952
    },
    {
      "epoch": 8.510403916768666,
      "grad_norm": 1.6297699448061456,
      "learning_rate": 1.9748935735128266e-06,
      "loss": 1.4047,
      "step": 6953
    },
    {
      "epoch": 8.511627906976745,
      "grad_norm": 2.9871413412687002,
      "learning_rate": 1.9742670075908353e-06,
      "loss": 0.6046,
      "step": 6954
    },
    {
      "epoch": 8.512851897184822,
      "grad_norm": 1.309861934293757,
      "learning_rate": 1.97364047621807e-06,
      "loss": 0.6985,
      "step": 6955
    },
    {
      "epoch": 8.5140758873929,
      "grad_norm": 1.2475183849935356,
      "learning_rate": 1.973013979435707e-06,
      "loss": 0.7786,
      "step": 6956
    },
    {
      "epoch": 8.51529987760098,
      "grad_norm": 1.683720266602124,
      "learning_rate": 1.972387517284916e-06,
      "loss": 1.4118,
      "step": 6957
    },
    {
      "epoch": 8.516523867809058,
      "grad_norm": 1.4700894603449188,
      "learning_rate": 1.9717610898068655e-06,
      "loss": 0.5771,
      "step": 6958
    },
    {
      "epoch": 8.517747858017136,
      "grad_norm": 1.5546460629328276,
      "learning_rate": 1.9711346970427227e-06,
      "loss": 1.2366,
      "step": 6959
    },
    {
      "epoch": 8.518971848225215,
      "grad_norm": 1.0445972967224164,
      "learning_rate": 1.970508339033651e-06,
      "loss": 0.6086,
      "step": 6960
    },
    {
      "epoch": 8.520195838433292,
      "grad_norm": 2.5897319070992877,
      "learning_rate": 1.969882015820813e-06,
      "loss": 0.3774,
      "step": 6961
    },
    {
      "epoch": 8.52141982864137,
      "grad_norm": 1.0624617437695234,
      "learning_rate": 1.969255727445368e-06,
      "loss": 0.5993,
      "step": 6962
    },
    {
      "epoch": 8.522643818849449,
      "grad_norm": 2.1278017626407775,
      "learning_rate": 1.968629473948474e-06,
      "loss": 0.3612,
      "step": 6963
    },
    {
      "epoch": 8.523867809057528,
      "grad_norm": 1.0893886143397888,
      "learning_rate": 1.9680032553712847e-06,
      "loss": 0.4488,
      "step": 6964
    },
    {
      "epoch": 8.525091799265606,
      "grad_norm": 1.2448772669332806,
      "learning_rate": 1.967377071754955e-06,
      "loss": 0.7217,
      "step": 6965
    },
    {
      "epoch": 8.526315789473685,
      "grad_norm": 1.8734994611788545,
      "learning_rate": 1.9667509231406332e-06,
      "loss": 1.0447,
      "step": 6966
    },
    {
      "epoch": 8.527539779681762,
      "grad_norm": 1.8564841346378953,
      "learning_rate": 1.966124809569469e-06,
      "loss": 0.5485,
      "step": 6967
    },
    {
      "epoch": 8.52876376988984,
      "grad_norm": 1.864999442228352,
      "learning_rate": 1.9654987310826067e-06,
      "loss": 0.4253,
      "step": 6968
    },
    {
      "epoch": 8.529987760097919,
      "grad_norm": 2.2516677305735264,
      "learning_rate": 1.964872687721191e-06,
      "loss": 0.5116,
      "step": 6969
    },
    {
      "epoch": 8.531211750305998,
      "grad_norm": 1.4414879105935106,
      "learning_rate": 1.964246679526364e-06,
      "loss": 0.412,
      "step": 6970
    },
    {
      "epoch": 8.532435740514076,
      "grad_norm": 1.5236935745854983,
      "learning_rate": 1.963620706539263e-06,
      "loss": 0.5546,
      "step": 6971
    },
    {
      "epoch": 8.533659730722155,
      "grad_norm": 2.0264787254515237,
      "learning_rate": 1.9629947688010253e-06,
      "loss": 0.5853,
      "step": 6972
    },
    {
      "epoch": 8.534883720930232,
      "grad_norm": 1.2899406145375123,
      "learning_rate": 1.962368866352784e-06,
      "loss": 1.5729,
      "step": 6973
    },
    {
      "epoch": 8.53610771113831,
      "grad_norm": 0.9917855599707611,
      "learning_rate": 1.9617429992356742e-06,
      "loss": 0.6343,
      "step": 6974
    },
    {
      "epoch": 8.537331701346389,
      "grad_norm": 2.13549473108488,
      "learning_rate": 1.9611171674908224e-06,
      "loss": 0.3213,
      "step": 6975
    },
    {
      "epoch": 8.538555691554468,
      "grad_norm": 1.5879604412863948,
      "learning_rate": 1.9604913711593577e-06,
      "loss": 0.9172,
      "step": 6976
    },
    {
      "epoch": 8.539779681762546,
      "grad_norm": 1.190114060866767,
      "learning_rate": 1.959865610282404e-06,
      "loss": 0.59,
      "step": 6977
    },
    {
      "epoch": 8.541003671970625,
      "grad_norm": 1.5341181966222488,
      "learning_rate": 1.9592398849010847e-06,
      "loss": 0.5287,
      "step": 6978
    },
    {
      "epoch": 8.542227662178703,
      "grad_norm": 1.7855740180502353,
      "learning_rate": 1.9586141950565205e-06,
      "loss": 1.0007,
      "step": 6979
    },
    {
      "epoch": 8.54345165238678,
      "grad_norm": 1.2107515043912824,
      "learning_rate": 1.957988540789829e-06,
      "loss": 0.5877,
      "step": 6980
    },
    {
      "epoch": 8.544675642594859,
      "grad_norm": 1.662352643095449,
      "learning_rate": 1.9573629221421246e-06,
      "loss": 0.5871,
      "step": 6981
    },
    {
      "epoch": 8.545899632802938,
      "grad_norm": 1.5743118417253978,
      "learning_rate": 1.956737339154523e-06,
      "loss": 0.5251,
      "step": 6982
    },
    {
      "epoch": 8.547123623011016,
      "grad_norm": 1.3388893345113348,
      "learning_rate": 1.9561117918681345e-06,
      "loss": 0.6286,
      "step": 6983
    },
    {
      "epoch": 8.548347613219095,
      "grad_norm": 1.6956677347821356,
      "learning_rate": 1.9554862803240667e-06,
      "loss": 0.593,
      "step": 6984
    },
    {
      "epoch": 8.549571603427173,
      "grad_norm": 1.7156477545231066,
      "learning_rate": 1.9548608045634267e-06,
      "loss": 0.5515,
      "step": 6985
    },
    {
      "epoch": 8.55079559363525,
      "grad_norm": 1.3581593877980909,
      "learning_rate": 1.9542353646273175e-06,
      "loss": 0.801,
      "step": 6986
    },
    {
      "epoch": 8.552019583843329,
      "grad_norm": 1.9786886043800822,
      "learning_rate": 1.9536099605568424e-06,
      "loss": 0.6175,
      "step": 6987
    },
    {
      "epoch": 8.553243574051407,
      "grad_norm": 1.0846975166652442,
      "learning_rate": 1.9529845923930997e-06,
      "loss": 0.505,
      "step": 6988
    },
    {
      "epoch": 8.554467564259486,
      "grad_norm": 1.6004140589306965,
      "learning_rate": 1.9523592601771863e-06,
      "loss": 1.5618,
      "step": 6989
    },
    {
      "epoch": 8.555691554467565,
      "grad_norm": 0.9480513133771694,
      "learning_rate": 1.9517339639501956e-06,
      "loss": 0.5154,
      "step": 6990
    },
    {
      "epoch": 8.556915544675643,
      "grad_norm": 1.1870889910233422,
      "learning_rate": 1.951108703753222e-06,
      "loss": 0.5246,
      "step": 6991
    },
    {
      "epoch": 8.55813953488372,
      "grad_norm": 2.397477385560184,
      "learning_rate": 1.9504834796273546e-06,
      "loss": 0.351,
      "step": 6992
    },
    {
      "epoch": 8.559363525091799,
      "grad_norm": 1.625215679643291,
      "learning_rate": 1.9498582916136798e-06,
      "loss": 0.4841,
      "step": 6993
    },
    {
      "epoch": 8.560587515299877,
      "grad_norm": 1.8104213337107102,
      "learning_rate": 1.949233139753283e-06,
      "loss": 0.5035,
      "step": 6994
    },
    {
      "epoch": 8.561811505507956,
      "grad_norm": 2.9058880199525237,
      "learning_rate": 1.9486080240872475e-06,
      "loss": 0.4709,
      "step": 6995
    },
    {
      "epoch": 8.563035495716035,
      "grad_norm": 1.3794313143383858,
      "learning_rate": 1.9479829446566528e-06,
      "loss": 0.3404,
      "step": 6996
    },
    {
      "epoch": 8.564259485924113,
      "grad_norm": 1.5714645157234262,
      "learning_rate": 1.9473579015025775e-06,
      "loss": 1.0895,
      "step": 6997
    },
    {
      "epoch": 8.56548347613219,
      "grad_norm": 1.4184598548313632,
      "learning_rate": 1.9467328946660974e-06,
      "loss": 0.3293,
      "step": 6998
    },
    {
      "epoch": 8.566707466340269,
      "grad_norm": 1.1664649432051177,
      "learning_rate": 1.9461079241882834e-06,
      "loss": 0.8242,
      "step": 6999
    },
    {
      "epoch": 8.567931456548347,
      "grad_norm": 2.127052968088281,
      "learning_rate": 1.9454829901102093e-06,
      "loss": 0.3611,
      "step": 7000
    },
    {
      "epoch": 8.569155446756426,
      "grad_norm": 1.5360334414169912,
      "learning_rate": 1.944858092472942e-06,
      "loss": 0.6065,
      "step": 7001
    },
    {
      "epoch": 8.570379436964505,
      "grad_norm": 2.0429365663111962,
      "learning_rate": 1.944233231317548e-06,
      "loss": 0.5486,
      "step": 7002
    },
    {
      "epoch": 8.571603427172583,
      "grad_norm": 2.2481252036061226,
      "learning_rate": 1.943608406685089e-06,
      "loss": 0.3751,
      "step": 7003
    },
    {
      "epoch": 8.572827417380662,
      "grad_norm": 1.4224850982577077,
      "learning_rate": 1.942983618616628e-06,
      "loss": 0.5685,
      "step": 7004
    },
    {
      "epoch": 8.574051407588739,
      "grad_norm": 2.027932625121946,
      "learning_rate": 1.942358867153224e-06,
      "loss": 0.3129,
      "step": 7005
    },
    {
      "epoch": 8.575275397796817,
      "grad_norm": 2.080973410913841,
      "learning_rate": 1.941734152335932e-06,
      "loss": 0.9716,
      "step": 7006
    },
    {
      "epoch": 8.576499388004896,
      "grad_norm": 1.3208582773207311,
      "learning_rate": 1.9411094742058063e-06,
      "loss": 0.6843,
      "step": 7007
    },
    {
      "epoch": 8.577723378212974,
      "grad_norm": 1.4115617657123003,
      "learning_rate": 1.9404848328039e-06,
      "loss": 0.6174,
      "step": 7008
    },
    {
      "epoch": 8.578947368421053,
      "grad_norm": 1.699165740286584,
      "learning_rate": 1.9398602281712607e-06,
      "loss": 0.5551,
      "step": 7009
    },
    {
      "epoch": 8.580171358629132,
      "grad_norm": 1.3790738655203627,
      "learning_rate": 1.939235660348935e-06,
      "loss": 1.4145,
      "step": 7010
    },
    {
      "epoch": 8.581395348837209,
      "grad_norm": 1.2866794876177157,
      "learning_rate": 1.9386111293779673e-06,
      "loss": 0.4922,
      "step": 7011
    },
    {
      "epoch": 8.582619339045287,
      "grad_norm": 1.2993168358965523,
      "learning_rate": 1.9379866352993997e-06,
      "loss": 0.4566,
      "step": 7012
    },
    {
      "epoch": 8.583843329253366,
      "grad_norm": 1.6774155201444798,
      "learning_rate": 1.9373621781542723e-06,
      "loss": 1.5073,
      "step": 7013
    },
    {
      "epoch": 8.585067319461444,
      "grad_norm": 1.7034041641804865,
      "learning_rate": 1.9367377579836213e-06,
      "loss": 1.0151,
      "step": 7014
    },
    {
      "epoch": 8.586291309669523,
      "grad_norm": 2.29651393733562,
      "learning_rate": 1.9361133748284812e-06,
      "loss": 0.4649,
      "step": 7015
    },
    {
      "epoch": 8.587515299877602,
      "grad_norm": 1.0265112025799277,
      "learning_rate": 1.9354890287298844e-06,
      "loss": 0.5127,
      "step": 7016
    },
    {
      "epoch": 8.588739290085678,
      "grad_norm": 1.4923692336409133,
      "learning_rate": 1.9348647197288608e-06,
      "loss": 0.5312,
      "step": 7017
    },
    {
      "epoch": 8.589963280293757,
      "grad_norm": 1.9711269523781725,
      "learning_rate": 1.9342404478664384e-06,
      "loss": 0.5563,
      "step": 7018
    },
    {
      "epoch": 8.591187270501836,
      "grad_norm": 1.224187828098308,
      "learning_rate": 1.9336162131836404e-06,
      "loss": 0.9876,
      "step": 7019
    },
    {
      "epoch": 8.592411260709914,
      "grad_norm": 1.1686508853068271,
      "learning_rate": 1.932992015721489e-06,
      "loss": 0.4749,
      "step": 7020
    },
    {
      "epoch": 8.593635250917993,
      "grad_norm": 1.743588962380306,
      "learning_rate": 1.9323678555210065e-06,
      "loss": 0.6186,
      "step": 7021
    },
    {
      "epoch": 8.594859241126072,
      "grad_norm": 1.5264901715173755,
      "learning_rate": 1.9317437326232086e-06,
      "loss": 0.3936,
      "step": 7022
    },
    {
      "epoch": 8.596083231334148,
      "grad_norm": 2.886225193117928,
      "learning_rate": 1.931119647069111e-06,
      "loss": 0.5895,
      "step": 7023
    },
    {
      "epoch": 8.597307221542227,
      "grad_norm": 3.2434324171743123,
      "learning_rate": 1.930495598899726e-06,
      "loss": 0.3668,
      "step": 7024
    },
    {
      "epoch": 8.598531211750306,
      "grad_norm": 1.412267348022248,
      "learning_rate": 1.9298715881560626e-06,
      "loss": 0.7561,
      "step": 7025
    },
    {
      "epoch": 8.599755201958384,
      "grad_norm": 1.4999659306686985,
      "learning_rate": 1.929247614879131e-06,
      "loss": 0.4722,
      "step": 7026
    },
    {
      "epoch": 8.600979192166463,
      "grad_norm": 1.2722238927932918,
      "learning_rate": 1.9286236791099352e-06,
      "loss": 0.47,
      "step": 7027
    },
    {
      "epoch": 8.602203182374542,
      "grad_norm": 2.8601166296548732,
      "learning_rate": 1.9279997808894774e-06,
      "loss": 0.387,
      "step": 7028
    },
    {
      "epoch": 8.60342717258262,
      "grad_norm": 2.988162430683071,
      "learning_rate": 1.927375920258757e-06,
      "loss": 0.3655,
      "step": 7029
    },
    {
      "epoch": 8.604651162790697,
      "grad_norm": 1.9041961352603067,
      "learning_rate": 1.9267520972587754e-06,
      "loss": 0.5144,
      "step": 7030
    },
    {
      "epoch": 8.605875152998776,
      "grad_norm": 1.0004572197871546,
      "learning_rate": 1.9261283119305245e-06,
      "loss": 0.5173,
      "step": 7031
    },
    {
      "epoch": 8.607099143206854,
      "grad_norm": 2.871630921248742,
      "learning_rate": 1.925504564314999e-06,
      "loss": 0.3169,
      "step": 7032
    },
    {
      "epoch": 8.608323133414933,
      "grad_norm": 2.3952329199061473,
      "learning_rate": 1.9248808544531874e-06,
      "loss": 0.5188,
      "step": 7033
    },
    {
      "epoch": 8.609547123623011,
      "grad_norm": 1.9647601353609718,
      "learning_rate": 1.9242571823860796e-06,
      "loss": 0.44,
      "step": 7034
    },
    {
      "epoch": 8.61077111383109,
      "grad_norm": 1.9466796840945508,
      "learning_rate": 1.9236335481546604e-06,
      "loss": 1.1592,
      "step": 7035
    },
    {
      "epoch": 8.611995104039167,
      "grad_norm": 1.014029076227133,
      "learning_rate": 1.9230099517999128e-06,
      "loss": 0.5347,
      "step": 7036
    },
    {
      "epoch": 8.613219094247246,
      "grad_norm": 1.7815667834562627,
      "learning_rate": 1.922386393362817e-06,
      "loss": 0.7368,
      "step": 7037
    },
    {
      "epoch": 8.614443084455324,
      "grad_norm": 1.4100560140789216,
      "learning_rate": 1.92176287288435e-06,
      "loss": 1.2198,
      "step": 7038
    },
    {
      "epoch": 8.615667074663403,
      "grad_norm": 1.4510869135240154,
      "learning_rate": 1.9211393904054887e-06,
      "loss": 0.4113,
      "step": 7039
    },
    {
      "epoch": 8.616891064871481,
      "grad_norm": 2.504149132648787,
      "learning_rate": 1.9205159459672055e-06,
      "loss": 0.8726,
      "step": 7040
    },
    {
      "epoch": 8.61811505507956,
      "grad_norm": 1.4407555093044733,
      "learning_rate": 1.9198925396104716e-06,
      "loss": 0.6108,
      "step": 7041
    },
    {
      "epoch": 8.619339045287639,
      "grad_norm": 2.0026325128611013,
      "learning_rate": 1.9192691713762534e-06,
      "loss": 0.7784,
      "step": 7042
    },
    {
      "epoch": 8.620563035495715,
      "grad_norm": 2.2659650231701973,
      "learning_rate": 1.9186458413055183e-06,
      "loss": 0.6164,
      "step": 7043
    },
    {
      "epoch": 8.621787025703794,
      "grad_norm": 2.810175391743407,
      "learning_rate": 1.9180225494392273e-06,
      "loss": 0.9746,
      "step": 7044
    },
    {
      "epoch": 8.623011015911873,
      "grad_norm": 1.8352839941725678,
      "learning_rate": 1.917399295818343e-06,
      "loss": 0.734,
      "step": 7045
    },
    {
      "epoch": 8.624235006119951,
      "grad_norm": 1.4823016670195748,
      "learning_rate": 1.91677608048382e-06,
      "loss": 1.1678,
      "step": 7046
    },
    {
      "epoch": 8.62545899632803,
      "grad_norm": 2.584698146955467,
      "learning_rate": 1.916152903476618e-06,
      "loss": 0.9663,
      "step": 7047
    },
    {
      "epoch": 8.626682986536107,
      "grad_norm": 1.6889411017291764,
      "learning_rate": 1.915529764837687e-06,
      "loss": 0.6955,
      "step": 7048
    },
    {
      "epoch": 8.627906976744185,
      "grad_norm": 1.4381027543449347,
      "learning_rate": 1.914906664607978e-06,
      "loss": 1.5759,
      "step": 7049
    },
    {
      "epoch": 8.629130966952264,
      "grad_norm": 1.2981824706348752,
      "learning_rate": 1.9142836028284385e-06,
      "loss": 0.5107,
      "step": 7050
    },
    {
      "epoch": 8.630354957160343,
      "grad_norm": 1.8903078351810934,
      "learning_rate": 1.913660579540014e-06,
      "loss": 0.6041,
      "step": 7051
    },
    {
      "epoch": 8.631578947368421,
      "grad_norm": 2.0930948149148474,
      "learning_rate": 1.913037594783648e-06,
      "loss": 0.5586,
      "step": 7052
    },
    {
      "epoch": 8.6328029375765,
      "grad_norm": 2.324881792910832,
      "learning_rate": 1.9124146486002803e-06,
      "loss": 0.6171,
      "step": 7053
    },
    {
      "epoch": 8.634026927784578,
      "grad_norm": 1.080089636661796,
      "learning_rate": 1.911791741030849e-06,
      "loss": 0.5399,
      "step": 7054
    },
    {
      "epoch": 8.635250917992655,
      "grad_norm": 1.1850243173915376,
      "learning_rate": 1.9111688721162872e-06,
      "loss": 0.5978,
      "step": 7055
    },
    {
      "epoch": 8.636474908200734,
      "grad_norm": 1.6580047350907374,
      "learning_rate": 1.9105460418975307e-06,
      "loss": 0.9607,
      "step": 7056
    },
    {
      "epoch": 8.637698898408813,
      "grad_norm": 1.4309727783931683,
      "learning_rate": 1.909923250415508e-06,
      "loss": 0.9225,
      "step": 7057
    },
    {
      "epoch": 8.638922888616891,
      "grad_norm": 1.1760131016164397,
      "learning_rate": 1.909300497711146e-06,
      "loss": 0.5133,
      "step": 7058
    },
    {
      "epoch": 8.64014687882497,
      "grad_norm": 0.9437675387366303,
      "learning_rate": 1.9086777838253714e-06,
      "loss": 0.5897,
      "step": 7059
    },
    {
      "epoch": 8.641370869033048,
      "grad_norm": 2.1794286002795964,
      "learning_rate": 1.9080551087991046e-06,
      "loss": 0.6794,
      "step": 7060
    },
    {
      "epoch": 8.642594859241125,
      "grad_norm": 1.8299791489657058,
      "learning_rate": 1.907432472673267e-06,
      "loss": 0.3692,
      "step": 7061
    },
    {
      "epoch": 8.643818849449204,
      "grad_norm": 3.041001436777335,
      "learning_rate": 1.906809875488776e-06,
      "loss": 0.6994,
      "step": 7062
    },
    {
      "epoch": 8.645042839657282,
      "grad_norm": 1.3097175843823354,
      "learning_rate": 1.906187317286546e-06,
      "loss": 0.6825,
      "step": 7063
    },
    {
      "epoch": 8.646266829865361,
      "grad_norm": 2.3533279964344627,
      "learning_rate": 1.9055647981074879e-06,
      "loss": 0.866,
      "step": 7064
    },
    {
      "epoch": 8.64749082007344,
      "grad_norm": 1.6535118934937274,
      "learning_rate": 1.904942317992514e-06,
      "loss": 0.9946,
      "step": 7065
    },
    {
      "epoch": 8.648714810281518,
      "grad_norm": 2.208882134733513,
      "learning_rate": 1.9043198769825294e-06,
      "loss": 0.5469,
      "step": 7066
    },
    {
      "epoch": 8.649938800489597,
      "grad_norm": 2.1936552903031044,
      "learning_rate": 1.9036974751184395e-06,
      "loss": 0.4327,
      "step": 7067
    },
    {
      "epoch": 8.651162790697674,
      "grad_norm": 1.8162583812433646,
      "learning_rate": 1.903075112441145e-06,
      "loss": 0.5547,
      "step": 7068
    },
    {
      "epoch": 8.652386780905752,
      "grad_norm": 1.5089522817213852,
      "learning_rate": 1.9024527889915472e-06,
      "loss": 1.0206,
      "step": 7069
    },
    {
      "epoch": 8.653610771113831,
      "grad_norm": 0.881211574787423,
      "learning_rate": 1.9018305048105418e-06,
      "loss": 0.4433,
      "step": 7070
    },
    {
      "epoch": 8.65483476132191,
      "grad_norm": 1.9622241437706243,
      "learning_rate": 1.901208259939023e-06,
      "loss": 0.6464,
      "step": 7071
    },
    {
      "epoch": 8.656058751529988,
      "grad_norm": 1.546410703153546,
      "learning_rate": 1.9005860544178834e-06,
      "loss": 0.9335,
      "step": 7072
    },
    {
      "epoch": 8.657282741738067,
      "grad_norm": 1.6934828056242963,
      "learning_rate": 1.8999638882880095e-06,
      "loss": 0.4794,
      "step": 7073
    },
    {
      "epoch": 8.658506731946144,
      "grad_norm": 1.448509313676436,
      "learning_rate": 1.8993417615902909e-06,
      "loss": 0.5971,
      "step": 7074
    },
    {
      "epoch": 8.659730722154222,
      "grad_norm": 1.3590215867979911,
      "learning_rate": 1.8987196743656096e-06,
      "loss": 0.7269,
      "step": 7075
    },
    {
      "epoch": 8.660954712362301,
      "grad_norm": 1.2993708905062111,
      "learning_rate": 1.8980976266548473e-06,
      "loss": 0.55,
      "step": 7076
    },
    {
      "epoch": 8.66217870257038,
      "grad_norm": 2.5270778953146955,
      "learning_rate": 1.8974756184988824e-06,
      "loss": 0.5117,
      "step": 7077
    },
    {
      "epoch": 8.663402692778458,
      "grad_norm": 1.832246629659369,
      "learning_rate": 1.8968536499385914e-06,
      "loss": 0.5853,
      "step": 7078
    },
    {
      "epoch": 8.664626682986537,
      "grad_norm": 2.0322226639541983,
      "learning_rate": 1.8962317210148482e-06,
      "loss": 0.5814,
      "step": 7079
    },
    {
      "epoch": 8.665850673194614,
      "grad_norm": 1.5499105365047605,
      "learning_rate": 1.8956098317685226e-06,
      "loss": 1.1808,
      "step": 7080
    },
    {
      "epoch": 8.667074663402692,
      "grad_norm": 2.5478562895980934,
      "learning_rate": 1.8949879822404831e-06,
      "loss": 0.4527,
      "step": 7081
    },
    {
      "epoch": 8.668298653610771,
      "grad_norm": 2.14225805637489,
      "learning_rate": 1.894366172471596e-06,
      "loss": 0.5423,
      "step": 7082
    },
    {
      "epoch": 8.66952264381885,
      "grad_norm": 1.6925124744282007,
      "learning_rate": 1.8937444025027245e-06,
      "loss": 0.7158,
      "step": 7083
    },
    {
      "epoch": 8.670746634026928,
      "grad_norm": 2.3335373341558423,
      "learning_rate": 1.8931226723747282e-06,
      "loss": 0.985,
      "step": 7084
    },
    {
      "epoch": 8.671970624235007,
      "grad_norm": 2.001227721487262,
      "learning_rate": 1.892500982128465e-06,
      "loss": 0.524,
      "step": 7085
    },
    {
      "epoch": 8.673194614443084,
      "grad_norm": 2.414808583681293,
      "learning_rate": 1.8918793318047896e-06,
      "loss": 0.3725,
      "step": 7086
    },
    {
      "epoch": 8.674418604651162,
      "grad_norm": 1.0461912660472263,
      "learning_rate": 1.891257721444556e-06,
      "loss": 0.5416,
      "step": 7087
    },
    {
      "epoch": 8.67564259485924,
      "grad_norm": 1.956498776744119,
      "learning_rate": 1.8906361510886135e-06,
      "loss": 0.3937,
      "step": 7088
    },
    {
      "epoch": 8.67686658506732,
      "grad_norm": 2.2709347688317894,
      "learning_rate": 1.8900146207778094e-06,
      "loss": 0.4722,
      "step": 7089
    },
    {
      "epoch": 8.678090575275398,
      "grad_norm": 1.3307463236237826,
      "learning_rate": 1.8893931305529867e-06,
      "loss": 0.7428,
      "step": 7090
    },
    {
      "epoch": 8.679314565483477,
      "grad_norm": 2.222697852002743,
      "learning_rate": 1.8887716804549903e-06,
      "loss": 0.484,
      "step": 7091
    },
    {
      "epoch": 8.680538555691555,
      "grad_norm": 1.9121330632124758,
      "learning_rate": 1.8881502705246586e-06,
      "loss": 0.663,
      "step": 7092
    },
    {
      "epoch": 8.681762545899632,
      "grad_norm": 1.5120190095954609,
      "learning_rate": 1.8875289008028275e-06,
      "loss": 0.5053,
      "step": 7093
    },
    {
      "epoch": 8.68298653610771,
      "grad_norm": 1.3906416561849189,
      "learning_rate": 1.8869075713303313e-06,
      "loss": 0.5087,
      "step": 7094
    },
    {
      "epoch": 8.68421052631579,
      "grad_norm": 1.731489166372191,
      "learning_rate": 1.8862862821480023e-06,
      "loss": 1.3131,
      "step": 7095
    },
    {
      "epoch": 8.685434516523868,
      "grad_norm": 1.906081391716827,
      "learning_rate": 1.885665033296669e-06,
      "loss": 0.5001,
      "step": 7096
    },
    {
      "epoch": 8.686658506731947,
      "grad_norm": 2.278492203909856,
      "learning_rate": 1.8850438248171571e-06,
      "loss": 0.7467,
      "step": 7097
    },
    {
      "epoch": 8.687882496940025,
      "grad_norm": 1.8290546291130974,
      "learning_rate": 1.8844226567502913e-06,
      "loss": 0.419,
      "step": 7098
    },
    {
      "epoch": 8.689106487148102,
      "grad_norm": 2.800739244346286,
      "learning_rate": 1.88380152913689e-06,
      "loss": 0.5425,
      "step": 7099
    },
    {
      "epoch": 8.69033047735618,
      "grad_norm": 2.0352202939219413,
      "learning_rate": 1.8831804420177742e-06,
      "loss": 0.4776,
      "step": 7100
    },
    {
      "epoch": 8.69155446756426,
      "grad_norm": 1.468113045108928,
      "learning_rate": 1.8825593954337583e-06,
      "loss": 0.6881,
      "step": 7101
    },
    {
      "epoch": 8.692778457772338,
      "grad_norm": 1.55666520701493,
      "learning_rate": 1.8819383894256554e-06,
      "loss": 0.6764,
      "step": 7102
    },
    {
      "epoch": 8.694002447980417,
      "grad_norm": 1.1295245974430754,
      "learning_rate": 1.8813174240342745e-06,
      "loss": 0.4784,
      "step": 7103
    },
    {
      "epoch": 8.695226438188495,
      "grad_norm": 2.1729655911827552,
      "learning_rate": 1.8806964993004248e-06,
      "loss": 0.5149,
      "step": 7104
    },
    {
      "epoch": 8.696450428396572,
      "grad_norm": 1.9107057761306587,
      "learning_rate": 1.8800756152649106e-06,
      "loss": 0.7419,
      "step": 7105
    },
    {
      "epoch": 8.69767441860465,
      "grad_norm": 1.678315526575965,
      "learning_rate": 1.8794547719685338e-06,
      "loss": 0.6393,
      "step": 7106
    },
    {
      "epoch": 8.69889840881273,
      "grad_norm": 1.352476619921566,
      "learning_rate": 1.8788339694520939e-06,
      "loss": 0.3411,
      "step": 7107
    },
    {
      "epoch": 8.700122399020808,
      "grad_norm": 0.8438920635287147,
      "learning_rate": 1.878213207756388e-06,
      "loss": 0.3543,
      "step": 7108
    },
    {
      "epoch": 8.701346389228886,
      "grad_norm": 2.2279804958428513,
      "learning_rate": 1.8775924869222106e-06,
      "loss": 0.4241,
      "step": 7109
    },
    {
      "epoch": 8.702570379436965,
      "grad_norm": 1.3833523162179864,
      "learning_rate": 1.8769718069903537e-06,
      "loss": 0.625,
      "step": 7110
    },
    {
      "epoch": 8.703794369645042,
      "grad_norm": 1.0861112155929005,
      "learning_rate": 1.8763511680016041e-06,
      "loss": 0.3129,
      "step": 7111
    },
    {
      "epoch": 8.70501835985312,
      "grad_norm": 1.8309394724905486,
      "learning_rate": 1.8757305699967487e-06,
      "loss": 1.0905,
      "step": 7112
    },
    {
      "epoch": 8.7062423500612,
      "grad_norm": 1.4891740286093416,
      "learning_rate": 1.875110013016572e-06,
      "loss": 0.5785,
      "step": 7113
    },
    {
      "epoch": 8.707466340269278,
      "grad_norm": 1.6970417809380671,
      "learning_rate": 1.8744894971018537e-06,
      "loss": 0.543,
      "step": 7114
    },
    {
      "epoch": 8.708690330477356,
      "grad_norm": 1.5826229899826285,
      "learning_rate": 1.873869022293372e-06,
      "loss": 0.6188,
      "step": 7115
    },
    {
      "epoch": 8.709914320685435,
      "grad_norm": 2.2050816206143606,
      "learning_rate": 1.8732485886319018e-06,
      "loss": 1.0478,
      "step": 7116
    },
    {
      "epoch": 8.711138310893514,
      "grad_norm": 1.4820074102937373,
      "learning_rate": 1.872628196158217e-06,
      "loss": 0.4746,
      "step": 7117
    },
    {
      "epoch": 8.71236230110159,
      "grad_norm": 1.5306508542614394,
      "learning_rate": 1.8720078449130868e-06,
      "loss": 0.4951,
      "step": 7118
    },
    {
      "epoch": 8.713586291309669,
      "grad_norm": 1.5863543778812912,
      "learning_rate": 1.8713875349372778e-06,
      "loss": 1.2335,
      "step": 7119
    },
    {
      "epoch": 8.714810281517748,
      "grad_norm": 1.299774848804881,
      "learning_rate": 1.8707672662715539e-06,
      "loss": 1.3366,
      "step": 7120
    },
    {
      "epoch": 8.716034271725826,
      "grad_norm": 1.7041710441511086,
      "learning_rate": 1.8701470389566795e-06,
      "loss": 0.4547,
      "step": 7121
    },
    {
      "epoch": 8.717258261933905,
      "grad_norm": 1.3966984748470117,
      "learning_rate": 1.8695268530334116e-06,
      "loss": 1.3126,
      "step": 7122
    },
    {
      "epoch": 8.718482252141984,
      "grad_norm": 2.486408542900024,
      "learning_rate": 1.8689067085425066e-06,
      "loss": 0.9259,
      "step": 7123
    },
    {
      "epoch": 8.71970624235006,
      "grad_norm": 1.2418564900002367,
      "learning_rate": 1.8682866055247189e-06,
      "loss": 0.5398,
      "step": 7124
    },
    {
      "epoch": 8.720930232558139,
      "grad_norm": 1.4797500347919583,
      "learning_rate": 1.8676665440207982e-06,
      "loss": 0.4738,
      "step": 7125
    },
    {
      "epoch": 8.722154222766218,
      "grad_norm": 0.7159341299437582,
      "learning_rate": 1.8670465240714939e-06,
      "loss": 0.2585,
      "step": 7126
    },
    {
      "epoch": 8.723378212974296,
      "grad_norm": 1.7216978262379328,
      "learning_rate": 1.8664265457175516e-06,
      "loss": 0.4474,
      "step": 7127
    },
    {
      "epoch": 8.724602203182375,
      "grad_norm": 1.3292998056026162,
      "learning_rate": 1.8658066089997124e-06,
      "loss": 0.4397,
      "step": 7128
    },
    {
      "epoch": 8.725826193390454,
      "grad_norm": 1.3515449009124987,
      "learning_rate": 1.8651867139587165e-06,
      "loss": 1.2867,
      "step": 7129
    },
    {
      "epoch": 8.727050183598532,
      "grad_norm": 2.7099167345348047,
      "learning_rate": 1.8645668606353028e-06,
      "loss": 0.4205,
      "step": 7130
    },
    {
      "epoch": 8.728274173806609,
      "grad_norm": 1.833455426510204,
      "learning_rate": 1.863947049070205e-06,
      "loss": 0.7673,
      "step": 7131
    },
    {
      "epoch": 8.729498164014688,
      "grad_norm": 1.9861925686613313,
      "learning_rate": 1.863327279304154e-06,
      "loss": 0.4487,
      "step": 7132
    },
    {
      "epoch": 8.730722154222766,
      "grad_norm": 1.257446200918336,
      "learning_rate": 1.8627075513778787e-06,
      "loss": 0.5154,
      "step": 7133
    },
    {
      "epoch": 8.731946144430845,
      "grad_norm": 2.0880211507297823,
      "learning_rate": 1.8620878653321066e-06,
      "loss": 0.498,
      "step": 7134
    },
    {
      "epoch": 8.733170134638923,
      "grad_norm": 1.1274470274667026,
      "learning_rate": 1.8614682212075605e-06,
      "loss": 0.5705,
      "step": 7135
    },
    {
      "epoch": 8.734394124847,
      "grad_norm": 1.0673109056126697,
      "learning_rate": 1.8608486190449617e-06,
      "loss": 0.6447,
      "step": 7136
    },
    {
      "epoch": 8.735618115055079,
      "grad_norm": 1.1760563568871727,
      "learning_rate": 1.860229058885027e-06,
      "loss": 0.6518,
      "step": 7137
    },
    {
      "epoch": 8.736842105263158,
      "grad_norm": 1.277679232812149,
      "learning_rate": 1.859609540768471e-06,
      "loss": 0.3144,
      "step": 7138
    },
    {
      "epoch": 8.738066095471236,
      "grad_norm": 1.360206804029494,
      "learning_rate": 1.8589900647360092e-06,
      "loss": 0.567,
      "step": 7139
    },
    {
      "epoch": 8.739290085679315,
      "grad_norm": 1.477765426240493,
      "learning_rate": 1.858370630828349e-06,
      "loss": 0.5191,
      "step": 7140
    },
    {
      "epoch": 8.740514075887393,
      "grad_norm": 1.9484798055456012,
      "learning_rate": 1.8577512390861977e-06,
      "loss": 0.4361,
      "step": 7141
    },
    {
      "epoch": 8.741738066095472,
      "grad_norm": 1.1573489726147173,
      "learning_rate": 1.8571318895502588e-06,
      "loss": 0.6884,
      "step": 7142
    },
    {
      "epoch": 8.742962056303549,
      "grad_norm": 1.9619957711431417,
      "learning_rate": 1.8565125822612345e-06,
      "loss": 0.8436,
      "step": 7143
    },
    {
      "epoch": 8.744186046511627,
      "grad_norm": 1.6908457730806874,
      "learning_rate": 1.8558933172598236e-06,
      "loss": 1.1104,
      "step": 7144
    },
    {
      "epoch": 8.745410036719706,
      "grad_norm": 1.175882831623166,
      "learning_rate": 1.855274094586722e-06,
      "loss": 0.5836,
      "step": 7145
    },
    {
      "epoch": 8.746634026927785,
      "grad_norm": 2.211198130511889,
      "learning_rate": 1.8546549142826205e-06,
      "loss": 1.0887,
      "step": 7146
    },
    {
      "epoch": 8.747858017135863,
      "grad_norm": 1.2544115939576916,
      "learning_rate": 1.8540357763882127e-06,
      "loss": 0.4482,
      "step": 7147
    },
    {
      "epoch": 8.749082007343942,
      "grad_norm": 1.4573301810335062,
      "learning_rate": 1.8534166809441843e-06,
      "loss": 0.5411,
      "step": 7148
    },
    {
      "epoch": 8.750305997552019,
      "grad_norm": 1.0865576561634696,
      "learning_rate": 1.8527976279912196e-06,
      "loss": 0.6975,
      "step": 7149
    },
    {
      "epoch": 8.751529987760097,
      "grad_norm": 1.0101926043841374,
      "learning_rate": 1.852178617570001e-06,
      "loss": 0.714,
      "step": 7150
    },
    {
      "epoch": 8.752753977968176,
      "grad_norm": 1.6715428794731262,
      "learning_rate": 1.851559649721207e-06,
      "loss": 0.5328,
      "step": 7151
    },
    {
      "epoch": 8.753977968176255,
      "grad_norm": 2.702937859280916,
      "learning_rate": 1.850940724485515e-06,
      "loss": 0.4088,
      "step": 7152
    },
    {
      "epoch": 8.755201958384333,
      "grad_norm": 2.0589486599672844,
      "learning_rate": 1.8503218419035978e-06,
      "loss": 0.4059,
      "step": 7153
    },
    {
      "epoch": 8.756425948592412,
      "grad_norm": 1.267654985290651,
      "learning_rate": 1.8497030020161268e-06,
      "loss": 1.0951,
      "step": 7154
    },
    {
      "epoch": 8.75764993880049,
      "grad_norm": 1.5032924902543292,
      "learning_rate": 1.8490842048637679e-06,
      "loss": 1.1757,
      "step": 7155
    },
    {
      "epoch": 8.758873929008567,
      "grad_norm": 1.806571878540636,
      "learning_rate": 1.8484654504871886e-06,
      "loss": 0.493,
      "step": 7156
    },
    {
      "epoch": 8.760097919216646,
      "grad_norm": 1.1445551377911019,
      "learning_rate": 1.8478467389270497e-06,
      "loss": 0.482,
      "step": 7157
    },
    {
      "epoch": 8.761321909424725,
      "grad_norm": 1.539866854129944,
      "learning_rate": 1.8472280702240108e-06,
      "loss": 0.6013,
      "step": 7158
    },
    {
      "epoch": 8.762545899632803,
      "grad_norm": 1.8846332217098338,
      "learning_rate": 1.8466094444187286e-06,
      "loss": 1.1837,
      "step": 7159
    },
    {
      "epoch": 8.763769889840882,
      "grad_norm": 1.8891854617880526,
      "learning_rate": 1.8459908615518572e-06,
      "loss": 0.7419,
      "step": 7160
    },
    {
      "epoch": 8.76499388004896,
      "grad_norm": 1.2568291827248856,
      "learning_rate": 1.8453723216640474e-06,
      "loss": 0.5579,
      "step": 7161
    },
    {
      "epoch": 8.766217870257037,
      "grad_norm": 2.4647931947772483,
      "learning_rate": 1.8447538247959474e-06,
      "loss": 0.9603,
      "step": 7162
    },
    {
      "epoch": 8.767441860465116,
      "grad_norm": 2.9163089604766266,
      "learning_rate": 1.8441353709882027e-06,
      "loss": 0.4196,
      "step": 7163
    },
    {
      "epoch": 8.768665850673194,
      "grad_norm": 1.8620220947996393,
      "learning_rate": 1.843516960281454e-06,
      "loss": 0.6367,
      "step": 7164
    },
    {
      "epoch": 8.769889840881273,
      "grad_norm": 1.1666398470039772,
      "learning_rate": 1.8428985927163446e-06,
      "loss": 0.4558,
      "step": 7165
    },
    {
      "epoch": 8.771113831089352,
      "grad_norm": 0.9346016007977423,
      "learning_rate": 1.8422802683335081e-06,
      "loss": 0.4454,
      "step": 7166
    },
    {
      "epoch": 8.77233782129743,
      "grad_norm": 2.473530626605038,
      "learning_rate": 1.8416619871735799e-06,
      "loss": 0.3684,
      "step": 7167
    },
    {
      "epoch": 8.773561811505507,
      "grad_norm": 2.2015132851805634,
      "learning_rate": 1.8410437492771905e-06,
      "loss": 0.4546,
      "step": 7168
    },
    {
      "epoch": 8.774785801713586,
      "grad_norm": 1.7290686434517775,
      "learning_rate": 1.8404255546849686e-06,
      "loss": 0.4614,
      "step": 7169
    },
    {
      "epoch": 8.776009791921664,
      "grad_norm": 2.3397374865324854,
      "learning_rate": 1.8398074034375402e-06,
      "loss": 1.291,
      "step": 7170
    },
    {
      "epoch": 8.777233782129743,
      "grad_norm": 1.3732711392126897,
      "learning_rate": 1.839189295575527e-06,
      "loss": 0.5644,
      "step": 7171
    },
    {
      "epoch": 8.778457772337822,
      "grad_norm": 1.241135423542788,
      "learning_rate": 1.8385712311395487e-06,
      "loss": 0.6451,
      "step": 7172
    },
    {
      "epoch": 8.7796817625459,
      "grad_norm": 1.5128611697132082,
      "learning_rate": 1.8379532101702235e-06,
      "loss": 1.1762,
      "step": 7173
    },
    {
      "epoch": 8.780905752753977,
      "grad_norm": 1.04134001338678,
      "learning_rate": 1.8373352327081647e-06,
      "loss": 0.5828,
      "step": 7174
    },
    {
      "epoch": 8.782129742962056,
      "grad_norm": 1.7826909947908924,
      "learning_rate": 1.836717298793983e-06,
      "loss": 0.4679,
      "step": 7175
    },
    {
      "epoch": 8.783353733170134,
      "grad_norm": 2.239704884397322,
      "learning_rate": 1.836099408468287e-06,
      "loss": 0.9065,
      "step": 7176
    },
    {
      "epoch": 8.784577723378213,
      "grad_norm": 1.5019365183523927,
      "learning_rate": 1.835481561771682e-06,
      "loss": 0.302,
      "step": 7177
    },
    {
      "epoch": 8.785801713586292,
      "grad_norm": 2.4964015522192544,
      "learning_rate": 1.8348637587447713e-06,
      "loss": 0.8045,
      "step": 7178
    },
    {
      "epoch": 8.78702570379437,
      "grad_norm": 1.4284307991833198,
      "learning_rate": 1.8342459994281542e-06,
      "loss": 1.2647,
      "step": 7179
    },
    {
      "epoch": 8.788249694002449,
      "grad_norm": 1.0687264779053247,
      "learning_rate": 1.8336282838624275e-06,
      "loss": 0.5414,
      "step": 7180
    },
    {
      "epoch": 8.789473684210526,
      "grad_norm": 1.6141424289599375,
      "learning_rate": 1.8330106120881846e-06,
      "loss": 0.5378,
      "step": 7181
    },
    {
      "epoch": 8.790697674418604,
      "grad_norm": 1.7506813768860336,
      "learning_rate": 1.832392984146018e-06,
      "loss": 0.3958,
      "step": 7182
    },
    {
      "epoch": 8.791921664626683,
      "grad_norm": 2.5273025164517517,
      "learning_rate": 1.8317754000765156e-06,
      "loss": 0.5252,
      "step": 7183
    },
    {
      "epoch": 8.793145654834762,
      "grad_norm": 1.710663631661679,
      "learning_rate": 1.8311578599202618e-06,
      "loss": 0.9181,
      "step": 7184
    },
    {
      "epoch": 8.79436964504284,
      "grad_norm": 2.1164883209008676,
      "learning_rate": 1.830540363717839e-06,
      "loss": 0.3851,
      "step": 7185
    },
    {
      "epoch": 8.795593635250919,
      "grad_norm": 1.4199559109988358,
      "learning_rate": 1.8299229115098277e-06,
      "loss": 0.9917,
      "step": 7186
    },
    {
      "epoch": 8.796817625458996,
      "grad_norm": 2.595606461658539,
      "learning_rate": 1.8293055033368046e-06,
      "loss": 0.4776,
      "step": 7187
    },
    {
      "epoch": 8.798041615667074,
      "grad_norm": 1.9423885480218512,
      "learning_rate": 1.828688139239343e-06,
      "loss": 1.0797,
      "step": 7188
    },
    {
      "epoch": 8.799265605875153,
      "grad_norm": 1.6567585372320848,
      "learning_rate": 1.8280708192580137e-06,
      "loss": 1.1398,
      "step": 7189
    },
    {
      "epoch": 8.800489596083231,
      "grad_norm": 3.1375980724372035,
      "learning_rate": 1.8274535434333846e-06,
      "loss": 0.3695,
      "step": 7190
    },
    {
      "epoch": 8.80171358629131,
      "grad_norm": 1.5862898723457155,
      "learning_rate": 1.8268363118060212e-06,
      "loss": 0.5799,
      "step": 7191
    },
    {
      "epoch": 8.802937576499389,
      "grad_norm": 1.3676519157134974,
      "learning_rate": 1.8262191244164862e-06,
      "loss": 0.6524,
      "step": 7192
    },
    {
      "epoch": 8.804161566707466,
      "grad_norm": 1.7147230856898554,
      "learning_rate": 1.8256019813053376e-06,
      "loss": 0.3599,
      "step": 7193
    },
    {
      "epoch": 8.805385556915544,
      "grad_norm": 2.079308442280361,
      "learning_rate": 1.8249848825131322e-06,
      "loss": 0.8289,
      "step": 7194
    },
    {
      "epoch": 8.806609547123623,
      "grad_norm": 2.725685570365817,
      "learning_rate": 1.824367828080424e-06,
      "loss": 0.8087,
      "step": 7195
    },
    {
      "epoch": 8.807833537331701,
      "grad_norm": 1.439739138924288,
      "learning_rate": 1.823750818047763e-06,
      "loss": 0.5342,
      "step": 7196
    },
    {
      "epoch": 8.80905752753978,
      "grad_norm": 1.5045913425571409,
      "learning_rate": 1.8231338524556969e-06,
      "loss": 1.3013,
      "step": 7197
    },
    {
      "epoch": 8.810281517747859,
      "grad_norm": 1.7923010071106449,
      "learning_rate": 1.8225169313447702e-06,
      "loss": 0.9697,
      "step": 7198
    },
    {
      "epoch": 8.811505507955935,
      "grad_norm": 1.8921176992985596,
      "learning_rate": 1.8219000547555255e-06,
      "loss": 0.5419,
      "step": 7199
    },
    {
      "epoch": 8.812729498164014,
      "grad_norm": 0.9442715469815652,
      "learning_rate": 1.8212832227285008e-06,
      "loss": 0.4446,
      "step": 7200
    },
    {
      "epoch": 8.813953488372093,
      "grad_norm": 1.6625531441359105,
      "learning_rate": 1.820666435304233e-06,
      "loss": 0.827,
      "step": 7201
    },
    {
      "epoch": 8.815177478580171,
      "grad_norm": 2.487102708946883,
      "learning_rate": 1.8200496925232535e-06,
      "loss": 0.9426,
      "step": 7202
    },
    {
      "epoch": 8.81640146878825,
      "grad_norm": 2.3946666545136415,
      "learning_rate": 1.819432994426093e-06,
      "loss": 0.4415,
      "step": 7203
    },
    {
      "epoch": 8.817625458996329,
      "grad_norm": 2.248999925297696,
      "learning_rate": 1.8188163410532794e-06,
      "loss": 0.5084,
      "step": 7204
    },
    {
      "epoch": 8.818849449204407,
      "grad_norm": 2.3902382707128838,
      "learning_rate": 1.8181997324453366e-06,
      "loss": 0.5885,
      "step": 7205
    },
    {
      "epoch": 8.820073439412484,
      "grad_norm": 1.8986904688918749,
      "learning_rate": 1.8175831686427853e-06,
      "loss": 0.7836,
      "step": 7206
    },
    {
      "epoch": 8.821297429620563,
      "grad_norm": 1.367353088257369,
      "learning_rate": 1.8169666496861435e-06,
      "loss": 0.6718,
      "step": 7207
    },
    {
      "epoch": 8.822521419828641,
      "grad_norm": 1.3886926710764194,
      "learning_rate": 1.8163501756159277e-06,
      "loss": 0.3617,
      "step": 7208
    },
    {
      "epoch": 8.82374541003672,
      "grad_norm": 1.56746505251863,
      "learning_rate": 1.81573374647265e-06,
      "loss": 0.5697,
      "step": 7209
    },
    {
      "epoch": 8.824969400244798,
      "grad_norm": 1.9368020544330675,
      "learning_rate": 1.8151173622968197e-06,
      "loss": 0.7308,
      "step": 7210
    },
    {
      "epoch": 8.826193390452877,
      "grad_norm": 1.9779152148425674,
      "learning_rate": 1.8145010231289417e-06,
      "loss": 0.8629,
      "step": 7211
    },
    {
      "epoch": 8.827417380660954,
      "grad_norm": 1.5850470641069907,
      "learning_rate": 1.8138847290095224e-06,
      "loss": 1.6291,
      "step": 7212
    },
    {
      "epoch": 8.828641370869033,
      "grad_norm": 1.5435676074218891,
      "learning_rate": 1.8132684799790606e-06,
      "loss": 0.4845,
      "step": 7213
    },
    {
      "epoch": 8.829865361077111,
      "grad_norm": 1.3845934019917574,
      "learning_rate": 1.812652276078054e-06,
      "loss": 0.4824,
      "step": 7214
    },
    {
      "epoch": 8.83108935128519,
      "grad_norm": 1.3899831059262633,
      "learning_rate": 1.812036117346998e-06,
      "loss": 0.2768,
      "step": 7215
    },
    {
      "epoch": 8.832313341493268,
      "grad_norm": 1.0226939603085714,
      "learning_rate": 1.8114200038263829e-06,
      "loss": 0.5515,
      "step": 7216
    },
    {
      "epoch": 8.833537331701347,
      "grad_norm": 1.2825198087961036,
      "learning_rate": 1.8108039355566987e-06,
      "loss": 0.4143,
      "step": 7217
    },
    {
      "epoch": 8.834761321909426,
      "grad_norm": 1.2665045733415046,
      "learning_rate": 1.8101879125784308e-06,
      "loss": 0.5123,
      "step": 7218
    },
    {
      "epoch": 8.835985312117502,
      "grad_norm": 1.4735652443833824,
      "learning_rate": 1.8095719349320622e-06,
      "loss": 1.1241,
      "step": 7219
    },
    {
      "epoch": 8.837209302325581,
      "grad_norm": 1.6412927499356138,
      "learning_rate": 1.808956002658071e-06,
      "loss": 0.9382,
      "step": 7220
    },
    {
      "epoch": 8.83843329253366,
      "grad_norm": 1.1432669225997005,
      "learning_rate": 1.8083401157969371e-06,
      "loss": 0.4349,
      "step": 7221
    },
    {
      "epoch": 8.839657282741738,
      "grad_norm": 1.4950016990776427,
      "learning_rate": 1.8077242743891316e-06,
      "loss": 0.5301,
      "step": 7222
    },
    {
      "epoch": 8.840881272949817,
      "grad_norm": 1.516791770456261,
      "learning_rate": 1.8071084784751265e-06,
      "loss": 1.5483,
      "step": 7223
    },
    {
      "epoch": 8.842105263157894,
      "grad_norm": 2.0752668112373147,
      "learning_rate": 1.8064927280953893e-06,
      "loss": 0.6116,
      "step": 7224
    },
    {
      "epoch": 8.843329253365972,
      "grad_norm": 0.7320055545097112,
      "learning_rate": 1.8058770232903842e-06,
      "loss": 0.221,
      "step": 7225
    },
    {
      "epoch": 8.844553243574051,
      "grad_norm": 2.675819183443763,
      "learning_rate": 1.8052613641005743e-06,
      "loss": 1.0372,
      "step": 7226
    },
    {
      "epoch": 8.84577723378213,
      "grad_norm": 1.9129526235716634,
      "learning_rate": 1.8046457505664183e-06,
      "loss": 0.4012,
      "step": 7227
    },
    {
      "epoch": 8.847001223990208,
      "grad_norm": 1.5379696399931382,
      "learning_rate": 1.8040301827283713e-06,
      "loss": 0.389,
      "step": 7228
    },
    {
      "epoch": 8.848225214198287,
      "grad_norm": 3.7082116491721386,
      "learning_rate": 1.8034146606268855e-06,
      "loss": 0.3999,
      "step": 7229
    },
    {
      "epoch": 8.849449204406366,
      "grad_norm": 2.0285320396078776,
      "learning_rate": 1.8027991843024133e-06,
      "loss": 1.138,
      "step": 7230
    },
    {
      "epoch": 8.850673194614442,
      "grad_norm": 2.2597987788465286,
      "learning_rate": 1.8021837537953992e-06,
      "loss": 0.4097,
      "step": 7231
    },
    {
      "epoch": 8.851897184822521,
      "grad_norm": 1.49872270747978,
      "learning_rate": 1.8015683691462878e-06,
      "loss": 0.5858,
      "step": 7232
    },
    {
      "epoch": 8.8531211750306,
      "grad_norm": 1.758351583936787,
      "learning_rate": 1.8009530303955198e-06,
      "loss": 0.7544,
      "step": 7233
    },
    {
      "epoch": 8.854345165238678,
      "grad_norm": 2.7330709056323474,
      "learning_rate": 1.800337737583533e-06,
      "loss": 0.6992,
      "step": 7234
    },
    {
      "epoch": 8.855569155446757,
      "grad_norm": 1.9129535256644716,
      "learning_rate": 1.7997224907507623e-06,
      "loss": 1.3,
      "step": 7235
    },
    {
      "epoch": 8.856793145654835,
      "grad_norm": 1.7408643695991997,
      "learning_rate": 1.79910728993764e-06,
      "loss": 0.7143,
      "step": 7236
    },
    {
      "epoch": 8.858017135862912,
      "grad_norm": 3.380495613609966,
      "learning_rate": 1.7984921351845934e-06,
      "loss": 0.4932,
      "step": 7237
    },
    {
      "epoch": 8.859241126070991,
      "grad_norm": 2.028477304708973,
      "learning_rate": 1.7978770265320484e-06,
      "loss": 0.4269,
      "step": 7238
    },
    {
      "epoch": 8.86046511627907,
      "grad_norm": 2.037640736284394,
      "learning_rate": 1.7972619640204298e-06,
      "loss": 0.6087,
      "step": 7239
    },
    {
      "epoch": 8.861689106487148,
      "grad_norm": 2.055975888507703,
      "learning_rate": 1.7966469476901548e-06,
      "loss": 0.578,
      "step": 7240
    },
    {
      "epoch": 8.862913096695227,
      "grad_norm": 1.376517422805062,
      "learning_rate": 1.796031977581641e-06,
      "loss": 0.8189,
      "step": 7241
    },
    {
      "epoch": 8.864137086903305,
      "grad_norm": 1.7973799773319954,
      "learning_rate": 1.7954170537353014e-06,
      "loss": 0.5919,
      "step": 7242
    },
    {
      "epoch": 8.865361077111384,
      "grad_norm": 2.8649783252672685,
      "learning_rate": 1.7948021761915474e-06,
      "loss": 0.4751,
      "step": 7243
    },
    {
      "epoch": 8.86658506731946,
      "grad_norm": 0.9235711792574973,
      "learning_rate": 1.794187344990786e-06,
      "loss": 0.4309,
      "step": 7244
    },
    {
      "epoch": 8.86780905752754,
      "grad_norm": 1.9403924437258337,
      "learning_rate": 1.7935725601734222e-06,
      "loss": 0.4163,
      "step": 7245
    },
    {
      "epoch": 8.869033047735618,
      "grad_norm": 2.3143938642765387,
      "learning_rate": 1.7929578217798554e-06,
      "loss": 0.5876,
      "step": 7246
    },
    {
      "epoch": 8.870257037943697,
      "grad_norm": 1.4686718580052012,
      "learning_rate": 1.7923431298504861e-06,
      "loss": 0.6289,
      "step": 7247
    },
    {
      "epoch": 8.871481028151775,
      "grad_norm": 1.750660590434502,
      "learning_rate": 1.79172848442571e-06,
      "loss": 1.3525,
      "step": 7248
    },
    {
      "epoch": 8.872705018359854,
      "grad_norm": 1.772536026065116,
      "learning_rate": 1.791113885545917e-06,
      "loss": 0.7032,
      "step": 7249
    },
    {
      "epoch": 8.87392900856793,
      "grad_norm": 2.0208851788886184,
      "learning_rate": 1.7904993332514981e-06,
      "loss": 1.1311,
      "step": 7250
    },
    {
      "epoch": 8.87515299877601,
      "grad_norm": 1.713137565098638,
      "learning_rate": 1.7898848275828378e-06,
      "loss": 1.199,
      "step": 7251
    },
    {
      "epoch": 8.876376988984088,
      "grad_norm": 1.9508795978593971,
      "learning_rate": 1.7892703685803205e-06,
      "loss": 0.7723,
      "step": 7252
    },
    {
      "epoch": 8.877600979192167,
      "grad_norm": 1.7784219426018923,
      "learning_rate": 1.788655956284326e-06,
      "loss": 0.6095,
      "step": 7253
    },
    {
      "epoch": 8.878824969400245,
      "grad_norm": 1.0942808753929916,
      "learning_rate": 1.7880415907352315e-06,
      "loss": 0.6364,
      "step": 7254
    },
    {
      "epoch": 8.880048959608324,
      "grad_norm": 1.7391020723418422,
      "learning_rate": 1.7874272719734087e-06,
      "loss": 0.4631,
      "step": 7255
    },
    {
      "epoch": 8.8812729498164,
      "grad_norm": 1.0284062942462464,
      "learning_rate": 1.7868130000392309e-06,
      "loss": 0.5969,
      "step": 7256
    },
    {
      "epoch": 8.88249694002448,
      "grad_norm": 1.578388022243378,
      "learning_rate": 1.7861987749730659e-06,
      "loss": 0.6467,
      "step": 7257
    },
    {
      "epoch": 8.883720930232558,
      "grad_norm": 2.6774575755430257,
      "learning_rate": 1.7855845968152761e-06,
      "loss": 0.3686,
      "step": 7258
    },
    {
      "epoch": 8.884944920440637,
      "grad_norm": 1.6665898068224396,
      "learning_rate": 1.784970465606224e-06,
      "loss": 0.7572,
      "step": 7259
    },
    {
      "epoch": 8.886168910648715,
      "grad_norm": 1.4796518506707077,
      "learning_rate": 1.7843563813862685e-06,
      "loss": 0.9351,
      "step": 7260
    },
    {
      "epoch": 8.887392900856794,
      "grad_norm": 0.5467381544691978,
      "learning_rate": 1.7837423441957646e-06,
      "loss": 0.1306,
      "step": 7261
    },
    {
      "epoch": 8.88861689106487,
      "grad_norm": 1.2477752434613232,
      "learning_rate": 1.7831283540750646e-06,
      "loss": 0.8996,
      "step": 7262
    },
    {
      "epoch": 8.88984088127295,
      "grad_norm": 1.5076713654083478,
      "learning_rate": 1.7825144110645182e-06,
      "loss": 0.9286,
      "step": 7263
    },
    {
      "epoch": 8.891064871481028,
      "grad_norm": 2.979574175669115,
      "learning_rate": 1.7819005152044694e-06,
      "loss": 0.424,
      "step": 7264
    },
    {
      "epoch": 8.892288861689106,
      "grad_norm": 1.4717015713662,
      "learning_rate": 1.781286666535264e-06,
      "loss": 0.574,
      "step": 7265
    },
    {
      "epoch": 8.893512851897185,
      "grad_norm": 1.502408973454318,
      "learning_rate": 1.7806728650972405e-06,
      "loss": 1.4075,
      "step": 7266
    },
    {
      "epoch": 8.894736842105264,
      "grad_norm": 2.226950508353413,
      "learning_rate": 1.7800591109307353e-06,
      "loss": 0.601,
      "step": 7267
    },
    {
      "epoch": 8.895960832313342,
      "grad_norm": 1.5704790991724589,
      "learning_rate": 1.7794454040760822e-06,
      "loss": 0.503,
      "step": 7268
    },
    {
      "epoch": 8.89718482252142,
      "grad_norm": 1.572585407924366,
      "learning_rate": 1.7788317445736126e-06,
      "loss": 0.5164,
      "step": 7269
    },
    {
      "epoch": 8.898408812729498,
      "grad_norm": 1.5469266223325504,
      "learning_rate": 1.7782181324636528e-06,
      "loss": 0.4624,
      "step": 7270
    },
    {
      "epoch": 8.899632802937576,
      "grad_norm": 1.5944443716307788,
      "learning_rate": 1.777604567786528e-06,
      "loss": 0.8594,
      "step": 7271
    },
    {
      "epoch": 8.900856793145655,
      "grad_norm": 2.4423840151601635,
      "learning_rate": 1.7769910505825586e-06,
      "loss": 0.4015,
      "step": 7272
    },
    {
      "epoch": 8.902080783353734,
      "grad_norm": 1.951278315470863,
      "learning_rate": 1.7763775808920636e-06,
      "loss": 1.3245,
      "step": 7273
    },
    {
      "epoch": 8.903304773561812,
      "grad_norm": 1.4755258663512056,
      "learning_rate": 1.775764158755358e-06,
      "loss": 0.6974,
      "step": 7274
    },
    {
      "epoch": 8.904528763769889,
      "grad_norm": 1.0436719703164428,
      "learning_rate": 1.7751507842127524e-06,
      "loss": 0.6846,
      "step": 7275
    },
    {
      "epoch": 8.905752753977968,
      "grad_norm": 2.4863360336753297,
      "learning_rate": 1.7745374573045566e-06,
      "loss": 0.4393,
      "step": 7276
    },
    {
      "epoch": 8.906976744186046,
      "grad_norm": 1.28380168324209,
      "learning_rate": 1.7739241780710749e-06,
      "loss": 0.3715,
      "step": 7277
    },
    {
      "epoch": 8.908200734394125,
      "grad_norm": 3.049118291345027,
      "learning_rate": 1.773310946552611e-06,
      "loss": 0.5353,
      "step": 7278
    },
    {
      "epoch": 8.909424724602204,
      "grad_norm": 1.525342834183679,
      "learning_rate": 1.7726977627894643e-06,
      "loss": 0.9874,
      "step": 7279
    },
    {
      "epoch": 8.910648714810282,
      "grad_norm": 1.2628119830494042,
      "learning_rate": 1.7720846268219306e-06,
      "loss": 0.7702,
      "step": 7280
    },
    {
      "epoch": 8.911872705018359,
      "grad_norm": 1.804141041345193,
      "learning_rate": 1.771471538690302e-06,
      "loss": 0.4062,
      "step": 7281
    },
    {
      "epoch": 8.913096695226438,
      "grad_norm": 1.1557754313848765,
      "learning_rate": 1.77085849843487e-06,
      "loss": 0.6631,
      "step": 7282
    },
    {
      "epoch": 8.914320685434516,
      "grad_norm": 1.8569190265225908,
      "learning_rate": 1.770245506095921e-06,
      "loss": 0.4885,
      "step": 7283
    },
    {
      "epoch": 8.915544675642595,
      "grad_norm": 1.6243052000935068,
      "learning_rate": 1.7696325617137377e-06,
      "loss": 0.4375,
      "step": 7284
    },
    {
      "epoch": 8.916768665850674,
      "grad_norm": 2.370303223623652,
      "learning_rate": 1.7690196653285999e-06,
      "loss": 1.0248,
      "step": 7285
    },
    {
      "epoch": 8.917992656058752,
      "grad_norm": 3.581657258254009,
      "learning_rate": 1.7684068169807875e-06,
      "loss": 0.3724,
      "step": 7286
    },
    {
      "epoch": 8.919216646266829,
      "grad_norm": 2.0289865612395723,
      "learning_rate": 1.767794016710573e-06,
      "loss": 0.3968,
      "step": 7287
    },
    {
      "epoch": 8.920440636474908,
      "grad_norm": 1.0229302016915696,
      "learning_rate": 1.7671812645582267e-06,
      "loss": 0.5125,
      "step": 7288
    },
    {
      "epoch": 8.921664626682986,
      "grad_norm": 3.2449769775471573,
      "learning_rate": 1.7665685605640177e-06,
      "loss": 0.4307,
      "step": 7289
    },
    {
      "epoch": 8.922888616891065,
      "grad_norm": 1.2679777739344953,
      "learning_rate": 1.7659559047682096e-06,
      "loss": 0.5293,
      "step": 7290
    },
    {
      "epoch": 8.924112607099143,
      "grad_norm": 1.6643903530350674,
      "learning_rate": 1.7653432972110646e-06,
      "loss": 1.5272,
      "step": 7291
    },
    {
      "epoch": 8.925336597307222,
      "grad_norm": 1.533170818512399,
      "learning_rate": 1.7647307379328416e-06,
      "loss": 0.7081,
      "step": 7292
    },
    {
      "epoch": 8.9265605875153,
      "grad_norm": 2.107621169466562,
      "learning_rate": 1.7641182269737944e-06,
      "loss": 0.6877,
      "step": 7293
    },
    {
      "epoch": 8.927784577723378,
      "grad_norm": 1.616720130796334,
      "learning_rate": 1.7635057643741748e-06,
      "loss": 0.9132,
      "step": 7294
    },
    {
      "epoch": 8.929008567931456,
      "grad_norm": 1.5971686331485757,
      "learning_rate": 1.7628933501742329e-06,
      "loss": 1.2728,
      "step": 7295
    },
    {
      "epoch": 8.930232558139535,
      "grad_norm": 1.82690413069777,
      "learning_rate": 1.7622809844142138e-06,
      "loss": 0.5623,
      "step": 7296
    },
    {
      "epoch": 8.931456548347613,
      "grad_norm": 1.647210268661482,
      "learning_rate": 1.7616686671343597e-06,
      "loss": 1.1804,
      "step": 7297
    },
    {
      "epoch": 8.932680538555692,
      "grad_norm": 2.7190606467067937,
      "learning_rate": 1.7610563983749093e-06,
      "loss": 1.0696,
      "step": 7298
    },
    {
      "epoch": 8.93390452876377,
      "grad_norm": 1.791917618280875,
      "learning_rate": 1.7604441781760997e-06,
      "loss": 0.5331,
      "step": 7299
    },
    {
      "epoch": 8.935128518971847,
      "grad_norm": 1.171288709421414,
      "learning_rate": 1.7598320065781634e-06,
      "loss": 0.4451,
      "step": 7300
    },
    {
      "epoch": 8.936352509179926,
      "grad_norm": 2.1333442598412713,
      "learning_rate": 1.7592198836213303e-06,
      "loss": 0.9304,
      "step": 7301
    },
    {
      "epoch": 8.937576499388005,
      "grad_norm": 2.459665837851935,
      "learning_rate": 1.7586078093458264e-06,
      "loss": 0.8271,
      "step": 7302
    },
    {
      "epoch": 8.938800489596083,
      "grad_norm": 2.598555636470584,
      "learning_rate": 1.757995783791874e-06,
      "loss": 0.4607,
      "step": 7303
    },
    {
      "epoch": 8.940024479804162,
      "grad_norm": 2.6726868483079484,
      "learning_rate": 1.7573838069996952e-06,
      "loss": 0.366,
      "step": 7304
    },
    {
      "epoch": 8.94124847001224,
      "grad_norm": 3.081633441370582,
      "learning_rate": 1.7567718790095054e-06,
      "loss": 0.3595,
      "step": 7305
    },
    {
      "epoch": 8.94247246022032,
      "grad_norm": 1.7195366255676325,
      "learning_rate": 1.756159999861519e-06,
      "loss": 0.4308,
      "step": 7306
    },
    {
      "epoch": 8.943696450428396,
      "grad_norm": 2.0725801412006954,
      "learning_rate": 1.7555481695959454e-06,
      "loss": 0.9615,
      "step": 7307
    },
    {
      "epoch": 8.944920440636475,
      "grad_norm": 1.6034859518174511,
      "learning_rate": 1.7549363882529933e-06,
      "loss": 0.5077,
      "step": 7308
    },
    {
      "epoch": 8.946144430844553,
      "grad_norm": 1.13764272081045,
      "learning_rate": 1.7543246558728658e-06,
      "loss": 0.5465,
      "step": 7309
    },
    {
      "epoch": 8.947368421052632,
      "grad_norm": 1.838306743231294,
      "learning_rate": 1.7537129724957644e-06,
      "loss": 0.9655,
      "step": 7310
    },
    {
      "epoch": 8.94859241126071,
      "grad_norm": 1.7071161011485083,
      "learning_rate": 1.7531013381618845e-06,
      "loss": 0.4746,
      "step": 7311
    },
    {
      "epoch": 8.949816401468787,
      "grad_norm": 2.2393829092126336,
      "learning_rate": 1.7524897529114238e-06,
      "loss": 1.2071,
      "step": 7312
    },
    {
      "epoch": 8.951040391676866,
      "grad_norm": 1.9619796232003515,
      "learning_rate": 1.7518782167845711e-06,
      "loss": 0.7218,
      "step": 7313
    },
    {
      "epoch": 8.952264381884945,
      "grad_norm": 1.3115222929669283,
      "learning_rate": 1.7512667298215148e-06,
      "loss": 0.5628,
      "step": 7314
    },
    {
      "epoch": 8.953488372093023,
      "grad_norm": 2.5344399473413377,
      "learning_rate": 1.7506552920624397e-06,
      "loss": 0.4139,
      "step": 7315
    },
    {
      "epoch": 8.954712362301102,
      "grad_norm": 2.1900013263305436,
      "learning_rate": 1.7500439035475266e-06,
      "loss": 0.5064,
      "step": 7316
    },
    {
      "epoch": 8.95593635250918,
      "grad_norm": 1.1581813379334414,
      "learning_rate": 1.749432564316955e-06,
      "loss": 0.5764,
      "step": 7317
    },
    {
      "epoch": 8.957160342717259,
      "grad_norm": 1.592425333078659,
      "learning_rate": 1.748821274410899e-06,
      "loss": 0.4619,
      "step": 7318
    },
    {
      "epoch": 8.958384332925336,
      "grad_norm": 1.3938741506356829,
      "learning_rate": 1.7482100338695307e-06,
      "loss": 0.7899,
      "step": 7319
    },
    {
      "epoch": 8.959608323133414,
      "grad_norm": 1.4368198683995814,
      "learning_rate": 1.747598842733017e-06,
      "loss": 0.5734,
      "step": 7320
    },
    {
      "epoch": 8.960832313341493,
      "grad_norm": 1.0838360185539475,
      "learning_rate": 1.746987701041526e-06,
      "loss": 0.5731,
      "step": 7321
    },
    {
      "epoch": 8.962056303549572,
      "grad_norm": 2.308754402059643,
      "learning_rate": 1.7463766088352174e-06,
      "loss": 0.7208,
      "step": 7322
    },
    {
      "epoch": 8.96328029375765,
      "grad_norm": 1.576790965673864,
      "learning_rate": 1.7457655661542507e-06,
      "loss": 0.8656,
      "step": 7323
    },
    {
      "epoch": 8.964504283965729,
      "grad_norm": 2.4277589496402623,
      "learning_rate": 1.7451545730387809e-06,
      "loss": 0.4746,
      "step": 7324
    },
    {
      "epoch": 8.965728274173806,
      "grad_norm": 1.2895734739086566,
      "learning_rate": 1.7445436295289608e-06,
      "loss": 0.5822,
      "step": 7325
    },
    {
      "epoch": 8.966952264381884,
      "grad_norm": 1.7934218937593864,
      "learning_rate": 1.7439327356649393e-06,
      "loss": 0.531,
      "step": 7326
    },
    {
      "epoch": 8.968176254589963,
      "grad_norm": 1.4644995218043975,
      "learning_rate": 1.7433218914868623e-06,
      "loss": 0.2291,
      "step": 7327
    },
    {
      "epoch": 8.969400244798042,
      "grad_norm": 2.4603890731752207,
      "learning_rate": 1.7427110970348715e-06,
      "loss": 0.2433,
      "step": 7328
    },
    {
      "epoch": 8.97062423500612,
      "grad_norm": 2.3298599839650733,
      "learning_rate": 1.7421003523491053e-06,
      "loss": 1.0318,
      "step": 7329
    },
    {
      "epoch": 8.971848225214199,
      "grad_norm": 1.0738048086629404,
      "learning_rate": 1.741489657469702e-06,
      "loss": 0.5765,
      "step": 7330
    },
    {
      "epoch": 8.973072215422278,
      "grad_norm": 2.5734639178488266,
      "learning_rate": 1.740879012436793e-06,
      "loss": 0.4886,
      "step": 7331
    },
    {
      "epoch": 8.974296205630354,
      "grad_norm": 2.6085480493635327,
      "learning_rate": 1.740268417290507e-06,
      "loss": 0.5175,
      "step": 7332
    },
    {
      "epoch": 8.975520195838433,
      "grad_norm": 1.3266427708898758,
      "learning_rate": 1.73965787207097e-06,
      "loss": 1.0491,
      "step": 7333
    },
    {
      "epoch": 8.976744186046512,
      "grad_norm": 2.480454393238444,
      "learning_rate": 1.7390473768183057e-06,
      "loss": 0.7954,
      "step": 7334
    },
    {
      "epoch": 8.97796817625459,
      "grad_norm": 1.4074068795036734,
      "learning_rate": 1.7384369315726339e-06,
      "loss": 1.0787,
      "step": 7335
    },
    {
      "epoch": 8.979192166462669,
      "grad_norm": 1.0356362334659857,
      "learning_rate": 1.73782653637407e-06,
      "loss": 0.5997,
      "step": 7336
    },
    {
      "epoch": 8.980416156670747,
      "grad_norm": 1.485679054577717,
      "learning_rate": 1.7372161912627257e-06,
      "loss": 0.5693,
      "step": 7337
    },
    {
      "epoch": 8.981640146878824,
      "grad_norm": 2.891961261565229,
      "learning_rate": 1.736605896278713e-06,
      "loss": 0.5569,
      "step": 7338
    },
    {
      "epoch": 8.982864137086903,
      "grad_norm": 1.2043327883313792,
      "learning_rate": 1.735995651462138e-06,
      "loss": 0.5626,
      "step": 7339
    },
    {
      "epoch": 8.984088127294982,
      "grad_norm": 1.3730325704176989,
      "learning_rate": 1.735385456853102e-06,
      "loss": 0.8025,
      "step": 7340
    },
    {
      "epoch": 8.98531211750306,
      "grad_norm": 2.4601944920817815,
      "learning_rate": 1.7347753124917061e-06,
      "loss": 0.4073,
      "step": 7341
    },
    {
      "epoch": 8.986536107711139,
      "grad_norm": 1.0942885526033632,
      "learning_rate": 1.7341652184180452e-06,
      "loss": 0.8489,
      "step": 7342
    },
    {
      "epoch": 8.987760097919217,
      "grad_norm": 1.5078588092896557,
      "learning_rate": 1.7335551746722145e-06,
      "loss": 0.3644,
      "step": 7343
    },
    {
      "epoch": 8.988984088127294,
      "grad_norm": 2.3235604598738044,
      "learning_rate": 1.7329451812943027e-06,
      "loss": 0.386,
      "step": 7344
    },
    {
      "epoch": 8.990208078335373,
      "grad_norm": 2.186779896769756,
      "learning_rate": 1.732335238324397e-06,
      "loss": 0.4365,
      "step": 7345
    },
    {
      "epoch": 8.991432068543451,
      "grad_norm": 2.135971929817507,
      "learning_rate": 1.7317253458025781e-06,
      "loss": 0.6278,
      "step": 7346
    },
    {
      "epoch": 8.99265605875153,
      "grad_norm": 1.7255863676093877,
      "learning_rate": 1.7311155037689291e-06,
      "loss": 0.5511,
      "step": 7347
    },
    {
      "epoch": 8.993880048959609,
      "grad_norm": 2.4747881333122552,
      "learning_rate": 1.730505712263526e-06,
      "loss": 0.2882,
      "step": 7348
    },
    {
      "epoch": 8.995104039167687,
      "grad_norm": 1.3995749897998755,
      "learning_rate": 1.7298959713264406e-06,
      "loss": 0.458,
      "step": 7349
    },
    {
      "epoch": 8.996328029375764,
      "grad_norm": 1.2088972642103417,
      "learning_rate": 1.7292862809977435e-06,
      "loss": 0.8983,
      "step": 7350
    },
    {
      "epoch": 8.997552019583843,
      "grad_norm": 1.148459420304103,
      "learning_rate": 1.7286766413175013e-06,
      "loss": 0.848,
      "step": 7351
    },
    {
      "epoch": 8.998776009791921,
      "grad_norm": 1.6300404356347673,
      "learning_rate": 1.7280670523257775e-06,
      "loss": 1.4589,
      "step": 7352
    },
    {
      "epoch": 9.0,
      "grad_norm": 2.309469555950492,
      "learning_rate": 1.7274575140626318e-06,
      "loss": 0.4989,
      "step": 7353
    },
    {
      "epoch": 9.001223990208079,
      "grad_norm": 2.4320292374327908,
      "learning_rate": 1.7268480265681214e-06,
      "loss": 0.5561,
      "step": 7354
    },
    {
      "epoch": 9.002447980416157,
      "grad_norm": 2.1936464845393497,
      "learning_rate": 1.7262385898822975e-06,
      "loss": 0.5535,
      "step": 7355
    },
    {
      "epoch": 9.003671970624236,
      "grad_norm": 1.8381751551558219,
      "learning_rate": 1.7256292040452122e-06,
      "loss": 0.4438,
      "step": 7356
    },
    {
      "epoch": 9.004895960832313,
      "grad_norm": 2.44716465069386,
      "learning_rate": 1.7250198690969122e-06,
      "loss": 0.3518,
      "step": 7357
    },
    {
      "epoch": 9.006119951040391,
      "grad_norm": 2.033779330329197,
      "learning_rate": 1.7244105850774395e-06,
      "loss": 1.1984,
      "step": 7358
    },
    {
      "epoch": 9.00734394124847,
      "grad_norm": 1.4259449873906966,
      "learning_rate": 1.723801352026834e-06,
      "loss": 1.5716,
      "step": 7359
    },
    {
      "epoch": 9.008567931456549,
      "grad_norm": 2.4819478339018897,
      "learning_rate": 1.7231921699851331e-06,
      "loss": 0.7305,
      "step": 7360
    },
    {
      "epoch": 9.009791921664627,
      "grad_norm": 1.3941192100423891,
      "learning_rate": 1.7225830389923695e-06,
      "loss": 0.357,
      "step": 7361
    },
    {
      "epoch": 9.011015911872706,
      "grad_norm": 3.002780147772951,
      "learning_rate": 1.721973959088573e-06,
      "loss": 0.7044,
      "step": 7362
    },
    {
      "epoch": 9.012239902080783,
      "grad_norm": 1.4639117390024883,
      "learning_rate": 1.72136493031377e-06,
      "loss": 0.9817,
      "step": 7363
    },
    {
      "epoch": 9.013463892288861,
      "grad_norm": 1.8798283869545693,
      "learning_rate": 1.7207559527079842e-06,
      "loss": 1.007,
      "step": 7364
    },
    {
      "epoch": 9.01468788249694,
      "grad_norm": 2.671469029121102,
      "learning_rate": 1.720147026311235e-06,
      "loss": 0.5882,
      "step": 7365
    },
    {
      "epoch": 9.015911872705018,
      "grad_norm": 1.60716243297815,
      "learning_rate": 1.719538151163539e-06,
      "loss": 0.432,
      "step": 7366
    },
    {
      "epoch": 9.017135862913097,
      "grad_norm": 1.3349995056224946,
      "learning_rate": 1.7189293273049087e-06,
      "loss": 0.68,
      "step": 7367
    },
    {
      "epoch": 9.018359853121176,
      "grad_norm": 1.5445956143643618,
      "learning_rate": 1.7183205547753534e-06,
      "loss": 0.4557,
      "step": 7368
    },
    {
      "epoch": 9.019583843329253,
      "grad_norm": 0.8974039351346341,
      "learning_rate": 1.7177118336148806e-06,
      "loss": 0.5246,
      "step": 7369
    },
    {
      "epoch": 9.020807833537331,
      "grad_norm": 1.0945899158253276,
      "learning_rate": 1.7171031638634927e-06,
      "loss": 0.5988,
      "step": 7370
    },
    {
      "epoch": 9.02203182374541,
      "grad_norm": 1.985047506625008,
      "learning_rate": 1.7164945455611892e-06,
      "loss": 0.3517,
      "step": 7371
    },
    {
      "epoch": 9.023255813953488,
      "grad_norm": 1.977454686395154,
      "learning_rate": 1.7158859787479654e-06,
      "loss": 0.386,
      "step": 7372
    },
    {
      "epoch": 9.024479804161567,
      "grad_norm": 1.7228349985905766,
      "learning_rate": 1.7152774634638153e-06,
      "loss": 1.0993,
      "step": 7373
    },
    {
      "epoch": 9.025703794369646,
      "grad_norm": 1.4462478558038456,
      "learning_rate": 1.7146689997487286e-06,
      "loss": 0.6078,
      "step": 7374
    },
    {
      "epoch": 9.026927784577722,
      "grad_norm": 2.3179122316069667,
      "learning_rate": 1.71406058764269e-06,
      "loss": 0.7379,
      "step": 7375
    },
    {
      "epoch": 9.028151774785801,
      "grad_norm": 1.3888921513914618,
      "learning_rate": 1.7134522271856823e-06,
      "loss": 0.6236,
      "step": 7376
    },
    {
      "epoch": 9.02937576499388,
      "grad_norm": 1.1096669564683088,
      "learning_rate": 1.7128439184176849e-06,
      "loss": 0.6139,
      "step": 7377
    },
    {
      "epoch": 9.030599755201958,
      "grad_norm": 1.7354039545722222,
      "learning_rate": 1.7122356613786739e-06,
      "loss": 0.9566,
      "step": 7378
    },
    {
      "epoch": 9.031823745410037,
      "grad_norm": 2.0265014556919145,
      "learning_rate": 1.7116274561086215e-06,
      "loss": 0.8138,
      "step": 7379
    },
    {
      "epoch": 9.033047735618116,
      "grad_norm": 1.4550687731848733,
      "learning_rate": 1.7110193026474967e-06,
      "loss": 0.6488,
      "step": 7380
    },
    {
      "epoch": 9.034271725826194,
      "grad_norm": 1.2314664462933818,
      "learning_rate": 1.7104112010352648e-06,
      "loss": 0.4319,
      "step": 7381
    },
    {
      "epoch": 9.035495716034271,
      "grad_norm": 2.148657200536083,
      "learning_rate": 1.7098031513118884e-06,
      "loss": 0.4501,
      "step": 7382
    },
    {
      "epoch": 9.03671970624235,
      "grad_norm": 0.9064463626245963,
      "learning_rate": 1.7091951535173268e-06,
      "loss": 0.3916,
      "step": 7383
    },
    {
      "epoch": 9.037943696450428,
      "grad_norm": 2.113452485343649,
      "learning_rate": 1.7085872076915343e-06,
      "loss": 0.5591,
      "step": 7384
    },
    {
      "epoch": 9.039167686658507,
      "grad_norm": 1.9414805894641267,
      "learning_rate": 1.7079793138744622e-06,
      "loss": 0.5511,
      "step": 7385
    },
    {
      "epoch": 9.040391676866586,
      "grad_norm": 1.0636713131653328,
      "learning_rate": 1.7073714721060619e-06,
      "loss": 0.6332,
      "step": 7386
    },
    {
      "epoch": 9.041615667074664,
      "grad_norm": 1.4125690729858618,
      "learning_rate": 1.706763682426276e-06,
      "loss": 0.53,
      "step": 7387
    },
    {
      "epoch": 9.042839657282741,
      "grad_norm": 1.1820816681765456,
      "learning_rate": 1.706155944875047e-06,
      "loss": 0.894,
      "step": 7388
    },
    {
      "epoch": 9.04406364749082,
      "grad_norm": 1.2946459628676847,
      "learning_rate": 1.7055482594923128e-06,
      "loss": 0.5779,
      "step": 7389
    },
    {
      "epoch": 9.045287637698898,
      "grad_norm": 1.428207256685282,
      "learning_rate": 1.7049406263180083e-06,
      "loss": 1.2311,
      "step": 7390
    },
    {
      "epoch": 9.046511627906977,
      "grad_norm": 3.3286952847981888,
      "learning_rate": 1.7043330453920653e-06,
      "loss": 0.3557,
      "step": 7391
    },
    {
      "epoch": 9.047735618115055,
      "grad_norm": 1.2241801016674074,
      "learning_rate": 1.7037255167544126e-06,
      "loss": 0.5943,
      "step": 7392
    },
    {
      "epoch": 9.048959608323134,
      "grad_norm": 2.2979281727308636,
      "learning_rate": 1.7031180404449727e-06,
      "loss": 0.4003,
      "step": 7393
    },
    {
      "epoch": 9.050183598531211,
      "grad_norm": 1.7279021369587342,
      "learning_rate": 1.702510616503667e-06,
      "loss": 1.1912,
      "step": 7394
    },
    {
      "epoch": 9.05140758873929,
      "grad_norm": 2.513210152125161,
      "learning_rate": 1.7019032449704154e-06,
      "loss": 0.3394,
      "step": 7395
    },
    {
      "epoch": 9.052631578947368,
      "grad_norm": 1.6443288371590166,
      "learning_rate": 1.70129592588513e-06,
      "loss": 0.8264,
      "step": 7396
    },
    {
      "epoch": 9.053855569155447,
      "grad_norm": 1.3364969871012475,
      "learning_rate": 1.7006886592877222e-06,
      "loss": 1.5693,
      "step": 7397
    },
    {
      "epoch": 9.055079559363525,
      "grad_norm": 1.5141677983562423,
      "learning_rate": 1.7000814452180987e-06,
      "loss": 0.82,
      "step": 7398
    },
    {
      "epoch": 9.056303549571604,
      "grad_norm": 2.483104312995922,
      "learning_rate": 1.6994742837161642e-06,
      "loss": 0.5391,
      "step": 7399
    },
    {
      "epoch": 9.057527539779683,
      "grad_norm": 1.2254782773621526,
      "learning_rate": 1.6988671748218194e-06,
      "loss": 0.5213,
      "step": 7400
    },
    {
      "epoch": 9.05875152998776,
      "grad_norm": 1.7507537023339734,
      "learning_rate": 1.698260118574961e-06,
      "loss": 1.0102,
      "step": 7401
    },
    {
      "epoch": 9.059975520195838,
      "grad_norm": 1.9723499197213978,
      "learning_rate": 1.6976531150154812e-06,
      "loss": 0.4368,
      "step": 7402
    },
    {
      "epoch": 9.061199510403917,
      "grad_norm": 2.1158641571342844,
      "learning_rate": 1.697046164183271e-06,
      "loss": 0.4802,
      "step": 7403
    },
    {
      "epoch": 9.062423500611995,
      "grad_norm": 2.4834183533399035,
      "learning_rate": 1.6964392661182171e-06,
      "loss": 0.6807,
      "step": 7404
    },
    {
      "epoch": 9.063647490820074,
      "grad_norm": 2.173938250173641,
      "learning_rate": 1.6958324208602028e-06,
      "loss": 0.6361,
      "step": 7405
    },
    {
      "epoch": 9.064871481028153,
      "grad_norm": 1.7283704997517901,
      "learning_rate": 1.6952256284491076e-06,
      "loss": 0.9332,
      "step": 7406
    },
    {
      "epoch": 9.06609547123623,
      "grad_norm": 1.4809577684031454,
      "learning_rate": 1.6946188889248066e-06,
      "loss": 0.57,
      "step": 7407
    },
    {
      "epoch": 9.067319461444308,
      "grad_norm": 1.4732728386203315,
      "learning_rate": 1.6940122023271737e-06,
      "loss": 0.7522,
      "step": 7408
    },
    {
      "epoch": 9.068543451652387,
      "grad_norm": 1.4631260871188463,
      "learning_rate": 1.693405568696078e-06,
      "loss": 0.4737,
      "step": 7409
    },
    {
      "epoch": 9.069767441860465,
      "grad_norm": 2.5344167705359073,
      "learning_rate": 1.6927989880713852e-06,
      "loss": 0.3958,
      "step": 7410
    },
    {
      "epoch": 9.070991432068544,
      "grad_norm": 2.6139084975741236,
      "learning_rate": 1.6921924604929556e-06,
      "loss": 0.9568,
      "step": 7411
    },
    {
      "epoch": 9.072215422276622,
      "grad_norm": 1.4331334036449888,
      "learning_rate": 1.6915859860006518e-06,
      "loss": 0.7967,
      "step": 7412
    },
    {
      "epoch": 9.0734394124847,
      "grad_norm": 1.2932872675392948,
      "learning_rate": 1.6909795646343258e-06,
      "loss": 0.5371,
      "step": 7413
    },
    {
      "epoch": 9.074663402692778,
      "grad_norm": 1.877414470536694,
      "learning_rate": 1.6903731964338307e-06,
      "loss": 0.393,
      "step": 7414
    },
    {
      "epoch": 9.075887392900857,
      "grad_norm": 1.9205248147961156,
      "learning_rate": 1.6897668814390146e-06,
      "loss": 0.5769,
      "step": 7415
    },
    {
      "epoch": 9.077111383108935,
      "grad_norm": 1.4565821586128898,
      "learning_rate": 1.6891606196897214e-06,
      "loss": 0.6139,
      "step": 7416
    },
    {
      "epoch": 9.078335373317014,
      "grad_norm": 1.4618410638503796,
      "learning_rate": 1.688554411225794e-06,
      "loss": 0.8109,
      "step": 7417
    },
    {
      "epoch": 9.079559363525092,
      "grad_norm": 1.8650891810955934,
      "learning_rate": 1.6879482560870692e-06,
      "loss": 0.6288,
      "step": 7418
    },
    {
      "epoch": 9.080783353733171,
      "grad_norm": 1.849012521862426,
      "learning_rate": 1.687342154313382e-06,
      "loss": 0.6264,
      "step": 7419
    },
    {
      "epoch": 9.082007343941248,
      "grad_norm": 1.673465420782349,
      "learning_rate": 1.6867361059445608e-06,
      "loss": 0.6815,
      "step": 7420
    },
    {
      "epoch": 9.083231334149326,
      "grad_norm": 1.481311999434326,
      "learning_rate": 1.6861301110204367e-06,
      "loss": 0.3823,
      "step": 7421
    },
    {
      "epoch": 9.084455324357405,
      "grad_norm": 1.1990501128762643,
      "learning_rate": 1.6855241695808305e-06,
      "loss": 0.6266,
      "step": 7422
    },
    {
      "epoch": 9.085679314565484,
      "grad_norm": 1.355299147917633,
      "learning_rate": 1.6849182816655634e-06,
      "loss": 0.4583,
      "step": 7423
    },
    {
      "epoch": 9.086903304773562,
      "grad_norm": 1.5845551257084098,
      "learning_rate": 1.6843124473144517e-06,
      "loss": 1.6407,
      "step": 7424
    },
    {
      "epoch": 9.088127294981641,
      "grad_norm": 2.30883691145928,
      "learning_rate": 1.6837066665673095e-06,
      "loss": 0.4642,
      "step": 7425
    },
    {
      "epoch": 9.089351285189718,
      "grad_norm": 1.8777367412475126,
      "learning_rate": 1.6831009394639457e-06,
      "loss": 0.96,
      "step": 7426
    },
    {
      "epoch": 9.090575275397796,
      "grad_norm": 1.0805342001267293,
      "learning_rate": 1.6824952660441668e-06,
      "loss": 0.581,
      "step": 7427
    },
    {
      "epoch": 9.091799265605875,
      "grad_norm": 1.7920189376951863,
      "learning_rate": 1.6818896463477757e-06,
      "loss": 1.2146,
      "step": 7428
    },
    {
      "epoch": 9.093023255813954,
      "grad_norm": 0.9466820138764676,
      "learning_rate": 1.6812840804145696e-06,
      "loss": 0.4487,
      "step": 7429
    },
    {
      "epoch": 9.094247246022032,
      "grad_norm": 1.994414379073264,
      "learning_rate": 1.6806785682843469e-06,
      "loss": 0.641,
      "step": 7430
    },
    {
      "epoch": 9.095471236230111,
      "grad_norm": 1.4433557806755262,
      "learning_rate": 1.680073109996898e-06,
      "loss": 0.868,
      "step": 7431
    },
    {
      "epoch": 9.096695226438188,
      "grad_norm": 1.5528384910429267,
      "learning_rate": 1.6794677055920115e-06,
      "loss": 0.6265,
      "step": 7432
    },
    {
      "epoch": 9.097919216646266,
      "grad_norm": 1.4563888367767,
      "learning_rate": 1.6788623551094719e-06,
      "loss": 0.6219,
      "step": 7433
    },
    {
      "epoch": 9.099143206854345,
      "grad_norm": 1.6216009605518524,
      "learning_rate": 1.6782570585890614e-06,
      "loss": 0.9079,
      "step": 7434
    },
    {
      "epoch": 9.100367197062424,
      "grad_norm": 1.0811948828613291,
      "learning_rate": 1.677651816070558e-06,
      "loss": 0.3719,
      "step": 7435
    },
    {
      "epoch": 9.101591187270502,
      "grad_norm": 2.512103991864088,
      "learning_rate": 1.6770466275937352e-06,
      "loss": 0.7876,
      "step": 7436
    },
    {
      "epoch": 9.10281517747858,
      "grad_norm": 1.6232227889531536,
      "learning_rate": 1.6764414931983641e-06,
      "loss": 0.7179,
      "step": 7437
    },
    {
      "epoch": 9.104039167686658,
      "grad_norm": 2.4135295157746848,
      "learning_rate": 1.6758364129242121e-06,
      "loss": 0.4662,
      "step": 7438
    },
    {
      "epoch": 9.105263157894736,
      "grad_norm": 1.8476128158723835,
      "learning_rate": 1.6752313868110432e-06,
      "loss": 0.3628,
      "step": 7439
    },
    {
      "epoch": 9.106487148102815,
      "grad_norm": 1.594114832721126,
      "learning_rate": 1.6746264148986165e-06,
      "loss": 0.641,
      "step": 7440
    },
    {
      "epoch": 9.107711138310894,
      "grad_norm": 2.5750320747101427,
      "learning_rate": 1.674021497226689e-06,
      "loss": 0.5945,
      "step": 7441
    },
    {
      "epoch": 9.108935128518972,
      "grad_norm": 2.450737319402239,
      "learning_rate": 1.6734166338350134e-06,
      "loss": 0.863,
      "step": 7442
    },
    {
      "epoch": 9.11015911872705,
      "grad_norm": 2.12673826122511,
      "learning_rate": 1.6728118247633393e-06,
      "loss": 0.4729,
      "step": 7443
    },
    {
      "epoch": 9.11138310893513,
      "grad_norm": 1.3155690434591856,
      "learning_rate": 1.6722070700514127e-06,
      "loss": 0.8674,
      "step": 7444
    },
    {
      "epoch": 9.112607099143206,
      "grad_norm": 2.8336592811884853,
      "learning_rate": 1.6716023697389755e-06,
      "loss": 0.3477,
      "step": 7445
    },
    {
      "epoch": 9.113831089351285,
      "grad_norm": 1.8540035009816502,
      "learning_rate": 1.6709977238657666e-06,
      "loss": 0.5226,
      "step": 7446
    },
    {
      "epoch": 9.115055079559363,
      "grad_norm": 3.3865994622373488,
      "learning_rate": 1.6703931324715208e-06,
      "loss": 0.4336,
      "step": 7447
    },
    {
      "epoch": 9.116279069767442,
      "grad_norm": 1.9578120569813284,
      "learning_rate": 1.6697885955959708e-06,
      "loss": 0.505,
      "step": 7448
    },
    {
      "epoch": 9.11750305997552,
      "grad_norm": 2.1043329784467875,
      "learning_rate": 1.6691841132788427e-06,
      "loss": 1.1792,
      "step": 7449
    },
    {
      "epoch": 9.1187270501836,
      "grad_norm": 1.3523302343157384,
      "learning_rate": 1.6685796855598618e-06,
      "loss": 0.6974,
      "step": 7450
    },
    {
      "epoch": 9.119951040391676,
      "grad_norm": 1.8931967385821504,
      "learning_rate": 1.6679753124787487e-06,
      "loss": 0.7305,
      "step": 7451
    },
    {
      "epoch": 9.121175030599755,
      "grad_norm": 1.6041782889438643,
      "learning_rate": 1.667370994075221e-06,
      "loss": 1.0846,
      "step": 7452
    },
    {
      "epoch": 9.122399020807833,
      "grad_norm": 1.5542174544576755,
      "learning_rate": 1.6667667303889915e-06,
      "loss": 0.7024,
      "step": 7453
    },
    {
      "epoch": 9.123623011015912,
      "grad_norm": 1.5976603224951254,
      "learning_rate": 1.6661625214597713e-06,
      "loss": 1.5584,
      "step": 7454
    },
    {
      "epoch": 9.12484700122399,
      "grad_norm": 0.8167979256575747,
      "learning_rate": 1.6655583673272646e-06,
      "loss": 0.3523,
      "step": 7455
    },
    {
      "epoch": 9.12607099143207,
      "grad_norm": 1.5282192978344455,
      "learning_rate": 1.6649542680311764e-06,
      "loss": 1.1666,
      "step": 7456
    },
    {
      "epoch": 9.127294981640146,
      "grad_norm": 3.47556122134679,
      "learning_rate": 1.6643502236112057e-06,
      "loss": 0.688,
      "step": 7457
    },
    {
      "epoch": 9.128518971848225,
      "grad_norm": 1.4809810862020516,
      "learning_rate": 1.663746234107047e-06,
      "loss": 0.6849,
      "step": 7458
    },
    {
      "epoch": 9.129742962056303,
      "grad_norm": 1.7869047861955456,
      "learning_rate": 1.6631422995583919e-06,
      "loss": 0.3146,
      "step": 7459
    },
    {
      "epoch": 9.130966952264382,
      "grad_norm": 1.5092463623738277,
      "learning_rate": 1.6625384200049306e-06,
      "loss": 0.468,
      "step": 7460
    },
    {
      "epoch": 9.13219094247246,
      "grad_norm": 1.3703307038467838,
      "learning_rate": 1.6619345954863464e-06,
      "loss": 1.1508,
      "step": 7461
    },
    {
      "epoch": 9.13341493268054,
      "grad_norm": 1.7546771127264036,
      "learning_rate": 1.6613308260423206e-06,
      "loss": 1.3086,
      "step": 7462
    },
    {
      "epoch": 9.134638922888616,
      "grad_norm": 2.5693560780419906,
      "learning_rate": 1.6607271117125307e-06,
      "loss": 0.7956,
      "step": 7463
    },
    {
      "epoch": 9.135862913096695,
      "grad_norm": 1.690079125697347,
      "learning_rate": 1.6601234525366514e-06,
      "loss": 0.6317,
      "step": 7464
    },
    {
      "epoch": 9.137086903304773,
      "grad_norm": 2.0519432348606195,
      "learning_rate": 1.659519848554352e-06,
      "loss": 0.3978,
      "step": 7465
    },
    {
      "epoch": 9.138310893512852,
      "grad_norm": 1.6848486984504072,
      "learning_rate": 1.6589162998053001e-06,
      "loss": 1.2311,
      "step": 7466
    },
    {
      "epoch": 9.13953488372093,
      "grad_norm": 1.9787035664555161,
      "learning_rate": 1.6583128063291576e-06,
      "loss": 0.412,
      "step": 7467
    },
    {
      "epoch": 9.140758873929009,
      "grad_norm": 1.2732037820252815,
      "learning_rate": 1.6577093681655836e-06,
      "loss": 0.6422,
      "step": 7468
    },
    {
      "epoch": 9.141982864137088,
      "grad_norm": 1.935440037801581,
      "learning_rate": 1.6571059853542349e-06,
      "loss": 0.9097,
      "step": 7469
    },
    {
      "epoch": 9.143206854345165,
      "grad_norm": 1.2810684817820306,
      "learning_rate": 1.6565026579347635e-06,
      "loss": 0.5547,
      "step": 7470
    },
    {
      "epoch": 9.144430844553243,
      "grad_norm": 0.9179909137125755,
      "learning_rate": 1.6558993859468175e-06,
      "loss": 0.4955,
      "step": 7471
    },
    {
      "epoch": 9.145654834761322,
      "grad_norm": 1.1190733425281176,
      "learning_rate": 1.6552961694300412e-06,
      "loss": 0.478,
      "step": 7472
    },
    {
      "epoch": 9.1468788249694,
      "grad_norm": 1.9055702894710678,
      "learning_rate": 1.6546930084240764e-06,
      "loss": 0.4305,
      "step": 7473
    },
    {
      "epoch": 9.148102815177479,
      "grad_norm": 1.5927696140868572,
      "learning_rate": 1.6540899029685612e-06,
      "loss": 1.2332,
      "step": 7474
    },
    {
      "epoch": 9.149326805385558,
      "grad_norm": 1.3861421229967714,
      "learning_rate": 1.653486853103129e-06,
      "loss": 0.6678,
      "step": 7475
    },
    {
      "epoch": 9.150550795593634,
      "grad_norm": 1.6672950290542983,
      "learning_rate": 1.6528838588674079e-06,
      "loss": 0.934,
      "step": 7476
    },
    {
      "epoch": 9.151774785801713,
      "grad_norm": 2.4262442547202876,
      "learning_rate": 1.6522809203010282e-06,
      "loss": 0.9524,
      "step": 7477
    },
    {
      "epoch": 9.152998776009792,
      "grad_norm": 1.3454839149488855,
      "learning_rate": 1.6516780374436103e-06,
      "loss": 1.1311,
      "step": 7478
    },
    {
      "epoch": 9.15422276621787,
      "grad_norm": 1.0676020976357605,
      "learning_rate": 1.6510752103347743e-06,
      "loss": 0.5386,
      "step": 7479
    },
    {
      "epoch": 9.155446756425949,
      "grad_norm": 1.2432936279532507,
      "learning_rate": 1.6504724390141353e-06,
      "loss": 0.718,
      "step": 7480
    },
    {
      "epoch": 9.156670746634028,
      "grad_norm": 1.970538873780607,
      "learning_rate": 1.6498697235213052e-06,
      "loss": 0.4105,
      "step": 7481
    },
    {
      "epoch": 9.157894736842104,
      "grad_norm": 1.1474040328986332,
      "learning_rate": 1.6492670638958924e-06,
      "loss": 0.5051,
      "step": 7482
    },
    {
      "epoch": 9.159118727050183,
      "grad_norm": 1.057407458276689,
      "learning_rate": 1.6486644601775026e-06,
      "loss": 0.6816,
      "step": 7483
    },
    {
      "epoch": 9.160342717258262,
      "grad_norm": 1.2656561484693738,
      "learning_rate": 1.6480619124057346e-06,
      "loss": 1.3327,
      "step": 7484
    },
    {
      "epoch": 9.16156670746634,
      "grad_norm": 1.0790203645680496,
      "learning_rate": 1.647459420620186e-06,
      "loss": 0.847,
      "step": 7485
    },
    {
      "epoch": 9.162790697674419,
      "grad_norm": 1.1764976300460082,
      "learning_rate": 1.6468569848604522e-06,
      "loss": 0.6854,
      "step": 7486
    },
    {
      "epoch": 9.164014687882498,
      "grad_norm": 1.496092009105994,
      "learning_rate": 1.6462546051661216e-06,
      "loss": 0.3599,
      "step": 7487
    },
    {
      "epoch": 9.165238678090576,
      "grad_norm": 1.4594527750186645,
      "learning_rate": 1.6456522815767803e-06,
      "loss": 0.4296,
      "step": 7488
    },
    {
      "epoch": 9.166462668298653,
      "grad_norm": 3.3226275455386824,
      "learning_rate": 1.6450500141320106e-06,
      "loss": 0.7665,
      "step": 7489
    },
    {
      "epoch": 9.167686658506732,
      "grad_norm": 2.2155004292618536,
      "learning_rate": 1.6444478028713922e-06,
      "loss": 1.0813,
      "step": 7490
    },
    {
      "epoch": 9.16891064871481,
      "grad_norm": 1.7777661521953578,
      "learning_rate": 1.6438456478345e-06,
      "loss": 0.7382,
      "step": 7491
    },
    {
      "epoch": 9.170134638922889,
      "grad_norm": 1.8536180788219556,
      "learning_rate": 1.643243549060905e-06,
      "loss": 0.4868,
      "step": 7492
    },
    {
      "epoch": 9.171358629130967,
      "grad_norm": 1.661181922160972,
      "learning_rate": 1.6426415065901752e-06,
      "loss": 1.3629,
      "step": 7493
    },
    {
      "epoch": 9.172582619339046,
      "grad_norm": 1.572638042219822,
      "learning_rate": 1.642039520461873e-06,
      "loss": 1.1767,
      "step": 7494
    },
    {
      "epoch": 9.173806609547123,
      "grad_norm": 2.841093339263793,
      "learning_rate": 1.6414375907155618e-06,
      "loss": 0.3269,
      "step": 7495
    },
    {
      "epoch": 9.175030599755202,
      "grad_norm": 1.4145488993828061,
      "learning_rate": 1.6408357173907963e-06,
      "loss": 0.7118,
      "step": 7496
    },
    {
      "epoch": 9.17625458996328,
      "grad_norm": 1.9912434461729371,
      "learning_rate": 1.6402339005271293e-06,
      "loss": 1.1239,
      "step": 7497
    },
    {
      "epoch": 9.177478580171359,
      "grad_norm": 1.1631664048328223,
      "learning_rate": 1.6396321401641097e-06,
      "loss": 0.6025,
      "step": 7498
    },
    {
      "epoch": 9.178702570379437,
      "grad_norm": 2.25539531702093,
      "learning_rate": 1.6390304363412844e-06,
      "loss": 0.4195,
      "step": 7499
    },
    {
      "epoch": 9.179926560587516,
      "grad_norm": 2.6269737645119995,
      "learning_rate": 1.638428789098194e-06,
      "loss": 0.4524,
      "step": 7500
    },
    {
      "epoch": 9.181150550795593,
      "grad_norm": 1.5935420819282342,
      "learning_rate": 1.6378271984743776e-06,
      "loss": 0.4588,
      "step": 7501
    },
    {
      "epoch": 9.182374541003671,
      "grad_norm": 2.1668346272573977,
      "learning_rate": 1.637225664509367e-06,
      "loss": 0.3788,
      "step": 7502
    },
    {
      "epoch": 9.18359853121175,
      "grad_norm": 1.4070613567009012,
      "learning_rate": 1.6366241872426958e-06,
      "loss": 0.8929,
      "step": 7503
    },
    {
      "epoch": 9.184822521419829,
      "grad_norm": 1.8299521014510098,
      "learning_rate": 1.63602276671389e-06,
      "loss": 1.1895,
      "step": 7504
    },
    {
      "epoch": 9.186046511627907,
      "grad_norm": 1.5112584911977351,
      "learning_rate": 1.6354214029624719e-06,
      "loss": 0.5261,
      "step": 7505
    },
    {
      "epoch": 9.187270501835986,
      "grad_norm": 1.7879417165329936,
      "learning_rate": 1.6348200960279615e-06,
      "loss": 0.6133,
      "step": 7506
    },
    {
      "epoch": 9.188494492044065,
      "grad_norm": 1.619224982556922,
      "learning_rate": 1.6342188459498734e-06,
      "loss": 0.6207,
      "step": 7507
    },
    {
      "epoch": 9.189718482252141,
      "grad_norm": 2.168237594354624,
      "learning_rate": 1.633617652767721e-06,
      "loss": 0.838,
      "step": 7508
    },
    {
      "epoch": 9.19094247246022,
      "grad_norm": 1.2735278005213808,
      "learning_rate": 1.6330165165210121e-06,
      "loss": 0.766,
      "step": 7509
    },
    {
      "epoch": 9.192166462668299,
      "grad_norm": 2.016123500966144,
      "learning_rate": 1.6324154372492517e-06,
      "loss": 0.3152,
      "step": 7510
    },
    {
      "epoch": 9.193390452876377,
      "grad_norm": 2.246387496156874,
      "learning_rate": 1.631814414991938e-06,
      "loss": 0.4233,
      "step": 7511
    },
    {
      "epoch": 9.194614443084456,
      "grad_norm": 1.214001098956161,
      "learning_rate": 1.6312134497885713e-06,
      "loss": 0.8433,
      "step": 7512
    },
    {
      "epoch": 9.195838433292534,
      "grad_norm": 1.4397466138779782,
      "learning_rate": 1.6306125416786428e-06,
      "loss": 0.4553,
      "step": 7513
    },
    {
      "epoch": 9.197062423500611,
      "grad_norm": 2.4739125519414435,
      "learning_rate": 1.6300116907016422e-06,
      "loss": 0.4343,
      "step": 7514
    },
    {
      "epoch": 9.19828641370869,
      "grad_norm": 0.9115821229459217,
      "learning_rate": 1.629410896897055e-06,
      "loss": 0.3468,
      "step": 7515
    },
    {
      "epoch": 9.199510403916769,
      "grad_norm": 2.274049998722299,
      "learning_rate": 1.6288101603043643e-06,
      "loss": 1.4134,
      "step": 7516
    },
    {
      "epoch": 9.200734394124847,
      "grad_norm": 1.3485728559554444,
      "learning_rate": 1.6282094809630475e-06,
      "loss": 0.7404,
      "step": 7517
    },
    {
      "epoch": 9.201958384332926,
      "grad_norm": 1.4903866773690893,
      "learning_rate": 1.6276088589125791e-06,
      "loss": 0.4067,
      "step": 7518
    },
    {
      "epoch": 9.203182374541004,
      "grad_norm": 2.0463047464912774,
      "learning_rate": 1.6270082941924303e-06,
      "loss": 0.9648,
      "step": 7519
    },
    {
      "epoch": 9.204406364749081,
      "grad_norm": 2.825413854895487,
      "learning_rate": 1.6264077868420658e-06,
      "loss": 0.4709,
      "step": 7520
    },
    {
      "epoch": 9.20563035495716,
      "grad_norm": 1.8617516938328105,
      "learning_rate": 1.6258073369009516e-06,
      "loss": 0.5198,
      "step": 7521
    },
    {
      "epoch": 9.206854345165238,
      "grad_norm": 2.301137925515693,
      "learning_rate": 1.6252069444085455e-06,
      "loss": 0.8644,
      "step": 7522
    },
    {
      "epoch": 9.208078335373317,
      "grad_norm": 2.832824186299885,
      "learning_rate": 1.6246066094043035e-06,
      "loss": 0.744,
      "step": 7523
    },
    {
      "epoch": 9.209302325581396,
      "grad_norm": 1.1655384786510916,
      "learning_rate": 1.6240063319276767e-06,
      "loss": 0.4835,
      "step": 7524
    },
    {
      "epoch": 9.210526315789474,
      "grad_norm": 1.8703431515203315,
      "learning_rate": 1.6234061120181144e-06,
      "loss": 0.4288,
      "step": 7525
    },
    {
      "epoch": 9.211750305997551,
      "grad_norm": 1.2251328249867697,
      "learning_rate": 1.6228059497150597e-06,
      "loss": 0.5594,
      "step": 7526
    },
    {
      "epoch": 9.21297429620563,
      "grad_norm": 1.2744250548482172,
      "learning_rate": 1.6222058450579536e-06,
      "loss": 1.0901,
      "step": 7527
    },
    {
      "epoch": 9.214198286413708,
      "grad_norm": 2.152058977683754,
      "learning_rate": 1.6216057980862332e-06,
      "loss": 1.3391,
      "step": 7528
    },
    {
      "epoch": 9.215422276621787,
      "grad_norm": 1.0959048755652003,
      "learning_rate": 1.6210058088393293e-06,
      "loss": 0.6638,
      "step": 7529
    },
    {
      "epoch": 9.216646266829866,
      "grad_norm": 1.8417302513053093,
      "learning_rate": 1.6204058773566742e-06,
      "loss": 0.483,
      "step": 7530
    },
    {
      "epoch": 9.217870257037944,
      "grad_norm": 1.8081453247114678,
      "learning_rate": 1.6198060036776907e-06,
      "loss": 0.6969,
      "step": 7531
    },
    {
      "epoch": 9.219094247246023,
      "grad_norm": 1.9393985557618256,
      "learning_rate": 1.6192061878418008e-06,
      "loss": 0.7779,
      "step": 7532
    },
    {
      "epoch": 9.2203182374541,
      "grad_norm": 1.530271845167981,
      "learning_rate": 1.6186064298884223e-06,
      "loss": 1.5569,
      "step": 7533
    },
    {
      "epoch": 9.221542227662178,
      "grad_norm": 2.015975167197436,
      "learning_rate": 1.6180067298569697e-06,
      "loss": 0.5764,
      "step": 7534
    },
    {
      "epoch": 9.222766217870257,
      "grad_norm": 1.7276944630303201,
      "learning_rate": 1.6174070877868523e-06,
      "loss": 0.6922,
      "step": 7535
    },
    {
      "epoch": 9.223990208078336,
      "grad_norm": 1.5077534472547494,
      "learning_rate": 1.6168075037174769e-06,
      "loss": 0.4993,
      "step": 7536
    },
    {
      "epoch": 9.225214198286414,
      "grad_norm": 2.5095281234550217,
      "learning_rate": 1.6162079776882447e-06,
      "loss": 0.5291,
      "step": 7537
    },
    {
      "epoch": 9.226438188494493,
      "grad_norm": 1.647393195631156,
      "learning_rate": 1.615608509738556e-06,
      "loss": 0.5555,
      "step": 7538
    },
    {
      "epoch": 9.22766217870257,
      "grad_norm": 1.149892454762325,
      "learning_rate": 1.6150090999078055e-06,
      "loss": 0.8865,
      "step": 7539
    },
    {
      "epoch": 9.228886168910648,
      "grad_norm": 1.1509723332673358,
      "learning_rate": 1.614409748235383e-06,
      "loss": 0.6732,
      "step": 7540
    },
    {
      "epoch": 9.230110159118727,
      "grad_norm": 1.387813686980432,
      "learning_rate": 1.6138104547606764e-06,
      "loss": 0.8144,
      "step": 7541
    },
    {
      "epoch": 9.231334149326806,
      "grad_norm": 2.134762082610977,
      "learning_rate": 1.613211219523068e-06,
      "loss": 0.4723,
      "step": 7542
    },
    {
      "epoch": 9.232558139534884,
      "grad_norm": 1.0567943190178304,
      "learning_rate": 1.6126120425619393e-06,
      "loss": 0.5656,
      "step": 7543
    },
    {
      "epoch": 9.233782129742963,
      "grad_norm": 1.620666753114307,
      "learning_rate": 1.612012923916665e-06,
      "loss": 0.854,
      "step": 7544
    },
    {
      "epoch": 9.23500611995104,
      "grad_norm": 2.757794154893919,
      "learning_rate": 1.6114138636266163e-06,
      "loss": 0.8297,
      "step": 7545
    },
    {
      "epoch": 9.236230110159118,
      "grad_norm": 1.6286892994301412,
      "learning_rate": 1.6108148617311615e-06,
      "loss": 0.3086,
      "step": 7546
    },
    {
      "epoch": 9.237454100367197,
      "grad_norm": 2.3588374642945147,
      "learning_rate": 1.6102159182696658e-06,
      "loss": 0.4605,
      "step": 7547
    },
    {
      "epoch": 9.238678090575275,
      "grad_norm": 1.6066203357204345,
      "learning_rate": 1.6096170332814895e-06,
      "loss": 0.5114,
      "step": 7548
    },
    {
      "epoch": 9.239902080783354,
      "grad_norm": 1.6109852211135527,
      "learning_rate": 1.6090182068059878e-06,
      "loss": 0.5553,
      "step": 7549
    },
    {
      "epoch": 9.241126070991433,
      "grad_norm": 2.9565928052739467,
      "learning_rate": 1.6084194388825136e-06,
      "loss": 0.4126,
      "step": 7550
    },
    {
      "epoch": 9.24235006119951,
      "grad_norm": 1.1458065398163393,
      "learning_rate": 1.6078207295504164e-06,
      "loss": 0.5709,
      "step": 7551
    },
    {
      "epoch": 9.243574051407588,
      "grad_norm": 1.941451908961006,
      "learning_rate": 1.6072220788490411e-06,
      "loss": 0.5367,
      "step": 7552
    },
    {
      "epoch": 9.244798041615667,
      "grad_norm": 1.4254214792588833,
      "learning_rate": 1.6066234868177289e-06,
      "loss": 0.5486,
      "step": 7553
    },
    {
      "epoch": 9.246022031823745,
      "grad_norm": 1.0491402201012068,
      "learning_rate": 1.6060249534958162e-06,
      "loss": 0.5489,
      "step": 7554
    },
    {
      "epoch": 9.247246022031824,
      "grad_norm": 1.0006747885443226,
      "learning_rate": 1.605426478922637e-06,
      "loss": 0.3996,
      "step": 7555
    },
    {
      "epoch": 9.248470012239903,
      "grad_norm": 1.3109338999615479,
      "learning_rate": 1.6048280631375208e-06,
      "loss": 0.6088,
      "step": 7556
    },
    {
      "epoch": 9.249694002447981,
      "grad_norm": 1.488472748994658,
      "learning_rate": 1.604229706179794e-06,
      "loss": 1.2609,
      "step": 7557
    },
    {
      "epoch": 9.250917992656058,
      "grad_norm": 1.454662542218424,
      "learning_rate": 1.6036314080887775e-06,
      "loss": 0.6536,
      "step": 7558
    },
    {
      "epoch": 9.252141982864137,
      "grad_norm": 1.5446260804534628,
      "learning_rate": 1.6030331689037886e-06,
      "loss": 0.516,
      "step": 7559
    },
    {
      "epoch": 9.253365973072215,
      "grad_norm": 2.257366593564765,
      "learning_rate": 1.6024349886641426e-06,
      "loss": 0.5224,
      "step": 7560
    },
    {
      "epoch": 9.254589963280294,
      "grad_norm": 1.932404285947136,
      "learning_rate": 1.6018368674091494e-06,
      "loss": 0.7649,
      "step": 7561
    },
    {
      "epoch": 9.255813953488373,
      "grad_norm": 2.619589126290217,
      "learning_rate": 1.6012388051781152e-06,
      "loss": 0.5088,
      "step": 7562
    },
    {
      "epoch": 9.257037943696451,
      "grad_norm": 2.151308208028464,
      "learning_rate": 1.600640802010342e-06,
      "loss": 0.5521,
      "step": 7563
    },
    {
      "epoch": 9.258261933904528,
      "grad_norm": 3.4344712032036613,
      "learning_rate": 1.6000428579451294e-06,
      "loss": 0.2819,
      "step": 7564
    },
    {
      "epoch": 9.259485924112607,
      "grad_norm": 2.303826968527441,
      "learning_rate": 1.5994449730217716e-06,
      "loss": 0.4927,
      "step": 7565
    },
    {
      "epoch": 9.260709914320685,
      "grad_norm": 1.8999382166168015,
      "learning_rate": 1.598847147279559e-06,
      "loss": 0.7327,
      "step": 7566
    },
    {
      "epoch": 9.261933904528764,
      "grad_norm": 1.4394114290656885,
      "learning_rate": 1.598249380757779e-06,
      "loss": 0.7858,
      "step": 7567
    },
    {
      "epoch": 9.263157894736842,
      "grad_norm": 2.980991453742353,
      "learning_rate": 1.5976516734957136e-06,
      "loss": 0.3119,
      "step": 7568
    },
    {
      "epoch": 9.264381884944921,
      "grad_norm": 3.037337445547191,
      "learning_rate": 1.597054025532643e-06,
      "loss": 0.4146,
      "step": 7569
    },
    {
      "epoch": 9.265605875152998,
      "grad_norm": 1.655967912464092,
      "learning_rate": 1.5964564369078423e-06,
      "loss": 1.5209,
      "step": 7570
    },
    {
      "epoch": 9.266829865361077,
      "grad_norm": 2.5232658104938235,
      "learning_rate": 1.5958589076605823e-06,
      "loss": 0.4786,
      "step": 7571
    },
    {
      "epoch": 9.268053855569155,
      "grad_norm": 1.930100952557493,
      "learning_rate": 1.5952614378301306e-06,
      "loss": 0.9969,
      "step": 7572
    },
    {
      "epoch": 9.269277845777234,
      "grad_norm": 1.5642594530504352,
      "learning_rate": 1.5946640274557512e-06,
      "loss": 0.5207,
      "step": 7573
    },
    {
      "epoch": 9.270501835985312,
      "grad_norm": 1.94343515820504,
      "learning_rate": 1.5940666765767032e-06,
      "loss": 1.0092,
      "step": 7574
    },
    {
      "epoch": 9.271725826193391,
      "grad_norm": 1.1717545439909676,
      "learning_rate": 1.593469385232243e-06,
      "loss": 0.5751,
      "step": 7575
    },
    {
      "epoch": 9.27294981640147,
      "grad_norm": 2.269651908477515,
      "learning_rate": 1.5928721534616197e-06,
      "loss": 0.4756,
      "step": 7576
    },
    {
      "epoch": 9.274173806609546,
      "grad_norm": 2.058980541305633,
      "learning_rate": 1.5922749813040854e-06,
      "loss": 0.5096,
      "step": 7577
    },
    {
      "epoch": 9.275397796817625,
      "grad_norm": 1.4028528032439145,
      "learning_rate": 1.5916778687988816e-06,
      "loss": 0.3625,
      "step": 7578
    },
    {
      "epoch": 9.276621787025704,
      "grad_norm": 1.7491620856876444,
      "learning_rate": 1.591080815985248e-06,
      "loss": 1.3473,
      "step": 7579
    },
    {
      "epoch": 9.277845777233782,
      "grad_norm": 1.79348763957945,
      "learning_rate": 1.5904838229024217e-06,
      "loss": 0.7476,
      "step": 7580
    },
    {
      "epoch": 9.279069767441861,
      "grad_norm": 1.7649929363549752,
      "learning_rate": 1.5898868895896336e-06,
      "loss": 0.9873,
      "step": 7581
    },
    {
      "epoch": 9.28029375764994,
      "grad_norm": 1.361732748978625,
      "learning_rate": 1.5892900160861136e-06,
      "loss": 0.4355,
      "step": 7582
    },
    {
      "epoch": 9.281517747858016,
      "grad_norm": 3.5326688660220604,
      "learning_rate": 1.5886932024310852e-06,
      "loss": 0.517,
      "step": 7583
    },
    {
      "epoch": 9.282741738066095,
      "grad_norm": 0.9090451174756919,
      "learning_rate": 1.5880964486637694e-06,
      "loss": 0.4141,
      "step": 7584
    },
    {
      "epoch": 9.283965728274174,
      "grad_norm": 2.339479188374513,
      "learning_rate": 1.5874997548233806e-06,
      "loss": 0.5988,
      "step": 7585
    },
    {
      "epoch": 9.285189718482252,
      "grad_norm": 2.1282053580621008,
      "learning_rate": 1.5869031209491342e-06,
      "loss": 0.5432,
      "step": 7586
    },
    {
      "epoch": 9.286413708690331,
      "grad_norm": 1.1175473069365263,
      "learning_rate": 1.5863065470802372e-06,
      "loss": 0.4507,
      "step": 7587
    },
    {
      "epoch": 9.28763769889841,
      "grad_norm": 2.135696524267453,
      "learning_rate": 1.5857100332558941e-06,
      "loss": 0.3071,
      "step": 7588
    },
    {
      "epoch": 9.288861689106486,
      "grad_norm": 2.162804181738649,
      "learning_rate": 1.5851135795153055e-06,
      "loss": 0.5154,
      "step": 7589
    },
    {
      "epoch": 9.290085679314565,
      "grad_norm": 3.252329033501468,
      "learning_rate": 1.584517185897669e-06,
      "loss": 0.3431,
      "step": 7590
    },
    {
      "epoch": 9.291309669522644,
      "grad_norm": 1.6688263547336875,
      "learning_rate": 1.583920852442177e-06,
      "loss": 0.4992,
      "step": 7591
    },
    {
      "epoch": 9.292533659730722,
      "grad_norm": 1.8547244979622948,
      "learning_rate": 1.5833245791880187e-06,
      "loss": 0.3426,
      "step": 7592
    },
    {
      "epoch": 9.2937576499388,
      "grad_norm": 1.5638078698115514,
      "learning_rate": 1.5827283661743777e-06,
      "loss": 0.4723,
      "step": 7593
    },
    {
      "epoch": 9.29498164014688,
      "grad_norm": 1.6968622122528758,
      "learning_rate": 1.5821322134404352e-06,
      "loss": 1.1942,
      "step": 7594
    },
    {
      "epoch": 9.296205630354958,
      "grad_norm": 1.342071760334371,
      "learning_rate": 1.58153612102537e-06,
      "loss": 0.5574,
      "step": 7595
    },
    {
      "epoch": 9.297429620563035,
      "grad_norm": 2.7046976116206185,
      "learning_rate": 1.580940088968353e-06,
      "loss": 0.3689,
      "step": 7596
    },
    {
      "epoch": 9.298653610771114,
      "grad_norm": 1.3212047157697402,
      "learning_rate": 1.5803441173085544e-06,
      "loss": 0.4671,
      "step": 7597
    },
    {
      "epoch": 9.299877600979192,
      "grad_norm": 1.4047847740027812,
      "learning_rate": 1.5797482060851379e-06,
      "loss": 0.4333,
      "step": 7598
    },
    {
      "epoch": 9.30110159118727,
      "grad_norm": 2.390689290822738,
      "learning_rate": 1.5791523553372662e-06,
      "loss": 1.0185,
      "step": 7599
    },
    {
      "epoch": 9.30232558139535,
      "grad_norm": 3.0831345022167844,
      "learning_rate": 1.5785565651040958e-06,
      "loss": 0.4624,
      "step": 7600
    },
    {
      "epoch": 9.303549571603428,
      "grad_norm": 2.138639495593019,
      "learning_rate": 1.5779608354247806e-06,
      "loss": 0.6065,
      "step": 7601
    },
    {
      "epoch": 9.304773561811505,
      "grad_norm": 2.1383068129112615,
      "learning_rate": 1.5773651663384667e-06,
      "loss": 1.0734,
      "step": 7602
    },
    {
      "epoch": 9.305997552019583,
      "grad_norm": 2.5168533224993315,
      "learning_rate": 1.576769557884303e-06,
      "loss": 0.4818,
      "step": 7603
    },
    {
      "epoch": 9.307221542227662,
      "grad_norm": 1.4050506152240345,
      "learning_rate": 1.5761740101014294e-06,
      "loss": 1.3135,
      "step": 7604
    },
    {
      "epoch": 9.30844553243574,
      "grad_norm": 1.2115716049683134,
      "learning_rate": 1.5755785230289824e-06,
      "loss": 0.5866,
      "step": 7605
    },
    {
      "epoch": 9.30966952264382,
      "grad_norm": 1.192425409171754,
      "learning_rate": 1.5749830967060958e-06,
      "loss": 1.0432,
      "step": 7606
    },
    {
      "epoch": 9.310893512851898,
      "grad_norm": 1.2759034800590605,
      "learning_rate": 1.5743877311718981e-06,
      "loss": 0.4446,
      "step": 7607
    },
    {
      "epoch": 9.312117503059975,
      "grad_norm": 1.7673772835308912,
      "learning_rate": 1.5737924264655155e-06,
      "loss": 0.9279,
      "step": 7608
    },
    {
      "epoch": 9.313341493268053,
      "grad_norm": 2.683291465314498,
      "learning_rate": 1.573197182626069e-06,
      "loss": 0.5062,
      "step": 7609
    },
    {
      "epoch": 9.314565483476132,
      "grad_norm": 2.016596997960212,
      "learning_rate": 1.5726019996926762e-06,
      "loss": 0.3911,
      "step": 7610
    },
    {
      "epoch": 9.31578947368421,
      "grad_norm": 1.712069436420586,
      "learning_rate": 1.5720068777044479e-06,
      "loss": 0.5497,
      "step": 7611
    },
    {
      "epoch": 9.31701346389229,
      "grad_norm": 1.4795882317143123,
      "learning_rate": 1.5714118167004967e-06,
      "loss": 0.5268,
      "step": 7612
    },
    {
      "epoch": 9.318237454100368,
      "grad_norm": 1.7267017365035329,
      "learning_rate": 1.5708168167199264e-06,
      "loss": 0.5818,
      "step": 7613
    },
    {
      "epoch": 9.319461444308445,
      "grad_norm": 2.34799663095906,
      "learning_rate": 1.5702218778018377e-06,
      "loss": 0.5213,
      "step": 7614
    },
    {
      "epoch": 9.320685434516523,
      "grad_norm": 1.9654454191284825,
      "learning_rate": 1.5696269999853278e-06,
      "loss": 1.1547,
      "step": 7615
    },
    {
      "epoch": 9.321909424724602,
      "grad_norm": 1.7540083018628057,
      "learning_rate": 1.5690321833094907e-06,
      "loss": 0.5873,
      "step": 7616
    },
    {
      "epoch": 9.32313341493268,
      "grad_norm": 2.2219159423055546,
      "learning_rate": 1.568437427813415e-06,
      "loss": 0.291,
      "step": 7617
    },
    {
      "epoch": 9.32435740514076,
      "grad_norm": 1.301144817391721,
      "learning_rate": 1.567842733536186e-06,
      "loss": 0.4112,
      "step": 7618
    },
    {
      "epoch": 9.325581395348838,
      "grad_norm": 2.3590202071159143,
      "learning_rate": 1.5672481005168851e-06,
      "loss": 0.7158,
      "step": 7619
    },
    {
      "epoch": 9.326805385556916,
      "grad_norm": 1.8427081582144975,
      "learning_rate": 1.5666535287945878e-06,
      "loss": 0.4008,
      "step": 7620
    },
    {
      "epoch": 9.328029375764993,
      "grad_norm": 1.808263878031781,
      "learning_rate": 1.5660590184083691e-06,
      "loss": 0.4607,
      "step": 7621
    },
    {
      "epoch": 9.329253365973072,
      "grad_norm": 0.8229265734992094,
      "learning_rate": 1.565464569397298e-06,
      "loss": 0.435,
      "step": 7622
    },
    {
      "epoch": 9.33047735618115,
      "grad_norm": 1.6558929204357908,
      "learning_rate": 1.5648701818004383e-06,
      "loss": 0.885,
      "step": 7623
    },
    {
      "epoch": 9.331701346389229,
      "grad_norm": 1.5215383471852133,
      "learning_rate": 1.564275855656851e-06,
      "loss": 0.5739,
      "step": 7624
    },
    {
      "epoch": 9.332925336597308,
      "grad_norm": 1.7930306291363125,
      "learning_rate": 1.5636815910055936e-06,
      "loss": 0.505,
      "step": 7625
    },
    {
      "epoch": 9.334149326805386,
      "grad_norm": 2.376361673609101,
      "learning_rate": 1.5630873878857193e-06,
      "loss": 0.6067,
      "step": 7626
    },
    {
      "epoch": 9.335373317013463,
      "grad_norm": 2.6036869894829824,
      "learning_rate": 1.5624932463362762e-06,
      "loss": 0.8309,
      "step": 7627
    },
    {
      "epoch": 9.336597307221542,
      "grad_norm": 2.7087013246130662,
      "learning_rate": 1.5618991663963091e-06,
      "loss": 1.0271,
      "step": 7628
    },
    {
      "epoch": 9.33782129742962,
      "grad_norm": 1.6421868824184356,
      "learning_rate": 1.561305148104859e-06,
      "loss": 1.3367,
      "step": 7629
    },
    {
      "epoch": 9.339045287637699,
      "grad_norm": 1.3553932528567398,
      "learning_rate": 1.5607111915009634e-06,
      "loss": 0.5097,
      "step": 7630
    },
    {
      "epoch": 9.340269277845778,
      "grad_norm": 1.0305316910906297,
      "learning_rate": 1.5601172966236536e-06,
      "loss": 0.5325,
      "step": 7631
    },
    {
      "epoch": 9.341493268053856,
      "grad_norm": 2.409073955333783,
      "learning_rate": 1.5595234635119587e-06,
      "loss": 0.6131,
      "step": 7632
    },
    {
      "epoch": 9.342717258261933,
      "grad_norm": 1.077794533422121,
      "learning_rate": 1.5589296922049025e-06,
      "loss": 0.6414,
      "step": 7633
    },
    {
      "epoch": 9.343941248470012,
      "grad_norm": 1.8536646905538017,
      "learning_rate": 1.5583359827415067e-06,
      "loss": 0.6058,
      "step": 7634
    },
    {
      "epoch": 9.34516523867809,
      "grad_norm": 2.8214767410696875,
      "learning_rate": 1.5577423351607873e-06,
      "loss": 0.4033,
      "step": 7635
    },
    {
      "epoch": 9.346389228886169,
      "grad_norm": 2.186990576758635,
      "learning_rate": 1.5571487495017565e-06,
      "loss": 0.9741,
      "step": 7636
    },
    {
      "epoch": 9.347613219094248,
      "grad_norm": 1.630383814889504,
      "learning_rate": 1.556555225803422e-06,
      "loss": 0.9119,
      "step": 7637
    },
    {
      "epoch": 9.348837209302326,
      "grad_norm": 1.8431071554710503,
      "learning_rate": 1.5559617641047886e-06,
      "loss": 0.9649,
      "step": 7638
    },
    {
      "epoch": 9.350061199510403,
      "grad_norm": 1.782870083823572,
      "learning_rate": 1.5553683644448574e-06,
      "loss": 0.7107,
      "step": 7639
    },
    {
      "epoch": 9.351285189718482,
      "grad_norm": 1.6951479853838531,
      "learning_rate": 1.5547750268626228e-06,
      "loss": 1.136,
      "step": 7640
    },
    {
      "epoch": 9.35250917992656,
      "grad_norm": 1.3836077206506656,
      "learning_rate": 1.5541817513970758e-06,
      "loss": 0.7244,
      "step": 7641
    },
    {
      "epoch": 9.353733170134639,
      "grad_norm": 1.769480744229076,
      "learning_rate": 1.5535885380872073e-06,
      "loss": 0.5297,
      "step": 7642
    },
    {
      "epoch": 9.354957160342718,
      "grad_norm": 1.896086784299193,
      "learning_rate": 1.5529953869719993e-06,
      "loss": 0.4831,
      "step": 7643
    },
    {
      "epoch": 9.356181150550796,
      "grad_norm": 1.2215784655024247,
      "learning_rate": 1.5524022980904318e-06,
      "loss": 0.7627,
      "step": 7644
    },
    {
      "epoch": 9.357405140758875,
      "grad_norm": 1.8307133723968905,
      "learning_rate": 1.5518092714814804e-06,
      "loss": 0.5854,
      "step": 7645
    },
    {
      "epoch": 9.358629130966952,
      "grad_norm": 1.9894250327069032,
      "learning_rate": 1.5512163071841158e-06,
      "loss": 0.6018,
      "step": 7646
    },
    {
      "epoch": 9.35985312117503,
      "grad_norm": 1.4087640779558612,
      "learning_rate": 1.5506234052373068e-06,
      "loss": 1.1184,
      "step": 7647
    },
    {
      "epoch": 9.361077111383109,
      "grad_norm": 1.5452525571186473,
      "learning_rate": 1.5500305656800167e-06,
      "loss": 0.4994,
      "step": 7648
    },
    {
      "epoch": 9.362301101591187,
      "grad_norm": 2.4439612013042398,
      "learning_rate": 1.5494377885512035e-06,
      "loss": 0.4702,
      "step": 7649
    },
    {
      "epoch": 9.363525091799266,
      "grad_norm": 2.3075053615597696,
      "learning_rate": 1.548845073889822e-06,
      "loss": 0.9011,
      "step": 7650
    },
    {
      "epoch": 9.364749082007345,
      "grad_norm": 1.6290569241892576,
      "learning_rate": 1.5482524217348254e-06,
      "loss": 0.5753,
      "step": 7651
    },
    {
      "epoch": 9.365973072215422,
      "grad_norm": 1.517145727666439,
      "learning_rate": 1.547659832125159e-06,
      "loss": 0.7207,
      "step": 7652
    },
    {
      "epoch": 9.3671970624235,
      "grad_norm": 2.4129736285474426,
      "learning_rate": 1.5470673050997664e-06,
      "loss": 0.9774,
      "step": 7653
    },
    {
      "epoch": 9.368421052631579,
      "grad_norm": 1.3316836891180373,
      "learning_rate": 1.546474840697585e-06,
      "loss": 0.362,
      "step": 7654
    },
    {
      "epoch": 9.369645042839657,
      "grad_norm": 1.3709060597558165,
      "learning_rate": 1.5458824389575505e-06,
      "loss": 0.5415,
      "step": 7655
    },
    {
      "epoch": 9.370869033047736,
      "grad_norm": 1.4913774635847235,
      "learning_rate": 1.5452900999185935e-06,
      "loss": 0.5649,
      "step": 7656
    },
    {
      "epoch": 9.372093023255815,
      "grad_norm": 1.4899986165819885,
      "learning_rate": 1.5446978236196402e-06,
      "loss": 0.576,
      "step": 7657
    },
    {
      "epoch": 9.373317013463891,
      "grad_norm": 1.7297256080676775,
      "learning_rate": 1.5441056100996121e-06,
      "loss": 1.805,
      "step": 7658
    },
    {
      "epoch": 9.37454100367197,
      "grad_norm": 1.3407580589767427,
      "learning_rate": 1.5435134593974272e-06,
      "loss": 0.5036,
      "step": 7659
    },
    {
      "epoch": 9.375764993880049,
      "grad_norm": 1.5638115718246837,
      "learning_rate": 1.5429213715520004e-06,
      "loss": 0.5027,
      "step": 7660
    },
    {
      "epoch": 9.376988984088127,
      "grad_norm": 2.393949686728973,
      "learning_rate": 1.542329346602241e-06,
      "loss": 0.4133,
      "step": 7661
    },
    {
      "epoch": 9.378212974296206,
      "grad_norm": 1.5164625647302858,
      "learning_rate": 1.541737384587055e-06,
      "loss": 1.1655,
      "step": 7662
    },
    {
      "epoch": 9.379436964504285,
      "grad_norm": 2.2362627623633293,
      "learning_rate": 1.5411454855453428e-06,
      "loss": 0.9901,
      "step": 7663
    },
    {
      "epoch": 9.380660954712361,
      "grad_norm": 1.9743227052685561,
      "learning_rate": 1.5405536495160036e-06,
      "loss": 0.536,
      "step": 7664
    },
    {
      "epoch": 9.38188494492044,
      "grad_norm": 1.6836518934439453,
      "learning_rate": 1.53996187653793e-06,
      "loss": 0.9997,
      "step": 7665
    },
    {
      "epoch": 9.383108935128519,
      "grad_norm": 1.777671918199954,
      "learning_rate": 1.539370166650011e-06,
      "loss": 0.5481,
      "step": 7666
    },
    {
      "epoch": 9.384332925336597,
      "grad_norm": 1.4381596955807274,
      "learning_rate": 1.5387785198911298e-06,
      "loss": 0.6725,
      "step": 7667
    },
    {
      "epoch": 9.385556915544676,
      "grad_norm": 1.8738116203087567,
      "learning_rate": 1.538186936300171e-06,
      "loss": 0.5502,
      "step": 7668
    },
    {
      "epoch": 9.386780905752754,
      "grad_norm": 1.7637689518555508,
      "learning_rate": 1.5375954159160086e-06,
      "loss": 0.6463,
      "step": 7669
    },
    {
      "epoch": 9.388004895960833,
      "grad_norm": 2.209167061231007,
      "learning_rate": 1.537003958777516e-06,
      "loss": 0.4308,
      "step": 7670
    },
    {
      "epoch": 9.38922888616891,
      "grad_norm": 1.3128577686424403,
      "learning_rate": 1.536412564923561e-06,
      "loss": 0.3686,
      "step": 7671
    },
    {
      "epoch": 9.390452876376989,
      "grad_norm": 3.136742353162124,
      "learning_rate": 1.5358212343930075e-06,
      "loss": 0.3841,
      "step": 7672
    },
    {
      "epoch": 9.391676866585067,
      "grad_norm": 2.031573890097241,
      "learning_rate": 1.5352299672247173e-06,
      "loss": 0.5504,
      "step": 7673
    },
    {
      "epoch": 9.392900856793146,
      "grad_norm": 3.0012185689367104,
      "learning_rate": 1.5346387634575447e-06,
      "loss": 0.4533,
      "step": 7674
    },
    {
      "epoch": 9.394124847001224,
      "grad_norm": 2.553692506369909,
      "learning_rate": 1.5340476231303426e-06,
      "loss": 0.3907,
      "step": 7675
    },
    {
      "epoch": 9.395348837209303,
      "grad_norm": 2.5661755241860575,
      "learning_rate": 1.5334565462819564e-06,
      "loss": 0.4247,
      "step": 7676
    },
    {
      "epoch": 9.39657282741738,
      "grad_norm": 1.4548819110936975,
      "learning_rate": 1.5328655329512327e-06,
      "loss": 0.5101,
      "step": 7677
    },
    {
      "epoch": 9.397796817625458,
      "grad_norm": 2.4242354771560484,
      "learning_rate": 1.532274583177008e-06,
      "loss": 0.3234,
      "step": 7678
    },
    {
      "epoch": 9.399020807833537,
      "grad_norm": 1.0300258741330313,
      "learning_rate": 1.5316836969981189e-06,
      "loss": 0.5972,
      "step": 7679
    },
    {
      "epoch": 9.400244798041616,
      "grad_norm": 1.788176614118708,
      "learning_rate": 1.5310928744533954e-06,
      "loss": 0.4572,
      "step": 7680
    },
    {
      "epoch": 9.401468788249694,
      "grad_norm": 1.0870691180630296,
      "learning_rate": 1.530502115581664e-06,
      "loss": 0.605,
      "step": 7681
    },
    {
      "epoch": 9.402692778457773,
      "grad_norm": 2.846313153703495,
      "learning_rate": 1.529911420421748e-06,
      "loss": 0.9662,
      "step": 7682
    },
    {
      "epoch": 9.403916768665852,
      "grad_norm": 3.160990339848278,
      "learning_rate": 1.5293207890124656e-06,
      "loss": 0.5279,
      "step": 7683
    },
    {
      "epoch": 9.405140758873928,
      "grad_norm": 2.5225510790438297,
      "learning_rate": 1.5287302213926308e-06,
      "loss": 0.5196,
      "step": 7684
    },
    {
      "epoch": 9.406364749082007,
      "grad_norm": 1.981582805860729,
      "learning_rate": 1.528139717601052e-06,
      "loss": 1.1424,
      "step": 7685
    },
    {
      "epoch": 9.407588739290086,
      "grad_norm": 1.8841424538471734,
      "learning_rate": 1.5275492776765383e-06,
      "loss": 0.4961,
      "step": 7686
    },
    {
      "epoch": 9.408812729498164,
      "grad_norm": 1.7537213207796096,
      "learning_rate": 1.5269589016578882e-06,
      "loss": 0.4706,
      "step": 7687
    },
    {
      "epoch": 9.410036719706243,
      "grad_norm": 1.5289965711741555,
      "learning_rate": 1.5263685895839e-06,
      "loss": 0.934,
      "step": 7688
    },
    {
      "epoch": 9.411260709914322,
      "grad_norm": 3.1428610249293643,
      "learning_rate": 1.5257783414933666e-06,
      "loss": 0.5966,
      "step": 7689
    },
    {
      "epoch": 9.412484700122398,
      "grad_norm": 1.863386261393905,
      "learning_rate": 1.5251881574250777e-06,
      "loss": 0.4617,
      "step": 7690
    },
    {
      "epoch": 9.413708690330477,
      "grad_norm": 3.404470780772266,
      "learning_rate": 1.5245980374178174e-06,
      "loss": 0.3512,
      "step": 7691
    },
    {
      "epoch": 9.414932680538556,
      "grad_norm": 1.492331858716318,
      "learning_rate": 1.5240079815103663e-06,
      "loss": 0.5537,
      "step": 7692
    },
    {
      "epoch": 9.416156670746634,
      "grad_norm": 2.851500793820335,
      "learning_rate": 1.5234179897415014e-06,
      "loss": 0.6928,
      "step": 7693
    },
    {
      "epoch": 9.417380660954713,
      "grad_norm": 1.5228350491719633,
      "learning_rate": 1.5228280621499924e-06,
      "loss": 0.573,
      "step": 7694
    },
    {
      "epoch": 9.418604651162791,
      "grad_norm": 1.5461196428676418,
      "learning_rate": 1.5222381987746104e-06,
      "loss": 0.4771,
      "step": 7695
    },
    {
      "epoch": 9.419828641370868,
      "grad_norm": 1.2905438081632337,
      "learning_rate": 1.521648399654117e-06,
      "loss": 0.6315,
      "step": 7696
    },
    {
      "epoch": 9.421052631578947,
      "grad_norm": 1.409768241260235,
      "learning_rate": 1.521058664827272e-06,
      "loss": 0.8005,
      "step": 7697
    },
    {
      "epoch": 9.422276621787026,
      "grad_norm": 1.5132861201688572,
      "learning_rate": 1.5204689943328304e-06,
      "loss": 0.4653,
      "step": 7698
    },
    {
      "epoch": 9.423500611995104,
      "grad_norm": 1.4788128604675161,
      "learning_rate": 1.519879388209544e-06,
      "loss": 0.3379,
      "step": 7699
    },
    {
      "epoch": 9.424724602203183,
      "grad_norm": 1.1102594406175743,
      "learning_rate": 1.5192898464961587e-06,
      "loss": 0.5969,
      "step": 7700
    },
    {
      "epoch": 9.425948592411261,
      "grad_norm": 1.2365524849132399,
      "learning_rate": 1.5187003692314174e-06,
      "loss": 0.6584,
      "step": 7701
    },
    {
      "epoch": 9.427172582619338,
      "grad_norm": 2.513447456755945,
      "learning_rate": 1.5181109564540578e-06,
      "loss": 0.3437,
      "step": 7702
    },
    {
      "epoch": 9.428396572827417,
      "grad_norm": 1.6502407109273025,
      "learning_rate": 1.517521608202815e-06,
      "loss": 0.5261,
      "step": 7703
    },
    {
      "epoch": 9.429620563035495,
      "grad_norm": 3.5585421741534047,
      "learning_rate": 1.5169323245164186e-06,
      "loss": 0.426,
      "step": 7704
    },
    {
      "epoch": 9.430844553243574,
      "grad_norm": 1.2309335169582556,
      "learning_rate": 1.5163431054335932e-06,
      "loss": 0.8207,
      "step": 7705
    },
    {
      "epoch": 9.432068543451653,
      "grad_norm": 1.6398812866529657,
      "learning_rate": 1.5157539509930601e-06,
      "loss": 0.4323,
      "step": 7706
    },
    {
      "epoch": 9.433292533659731,
      "grad_norm": 1.2429630779294745,
      "learning_rate": 1.5151648612335374e-06,
      "loss": 0.6499,
      "step": 7707
    },
    {
      "epoch": 9.43451652386781,
      "grad_norm": 1.63963394427148,
      "learning_rate": 1.514575836193737e-06,
      "loss": 0.5118,
      "step": 7708
    },
    {
      "epoch": 9.435740514075887,
      "grad_norm": 2.426609888425431,
      "learning_rate": 1.5139868759123682e-06,
      "loss": 0.3825,
      "step": 7709
    },
    {
      "epoch": 9.436964504283965,
      "grad_norm": 2.859314333428318,
      "learning_rate": 1.5133979804281354e-06,
      "loss": 0.4099,
      "step": 7710
    },
    {
      "epoch": 9.438188494492044,
      "grad_norm": 2.825770505726081,
      "learning_rate": 1.5128091497797364e-06,
      "loss": 0.8036,
      "step": 7711
    },
    {
      "epoch": 9.439412484700123,
      "grad_norm": 1.6209661009308896,
      "learning_rate": 1.5122203840058699e-06,
      "loss": 0.5609,
      "step": 7712
    },
    {
      "epoch": 9.440636474908201,
      "grad_norm": 2.0004031004642577,
      "learning_rate": 1.5116316831452269e-06,
      "loss": 0.771,
      "step": 7713
    },
    {
      "epoch": 9.44186046511628,
      "grad_norm": 1.42012215249765,
      "learning_rate": 1.511043047236493e-06,
      "loss": 0.5631,
      "step": 7714
    },
    {
      "epoch": 9.443084455324357,
      "grad_norm": 2.935060318438459,
      "learning_rate": 1.510454476318352e-06,
      "loss": 0.5333,
      "step": 7715
    },
    {
      "epoch": 9.444308445532435,
      "grad_norm": 1.9731980192439802,
      "learning_rate": 1.5098659704294833e-06,
      "loss": 0.5312,
      "step": 7716
    },
    {
      "epoch": 9.445532435740514,
      "grad_norm": 2.1999761331452032,
      "learning_rate": 1.5092775296085604e-06,
      "loss": 0.3559,
      "step": 7717
    },
    {
      "epoch": 9.446756425948593,
      "grad_norm": 1.5051226361301662,
      "learning_rate": 1.5086891538942542e-06,
      "loss": 1.2173,
      "step": 7718
    },
    {
      "epoch": 9.447980416156671,
      "grad_norm": 1.1944251790934965,
      "learning_rate": 1.5081008433252306e-06,
      "loss": 0.5812,
      "step": 7719
    },
    {
      "epoch": 9.44920440636475,
      "grad_norm": 1.3513047029674243,
      "learning_rate": 1.5075125979401492e-06,
      "loss": 0.5594,
      "step": 7720
    },
    {
      "epoch": 9.450428396572827,
      "grad_norm": 1.5929505799460393,
      "learning_rate": 1.50692441777767e-06,
      "loss": 0.6722,
      "step": 7721
    },
    {
      "epoch": 9.451652386780905,
      "grad_norm": 3.017591693603603,
      "learning_rate": 1.5063363028764456e-06,
      "loss": 0.3501,
      "step": 7722
    },
    {
      "epoch": 9.452876376988984,
      "grad_norm": 1.2431950018983517,
      "learning_rate": 1.505748253275123e-06,
      "loss": 0.3896,
      "step": 7723
    },
    {
      "epoch": 9.454100367197062,
      "grad_norm": 1.8155434298331987,
      "learning_rate": 1.5051602690123476e-06,
      "loss": 0.4882,
      "step": 7724
    },
    {
      "epoch": 9.455324357405141,
      "grad_norm": 2.942920789026892,
      "learning_rate": 1.50457235012676e-06,
      "loss": 0.4537,
      "step": 7725
    },
    {
      "epoch": 9.45654834761322,
      "grad_norm": 1.422440232204843,
      "learning_rate": 1.503984496656996e-06,
      "loss": 0.56,
      "step": 7726
    },
    {
      "epoch": 9.457772337821297,
      "grad_norm": 1.2222573756554564,
      "learning_rate": 1.5033967086416867e-06,
      "loss": 0.7513,
      "step": 7727
    },
    {
      "epoch": 9.458996328029375,
      "grad_norm": 1.6524463196543624,
      "learning_rate": 1.502808986119459e-06,
      "loss": 0.6617,
      "step": 7728
    },
    {
      "epoch": 9.460220318237454,
      "grad_norm": 2.6200365359053346,
      "learning_rate": 1.5022213291289368e-06,
      "loss": 0.8998,
      "step": 7729
    },
    {
      "epoch": 9.461444308445532,
      "grad_norm": 1.4110726999801746,
      "learning_rate": 1.5016337377087388e-06,
      "loss": 0.5035,
      "step": 7730
    },
    {
      "epoch": 9.462668298653611,
      "grad_norm": 1.9901968175959863,
      "learning_rate": 1.501046211897479e-06,
      "loss": 0.4451,
      "step": 7731
    },
    {
      "epoch": 9.46389228886169,
      "grad_norm": 1.93526354345633,
      "learning_rate": 1.5004587517337665e-06,
      "loss": 0.4944,
      "step": 7732
    },
    {
      "epoch": 9.465116279069768,
      "grad_norm": 1.9819462865222912,
      "learning_rate": 1.4998713572562076e-06,
      "loss": 0.421,
      "step": 7733
    },
    {
      "epoch": 9.466340269277845,
      "grad_norm": 1.330597474131783,
      "learning_rate": 1.4992840285034045e-06,
      "loss": 0.4537,
      "step": 7734
    },
    {
      "epoch": 9.467564259485924,
      "grad_norm": 1.0316442455516797,
      "learning_rate": 1.4986967655139533e-06,
      "loss": 0.514,
      "step": 7735
    },
    {
      "epoch": 9.468788249694002,
      "grad_norm": 1.0331210598123945,
      "learning_rate": 1.4981095683264474e-06,
      "loss": 0.5484,
      "step": 7736
    },
    {
      "epoch": 9.470012239902081,
      "grad_norm": 1.6280672085391878,
      "learning_rate": 1.4975224369794745e-06,
      "loss": 0.3752,
      "step": 7737
    },
    {
      "epoch": 9.47123623011016,
      "grad_norm": 2.3387651939120593,
      "learning_rate": 1.4969353715116192e-06,
      "loss": 1.2032,
      "step": 7738
    },
    {
      "epoch": 9.472460220318238,
      "grad_norm": 1.1533121805740116,
      "learning_rate": 1.4963483719614618e-06,
      "loss": 0.511,
      "step": 7739
    },
    {
      "epoch": 9.473684210526315,
      "grad_norm": 2.221493511139791,
      "learning_rate": 1.495761438367577e-06,
      "loss": 1.0364,
      "step": 7740
    },
    {
      "epoch": 9.474908200734394,
      "grad_norm": 1.7847144407512583,
      "learning_rate": 1.495174570768535e-06,
      "loss": 0.4644,
      "step": 7741
    },
    {
      "epoch": 9.476132190942472,
      "grad_norm": 3.0370539710921345,
      "learning_rate": 1.4945877692029049e-06,
      "loss": 0.5482,
      "step": 7742
    },
    {
      "epoch": 9.477356181150551,
      "grad_norm": 2.8287731391399067,
      "learning_rate": 1.4940010337092476e-06,
      "loss": 0.4969,
      "step": 7743
    },
    {
      "epoch": 9.47858017135863,
      "grad_norm": 1.297409915256897,
      "learning_rate": 1.4934143643261212e-06,
      "loss": 0.6743,
      "step": 7744
    },
    {
      "epoch": 9.479804161566708,
      "grad_norm": 1.7222407827546702,
      "learning_rate": 1.4928277610920796e-06,
      "loss": 1.162,
      "step": 7745
    },
    {
      "epoch": 9.481028151774785,
      "grad_norm": 1.9335693017029854,
      "learning_rate": 1.492241224045672e-06,
      "loss": 0.4156,
      "step": 7746
    },
    {
      "epoch": 9.482252141982864,
      "grad_norm": 1.7646391255095528,
      "learning_rate": 1.4916547532254439e-06,
      "loss": 0.5465,
      "step": 7747
    },
    {
      "epoch": 9.483476132190942,
      "grad_norm": 1.7079652169095414,
      "learning_rate": 1.4910683486699367e-06,
      "loss": 0.4353,
      "step": 7748
    },
    {
      "epoch": 9.48470012239902,
      "grad_norm": 2.3140826602586477,
      "learning_rate": 1.490482010417685e-06,
      "loss": 0.4498,
      "step": 7749
    },
    {
      "epoch": 9.4859241126071,
      "grad_norm": 2.2335820041641044,
      "learning_rate": 1.489895738507221e-06,
      "loss": 0.4864,
      "step": 7750
    },
    {
      "epoch": 9.487148102815178,
      "grad_norm": 1.0511851714779545,
      "learning_rate": 1.489309532977074e-06,
      "loss": 0.5102,
      "step": 7751
    },
    {
      "epoch": 9.488372093023255,
      "grad_norm": 1.0961443961737851,
      "learning_rate": 1.488723393865766e-06,
      "loss": 0.4195,
      "step": 7752
    },
    {
      "epoch": 9.489596083231334,
      "grad_norm": 1.7528833543708489,
      "learning_rate": 1.488137321211816e-06,
      "loss": 0.4185,
      "step": 7753
    },
    {
      "epoch": 9.490820073439412,
      "grad_norm": 1.7706910788916235,
      "learning_rate": 1.4875513150537386e-06,
      "loss": 0.71,
      "step": 7754
    },
    {
      "epoch": 9.49204406364749,
      "grad_norm": 1.3084633680813926,
      "learning_rate": 1.486965375430044e-06,
      "loss": 0.3913,
      "step": 7755
    },
    {
      "epoch": 9.49326805385557,
      "grad_norm": 3.245317618871229,
      "learning_rate": 1.4863795023792379e-06,
      "loss": 0.478,
      "step": 7756
    },
    {
      "epoch": 9.494492044063648,
      "grad_norm": 1.4586383337078286,
      "learning_rate": 1.4857936959398222e-06,
      "loss": 1.0855,
      "step": 7757
    },
    {
      "epoch": 9.495716034271727,
      "grad_norm": 2.546193504330229,
      "learning_rate": 1.4852079561502935e-06,
      "loss": 0.468,
      "step": 7758
    },
    {
      "epoch": 9.496940024479803,
      "grad_norm": 1.9592531389716572,
      "learning_rate": 1.484622283049143e-06,
      "loss": 0.5989,
      "step": 7759
    },
    {
      "epoch": 9.498164014687882,
      "grad_norm": 1.5683182830050475,
      "learning_rate": 1.4840366766748624e-06,
      "loss": 1.4866,
      "step": 7760
    },
    {
      "epoch": 9.49938800489596,
      "grad_norm": 2.263452282701963,
      "learning_rate": 1.4834511370659328e-06,
      "loss": 0.6095,
      "step": 7761
    },
    {
      "epoch": 9.50061199510404,
      "grad_norm": 1.0469385498693353,
      "learning_rate": 1.4828656642608347e-06,
      "loss": 0.7109,
      "step": 7762
    },
    {
      "epoch": 9.501835985312118,
      "grad_norm": 1.6884831926260175,
      "learning_rate": 1.4822802582980422e-06,
      "loss": 1.3635,
      "step": 7763
    },
    {
      "epoch": 9.503059975520197,
      "grad_norm": 2.372569541218042,
      "learning_rate": 1.4816949192160277e-06,
      "loss": 0.6124,
      "step": 7764
    },
    {
      "epoch": 9.504283965728273,
      "grad_norm": 2.368780894029947,
      "learning_rate": 1.4811096470532566e-06,
      "loss": 0.5068,
      "step": 7765
    },
    {
      "epoch": 9.505507955936352,
      "grad_norm": 1.4449189639629425,
      "learning_rate": 1.4805244418481911e-06,
      "loss": 0.568,
      "step": 7766
    },
    {
      "epoch": 9.50673194614443,
      "grad_norm": 2.43495043758298,
      "learning_rate": 1.4799393036392873e-06,
      "loss": 0.4837,
      "step": 7767
    },
    {
      "epoch": 9.50795593635251,
      "grad_norm": 1.6514222833394305,
      "learning_rate": 1.4793542324650007e-06,
      "loss": 0.4443,
      "step": 7768
    },
    {
      "epoch": 9.509179926560588,
      "grad_norm": 1.4303365639859202,
      "learning_rate": 1.4787692283637785e-06,
      "loss": 0.6209,
      "step": 7769
    },
    {
      "epoch": 9.510403916768666,
      "grad_norm": 1.0774641623553034,
      "learning_rate": 1.4781842913740654e-06,
      "loss": 0.389,
      "step": 7770
    },
    {
      "epoch": 9.511627906976745,
      "grad_norm": 1.6361367762109296,
      "learning_rate": 1.4775994215343016e-06,
      "loss": 1.4543,
      "step": 7771
    },
    {
      "epoch": 9.512851897184822,
      "grad_norm": 1.2927315909019155,
      "learning_rate": 1.477014618882921e-06,
      "loss": 0.7205,
      "step": 7772
    },
    {
      "epoch": 9.5140758873929,
      "grad_norm": 1.102163354968593,
      "learning_rate": 1.4764298834583569e-06,
      "loss": 0.5091,
      "step": 7773
    },
    {
      "epoch": 9.51529987760098,
      "grad_norm": 1.7697297329569817,
      "learning_rate": 1.4758452152990345e-06,
      "loss": 1.503,
      "step": 7774
    },
    {
      "epoch": 9.516523867809058,
      "grad_norm": 2.1967312264212513,
      "learning_rate": 1.4752606144433772e-06,
      "loss": 0.5829,
      "step": 7775
    },
    {
      "epoch": 9.517747858017136,
      "grad_norm": 1.7306170689814768,
      "learning_rate": 1.4746760809298009e-06,
      "loss": 0.4666,
      "step": 7776
    },
    {
      "epoch": 9.518971848225215,
      "grad_norm": 3.2541631665038144,
      "learning_rate": 1.474091614796721e-06,
      "loss": 0.3599,
      "step": 7777
    },
    {
      "epoch": 9.520195838433292,
      "grad_norm": 1.9696820920200477,
      "learning_rate": 1.4735072160825456e-06,
      "loss": 0.298,
      "step": 7778
    },
    {
      "epoch": 9.52141982864137,
      "grad_norm": 1.7624668221618032,
      "learning_rate": 1.4729228848256792e-06,
      "loss": 0.4505,
      "step": 7779
    },
    {
      "epoch": 9.522643818849449,
      "grad_norm": 1.6456610027621148,
      "learning_rate": 1.4723386210645213e-06,
      "loss": 0.4883,
      "step": 7780
    },
    {
      "epoch": 9.523867809057528,
      "grad_norm": 1.7661337505400532,
      "learning_rate": 1.4717544248374686e-06,
      "loss": 1.3847,
      "step": 7781
    },
    {
      "epoch": 9.525091799265606,
      "grad_norm": 2.324209520682565,
      "learning_rate": 1.471170296182912e-06,
      "loss": 0.5952,
      "step": 7782
    },
    {
      "epoch": 9.526315789473685,
      "grad_norm": 1.9447894510856336,
      "learning_rate": 1.4705862351392381e-06,
      "loss": 1.0379,
      "step": 7783
    },
    {
      "epoch": 9.527539779681762,
      "grad_norm": 2.7210739070395613,
      "learning_rate": 1.4700022417448296e-06,
      "loss": 0.3587,
      "step": 7784
    },
    {
      "epoch": 9.52876376988984,
      "grad_norm": 1.4361038823969978,
      "learning_rate": 1.4694183160380627e-06,
      "loss": 1.1598,
      "step": 7785
    },
    {
      "epoch": 9.529987760097919,
      "grad_norm": 1.020347335262806,
      "learning_rate": 1.468834458057314e-06,
      "loss": 0.5193,
      "step": 7786
    },
    {
      "epoch": 9.531211750305998,
      "grad_norm": 1.1049471167926914,
      "learning_rate": 1.4682506678409496e-06,
      "loss": 0.5854,
      "step": 7787
    },
    {
      "epoch": 9.532435740514076,
      "grad_norm": 3.6200201669649217,
      "learning_rate": 1.4676669454273357e-06,
      "loss": 0.3592,
      "step": 7788
    },
    {
      "epoch": 9.533659730722155,
      "grad_norm": 1.0423647094234372,
      "learning_rate": 1.467083290854831e-06,
      "loss": 0.5394,
      "step": 7789
    },
    {
      "epoch": 9.534883720930232,
      "grad_norm": 2.3141292887706686,
      "learning_rate": 1.4664997041617929e-06,
      "loss": 0.4518,
      "step": 7790
    },
    {
      "epoch": 9.53610771113831,
      "grad_norm": 1.1624547058258448,
      "learning_rate": 1.4659161853865712e-06,
      "loss": 0.5021,
      "step": 7791
    },
    {
      "epoch": 9.537331701346389,
      "grad_norm": 1.5464655430985592,
      "learning_rate": 1.4653327345675128e-06,
      "loss": 0.7497,
      "step": 7792
    },
    {
      "epoch": 9.538555691554468,
      "grad_norm": 1.5404888692298353,
      "learning_rate": 1.46474935174296e-06,
      "loss": 0.4797,
      "step": 7793
    },
    {
      "epoch": 9.539779681762546,
      "grad_norm": 1.4484178421911822,
      "learning_rate": 1.464166036951251e-06,
      "loss": 0.3357,
      "step": 7794
    },
    {
      "epoch": 9.541003671970625,
      "grad_norm": 2.327931362438206,
      "learning_rate": 1.4635827902307192e-06,
      "loss": 0.5821,
      "step": 7795
    },
    {
      "epoch": 9.542227662178703,
      "grad_norm": 2.3369418641811994,
      "learning_rate": 1.4629996116196927e-06,
      "loss": 0.4664,
      "step": 7796
    },
    {
      "epoch": 9.54345165238678,
      "grad_norm": 1.5895154356358212,
      "learning_rate": 1.462416501156496e-06,
      "loss": 0.5973,
      "step": 7797
    },
    {
      "epoch": 9.544675642594859,
      "grad_norm": 2.725269090929542,
      "learning_rate": 1.4618334588794484e-06,
      "loss": 0.3605,
      "step": 7798
    },
    {
      "epoch": 9.545899632802938,
      "grad_norm": 1.9165297045401903,
      "learning_rate": 1.4612504848268666e-06,
      "loss": 0.5589,
      "step": 7799
    },
    {
      "epoch": 9.547123623011016,
      "grad_norm": 2.0179432226742255,
      "learning_rate": 1.4606675790370612e-06,
      "loss": 0.7688,
      "step": 7800
    },
    {
      "epoch": 9.548347613219095,
      "grad_norm": 1.9835961588400162,
      "learning_rate": 1.4600847415483376e-06,
      "loss": 0.5966,
      "step": 7801
    },
    {
      "epoch": 9.549571603427173,
      "grad_norm": 1.5705094145015566,
      "learning_rate": 1.459501972398999e-06,
      "loss": 1.2966,
      "step": 7802
    },
    {
      "epoch": 9.55079559363525,
      "grad_norm": 1.4413121268827684,
      "learning_rate": 1.458919271627342e-06,
      "loss": 1.276,
      "step": 7803
    },
    {
      "epoch": 9.552019583843329,
      "grad_norm": 1.995387636502786,
      "learning_rate": 1.4583366392716597e-06,
      "loss": 1.3154,
      "step": 7804
    },
    {
      "epoch": 9.553243574051407,
      "grad_norm": 1.8011023225372598,
      "learning_rate": 1.4577540753702406e-06,
      "loss": 1.4084,
      "step": 7805
    },
    {
      "epoch": 9.554467564259486,
      "grad_norm": 1.3834294944255765,
      "learning_rate": 1.457171579961368e-06,
      "loss": 0.8463,
      "step": 7806
    },
    {
      "epoch": 9.555691554467565,
      "grad_norm": 2.5550572126455062,
      "learning_rate": 1.456589153083323e-06,
      "loss": 0.4073,
      "step": 7807
    },
    {
      "epoch": 9.556915544675643,
      "grad_norm": 2.011819189716954,
      "learning_rate": 1.45600679477438e-06,
      "loss": 0.5108,
      "step": 7808
    },
    {
      "epoch": 9.55813953488372,
      "grad_norm": 1.3136719539396435,
      "learning_rate": 1.4554245050728085e-06,
      "loss": 0.5299,
      "step": 7809
    },
    {
      "epoch": 9.559363525091799,
      "grad_norm": 2.184789209002383,
      "learning_rate": 1.4548422840168747e-06,
      "loss": 0.6223,
      "step": 7810
    },
    {
      "epoch": 9.560587515299877,
      "grad_norm": 1.7667545941254406,
      "learning_rate": 1.4542601316448393e-06,
      "loss": 0.5373,
      "step": 7811
    },
    {
      "epoch": 9.561811505507956,
      "grad_norm": 2.103308800745441,
      "learning_rate": 1.4536780479949614e-06,
      "loss": 1.108,
      "step": 7812
    },
    {
      "epoch": 9.563035495716035,
      "grad_norm": 2.075652930487441,
      "learning_rate": 1.4530960331054916e-06,
      "loss": 1.123,
      "step": 7813
    },
    {
      "epoch": 9.564259485924113,
      "grad_norm": 1.3458660346917939,
      "learning_rate": 1.4525140870146787e-06,
      "loss": 0.7404,
      "step": 7814
    },
    {
      "epoch": 9.56548347613219,
      "grad_norm": 1.7698614820452299,
      "learning_rate": 1.4519322097607653e-06,
      "loss": 1.5636,
      "step": 7815
    },
    {
      "epoch": 9.566707466340269,
      "grad_norm": 2.2204272376657808,
      "learning_rate": 1.4513504013819906e-06,
      "loss": 0.3771,
      "step": 7816
    },
    {
      "epoch": 9.567931456548347,
      "grad_norm": 2.266406691502686,
      "learning_rate": 1.450768661916589e-06,
      "loss": 0.6446,
      "step": 7817
    },
    {
      "epoch": 9.569155446756426,
      "grad_norm": 1.8866655263899608,
      "learning_rate": 1.45018699140279e-06,
      "loss": 1.1115,
      "step": 7818
    },
    {
      "epoch": 9.570379436964505,
      "grad_norm": 1.6348100183288583,
      "learning_rate": 1.4496053898788177e-06,
      "loss": 1.6255,
      "step": 7819
    },
    {
      "epoch": 9.571603427172583,
      "grad_norm": 1.161287885690219,
      "learning_rate": 1.4490238573828962e-06,
      "loss": 0.3849,
      "step": 7820
    },
    {
      "epoch": 9.572827417380662,
      "grad_norm": 1.1336257176363878,
      "learning_rate": 1.4484423939532385e-06,
      "loss": 0.6221,
      "step": 7821
    },
    {
      "epoch": 9.574051407588739,
      "grad_norm": 1.4662963054403146,
      "learning_rate": 1.4478609996280574e-06,
      "loss": 1.3097,
      "step": 7822
    },
    {
      "epoch": 9.575275397796817,
      "grad_norm": 2.085704359531065,
      "learning_rate": 1.4472796744455597e-06,
      "loss": 0.572,
      "step": 7823
    },
    {
      "epoch": 9.576499388004896,
      "grad_norm": 2.2412093208184256,
      "learning_rate": 1.446698418443947e-06,
      "loss": 0.4561,
      "step": 7824
    },
    {
      "epoch": 9.577723378212974,
      "grad_norm": 2.511733856993857,
      "learning_rate": 1.4461172316614194e-06,
      "loss": 0.6253,
      "step": 7825
    },
    {
      "epoch": 9.578947368421053,
      "grad_norm": 1.1272914598273136,
      "learning_rate": 1.4455361141361702e-06,
      "loss": 0.498,
      "step": 7826
    },
    {
      "epoch": 9.580171358629132,
      "grad_norm": 1.7850858120185553,
      "learning_rate": 1.4449550659063865e-06,
      "loss": 0.4749,
      "step": 7827
    },
    {
      "epoch": 9.581395348837209,
      "grad_norm": 2.5387154750620757,
      "learning_rate": 1.4443740870102523e-06,
      "loss": 0.9372,
      "step": 7828
    },
    {
      "epoch": 9.582619339045287,
      "grad_norm": 1.792417429866816,
      "learning_rate": 1.4437931774859498e-06,
      "loss": 1.5951,
      "step": 7829
    },
    {
      "epoch": 9.583843329253366,
      "grad_norm": 1.8055540216787278,
      "learning_rate": 1.4432123373716533e-06,
      "loss": 1.0893,
      "step": 7830
    },
    {
      "epoch": 9.585067319461444,
      "grad_norm": 2.2656926473564174,
      "learning_rate": 1.442631566705533e-06,
      "loss": 0.9082,
      "step": 7831
    },
    {
      "epoch": 9.586291309669523,
      "grad_norm": 1.8209217211503437,
      "learning_rate": 1.4420508655257551e-06,
      "loss": 0.5269,
      "step": 7832
    },
    {
      "epoch": 9.587515299877602,
      "grad_norm": 1.730461444608734,
      "learning_rate": 1.4414702338704811e-06,
      "loss": 1.1031,
      "step": 7833
    },
    {
      "epoch": 9.588739290085678,
      "grad_norm": 1.918957433846606,
      "learning_rate": 1.4408896717778684e-06,
      "loss": 0.3698,
      "step": 7834
    },
    {
      "epoch": 9.589963280293757,
      "grad_norm": 1.6762879847138858,
      "learning_rate": 1.440309179286069e-06,
      "loss": 1.2683,
      "step": 7835
    },
    {
      "epoch": 9.591187270501836,
      "grad_norm": 1.409427507958619,
      "learning_rate": 1.4397287564332308e-06,
      "loss": 0.5937,
      "step": 7836
    },
    {
      "epoch": 9.592411260709914,
      "grad_norm": 1.0944849049835192,
      "learning_rate": 1.439148403257497e-06,
      "loss": 0.6161,
      "step": 7837
    },
    {
      "epoch": 9.593635250917993,
      "grad_norm": 1.541258941020548,
      "learning_rate": 1.4385681197970065e-06,
      "loss": 0.5385,
      "step": 7838
    },
    {
      "epoch": 9.594859241126072,
      "grad_norm": 1.867668374573175,
      "learning_rate": 1.437987906089893e-06,
      "loss": 0.3655,
      "step": 7839
    },
    {
      "epoch": 9.596083231334148,
      "grad_norm": 2.1522802541860564,
      "learning_rate": 1.4374077621742863e-06,
      "loss": 0.8251,
      "step": 7840
    },
    {
      "epoch": 9.597307221542227,
      "grad_norm": 1.8006727526273558,
      "learning_rate": 1.4368276880883104e-06,
      "loss": 0.2976,
      "step": 7841
    },
    {
      "epoch": 9.598531211750306,
      "grad_norm": 1.2712864487880309,
      "learning_rate": 1.4362476838700868e-06,
      "loss": 0.9865,
      "step": 7842
    },
    {
      "epoch": 9.599755201958384,
      "grad_norm": 1.1284615046478494,
      "learning_rate": 1.4356677495577315e-06,
      "loss": 0.6061,
      "step": 7843
    },
    {
      "epoch": 9.600979192166463,
      "grad_norm": 1.2301289040702679,
      "learning_rate": 1.4350878851893557e-06,
      "loss": 0.8472,
      "step": 7844
    },
    {
      "epoch": 9.602203182374542,
      "grad_norm": 2.827472499302893,
      "learning_rate": 1.4345080908030644e-06,
      "loss": 0.4197,
      "step": 7845
    },
    {
      "epoch": 9.60342717258262,
      "grad_norm": 2.380886332457003,
      "learning_rate": 1.433928366436959e-06,
      "loss": 0.5406,
      "step": 7846
    },
    {
      "epoch": 9.604651162790697,
      "grad_norm": 1.8523805583802497,
      "learning_rate": 1.4333487121291397e-06,
      "loss": 0.4624,
      "step": 7847
    },
    {
      "epoch": 9.605875152998776,
      "grad_norm": 1.81005603652633,
      "learning_rate": 1.4327691279176977e-06,
      "loss": 0.4791,
      "step": 7848
    },
    {
      "epoch": 9.607099143206854,
      "grad_norm": 2.2833754692898514,
      "learning_rate": 1.432189613840721e-06,
      "loss": 0.6727,
      "step": 7849
    },
    {
      "epoch": 9.608323133414933,
      "grad_norm": 1.4962475092410927,
      "learning_rate": 1.4316101699362936e-06,
      "loss": 0.9147,
      "step": 7850
    },
    {
      "epoch": 9.609547123623011,
      "grad_norm": 1.5586761627539671,
      "learning_rate": 1.4310307962424941e-06,
      "loss": 0.9382,
      "step": 7851
    },
    {
      "epoch": 9.61077111383109,
      "grad_norm": 2.090053169462759,
      "learning_rate": 1.4304514927973967e-06,
      "loss": 0.3989,
      "step": 7852
    },
    {
      "epoch": 9.611995104039167,
      "grad_norm": 2.2057025175042533,
      "learning_rate": 1.4298722596390712e-06,
      "loss": 0.822,
      "step": 7853
    },
    {
      "epoch": 9.613219094247246,
      "grad_norm": 1.3573511614005591,
      "learning_rate": 1.4292930968055814e-06,
      "loss": 0.4382,
      "step": 7854
    },
    {
      "epoch": 9.614443084455324,
      "grad_norm": 1.33405331022911,
      "learning_rate": 1.4287140043349915e-06,
      "loss": 0.3115,
      "step": 7855
    },
    {
      "epoch": 9.615667074663403,
      "grad_norm": 2.069477248794864,
      "learning_rate": 1.4281349822653534e-06,
      "loss": 0.6041,
      "step": 7856
    },
    {
      "epoch": 9.616891064871481,
      "grad_norm": 1.5627865517408002,
      "learning_rate": 1.4275560306347202e-06,
      "loss": 0.4697,
      "step": 7857
    },
    {
      "epoch": 9.61811505507956,
      "grad_norm": 1.8510052643401809,
      "learning_rate": 1.4269771494811374e-06,
      "loss": 0.5335,
      "step": 7858
    },
    {
      "epoch": 9.619339045287639,
      "grad_norm": 1.5917315685237576,
      "learning_rate": 1.426398338842647e-06,
      "loss": 0.5644,
      "step": 7859
    },
    {
      "epoch": 9.620563035495715,
      "grad_norm": 2.0837222569498297,
      "learning_rate": 1.4258195987572876e-06,
      "loss": 0.5179,
      "step": 7860
    },
    {
      "epoch": 9.621787025703794,
      "grad_norm": 1.2254178811216514,
      "learning_rate": 1.4252409292630908e-06,
      "loss": 0.7708,
      "step": 7861
    },
    {
      "epoch": 9.623011015911873,
      "grad_norm": 2.9036862784295114,
      "learning_rate": 1.4246623303980865e-06,
      "loss": 0.9177,
      "step": 7862
    },
    {
      "epoch": 9.624235006119951,
      "grad_norm": 1.9195428132354044,
      "learning_rate": 1.424083802200294e-06,
      "loss": 0.5815,
      "step": 7863
    },
    {
      "epoch": 9.62545899632803,
      "grad_norm": 1.1075328625690877,
      "learning_rate": 1.4235053447077355e-06,
      "loss": 0.4731,
      "step": 7864
    },
    {
      "epoch": 9.626682986536107,
      "grad_norm": 2.4820894656888863,
      "learning_rate": 1.422926957958424e-06,
      "loss": 0.582,
      "step": 7865
    },
    {
      "epoch": 9.627906976744185,
      "grad_norm": 1.4658961978071154,
      "learning_rate": 1.4223486419903692e-06,
      "loss": 0.716,
      "step": 7866
    },
    {
      "epoch": 9.629130966952264,
      "grad_norm": 1.6589283452960928,
      "learning_rate": 1.4217703968415756e-06,
      "loss": 0.5043,
      "step": 7867
    },
    {
      "epoch": 9.630354957160343,
      "grad_norm": 1.3957420783107903,
      "learning_rate": 1.4211922225500435e-06,
      "loss": 0.9504,
      "step": 7868
    },
    {
      "epoch": 9.631578947368421,
      "grad_norm": 1.059235944846936,
      "learning_rate": 1.4206141191537681e-06,
      "loss": 0.5098,
      "step": 7869
    },
    {
      "epoch": 9.6328029375765,
      "grad_norm": 2.1807055266898696,
      "learning_rate": 1.4200360866907405e-06,
      "loss": 0.5276,
      "step": 7870
    },
    {
      "epoch": 9.634026927784578,
      "grad_norm": 1.8256361474364171,
      "learning_rate": 1.419458125198947e-06,
      "loss": 0.6137,
      "step": 7871
    },
    {
      "epoch": 9.635250917992655,
      "grad_norm": 2.0242362764237356,
      "learning_rate": 1.4188802347163677e-06,
      "loss": 0.6728,
      "step": 7872
    },
    {
      "epoch": 9.636474908200734,
      "grad_norm": 1.9362464625293698,
      "learning_rate": 1.4183024152809824e-06,
      "loss": 0.7264,
      "step": 7873
    },
    {
      "epoch": 9.637698898408813,
      "grad_norm": 1.2072852083308974,
      "learning_rate": 1.4177246669307606e-06,
      "loss": 0.5321,
      "step": 7874
    },
    {
      "epoch": 9.638922888616891,
      "grad_norm": 1.0470116355037258,
      "learning_rate": 1.4171469897036705e-06,
      "loss": 0.3959,
      "step": 7875
    },
    {
      "epoch": 9.64014687882497,
      "grad_norm": 1.5675033899924284,
      "learning_rate": 1.4165693836376743e-06,
      "loss": 1.1726,
      "step": 7876
    },
    {
      "epoch": 9.641370869033048,
      "grad_norm": 1.8090548772907329,
      "learning_rate": 1.415991848770732e-06,
      "loss": 1.4908,
      "step": 7877
    },
    {
      "epoch": 9.642594859241125,
      "grad_norm": 1.53413849233614,
      "learning_rate": 1.4154143851407953e-06,
      "loss": 0.6937,
      "step": 7878
    },
    {
      "epoch": 9.643818849449204,
      "grad_norm": 1.6542013259840307,
      "learning_rate": 1.4148369927858142e-06,
      "loss": 0.471,
      "step": 7879
    },
    {
      "epoch": 9.645042839657282,
      "grad_norm": 1.4517797336792546,
      "learning_rate": 1.4142596717437319e-06,
      "loss": 1.0757,
      "step": 7880
    },
    {
      "epoch": 9.646266829865361,
      "grad_norm": 1.5102948137964847,
      "learning_rate": 1.4136824220524884e-06,
      "loss": 0.5947,
      "step": 7881
    },
    {
      "epoch": 9.64749082007344,
      "grad_norm": 3.1325541856365744,
      "learning_rate": 1.4131052437500177e-06,
      "loss": 0.4458,
      "step": 7882
    },
    {
      "epoch": 9.648714810281518,
      "grad_norm": 1.207346392395892,
      "learning_rate": 1.4125281368742505e-06,
      "loss": 0.4517,
      "step": 7883
    },
    {
      "epoch": 9.649938800489597,
      "grad_norm": 1.179459796599702,
      "learning_rate": 1.4119511014631118e-06,
      "loss": 0.5734,
      "step": 7884
    },
    {
      "epoch": 9.651162790697674,
      "grad_norm": 2.384807294317284,
      "learning_rate": 1.4113741375545222e-06,
      "loss": 0.4529,
      "step": 7885
    },
    {
      "epoch": 9.652386780905752,
      "grad_norm": 1.5995875059351827,
      "learning_rate": 1.410797245186398e-06,
      "loss": 0.4625,
      "step": 7886
    },
    {
      "epoch": 9.653610771113831,
      "grad_norm": 0.5607521498322625,
      "learning_rate": 1.41022042439665e-06,
      "loss": 0.1298,
      "step": 7887
    },
    {
      "epoch": 9.65483476132191,
      "grad_norm": 1.4150962882603404,
      "learning_rate": 1.4096436752231849e-06,
      "loss": 1.0028,
      "step": 7888
    },
    {
      "epoch": 9.656058751529988,
      "grad_norm": 2.8092459572139545,
      "learning_rate": 1.4090669977039035e-06,
      "loss": 0.5353,
      "step": 7889
    },
    {
      "epoch": 9.657282741738067,
      "grad_norm": 3.114554793643203,
      "learning_rate": 1.4084903918767049e-06,
      "loss": 0.3566,
      "step": 7890
    },
    {
      "epoch": 9.658506731946144,
      "grad_norm": 1.340799385582237,
      "learning_rate": 1.4079138577794815e-06,
      "loss": 1.0899,
      "step": 7891
    },
    {
      "epoch": 9.659730722154222,
      "grad_norm": 2.5873698275307686,
      "learning_rate": 1.4073373954501191e-06,
      "loss": 0.663,
      "step": 7892
    },
    {
      "epoch": 9.660954712362301,
      "grad_norm": 1.4312496939337525,
      "learning_rate": 1.4067610049265008e-06,
      "loss": 0.5915,
      "step": 7893
    },
    {
      "epoch": 9.66217870257038,
      "grad_norm": 1.7395484822956622,
      "learning_rate": 1.4061846862465062e-06,
      "loss": 0.4119,
      "step": 7894
    },
    {
      "epoch": 9.663402692778458,
      "grad_norm": 1.121834135546993,
      "learning_rate": 1.4056084394480086e-06,
      "loss": 0.3102,
      "step": 7895
    },
    {
      "epoch": 9.664626682986537,
      "grad_norm": 1.876224009176409,
      "learning_rate": 1.4050322645688763e-06,
      "loss": 1.1101,
      "step": 7896
    },
    {
      "epoch": 9.665850673194614,
      "grad_norm": 1.4410877786494556,
      "learning_rate": 1.4044561616469745e-06,
      "loss": 0.9865,
      "step": 7897
    },
    {
      "epoch": 9.667074663402692,
      "grad_norm": 1.5269398591140442,
      "learning_rate": 1.4038801307201598e-06,
      "loss": 0.6927,
      "step": 7898
    },
    {
      "epoch": 9.668298653610771,
      "grad_norm": 1.420420711175893,
      "learning_rate": 1.4033041718262897e-06,
      "loss": 0.6192,
      "step": 7899
    },
    {
      "epoch": 9.66952264381885,
      "grad_norm": 1.8965038201077495,
      "learning_rate": 1.402728285003213e-06,
      "loss": 0.5421,
      "step": 7900
    },
    {
      "epoch": 9.670746634026928,
      "grad_norm": 1.6404664777070186,
      "learning_rate": 1.4021524702887746e-06,
      "loss": 0.9849,
      "step": 7901
    },
    {
      "epoch": 9.671970624235007,
      "grad_norm": 1.5174143165895246,
      "learning_rate": 1.4015767277208155e-06,
      "loss": 0.3255,
      "step": 7902
    },
    {
      "epoch": 9.673194614443084,
      "grad_norm": 2.534897772113679,
      "learning_rate": 1.401001057337171e-06,
      "loss": 0.4046,
      "step": 7903
    },
    {
      "epoch": 9.674418604651162,
      "grad_norm": 1.9596061033538106,
      "learning_rate": 1.4004254591756722e-06,
      "loss": 0.4974,
      "step": 7904
    },
    {
      "epoch": 9.67564259485924,
      "grad_norm": 1.8112417314344553,
      "learning_rate": 1.399849933274145e-06,
      "loss": 0.3541,
      "step": 7905
    },
    {
      "epoch": 9.67686658506732,
      "grad_norm": 2.019690822404401,
      "learning_rate": 1.39927447967041e-06,
      "loss": 0.7365,
      "step": 7906
    },
    {
      "epoch": 9.678090575275398,
      "grad_norm": 2.4389126154149894,
      "learning_rate": 1.398699098402286e-06,
      "loss": 1.0279,
      "step": 7907
    },
    {
      "epoch": 9.679314565483477,
      "grad_norm": 1.806026717270288,
      "learning_rate": 1.3981237895075839e-06,
      "loss": 0.4745,
      "step": 7908
    },
    {
      "epoch": 9.680538555691555,
      "grad_norm": 2.5704059657017084,
      "learning_rate": 1.3975485530241116e-06,
      "loss": 0.5146,
      "step": 7909
    },
    {
      "epoch": 9.681762545899632,
      "grad_norm": 1.5269557164089338,
      "learning_rate": 1.3969733889896696e-06,
      "loss": 0.4089,
      "step": 7910
    },
    {
      "epoch": 9.68298653610771,
      "grad_norm": 1.2741160150708244,
      "learning_rate": 1.3963982974420559e-06,
      "loss": 0.6047,
      "step": 7911
    },
    {
      "epoch": 9.68421052631579,
      "grad_norm": 2.0035054042985236,
      "learning_rate": 1.395823278419065e-06,
      "loss": 0.7365,
      "step": 7912
    },
    {
      "epoch": 9.685434516523868,
      "grad_norm": 1.7814404376258857,
      "learning_rate": 1.3952483319584846e-06,
      "loss": 0.5708,
      "step": 7913
    },
    {
      "epoch": 9.686658506731947,
      "grad_norm": 2.2125335872229916,
      "learning_rate": 1.3946734580980972e-06,
      "loss": 0.9677,
      "step": 7914
    },
    {
      "epoch": 9.687882496940025,
      "grad_norm": 2.4284247779225923,
      "learning_rate": 1.394098656875682e-06,
      "loss": 0.4882,
      "step": 7915
    },
    {
      "epoch": 9.689106487148102,
      "grad_norm": 1.5066525532846922,
      "learning_rate": 1.3935239283290124e-06,
      "loss": 1.1195,
      "step": 7916
    },
    {
      "epoch": 9.69033047735618,
      "grad_norm": 1.7739658544798638,
      "learning_rate": 1.392949272495858e-06,
      "loss": 0.9136,
      "step": 7917
    },
    {
      "epoch": 9.69155446756426,
      "grad_norm": 1.34686496683287,
      "learning_rate": 1.3923746894139823e-06,
      "loss": 0.6485,
      "step": 7918
    },
    {
      "epoch": 9.692778457772338,
      "grad_norm": 2.573370594637822,
      "learning_rate": 1.3918001791211444e-06,
      "loss": 0.3989,
      "step": 7919
    },
    {
      "epoch": 9.694002447980417,
      "grad_norm": 1.5362361724905673,
      "learning_rate": 1.391225741655102e-06,
      "loss": 0.2251,
      "step": 7920
    },
    {
      "epoch": 9.695226438188495,
      "grad_norm": 1.5949376271782958,
      "learning_rate": 1.3906513770536015e-06,
      "loss": 1.2478,
      "step": 7921
    },
    {
      "epoch": 9.696450428396572,
      "grad_norm": 1.7036609886837253,
      "learning_rate": 1.3900770853543893e-06,
      "loss": 0.7342,
      "step": 7922
    },
    {
      "epoch": 9.69767441860465,
      "grad_norm": 1.9481514454936066,
      "learning_rate": 1.389502866595206e-06,
      "loss": 0.5316,
      "step": 7923
    },
    {
      "epoch": 9.69889840881273,
      "grad_norm": 1.5251403320729136,
      "learning_rate": 1.3889287208137856e-06,
      "loss": 0.9237,
      "step": 7924
    },
    {
      "epoch": 9.700122399020808,
      "grad_norm": 1.4005386934755517,
      "learning_rate": 1.3883546480478611e-06,
      "loss": 0.5344,
      "step": 7925
    },
    {
      "epoch": 9.701346389228886,
      "grad_norm": 0.8040057559797467,
      "learning_rate": 1.3877806483351586e-06,
      "loss": 0.2191,
      "step": 7926
    },
    {
      "epoch": 9.702570379436965,
      "grad_norm": 1.5271536758140112,
      "learning_rate": 1.387206721713397e-06,
      "loss": 0.8235,
      "step": 7927
    },
    {
      "epoch": 9.703794369645042,
      "grad_norm": 1.1375149961408149,
      "learning_rate": 1.386632868220293e-06,
      "loss": 0.5374,
      "step": 7928
    },
    {
      "epoch": 9.70501835985312,
      "grad_norm": 2.7828863520663134,
      "learning_rate": 1.3860590878935593e-06,
      "loss": 0.4272,
      "step": 7929
    },
    {
      "epoch": 9.7062423500612,
      "grad_norm": 1.0195635340317435,
      "learning_rate": 1.3854853807709023e-06,
      "loss": 0.5874,
      "step": 7930
    },
    {
      "epoch": 9.707466340269278,
      "grad_norm": 2.7226350872138916,
      "learning_rate": 1.3849117468900238e-06,
      "loss": 0.5689,
      "step": 7931
    },
    {
      "epoch": 9.708690330477356,
      "grad_norm": 1.5618211395494557,
      "learning_rate": 1.384338186288621e-06,
      "loss": 0.4781,
      "step": 7932
    },
    {
      "epoch": 9.709914320685435,
      "grad_norm": 1.1922584984490505,
      "learning_rate": 1.3837646990043859e-06,
      "loss": 0.5401,
      "step": 7933
    },
    {
      "epoch": 9.711138310893514,
      "grad_norm": 2.3285419877722244,
      "learning_rate": 1.3831912850750062e-06,
      "loss": 0.5485,
      "step": 7934
    },
    {
      "epoch": 9.71236230110159,
      "grad_norm": 1.7883402571322842,
      "learning_rate": 1.3826179445381647e-06,
      "loss": 0.549,
      "step": 7935
    },
    {
      "epoch": 9.713586291309669,
      "grad_norm": 1.1940255368098498,
      "learning_rate": 1.3820446774315388e-06,
      "loss": 0.4748,
      "step": 7936
    },
    {
      "epoch": 9.714810281517748,
      "grad_norm": 1.575311527524914,
      "learning_rate": 1.3814714837928006e-06,
      "loss": 0.5813,
      "step": 7937
    },
    {
      "epoch": 9.716034271725826,
      "grad_norm": 1.920648392276851,
      "learning_rate": 1.3808983636596213e-06,
      "loss": 0.7495,
      "step": 7938
    },
    {
      "epoch": 9.717258261933905,
      "grad_norm": 1.9122446903195498,
      "learning_rate": 1.3803253170696612e-06,
      "loss": 0.4204,
      "step": 7939
    },
    {
      "epoch": 9.718482252141984,
      "grad_norm": 1.4910147609659479,
      "learning_rate": 1.3797523440605798e-06,
      "loss": 0.5301,
      "step": 7940
    },
    {
      "epoch": 9.71970624235006,
      "grad_norm": 1.5586942573870273,
      "learning_rate": 1.3791794446700302e-06,
      "loss": 1.0693,
      "step": 7941
    },
    {
      "epoch": 9.720930232558139,
      "grad_norm": 1.062781783465664,
      "learning_rate": 1.3786066189356628e-06,
      "loss": 0.5968,
      "step": 7942
    },
    {
      "epoch": 9.722154222766218,
      "grad_norm": 1.3869127936036552,
      "learning_rate": 1.3780338668951205e-06,
      "loss": 0.697,
      "step": 7943
    },
    {
      "epoch": 9.723378212974296,
      "grad_norm": 3.754608410003829,
      "learning_rate": 1.3774611885860434e-06,
      "loss": 0.3876,
      "step": 7944
    },
    {
      "epoch": 9.724602203182375,
      "grad_norm": 1.3922848199268503,
      "learning_rate": 1.3768885840460633e-06,
      "loss": 0.4031,
      "step": 7945
    },
    {
      "epoch": 9.725826193390454,
      "grad_norm": 1.5642192514310513,
      "learning_rate": 1.3763160533128122e-06,
      "loss": 0.2973,
      "step": 7946
    },
    {
      "epoch": 9.727050183598532,
      "grad_norm": 3.153555783694688,
      "learning_rate": 1.3757435964239141e-06,
      "loss": 0.3213,
      "step": 7947
    },
    {
      "epoch": 9.728274173806609,
      "grad_norm": 2.1682596151663174,
      "learning_rate": 1.3751712134169887e-06,
      "loss": 0.9232,
      "step": 7948
    },
    {
      "epoch": 9.729498164014688,
      "grad_norm": 2.0092496840646845,
      "learning_rate": 1.3745989043296502e-06,
      "loss": 0.6535,
      "step": 7949
    },
    {
      "epoch": 9.730722154222766,
      "grad_norm": 2.7421839687460907,
      "learning_rate": 1.374026669199509e-06,
      "loss": 0.4713,
      "step": 7950
    },
    {
      "epoch": 9.731946144430845,
      "grad_norm": 1.4311792871406153,
      "learning_rate": 1.373454508064171e-06,
      "loss": 0.9699,
      "step": 7951
    },
    {
      "epoch": 9.733170134638923,
      "grad_norm": 2.6108756316929513,
      "learning_rate": 1.3728824209612356e-06,
      "loss": 1.1841,
      "step": 7952
    },
    {
      "epoch": 9.734394124847,
      "grad_norm": 1.3649728677935322,
      "learning_rate": 1.3723104079282987e-06,
      "loss": 1.0468,
      "step": 7953
    },
    {
      "epoch": 9.735618115055079,
      "grad_norm": 1.2286292938480798,
      "learning_rate": 1.3717384690029494e-06,
      "loss": 0.4803,
      "step": 7954
    },
    {
      "epoch": 9.736842105263158,
      "grad_norm": 1.2695329685799752,
      "learning_rate": 1.3711666042227772e-06,
      "loss": 0.5109,
      "step": 7955
    },
    {
      "epoch": 9.738066095471236,
      "grad_norm": 2.25379087666454,
      "learning_rate": 1.3705948136253593e-06,
      "loss": 0.4536,
      "step": 7956
    },
    {
      "epoch": 9.739290085679315,
      "grad_norm": 2.651539324117214,
      "learning_rate": 1.370023097248273e-06,
      "loss": 0.5837,
      "step": 7957
    },
    {
      "epoch": 9.740514075887393,
      "grad_norm": 1.1657780094027623,
      "learning_rate": 1.3694514551290882e-06,
      "loss": 0.543,
      "step": 7958
    },
    {
      "epoch": 9.741738066095472,
      "grad_norm": 2.7753768525757496,
      "learning_rate": 1.3688798873053733e-06,
      "loss": 0.3606,
      "step": 7959
    },
    {
      "epoch": 9.742962056303549,
      "grad_norm": 1.712212797418991,
      "learning_rate": 1.3683083938146879e-06,
      "loss": 0.5202,
      "step": 7960
    },
    {
      "epoch": 9.744186046511627,
      "grad_norm": 1.6035053031878594,
      "learning_rate": 1.3677369746945896e-06,
      "loss": 0.4578,
      "step": 7961
    },
    {
      "epoch": 9.745410036719706,
      "grad_norm": 1.369445377042963,
      "learning_rate": 1.3671656299826302e-06,
      "loss": 0.4889,
      "step": 7962
    },
    {
      "epoch": 9.746634026927785,
      "grad_norm": 1.3119321635290635,
      "learning_rate": 1.3665943597163537e-06,
      "loss": 0.6274,
      "step": 7963
    },
    {
      "epoch": 9.747858017135863,
      "grad_norm": 2.5926756422167996,
      "learning_rate": 1.3660231639333043e-06,
      "loss": 0.9545,
      "step": 7964
    },
    {
      "epoch": 9.749082007343942,
      "grad_norm": 1.4869742578345693,
      "learning_rate": 1.3654520426710188e-06,
      "loss": 1.4116,
      "step": 7965
    },
    {
      "epoch": 9.750305997552019,
      "grad_norm": 2.558374331211004,
      "learning_rate": 1.3648809959670283e-06,
      "loss": 0.2816,
      "step": 7966
    },
    {
      "epoch": 9.751529987760097,
      "grad_norm": 1.8324285470064825,
      "learning_rate": 1.3643100238588603e-06,
      "loss": 0.6534,
      "step": 7967
    },
    {
      "epoch": 9.752753977968176,
      "grad_norm": 2.1617208159238888,
      "learning_rate": 1.363739126384037e-06,
      "loss": 0.6807,
      "step": 7968
    },
    {
      "epoch": 9.753977968176255,
      "grad_norm": 2.280708071464484,
      "learning_rate": 1.3631683035800753e-06,
      "loss": 0.5004,
      "step": 7969
    },
    {
      "epoch": 9.755201958384333,
      "grad_norm": 2.2654461758225106,
      "learning_rate": 1.362597555484488e-06,
      "loss": 0.4941,
      "step": 7970
    },
    {
      "epoch": 9.756425948592412,
      "grad_norm": 1.577259656422622,
      "learning_rate": 1.3620268821347815e-06,
      "loss": 1.0454,
      "step": 7971
    },
    {
      "epoch": 9.75764993880049,
      "grad_norm": 1.6850722920389036,
      "learning_rate": 1.3614562835684603e-06,
      "loss": 1.4016,
      "step": 7972
    },
    {
      "epoch": 9.758873929008567,
      "grad_norm": 1.8050305245592908,
      "learning_rate": 1.3608857598230215e-06,
      "loss": 0.491,
      "step": 7973
    },
    {
      "epoch": 9.760097919216646,
      "grad_norm": 2.7753517712366595,
      "learning_rate": 1.3603153109359568e-06,
      "loss": 1.0649,
      "step": 7974
    },
    {
      "epoch": 9.761321909424725,
      "grad_norm": 3.0081974476209443,
      "learning_rate": 1.3597449369447541e-06,
      "loss": 0.4678,
      "step": 7975
    },
    {
      "epoch": 9.762545899632803,
      "grad_norm": 1.3790644568257628,
      "learning_rate": 1.3591746378868959e-06,
      "loss": 0.4366,
      "step": 7976
    },
    {
      "epoch": 9.763769889840882,
      "grad_norm": 1.6788270304581494,
      "learning_rate": 1.3586044137998617e-06,
      "loss": 1.1759,
      "step": 7977
    },
    {
      "epoch": 9.76499388004896,
      "grad_norm": 1.5318741515706569,
      "learning_rate": 1.3580342647211236e-06,
      "loss": 2.2828,
      "step": 7978
    },
    {
      "epoch": 9.766217870257037,
      "grad_norm": 2.132443454682794,
      "learning_rate": 1.3574641906881504e-06,
      "loss": 0.5771,
      "step": 7979
    },
    {
      "epoch": 9.767441860465116,
      "grad_norm": 1.8681997852020857,
      "learning_rate": 1.3568941917384038e-06,
      "loss": 0.5884,
      "step": 7980
    },
    {
      "epoch": 9.768665850673194,
      "grad_norm": 2.483442977222384,
      "learning_rate": 1.3563242679093433e-06,
      "loss": 0.2374,
      "step": 7981
    },
    {
      "epoch": 9.769889840881273,
      "grad_norm": 1.792394140597607,
      "learning_rate": 1.3557544192384217e-06,
      "loss": 0.5337,
      "step": 7982
    },
    {
      "epoch": 9.771113831089352,
      "grad_norm": 1.871923691677858,
      "learning_rate": 1.3551846457630874e-06,
      "loss": 1.3919,
      "step": 7983
    },
    {
      "epoch": 9.77233782129743,
      "grad_norm": 2.553470644664137,
      "learning_rate": 1.3546149475207836e-06,
      "loss": 0.9168,
      "step": 7984
    },
    {
      "epoch": 9.773561811505507,
      "grad_norm": 1.9195844527159664,
      "learning_rate": 1.3540453245489493e-06,
      "loss": 0.3978,
      "step": 7985
    },
    {
      "epoch": 9.774785801713586,
      "grad_norm": 2.5286894843545085,
      "learning_rate": 1.3534757768850176e-06,
      "loss": 0.6752,
      "step": 7986
    },
    {
      "epoch": 9.776009791921664,
      "grad_norm": 1.7435202679144155,
      "learning_rate": 1.352906304566417e-06,
      "loss": 0.4211,
      "step": 7987
    },
    {
      "epoch": 9.777233782129743,
      "grad_norm": 2.233622415336265,
      "learning_rate": 1.3523369076305715e-06,
      "loss": 0.3543,
      "step": 7988
    },
    {
      "epoch": 9.778457772337822,
      "grad_norm": 1.4218462622780295,
      "learning_rate": 1.3517675861148982e-06,
      "loss": 1.3202,
      "step": 7989
    },
    {
      "epoch": 9.7796817625459,
      "grad_norm": 1.9995611685756287,
      "learning_rate": 1.3511983400568132e-06,
      "loss": 0.4402,
      "step": 7990
    },
    {
      "epoch": 9.780905752753977,
      "grad_norm": 1.3755761378221443,
      "learning_rate": 1.3506291694937252e-06,
      "loss": 0.368,
      "step": 7991
    },
    {
      "epoch": 9.782129742962056,
      "grad_norm": 1.7211652047967565,
      "learning_rate": 1.350060074463036e-06,
      "loss": 0.4515,
      "step": 7992
    },
    {
      "epoch": 9.783353733170134,
      "grad_norm": 3.0247166437372845,
      "learning_rate": 1.3494910550021445e-06,
      "loss": 0.5809,
      "step": 7993
    },
    {
      "epoch": 9.784577723378213,
      "grad_norm": 3.3940690297809146,
      "learning_rate": 1.3489221111484463e-06,
      "loss": 0.4208,
      "step": 7994
    },
    {
      "epoch": 9.785801713586292,
      "grad_norm": 1.3294161011406622,
      "learning_rate": 1.3483532429393291e-06,
      "loss": 0.68,
      "step": 7995
    },
    {
      "epoch": 9.78702570379437,
      "grad_norm": 1.9862537328369367,
      "learning_rate": 1.3477844504121776e-06,
      "loss": 0.9301,
      "step": 7996
    },
    {
      "epoch": 9.788249694002449,
      "grad_norm": 2.0135147164123515,
      "learning_rate": 1.3472157336043711e-06,
      "loss": 1.074,
      "step": 7997
    },
    {
      "epoch": 9.789473684210526,
      "grad_norm": 2.3409010296086077,
      "learning_rate": 1.346647092553281e-06,
      "loss": 0.4746,
      "step": 7998
    },
    {
      "epoch": 9.790697674418604,
      "grad_norm": 1.8285893979473264,
      "learning_rate": 1.3460785272962784e-06,
      "loss": 0.4588,
      "step": 7999
    },
    {
      "epoch": 9.791921664626683,
      "grad_norm": 1.6193666017388635,
      "learning_rate": 1.3455100378707276e-06,
      "loss": 0.4912,
      "step": 8000
    },
    {
      "epoch": 9.793145654834762,
      "grad_norm": 1.4663630193535033,
      "learning_rate": 1.3449416243139863e-06,
      "loss": 0.6176,
      "step": 8001
    },
    {
      "epoch": 9.79436964504284,
      "grad_norm": 1.847407936984587,
      "learning_rate": 1.3443732866634096e-06,
      "loss": 1.0126,
      "step": 8002
    },
    {
      "epoch": 9.795593635250919,
      "grad_norm": 1.2915425901654443,
      "learning_rate": 1.3438050249563461e-06,
      "loss": 0.977,
      "step": 8003
    },
    {
      "epoch": 9.796817625458996,
      "grad_norm": 1.0879716874868688,
      "learning_rate": 1.3432368392301397e-06,
      "loss": 0.5391,
      "step": 8004
    },
    {
      "epoch": 9.798041615667074,
      "grad_norm": 1.4768485969881757,
      "learning_rate": 1.34266872952213e-06,
      "loss": 0.6308,
      "step": 8005
    },
    {
      "epoch": 9.799265605875153,
      "grad_norm": 2.442793213840337,
      "learning_rate": 1.3421006958696495e-06,
      "loss": 0.5038,
      "step": 8006
    },
    {
      "epoch": 9.800489596083231,
      "grad_norm": 3.551107960139448,
      "learning_rate": 1.3415327383100297e-06,
      "loss": 0.4807,
      "step": 8007
    },
    {
      "epoch": 9.80171358629131,
      "grad_norm": 1.2262453504172577,
      "learning_rate": 1.340964856880593e-06,
      "loss": 0.4716,
      "step": 8008
    },
    {
      "epoch": 9.802937576499389,
      "grad_norm": 1.6222198961968428,
      "learning_rate": 1.34039705161866e-06,
      "loss": 0.5496,
      "step": 8009
    },
    {
      "epoch": 9.804161566707466,
      "grad_norm": 2.3803470921417484,
      "learning_rate": 1.339829322561543e-06,
      "loss": 0.4025,
      "step": 8010
    },
    {
      "epoch": 9.805385556915544,
      "grad_norm": 0.997432480439704,
      "learning_rate": 1.3392616697465507e-06,
      "loss": 0.4412,
      "step": 8011
    },
    {
      "epoch": 9.806609547123623,
      "grad_norm": 2.6626344846381533,
      "learning_rate": 1.3386940932109894e-06,
      "loss": 0.3651,
      "step": 8012
    },
    {
      "epoch": 9.807833537331701,
      "grad_norm": 2.1076255856199118,
      "learning_rate": 1.3381265929921566e-06,
      "loss": 0.423,
      "step": 8013
    },
    {
      "epoch": 9.80905752753978,
      "grad_norm": 2.155288162424569,
      "learning_rate": 1.3375591691273465e-06,
      "loss": 0.6179,
      "step": 8014
    },
    {
      "epoch": 9.810281517747859,
      "grad_norm": 1.020988069391773,
      "learning_rate": 1.3369918216538486e-06,
      "loss": 0.5283,
      "step": 8015
    },
    {
      "epoch": 9.811505507955935,
      "grad_norm": 1.0531956783136196,
      "learning_rate": 1.3364245506089463e-06,
      "loss": 0.6314,
      "step": 8016
    },
    {
      "epoch": 9.812729498164014,
      "grad_norm": 2.256314816581433,
      "learning_rate": 1.3358573560299188e-06,
      "loss": 0.3309,
      "step": 8017
    },
    {
      "epoch": 9.813953488372093,
      "grad_norm": 1.7406861177445812,
      "learning_rate": 1.33529023795404e-06,
      "loss": 0.6584,
      "step": 8018
    },
    {
      "epoch": 9.815177478580171,
      "grad_norm": 1.729475474888018,
      "learning_rate": 1.3347231964185774e-06,
      "loss": 0.7518,
      "step": 8019
    },
    {
      "epoch": 9.81640146878825,
      "grad_norm": 1.4813776660514777,
      "learning_rate": 1.3341562314607989e-06,
      "loss": 0.8742,
      "step": 8020
    },
    {
      "epoch": 9.817625458996329,
      "grad_norm": 1.815358076413664,
      "learning_rate": 1.3335893431179595e-06,
      "loss": 0.2987,
      "step": 8021
    },
    {
      "epoch": 9.818849449204407,
      "grad_norm": 1.1101863653528696,
      "learning_rate": 1.3330225314273142e-06,
      "loss": 0.6421,
      "step": 8022
    },
    {
      "epoch": 9.820073439412484,
      "grad_norm": 1.7811367614170501,
      "learning_rate": 1.3324557964261117e-06,
      "loss": 0.5298,
      "step": 8023
    },
    {
      "epoch": 9.821297429620563,
      "grad_norm": 2.1466995107353455,
      "learning_rate": 1.331889138151595e-06,
      "loss": 0.3794,
      "step": 8024
    },
    {
      "epoch": 9.822521419828641,
      "grad_norm": 1.338826732279015,
      "learning_rate": 1.3313225566410042e-06,
      "loss": 0.7136,
      "step": 8025
    },
    {
      "epoch": 9.82374541003672,
      "grad_norm": 1.8117744573329801,
      "learning_rate": 1.330756051931573e-06,
      "loss": 1.0233,
      "step": 8026
    },
    {
      "epoch": 9.824969400244798,
      "grad_norm": 2.558987007209238,
      "learning_rate": 1.33018962406053e-06,
      "loss": 0.8581,
      "step": 8027
    },
    {
      "epoch": 9.826193390452877,
      "grad_norm": 1.7708616995895166,
      "learning_rate": 1.329623273065096e-06,
      "loss": 1.4242,
      "step": 8028
    },
    {
      "epoch": 9.827417380660954,
      "grad_norm": 1.5961587348581474,
      "learning_rate": 1.3290569989824926e-06,
      "loss": 1.1718,
      "step": 8029
    },
    {
      "epoch": 9.828641370869033,
      "grad_norm": 1.6152438610091786,
      "learning_rate": 1.328490801849932e-06,
      "loss": 1.0163,
      "step": 8030
    },
    {
      "epoch": 9.829865361077111,
      "grad_norm": 2.2100971686955884,
      "learning_rate": 1.3279246817046233e-06,
      "loss": 0.5333,
      "step": 8031
    },
    {
      "epoch": 9.83108935128519,
      "grad_norm": 2.09330299814667,
      "learning_rate": 1.3273586385837688e-06,
      "loss": 0.5388,
      "step": 8032
    },
    {
      "epoch": 9.832313341493268,
      "grad_norm": 1.2913438100053727,
      "learning_rate": 1.3267926725245675e-06,
      "loss": 0.7753,
      "step": 8033
    },
    {
      "epoch": 9.833537331701347,
      "grad_norm": 2.5680303976268695,
      "learning_rate": 1.3262267835642122e-06,
      "loss": 0.4297,
      "step": 8034
    },
    {
      "epoch": 9.834761321909426,
      "grad_norm": 1.235802409813482,
      "learning_rate": 1.3256609717398916e-06,
      "loss": 0.5568,
      "step": 8035
    },
    {
      "epoch": 9.835985312117502,
      "grad_norm": 1.517519334032845,
      "learning_rate": 1.3250952370887881e-06,
      "loss": 1.4044,
      "step": 8036
    },
    {
      "epoch": 9.837209302325581,
      "grad_norm": 1.5942145995008146,
      "learning_rate": 1.324529579648079e-06,
      "loss": 0.929,
      "step": 8037
    },
    {
      "epoch": 9.83843329253366,
      "grad_norm": 1.5594342229237552,
      "learning_rate": 1.3239639994549402e-06,
      "loss": 0.6025,
      "step": 8038
    },
    {
      "epoch": 9.839657282741738,
      "grad_norm": 2.085534351577223,
      "learning_rate": 1.3233984965465363e-06,
      "loss": 0.8503,
      "step": 8039
    },
    {
      "epoch": 9.840881272949817,
      "grad_norm": 2.340798645266233,
      "learning_rate": 1.3228330709600311e-06,
      "loss": 0.5028,
      "step": 8040
    },
    {
      "epoch": 9.842105263157894,
      "grad_norm": 0.9059008798024042,
      "learning_rate": 1.322267722732582e-06,
      "loss": 0.4654,
      "step": 8041
    },
    {
      "epoch": 9.843329253365972,
      "grad_norm": 1.356303355752724,
      "learning_rate": 1.3217024519013427e-06,
      "loss": 0.7781,
      "step": 8042
    },
    {
      "epoch": 9.844553243574051,
      "grad_norm": 1.1116084051416137,
      "learning_rate": 1.32113725850346e-06,
      "loss": 0.5605,
      "step": 8043
    },
    {
      "epoch": 9.84577723378213,
      "grad_norm": 1.0379963082103314,
      "learning_rate": 1.3205721425760775e-06,
      "loss": 0.5931,
      "step": 8044
    },
    {
      "epoch": 9.847001223990208,
      "grad_norm": 3.014244156687469,
      "learning_rate": 1.3200071041563295e-06,
      "loss": 0.3801,
      "step": 8045
    },
    {
      "epoch": 9.848225214198287,
      "grad_norm": 1.5970972817302973,
      "learning_rate": 1.319442143281351e-06,
      "loss": 0.8606,
      "step": 8046
    },
    {
      "epoch": 9.849449204406366,
      "grad_norm": 1.6243014755566054,
      "learning_rate": 1.3188772599882683e-06,
      "loss": 0.4926,
      "step": 8047
    },
    {
      "epoch": 9.850673194614442,
      "grad_norm": 1.3576925305407561,
      "learning_rate": 1.318312454314203e-06,
      "loss": 0.5479,
      "step": 8048
    },
    {
      "epoch": 9.851897184822521,
      "grad_norm": 2.07896350509473,
      "learning_rate": 1.3177477262962729e-06,
      "loss": 0.5163,
      "step": 8049
    },
    {
      "epoch": 9.8531211750306,
      "grad_norm": 1.1742384952908267,
      "learning_rate": 1.3171830759715893e-06,
      "loss": 0.6611,
      "step": 8050
    },
    {
      "epoch": 9.854345165238678,
      "grad_norm": 1.3390240074058366,
      "learning_rate": 1.3166185033772589e-06,
      "loss": 0.5183,
      "step": 8051
    },
    {
      "epoch": 9.855569155446757,
      "grad_norm": 2.1154485430279633,
      "learning_rate": 1.3160540085503836e-06,
      "loss": 1.0176,
      "step": 8052
    },
    {
      "epoch": 9.856793145654835,
      "grad_norm": 1.6772393558237146,
      "learning_rate": 1.3154895915280596e-06,
      "loss": 1.6168,
      "step": 8053
    },
    {
      "epoch": 9.858017135862912,
      "grad_norm": 1.8609102070246282,
      "learning_rate": 1.3149252523473773e-06,
      "loss": 0.5943,
      "step": 8054
    },
    {
      "epoch": 9.859241126070991,
      "grad_norm": 1.6971387315376707,
      "learning_rate": 1.3143609910454252e-06,
      "loss": 0.5034,
      "step": 8055
    },
    {
      "epoch": 9.86046511627907,
      "grad_norm": 2.590757944734166,
      "learning_rate": 1.3137968076592847e-06,
      "loss": 0.4953,
      "step": 8056
    },
    {
      "epoch": 9.861689106487148,
      "grad_norm": 1.1483565394422681,
      "learning_rate": 1.3132327022260295e-06,
      "loss": 0.6951,
      "step": 8057
    },
    {
      "epoch": 9.862913096695227,
      "grad_norm": 1.3715307972084403,
      "learning_rate": 1.3126686747827303e-06,
      "loss": 0.5065,
      "step": 8058
    },
    {
      "epoch": 9.864137086903305,
      "grad_norm": 1.020385223578941,
      "learning_rate": 1.3121047253664555e-06,
      "loss": 0.5123,
      "step": 8059
    },
    {
      "epoch": 9.865361077111384,
      "grad_norm": 2.0791068406193873,
      "learning_rate": 1.3115408540142639e-06,
      "loss": 0.9728,
      "step": 8060
    },
    {
      "epoch": 9.86658506731946,
      "grad_norm": 1.291137886221936,
      "learning_rate": 1.3109770607632119e-06,
      "loss": 0.5921,
      "step": 8061
    },
    {
      "epoch": 9.86780905752754,
      "grad_norm": 1.995161090573588,
      "learning_rate": 1.3104133456503504e-06,
      "loss": 0.4117,
      "step": 8062
    },
    {
      "epoch": 9.869033047735618,
      "grad_norm": 2.142027805824703,
      "learning_rate": 1.3098497087127222e-06,
      "loss": 0.5806,
      "step": 8063
    },
    {
      "epoch": 9.870257037943697,
      "grad_norm": 1.8066999786471458,
      "learning_rate": 1.30928614998737e-06,
      "loss": 0.3917,
      "step": 8064
    },
    {
      "epoch": 9.871481028151775,
      "grad_norm": 1.6420458653040844,
      "learning_rate": 1.3087226695113276e-06,
      "loss": 0.5743,
      "step": 8065
    },
    {
      "epoch": 9.872705018359854,
      "grad_norm": 1.8625882107139966,
      "learning_rate": 1.3081592673216252e-06,
      "loss": 0.5333,
      "step": 8066
    },
    {
      "epoch": 9.87392900856793,
      "grad_norm": 2.559882595034093,
      "learning_rate": 1.3075959434552876e-06,
      "loss": 0.4058,
      "step": 8067
    },
    {
      "epoch": 9.87515299877601,
      "grad_norm": 1.7933464243498776,
      "learning_rate": 1.3070326979493338e-06,
      "loss": 1.0853,
      "step": 8068
    },
    {
      "epoch": 9.876376988984088,
      "grad_norm": 1.7445095797078758,
      "learning_rate": 1.306469530840779e-06,
      "loss": 0.9882,
      "step": 8069
    },
    {
      "epoch": 9.877600979192167,
      "grad_norm": 2.056815653286675,
      "learning_rate": 1.305906442166632e-06,
      "loss": 0.7186,
      "step": 8070
    },
    {
      "epoch": 9.878824969400245,
      "grad_norm": 1.2407986748266806,
      "learning_rate": 1.305343431963896e-06,
      "loss": 0.8955,
      "step": 8071
    },
    {
      "epoch": 9.880048959608324,
      "grad_norm": 2.759552075835713,
      "learning_rate": 1.3047805002695718e-06,
      "loss": 0.4134,
      "step": 8072
    },
    {
      "epoch": 9.8812729498164,
      "grad_norm": 1.5772786785977546,
      "learning_rate": 1.3042176471206536e-06,
      "loss": 0.7017,
      "step": 8073
    },
    {
      "epoch": 9.88249694002448,
      "grad_norm": 1.12952913792466,
      "learning_rate": 1.3036548725541274e-06,
      "loss": 0.5631,
      "step": 8074
    },
    {
      "epoch": 9.883720930232558,
      "grad_norm": 1.8853864236805709,
      "learning_rate": 1.3030921766069782e-06,
      "loss": 0.3928,
      "step": 8075
    },
    {
      "epoch": 9.884944920440637,
      "grad_norm": 1.5306608325133928,
      "learning_rate": 1.302529559316183e-06,
      "loss": 0.4799,
      "step": 8076
    },
    {
      "epoch": 9.886168910648715,
      "grad_norm": 1.3990669146836523,
      "learning_rate": 1.3019670207187168e-06,
      "loss": 0.5331,
      "step": 8077
    },
    {
      "epoch": 9.887392900856794,
      "grad_norm": 2.73480818433506,
      "learning_rate": 1.3014045608515474e-06,
      "loss": 0.4548,
      "step": 8078
    },
    {
      "epoch": 9.88861689106487,
      "grad_norm": 1.9907632264119175,
      "learning_rate": 1.3008421797516362e-06,
      "loss": 1.3199,
      "step": 8079
    },
    {
      "epoch": 9.88984088127295,
      "grad_norm": 1.0702868473666032,
      "learning_rate": 1.3002798774559422e-06,
      "loss": 0.3641,
      "step": 8080
    },
    {
      "epoch": 9.891064871481028,
      "grad_norm": 2.450308243490088,
      "learning_rate": 1.299717654001417e-06,
      "loss": 0.4247,
      "step": 8081
    },
    {
      "epoch": 9.892288861689106,
      "grad_norm": 0.9317481090065809,
      "learning_rate": 1.299155509425008e-06,
      "loss": 0.4284,
      "step": 8082
    },
    {
      "epoch": 9.893512851897185,
      "grad_norm": 1.446039585315098,
      "learning_rate": 1.2985934437636571e-06,
      "loss": 0.5624,
      "step": 8083
    },
    {
      "epoch": 9.894736842105264,
      "grad_norm": 2.0869902215945944,
      "learning_rate": 1.2980314570543007e-06,
      "loss": 0.6139,
      "step": 8084
    },
    {
      "epoch": 9.895960832313342,
      "grad_norm": 2.371140325055438,
      "learning_rate": 1.297469549333873e-06,
      "loss": 0.4909,
      "step": 8085
    },
    {
      "epoch": 9.89718482252142,
      "grad_norm": 2.0471134093654477,
      "learning_rate": 1.2969077206392978e-06,
      "loss": 0.8582,
      "step": 8086
    },
    {
      "epoch": 9.898408812729498,
      "grad_norm": 1.1215147470462608,
      "learning_rate": 1.296345971007497e-06,
      "loss": 1.2,
      "step": 8087
    },
    {
      "epoch": 9.899632802937576,
      "grad_norm": 1.9422197951080777,
      "learning_rate": 1.2957843004753872e-06,
      "loss": 1.2937,
      "step": 8088
    },
    {
      "epoch": 9.900856793145655,
      "grad_norm": 2.544173687299527,
      "learning_rate": 1.2952227090798784e-06,
      "loss": 0.4012,
      "step": 8089
    },
    {
      "epoch": 9.902080783353734,
      "grad_norm": 2.105821548421532,
      "learning_rate": 1.2946611968578776e-06,
      "loss": 1.1894,
      "step": 8090
    },
    {
      "epoch": 9.903304773561812,
      "grad_norm": 1.5427227474288623,
      "learning_rate": 1.2940997638462858e-06,
      "loss": 0.3576,
      "step": 8091
    },
    {
      "epoch": 9.904528763769889,
      "grad_norm": 1.6426363785752642,
      "learning_rate": 1.293538410081996e-06,
      "loss": 0.7311,
      "step": 8092
    },
    {
      "epoch": 9.905752753977968,
      "grad_norm": 2.3765374565375836,
      "learning_rate": 1.2929771356018988e-06,
      "loss": 0.4657,
      "step": 8093
    },
    {
      "epoch": 9.906976744186046,
      "grad_norm": 1.8893264336677156,
      "learning_rate": 1.2924159404428804e-06,
      "loss": 0.5267,
      "step": 8094
    },
    {
      "epoch": 9.908200734394125,
      "grad_norm": 1.8785081050415589,
      "learning_rate": 1.2918548246418203e-06,
      "loss": 0.6874,
      "step": 8095
    },
    {
      "epoch": 9.909424724602204,
      "grad_norm": 2.106446790910935,
      "learning_rate": 1.2912937882355922e-06,
      "loss": 1.134,
      "step": 8096
    },
    {
      "epoch": 9.910648714810282,
      "grad_norm": 0.9121897954427349,
      "learning_rate": 1.2907328312610657e-06,
      "loss": 0.4405,
      "step": 8097
    },
    {
      "epoch": 9.911872705018359,
      "grad_norm": 1.5801891280845868,
      "learning_rate": 1.290171953755105e-06,
      "loss": 0.6424,
      "step": 8098
    },
    {
      "epoch": 9.913096695226438,
      "grad_norm": 1.5312489742730087,
      "learning_rate": 1.289611155754568e-06,
      "loss": 0.4663,
      "step": 8099
    },
    {
      "epoch": 9.914320685434516,
      "grad_norm": 1.832173899501259,
      "learning_rate": 1.2890504372963097e-06,
      "loss": 0.7327,
      "step": 8100
    },
    {
      "epoch": 9.915544675642595,
      "grad_norm": 1.4342790719848306,
      "learning_rate": 1.2884897984171774e-06,
      "loss": 0.6887,
      "step": 8101
    },
    {
      "epoch": 9.916768665850674,
      "grad_norm": 1.5407238468634554,
      "learning_rate": 1.2879292391540144e-06,
      "loss": 0.9194,
      "step": 8102
    },
    {
      "epoch": 9.917992656058752,
      "grad_norm": 0.7183188872005705,
      "learning_rate": 1.2873687595436584e-06,
      "loss": 0.2571,
      "step": 8103
    },
    {
      "epoch": 9.919216646266829,
      "grad_norm": 2.240760802304668,
      "learning_rate": 1.286808359622943e-06,
      "loss": 1.0402,
      "step": 8104
    },
    {
      "epoch": 9.920440636474908,
      "grad_norm": 2.2041270011574223,
      "learning_rate": 1.2862480394286943e-06,
      "loss": 0.4201,
      "step": 8105
    },
    {
      "epoch": 9.921664626682986,
      "grad_norm": 1.9233089290326,
      "learning_rate": 1.2856877989977345e-06,
      "loss": 1.0515,
      "step": 8106
    },
    {
      "epoch": 9.922888616891065,
      "grad_norm": 2.066078765763395,
      "learning_rate": 1.2851276383668818e-06,
      "loss": 1.2067,
      "step": 8107
    },
    {
      "epoch": 9.924112607099143,
      "grad_norm": 1.72009800882765,
      "learning_rate": 1.2845675575729472e-06,
      "loss": 0.4748,
      "step": 8108
    },
    {
      "epoch": 9.925336597307222,
      "grad_norm": 1.4632596918282041,
      "learning_rate": 1.2840075566527382e-06,
      "loss": 0.2737,
      "step": 8109
    },
    {
      "epoch": 9.9265605875153,
      "grad_norm": 2.2461122903742936,
      "learning_rate": 1.283447635643053e-06,
      "loss": 1.021,
      "step": 8110
    },
    {
      "epoch": 9.927784577723378,
      "grad_norm": 2.640656277716113,
      "learning_rate": 1.2828877945806906e-06,
      "loss": 0.5183,
      "step": 8111
    },
    {
      "epoch": 9.929008567931456,
      "grad_norm": 2.355921390401225,
      "learning_rate": 1.2823280335024408e-06,
      "loss": 0.3703,
      "step": 8112
    },
    {
      "epoch": 9.930232558139535,
      "grad_norm": 1.3900421040770787,
      "learning_rate": 1.281768352445088e-06,
      "loss": 1.2829,
      "step": 8113
    },
    {
      "epoch": 9.931456548347613,
      "grad_norm": 1.1959243631425074,
      "learning_rate": 1.2812087514454135e-06,
      "loss": 0.4423,
      "step": 8114
    },
    {
      "epoch": 9.932680538555692,
      "grad_norm": 0.8582855255436878,
      "learning_rate": 1.2806492305401922e-06,
      "loss": 0.3505,
      "step": 8115
    },
    {
      "epoch": 9.93390452876377,
      "grad_norm": 1.3458561371163833,
      "learning_rate": 1.280089789766193e-06,
      "loss": 0.8735,
      "step": 8116
    },
    {
      "epoch": 9.935128518971847,
      "grad_norm": 1.2534729354487257,
      "learning_rate": 1.2795304291601806e-06,
      "loss": 0.4947,
      "step": 8117
    },
    {
      "epoch": 9.936352509179926,
      "grad_norm": 1.2740451163117554,
      "learning_rate": 1.2789711487589145e-06,
      "loss": 0.7332,
      "step": 8118
    },
    {
      "epoch": 9.937576499388005,
      "grad_norm": 0.9986660541729924,
      "learning_rate": 1.2784119485991475e-06,
      "loss": 0.4433,
      "step": 8119
    },
    {
      "epoch": 9.938800489596083,
      "grad_norm": 1.7607862007543333,
      "learning_rate": 1.2778528287176303e-06,
      "loss": 0.4632,
      "step": 8120
    },
    {
      "epoch": 9.940024479804162,
      "grad_norm": 2.5198665226735155,
      "learning_rate": 1.277293789151104e-06,
      "loss": 0.3638,
      "step": 8121
    },
    {
      "epoch": 9.94124847001224,
      "grad_norm": 1.7615670651280837,
      "learning_rate": 1.2767348299363075e-06,
      "loss": 1.4606,
      "step": 8122
    },
    {
      "epoch": 9.94247246022032,
      "grad_norm": 1.5351733186565144,
      "learning_rate": 1.2761759511099725e-06,
      "loss": 1.5444,
      "step": 8123
    },
    {
      "epoch": 9.943696450428396,
      "grad_norm": 1.1190199926485953,
      "learning_rate": 1.275617152708828e-06,
      "loss": 0.5728,
      "step": 8124
    },
    {
      "epoch": 9.944920440636475,
      "grad_norm": 1.7935902615294652,
      "learning_rate": 1.2750584347695958e-06,
      "loss": 0.4428,
      "step": 8125
    },
    {
      "epoch": 9.946144430844553,
      "grad_norm": 1.6938499208156552,
      "learning_rate": 1.2744997973289925e-06,
      "loss": 0.6233,
      "step": 8126
    },
    {
      "epoch": 9.947368421052632,
      "grad_norm": 2.9855230531314922,
      "learning_rate": 1.2739412404237306e-06,
      "loss": 0.4046,
      "step": 8127
    },
    {
      "epoch": 9.94859241126071,
      "grad_norm": 1.707416395903036,
      "learning_rate": 1.2733827640905138e-06,
      "loss": 0.5271,
      "step": 8128
    },
    {
      "epoch": 9.949816401468787,
      "grad_norm": 1.8332145661999193,
      "learning_rate": 1.2728243683660455e-06,
      "loss": 0.6163,
      "step": 8129
    },
    {
      "epoch": 9.951040391676866,
      "grad_norm": 2.73647573936174,
      "learning_rate": 1.2722660532870206e-06,
      "loss": 0.3684,
      "step": 8130
    },
    {
      "epoch": 9.952264381884945,
      "grad_norm": 2.522206804658083,
      "learning_rate": 1.27170781889013e-06,
      "loss": 0.8212,
      "step": 8131
    },
    {
      "epoch": 9.953488372093023,
      "grad_norm": 1.725992155644301,
      "learning_rate": 1.2711496652120581e-06,
      "loss": 0.632,
      "step": 8132
    },
    {
      "epoch": 9.954712362301102,
      "grad_norm": 1.8060124171612597,
      "learning_rate": 1.270591592289485e-06,
      "loss": 1.1746,
      "step": 8133
    },
    {
      "epoch": 9.95593635250918,
      "grad_norm": 1.9068337294165791,
      "learning_rate": 1.2700336001590855e-06,
      "loss": 0.4862,
      "step": 8134
    },
    {
      "epoch": 9.957160342717259,
      "grad_norm": 1.9872543582078661,
      "learning_rate": 1.2694756888575283e-06,
      "loss": 0.6593,
      "step": 8135
    },
    {
      "epoch": 9.958384332925336,
      "grad_norm": 1.8445177704521656,
      "learning_rate": 1.2689178584214764e-06,
      "loss": 0.5101,
      "step": 8136
    },
    {
      "epoch": 9.959608323133414,
      "grad_norm": 2.615276944257637,
      "learning_rate": 1.2683601088875903e-06,
      "loss": 1.2841,
      "step": 8137
    },
    {
      "epoch": 9.960832313341493,
      "grad_norm": 2.739393669596744,
      "learning_rate": 1.2678024402925234e-06,
      "loss": 0.446,
      "step": 8138
    },
    {
      "epoch": 9.962056303549572,
      "grad_norm": 1.8035775616269512,
      "learning_rate": 1.2672448526729217e-06,
      "loss": 0.4265,
      "step": 8139
    },
    {
      "epoch": 9.96328029375765,
      "grad_norm": 1.3163824057565425,
      "learning_rate": 1.2666873460654283e-06,
      "loss": 0.3976,
      "step": 8140
    },
    {
      "epoch": 9.964504283965729,
      "grad_norm": 2.598975873256679,
      "learning_rate": 1.2661299205066802e-06,
      "loss": 0.4378,
      "step": 8141
    },
    {
      "epoch": 9.965728274173806,
      "grad_norm": 1.6961857512775436,
      "learning_rate": 1.2655725760333107e-06,
      "loss": 0.5427,
      "step": 8142
    },
    {
      "epoch": 9.966952264381884,
      "grad_norm": 1.3930666744594422,
      "learning_rate": 1.2650153126819458e-06,
      "loss": 0.5461,
      "step": 8143
    },
    {
      "epoch": 9.968176254589963,
      "grad_norm": 1.948271497494847,
      "learning_rate": 1.2644581304892073e-06,
      "loss": 0.6331,
      "step": 8144
    },
    {
      "epoch": 9.969400244798042,
      "grad_norm": 1.9450647135590244,
      "learning_rate": 1.2639010294917089e-06,
      "loss": 0.9635,
      "step": 8145
    },
    {
      "epoch": 9.97062423500612,
      "grad_norm": 2.3478983003175617,
      "learning_rate": 1.2633440097260639e-06,
      "loss": 0.5526,
      "step": 8146
    },
    {
      "epoch": 9.971848225214199,
      "grad_norm": 1.5092468577056113,
      "learning_rate": 1.2627870712288761e-06,
      "loss": 0.494,
      "step": 8147
    },
    {
      "epoch": 9.973072215422278,
      "grad_norm": 2.4133926173573963,
      "learning_rate": 1.2622302140367463e-06,
      "loss": 0.5415,
      "step": 8148
    },
    {
      "epoch": 9.974296205630354,
      "grad_norm": 1.288738391285761,
      "learning_rate": 1.2616734381862684e-06,
      "loss": 0.5845,
      "step": 8149
    },
    {
      "epoch": 9.975520195838433,
      "grad_norm": 2.232398482721865,
      "learning_rate": 1.2611167437140318e-06,
      "loss": 0.676,
      "step": 8150
    },
    {
      "epoch": 9.976744186046512,
      "grad_norm": 2.376702996911529,
      "learning_rate": 1.2605601306566206e-06,
      "loss": 0.4283,
      "step": 8151
    },
    {
      "epoch": 9.97796817625459,
      "grad_norm": 1.15348726919268,
      "learning_rate": 1.2600035990506132e-06,
      "loss": 0.4457,
      "step": 8152
    },
    {
      "epoch": 9.979192166462669,
      "grad_norm": 1.514528109000496,
      "learning_rate": 1.259447148932583e-06,
      "loss": 0.6075,
      "step": 8153
    },
    {
      "epoch": 9.980416156670747,
      "grad_norm": 1.3495700232353645,
      "learning_rate": 1.2588907803390964e-06,
      "loss": 0.4051,
      "step": 8154
    },
    {
      "epoch": 9.981640146878824,
      "grad_norm": 1.133810049852892,
      "learning_rate": 1.2583344933067184e-06,
      "loss": 0.5689,
      "step": 8155
    },
    {
      "epoch": 9.982864137086903,
      "grad_norm": 1.5252738752754431,
      "learning_rate": 1.2577782878720058e-06,
      "loss": 0.6255,
      "step": 8156
    },
    {
      "epoch": 9.984088127294982,
      "grad_norm": 2.1745258972358594,
      "learning_rate": 1.2572221640715086e-06,
      "loss": 0.4886,
      "step": 8157
    },
    {
      "epoch": 9.98531211750306,
      "grad_norm": 1.8286749212242521,
      "learning_rate": 1.2566661219417732e-06,
      "loss": 1.8838,
      "step": 8158
    },
    {
      "epoch": 9.986536107711139,
      "grad_norm": 1.6495961404778083,
      "learning_rate": 1.256110161519342e-06,
      "loss": 0.3891,
      "step": 8159
    },
    {
      "epoch": 9.987760097919217,
      "grad_norm": 1.3911984108296396,
      "learning_rate": 1.2555542828407501e-06,
      "loss": 1.5603,
      "step": 8160
    },
    {
      "epoch": 9.988984088127294,
      "grad_norm": 1.1510034179347792,
      "learning_rate": 1.2549984859425285e-06,
      "loss": 0.7474,
      "step": 8161
    },
    {
      "epoch": 9.990208078335373,
      "grad_norm": 1.8815924038221463,
      "learning_rate": 1.254442770861202e-06,
      "loss": 0.4201,
      "step": 8162
    },
    {
      "epoch": 9.991432068543451,
      "grad_norm": 1.9741281136590338,
      "learning_rate": 1.2538871376332878e-06,
      "loss": 0.3242,
      "step": 8163
    },
    {
      "epoch": 9.99265605875153,
      "grad_norm": 1.3254113287515683,
      "learning_rate": 1.2533315862953027e-06,
      "loss": 0.5261,
      "step": 8164
    },
    {
      "epoch": 9.993880048959609,
      "grad_norm": 2.1754619356017253,
      "learning_rate": 1.2527761168837546e-06,
      "loss": 0.9569,
      "step": 8165
    },
    {
      "epoch": 9.995104039167687,
      "grad_norm": 2.4335911602722717,
      "learning_rate": 1.252220729435147e-06,
      "loss": 0.5612,
      "step": 8166
    },
    {
      "epoch": 9.996328029375764,
      "grad_norm": 2.0015933829808867,
      "learning_rate": 1.251665423985978e-06,
      "loss": 0.4919,
      "step": 8167
    },
    {
      "epoch": 9.997552019583843,
      "grad_norm": 1.4686481146738966,
      "learning_rate": 1.2511102005727399e-06,
      "loss": 1.0264,
      "step": 8168
    },
    {
      "epoch": 9.998776009791921,
      "grad_norm": 2.749867258845465,
      "learning_rate": 1.2505550592319202e-06,
      "loss": 0.2142,
      "step": 8169
    },
    {
      "epoch": 10.0,
      "grad_norm": 2.6529243501104967,
      "learning_rate": 1.2500000000000007e-06,
      "loss": 0.5056,
      "step": 8170
    },
    {
      "epoch": 10.001223990208079,
      "grad_norm": 1.5272856604806633,
      "learning_rate": 1.2494450229134567e-06,
      "loss": 0.5623,
      "step": 8171
    },
    {
      "epoch": 10.002447980416157,
      "grad_norm": 1.5543008482735035,
      "learning_rate": 1.2488901280087613e-06,
      "loss": 0.5772,
      "step": 8172
    },
    {
      "epoch": 10.003671970624236,
      "grad_norm": 1.5202782545636946,
      "learning_rate": 1.2483353153223798e-06,
      "loss": 0.9298,
      "step": 8173
    },
    {
      "epoch": 10.004895960832313,
      "grad_norm": 1.5064042870638137,
      "learning_rate": 1.2477805848907712e-06,
      "loss": 0.4899,
      "step": 8174
    },
    {
      "epoch": 10.006119951040391,
      "grad_norm": 1.1641299581266906,
      "learning_rate": 1.2472259367503909e-06,
      "loss": 0.4808,
      "step": 8175
    },
    {
      "epoch": 10.00734394124847,
      "grad_norm": 2.5686894202595267,
      "learning_rate": 1.2466713709376875e-06,
      "loss": 0.3473,
      "step": 8176
    },
    {
      "epoch": 10.008567931456549,
      "grad_norm": 1.4986064961233794,
      "learning_rate": 1.2461168874891064e-06,
      "loss": 1.2136,
      "step": 8177
    },
    {
      "epoch": 10.009791921664627,
      "grad_norm": 2.012337930389982,
      "learning_rate": 1.245562486441086e-06,
      "loss": 0.4355,
      "step": 8178
    },
    {
      "epoch": 10.011015911872706,
      "grad_norm": 2.451759957510325,
      "learning_rate": 1.245008167830059e-06,
      "loss": 0.2295,
      "step": 8179
    },
    {
      "epoch": 10.012239902080783,
      "grad_norm": 1.9988336048301478,
      "learning_rate": 1.2444539316924534e-06,
      "loss": 1.138,
      "step": 8180
    },
    {
      "epoch": 10.013463892288861,
      "grad_norm": 1.385858015401202,
      "learning_rate": 1.2438997780646914e-06,
      "loss": 0.5383,
      "step": 8181
    },
    {
      "epoch": 10.01468788249694,
      "grad_norm": 1.7335782221932536,
      "learning_rate": 1.2433457069831901e-06,
      "loss": 0.5225,
      "step": 8182
    },
    {
      "epoch": 10.015911872705018,
      "grad_norm": 1.2246282604725975,
      "learning_rate": 1.2427917184843608e-06,
      "loss": 0.5791,
      "step": 8183
    },
    {
      "epoch": 10.017135862913097,
      "grad_norm": 1.932138429942333,
      "learning_rate": 1.2422378126046086e-06,
      "loss": 0.4143,
      "step": 8184
    },
    {
      "epoch": 10.018359853121176,
      "grad_norm": 1.0066729571219584,
      "learning_rate": 1.2416839893803372e-06,
      "loss": 0.5843,
      "step": 8185
    },
    {
      "epoch": 10.019583843329253,
      "grad_norm": 2.4721424105251684,
      "learning_rate": 1.241130248847939e-06,
      "loss": 0.4168,
      "step": 8186
    },
    {
      "epoch": 10.020807833537331,
      "grad_norm": 2.1850229102350776,
      "learning_rate": 1.2405765910438044e-06,
      "loss": 0.3745,
      "step": 8187
    },
    {
      "epoch": 10.02203182374541,
      "grad_norm": 1.7509585065173705,
      "learning_rate": 1.240023016004318e-06,
      "loss": 1.4018,
      "step": 8188
    },
    {
      "epoch": 10.023255813953488,
      "grad_norm": 1.4657240349013696,
      "learning_rate": 1.239469523765858e-06,
      "loss": 0.453,
      "step": 8189
    },
    {
      "epoch": 10.024479804161567,
      "grad_norm": 1.622072845639361,
      "learning_rate": 1.2389161143647992e-06,
      "loss": 1.6222,
      "step": 8190
    },
    {
      "epoch": 10.025703794369646,
      "grad_norm": 2.6222760738247426,
      "learning_rate": 1.2383627878375098e-06,
      "loss": 0.8949,
      "step": 8191
    },
    {
      "epoch": 10.026927784577722,
      "grad_norm": 1.787674520191941,
      "learning_rate": 1.2378095442203509e-06,
      "loss": 1.5911,
      "step": 8192
    },
    {
      "epoch": 10.028151774785801,
      "grad_norm": 2.013038694250396,
      "learning_rate": 1.237256383549679e-06,
      "loss": 0.3099,
      "step": 8193
    },
    {
      "epoch": 10.02937576499388,
      "grad_norm": 1.7565197563065815,
      "learning_rate": 1.236703305861848e-06,
      "loss": 1.1584,
      "step": 8194
    },
    {
      "epoch": 10.030599755201958,
      "grad_norm": 3.025333169232765,
      "learning_rate": 1.2361503111932033e-06,
      "loss": 0.5436,
      "step": 8195
    },
    {
      "epoch": 10.031823745410037,
      "grad_norm": 1.1840975399804163,
      "learning_rate": 1.2355973995800855e-06,
      "loss": 0.583,
      "step": 8196
    },
    {
      "epoch": 10.033047735618116,
      "grad_norm": 0.9361176290479428,
      "learning_rate": 1.2350445710588297e-06,
      "loss": 0.3903,
      "step": 8197
    },
    {
      "epoch": 10.034271725826194,
      "grad_norm": 1.2009336767901206,
      "learning_rate": 1.2344918256657659e-06,
      "loss": 0.4389,
      "step": 8198
    },
    {
      "epoch": 10.035495716034271,
      "grad_norm": 2.1754870362520475,
      "learning_rate": 1.2339391634372185e-06,
      "loss": 0.554,
      "step": 8199
    },
    {
      "epoch": 10.03671970624235,
      "grad_norm": 1.9482305066046048,
      "learning_rate": 1.2333865844095067e-06,
      "loss": 0.4277,
      "step": 8200
    },
    {
      "epoch": 10.037943696450428,
      "grad_norm": 1.3754165781652723,
      "learning_rate": 1.2328340886189436e-06,
      "loss": 0.4338,
      "step": 8201
    },
    {
      "epoch": 10.039167686658507,
      "grad_norm": 1.7629831464277703,
      "learning_rate": 1.2322816761018371e-06,
      "loss": 0.5256,
      "step": 8202
    },
    {
      "epoch": 10.040391676866586,
      "grad_norm": 1.3989393514925506,
      "learning_rate": 1.2317293468944902e-06,
      "loss": 0.6196,
      "step": 8203
    },
    {
      "epoch": 10.041615667074664,
      "grad_norm": 2.4474447861156077,
      "learning_rate": 1.2311771010331995e-06,
      "loss": 0.6019,
      "step": 8204
    },
    {
      "epoch": 10.042839657282741,
      "grad_norm": 1.99594135717264,
      "learning_rate": 1.2306249385542566e-06,
      "loss": 0.4061,
      "step": 8205
    },
    {
      "epoch": 10.04406364749082,
      "grad_norm": 1.3853867474988133,
      "learning_rate": 1.2300728594939468e-06,
      "loss": 0.5548,
      "step": 8206
    },
    {
      "epoch": 10.045287637698898,
      "grad_norm": 2.0816188342711506,
      "learning_rate": 1.2295208638885522e-06,
      "loss": 0.5453,
      "step": 8207
    },
    {
      "epoch": 10.046511627906977,
      "grad_norm": 2.491426443448446,
      "learning_rate": 1.2289689517743475e-06,
      "loss": 0.3207,
      "step": 8208
    },
    {
      "epoch": 10.047735618115055,
      "grad_norm": 2.216455255875227,
      "learning_rate": 1.228417123187603e-06,
      "loss": 0.9833,
      "step": 8209
    },
    {
      "epoch": 10.048959608323134,
      "grad_norm": 0.7265104739593882,
      "learning_rate": 1.2278653781645799e-06,
      "loss": 0.2565,
      "step": 8210
    },
    {
      "epoch": 10.050183598531211,
      "grad_norm": 1.6455866797763894,
      "learning_rate": 1.22731371674154e-06,
      "loss": 0.7274,
      "step": 8211
    },
    {
      "epoch": 10.05140758873929,
      "grad_norm": 1.0537036040994,
      "learning_rate": 1.2267621389547352e-06,
      "loss": 0.594,
      "step": 8212
    },
    {
      "epoch": 10.052631578947368,
      "grad_norm": 1.442942957641787,
      "learning_rate": 1.2262106448404133e-06,
      "loss": 1.3063,
      "step": 8213
    },
    {
      "epoch": 10.053855569155447,
      "grad_norm": 1.673018287290156,
      "learning_rate": 1.2256592344348162e-06,
      "loss": 0.5001,
      "step": 8214
    },
    {
      "epoch": 10.055079559363525,
      "grad_norm": 1.9655095847721382,
      "learning_rate": 1.2251079077741808e-06,
      "loss": 0.7747,
      "step": 8215
    },
    {
      "epoch": 10.056303549571604,
      "grad_norm": 2.0685399088264167,
      "learning_rate": 1.2245566648947383e-06,
      "loss": 0.4333,
      "step": 8216
    },
    {
      "epoch": 10.057527539779683,
      "grad_norm": 0.9206153331720969,
      "learning_rate": 1.224005505832714e-06,
      "loss": 0.4128,
      "step": 8217
    },
    {
      "epoch": 10.05875152998776,
      "grad_norm": 1.1826377781750672,
      "learning_rate": 1.2234544306243288e-06,
      "loss": 0.8173,
      "step": 8218
    },
    {
      "epoch": 10.059975520195838,
      "grad_norm": 2.000236319146925,
      "learning_rate": 1.2229034393057953e-06,
      "loss": 1.0052,
      "step": 8219
    },
    {
      "epoch": 10.061199510403917,
      "grad_norm": 2.4806470268029908,
      "learning_rate": 1.2223525319133264e-06,
      "loss": 0.4018,
      "step": 8220
    },
    {
      "epoch": 10.062423500611995,
      "grad_norm": 1.9793296454222111,
      "learning_rate": 1.2218017084831223e-06,
      "loss": 0.4916,
      "step": 8221
    },
    {
      "epoch": 10.063647490820074,
      "grad_norm": 2.252426027766268,
      "learning_rate": 1.221250969051382e-06,
      "loss": 1.0335,
      "step": 8222
    },
    {
      "epoch": 10.064871481028153,
      "grad_norm": 1.0315081804434139,
      "learning_rate": 1.2207003136542978e-06,
      "loss": 0.5249,
      "step": 8223
    },
    {
      "epoch": 10.06609547123623,
      "grad_norm": 1.7496773267011276,
      "learning_rate": 1.2201497423280578e-06,
      "loss": 1.1004,
      "step": 8224
    },
    {
      "epoch": 10.067319461444308,
      "grad_norm": 1.7127974635210463,
      "learning_rate": 1.2195992551088428e-06,
      "loss": 0.9317,
      "step": 8225
    },
    {
      "epoch": 10.068543451652387,
      "grad_norm": 3.0772592770699694,
      "learning_rate": 1.2190488520328292e-06,
      "loss": 0.4501,
      "step": 8226
    },
    {
      "epoch": 10.069767441860465,
      "grad_norm": 1.677476660950609,
      "learning_rate": 1.2184985331361878e-06,
      "loss": 0.5169,
      "step": 8227
    },
    {
      "epoch": 10.070991432068544,
      "grad_norm": 2.6887453707222195,
      "learning_rate": 1.217948298455081e-06,
      "loss": 0.4996,
      "step": 8228
    },
    {
      "epoch": 10.072215422276622,
      "grad_norm": 1.5648063566456674,
      "learning_rate": 1.217398148025671e-06,
      "loss": 0.4967,
      "step": 8229
    },
    {
      "epoch": 10.0734394124847,
      "grad_norm": 1.8136752847791946,
      "learning_rate": 1.2168480818841107e-06,
      "loss": 0.4374,
      "step": 8230
    },
    {
      "epoch": 10.074663402692778,
      "grad_norm": 2.2246621789419283,
      "learning_rate": 1.2162981000665486e-06,
      "loss": 0.6307,
      "step": 8231
    },
    {
      "epoch": 10.075887392900857,
      "grad_norm": 1.825704109735742,
      "learning_rate": 1.2157482026091272e-06,
      "loss": 1.085,
      "step": 8232
    },
    {
      "epoch": 10.077111383108935,
      "grad_norm": 2.4097287098600786,
      "learning_rate": 1.2151983895479837e-06,
      "loss": 0.4595,
      "step": 8233
    },
    {
      "epoch": 10.078335373317014,
      "grad_norm": 1.9328201932020421,
      "learning_rate": 1.2146486609192498e-06,
      "loss": 1.1061,
      "step": 8234
    },
    {
      "epoch": 10.079559363525092,
      "grad_norm": 1.8236755858004243,
      "learning_rate": 1.2140990167590518e-06,
      "loss": 0.5444,
      "step": 8235
    },
    {
      "epoch": 10.080783353733171,
      "grad_norm": 2.1869585634303372,
      "learning_rate": 1.2135494571035094e-06,
      "loss": 0.3034,
      "step": 8236
    },
    {
      "epoch": 10.082007343941248,
      "grad_norm": 3.0280838363090163,
      "learning_rate": 1.2129999819887396e-06,
      "loss": 0.6973,
      "step": 8237
    },
    {
      "epoch": 10.083231334149326,
      "grad_norm": 1.647190943979056,
      "learning_rate": 1.2124505914508511e-06,
      "loss": 1.3327,
      "step": 8238
    },
    {
      "epoch": 10.084455324357405,
      "grad_norm": 2.702466857567919,
      "learning_rate": 1.2119012855259466e-06,
      "loss": 0.3638,
      "step": 8239
    },
    {
      "epoch": 10.085679314565484,
      "grad_norm": 1.7632375628661545,
      "learning_rate": 1.2113520642501254e-06,
      "loss": 1.3602,
      "step": 8240
    },
    {
      "epoch": 10.086903304773562,
      "grad_norm": 2.2592475324561976,
      "learning_rate": 1.2108029276594794e-06,
      "loss": 0.462,
      "step": 8241
    },
    {
      "epoch": 10.088127294981641,
      "grad_norm": 1.7987748992805637,
      "learning_rate": 1.210253875790097e-06,
      "loss": 0.728,
      "step": 8242
    },
    {
      "epoch": 10.089351285189718,
      "grad_norm": 3.519550948084943,
      "learning_rate": 1.2097049086780599e-06,
      "loss": 0.4761,
      "step": 8243
    },
    {
      "epoch": 10.090575275397796,
      "grad_norm": 1.9751605780177988,
      "learning_rate": 1.2091560263594435e-06,
      "loss": 0.3184,
      "step": 8244
    },
    {
      "epoch": 10.091799265605875,
      "grad_norm": 1.6908781106123083,
      "learning_rate": 1.2086072288703184e-06,
      "loss": 0.4175,
      "step": 8245
    },
    {
      "epoch": 10.093023255813954,
      "grad_norm": 1.943353057579902,
      "learning_rate": 1.20805851624675e-06,
      "loss": 1.31,
      "step": 8246
    },
    {
      "epoch": 10.094247246022032,
      "grad_norm": 2.0052688699310344,
      "learning_rate": 1.2075098885247974e-06,
      "loss": 0.44,
      "step": 8247
    },
    {
      "epoch": 10.095471236230111,
      "grad_norm": 1.4785465428453508,
      "learning_rate": 1.206961345740514e-06,
      "loss": 1.0817,
      "step": 8248
    },
    {
      "epoch": 10.096695226438188,
      "grad_norm": 1.490436442253079,
      "learning_rate": 1.2064128879299486e-06,
      "loss": 0.6816,
      "step": 8249
    },
    {
      "epoch": 10.097919216646266,
      "grad_norm": 2.370235088005835,
      "learning_rate": 1.2058645151291436e-06,
      "loss": 0.4538,
      "step": 8250
    },
    {
      "epoch": 10.099143206854345,
      "grad_norm": 2.773333748665765,
      "learning_rate": 1.205316227374136e-06,
      "loss": 0.3628,
      "step": 8251
    },
    {
      "epoch": 10.100367197062424,
      "grad_norm": 2.070182061908845,
      "learning_rate": 1.204768024700957e-06,
      "loss": 0.6366,
      "step": 8252
    },
    {
      "epoch": 10.101591187270502,
      "grad_norm": 1.2789623411138893,
      "learning_rate": 1.2042199071456329e-06,
      "loss": 0.7162,
      "step": 8253
    },
    {
      "epoch": 10.10281517747858,
      "grad_norm": 1.8348475575505871,
      "learning_rate": 1.2036718747441824e-06,
      "loss": 0.4543,
      "step": 8254
    },
    {
      "epoch": 10.104039167686658,
      "grad_norm": 2.3955825246976454,
      "learning_rate": 1.2031239275326228e-06,
      "loss": 0.3778,
      "step": 8255
    },
    {
      "epoch": 10.105263157894736,
      "grad_norm": 1.7767644922741932,
      "learning_rate": 1.2025760655469629e-06,
      "loss": 1.3813,
      "step": 8256
    },
    {
      "epoch": 10.106487148102815,
      "grad_norm": 2.0748755502066083,
      "learning_rate": 1.202028288823204e-06,
      "loss": 0.5967,
      "step": 8257
    },
    {
      "epoch": 10.107711138310894,
      "grad_norm": 1.7294531377421545,
      "learning_rate": 1.2014805973973443e-06,
      "loss": 0.8245,
      "step": 8258
    },
    {
      "epoch": 10.108935128518972,
      "grad_norm": 1.064669276497954,
      "learning_rate": 1.2009329913053778e-06,
      "loss": 0.508,
      "step": 8259
    },
    {
      "epoch": 10.11015911872705,
      "grad_norm": 1.4655805016074455,
      "learning_rate": 1.2003854705832898e-06,
      "loss": 0.3339,
      "step": 8260
    },
    {
      "epoch": 10.11138310893513,
      "grad_norm": 1.1222544637606002,
      "learning_rate": 1.1998380352670622e-06,
      "loss": 0.5782,
      "step": 8261
    },
    {
      "epoch": 10.112607099143206,
      "grad_norm": 1.541785995963906,
      "learning_rate": 1.1992906853926697e-06,
      "loss": 0.6895,
      "step": 8262
    },
    {
      "epoch": 10.113831089351285,
      "grad_norm": 2.450949135177737,
      "learning_rate": 1.1987434209960824e-06,
      "loss": 0.973,
      "step": 8263
    },
    {
      "epoch": 10.115055079559363,
      "grad_norm": 2.0126933828660314,
      "learning_rate": 1.1981962421132646e-06,
      "loss": 0.5913,
      "step": 8264
    },
    {
      "epoch": 10.116279069767442,
      "grad_norm": 1.1095219275690256,
      "learning_rate": 1.1976491487801747e-06,
      "loss": 0.5076,
      "step": 8265
    },
    {
      "epoch": 10.11750305997552,
      "grad_norm": 2.5256076999654327,
      "learning_rate": 1.1971021410327655e-06,
      "loss": 0.8521,
      "step": 8266
    },
    {
      "epoch": 10.1187270501836,
      "grad_norm": 2.1571133615491456,
      "learning_rate": 1.1965552189069845e-06,
      "loss": 0.5099,
      "step": 8267
    },
    {
      "epoch": 10.119951040391676,
      "grad_norm": 3.0704963594359054,
      "learning_rate": 1.1960083824387736e-06,
      "loss": 0.4063,
      "step": 8268
    },
    {
      "epoch": 10.121175030599755,
      "grad_norm": 1.5745353491333403,
      "learning_rate": 1.1954616316640686e-06,
      "loss": 0.4764,
      "step": 8269
    },
    {
      "epoch": 10.122399020807833,
      "grad_norm": 1.380129608422423,
      "learning_rate": 1.1949149666188e-06,
      "loss": 0.5026,
      "step": 8270
    },
    {
      "epoch": 10.123623011015912,
      "grad_norm": 1.2596891997770223,
      "learning_rate": 1.1943683873388917e-06,
      "loss": 0.8945,
      "step": 8271
    },
    {
      "epoch": 10.12484700122399,
      "grad_norm": 2.501873461919243,
      "learning_rate": 1.1938218938602647e-06,
      "loss": 0.6069,
      "step": 8272
    },
    {
      "epoch": 10.12607099143207,
      "grad_norm": 1.8309436780732844,
      "learning_rate": 1.1932754862188317e-06,
      "loss": 0.293,
      "step": 8273
    },
    {
      "epoch": 10.127294981640146,
      "grad_norm": 1.4932878912255487,
      "learning_rate": 1.1927291644505012e-06,
      "loss": 0.7847,
      "step": 8274
    },
    {
      "epoch": 10.128518971848225,
      "grad_norm": 1.8103390603412906,
      "learning_rate": 1.1921829285911732e-06,
      "loss": 0.5439,
      "step": 8275
    },
    {
      "epoch": 10.129742962056303,
      "grad_norm": 3.0065891509859193,
      "learning_rate": 1.1916367786767467e-06,
      "loss": 0.3207,
      "step": 8276
    },
    {
      "epoch": 10.130966952264382,
      "grad_norm": 1.7525682074132436,
      "learning_rate": 1.1910907147431116e-06,
      "loss": 0.4537,
      "step": 8277
    },
    {
      "epoch": 10.13219094247246,
      "grad_norm": 2.80797548226728,
      "learning_rate": 1.1905447368261535e-06,
      "loss": 0.5289,
      "step": 8278
    },
    {
      "epoch": 10.13341493268054,
      "grad_norm": 1.9902141688267647,
      "learning_rate": 1.1899988449617522e-06,
      "loss": 0.4123,
      "step": 8279
    },
    {
      "epoch": 10.134638922888616,
      "grad_norm": 2.500320404453797,
      "learning_rate": 1.1894530391857812e-06,
      "loss": 0.4628,
      "step": 8280
    },
    {
      "epoch": 10.135862913096695,
      "grad_norm": 1.5155202327501696,
      "learning_rate": 1.188907319534109e-06,
      "loss": 0.6703,
      "step": 8281
    },
    {
      "epoch": 10.137086903304773,
      "grad_norm": 2.7700567278648305,
      "learning_rate": 1.1883616860425986e-06,
      "loss": 1.0568,
      "step": 8282
    },
    {
      "epoch": 10.138310893512852,
      "grad_norm": 1.2351288809203242,
      "learning_rate": 1.1878161387471067e-06,
      "loss": 0.7676,
      "step": 8283
    },
    {
      "epoch": 10.13953488372093,
      "grad_norm": 1.6839217340996075,
      "learning_rate": 1.1872706776834839e-06,
      "loss": 0.6129,
      "step": 8284
    },
    {
      "epoch": 10.140758873929009,
      "grad_norm": 1.6982600803549823,
      "learning_rate": 1.1867253028875784e-06,
      "loss": 0.4296,
      "step": 8285
    },
    {
      "epoch": 10.141982864137088,
      "grad_norm": 2.4815063342210366,
      "learning_rate": 1.1861800143952276e-06,
      "loss": 0.4088,
      "step": 8286
    },
    {
      "epoch": 10.143206854345165,
      "grad_norm": 3.1407232694985896,
      "learning_rate": 1.185634812242267e-06,
      "loss": 0.3108,
      "step": 8287
    },
    {
      "epoch": 10.144430844553243,
      "grad_norm": 2.4330693436305153,
      "learning_rate": 1.1850896964645239e-06,
      "loss": 0.4856,
      "step": 8288
    },
    {
      "epoch": 10.145654834761322,
      "grad_norm": 1.0485110549815666,
      "learning_rate": 1.1845446670978234e-06,
      "loss": 0.5467,
      "step": 8289
    },
    {
      "epoch": 10.1468788249694,
      "grad_norm": 1.92814638414238,
      "learning_rate": 1.1839997241779818e-06,
      "loss": 0.5174,
      "step": 8290
    },
    {
      "epoch": 10.148102815177479,
      "grad_norm": 1.1528713881970116,
      "learning_rate": 1.1834548677408117e-06,
      "loss": 0.3078,
      "step": 8291
    },
    {
      "epoch": 10.149326805385558,
      "grad_norm": 2.1399883054346422,
      "learning_rate": 1.1829100978221173e-06,
      "loss": 1.1287,
      "step": 8292
    },
    {
      "epoch": 10.150550795593634,
      "grad_norm": 1.200249753807425,
      "learning_rate": 1.1823654144576987e-06,
      "loss": 0.6247,
      "step": 8293
    },
    {
      "epoch": 10.151774785801713,
      "grad_norm": 2.5688677097325536,
      "learning_rate": 1.1818208176833524e-06,
      "loss": 0.9486,
      "step": 8294
    },
    {
      "epoch": 10.152998776009792,
      "grad_norm": 1.8820273375769543,
      "learning_rate": 1.1812763075348663e-06,
      "loss": 0.4824,
      "step": 8295
    },
    {
      "epoch": 10.15422276621787,
      "grad_norm": 1.3585315496956756,
      "learning_rate": 1.1807318840480238e-06,
      "loss": 0.5342,
      "step": 8296
    },
    {
      "epoch": 10.155446756425949,
      "grad_norm": 2.7689388741900696,
      "learning_rate": 1.180187547258602e-06,
      "loss": 0.5641,
      "step": 8297
    },
    {
      "epoch": 10.156670746634028,
      "grad_norm": 2.6318550499052824,
      "learning_rate": 1.1796432972023733e-06,
      "loss": 0.3913,
      "step": 8298
    },
    {
      "epoch": 10.157894736842104,
      "grad_norm": 1.6261943813738844,
      "learning_rate": 1.179099133915103e-06,
      "loss": 0.8579,
      "step": 8299
    },
    {
      "epoch": 10.159118727050183,
      "grad_norm": 2.680480454729199,
      "learning_rate": 1.1785550574325524e-06,
      "loss": 0.4338,
      "step": 8300
    },
    {
      "epoch": 10.160342717258262,
      "grad_norm": 0.8241020161653674,
      "learning_rate": 1.1780110677904755e-06,
      "loss": 0.4332,
      "step": 8301
    },
    {
      "epoch": 10.16156670746634,
      "grad_norm": 1.631738735687105,
      "learning_rate": 1.1774671650246204e-06,
      "loss": 0.5582,
      "step": 8302
    },
    {
      "epoch": 10.162790697674419,
      "grad_norm": 1.4864583142816723,
      "learning_rate": 1.176923349170734e-06,
      "loss": 0.5071,
      "step": 8303
    },
    {
      "epoch": 10.164014687882498,
      "grad_norm": 2.632384294296717,
      "learning_rate": 1.17637962026455e-06,
      "loss": 0.3834,
      "step": 8304
    },
    {
      "epoch": 10.165238678090576,
      "grad_norm": 1.8159657192021468,
      "learning_rate": 1.1758359783418016e-06,
      "loss": 0.9095,
      "step": 8305
    },
    {
      "epoch": 10.166462668298653,
      "grad_norm": 2.087643538320378,
      "learning_rate": 1.1752924234382141e-06,
      "loss": 0.3477,
      "step": 8306
    },
    {
      "epoch": 10.167686658506732,
      "grad_norm": 2.129387787716989,
      "learning_rate": 1.1747489555895097e-06,
      "loss": 0.961,
      "step": 8307
    },
    {
      "epoch": 10.16891064871481,
      "grad_norm": 1.5299856998826533,
      "learning_rate": 1.174205574831402e-06,
      "loss": 0.8902,
      "step": 8308
    },
    {
      "epoch": 10.170134638922889,
      "grad_norm": 2.527984789059689,
      "learning_rate": 1.1736622811996016e-06,
      "loss": 0.4595,
      "step": 8309
    },
    {
      "epoch": 10.171358629130967,
      "grad_norm": 2.6295948047871867,
      "learning_rate": 1.173119074729808e-06,
      "loss": 0.5897,
      "step": 8310
    },
    {
      "epoch": 10.172582619339046,
      "grad_norm": 1.1382230550274333,
      "learning_rate": 1.1725759554577217e-06,
      "loss": 0.4959,
      "step": 8311
    },
    {
      "epoch": 10.173806609547123,
      "grad_norm": 2.2634845601624667,
      "learning_rate": 1.1720329234190345e-06,
      "loss": 0.5457,
      "step": 8312
    },
    {
      "epoch": 10.175030599755202,
      "grad_norm": 1.7138726695699769,
      "learning_rate": 1.1714899786494314e-06,
      "loss": 0.4313,
      "step": 8313
    },
    {
      "epoch": 10.17625458996328,
      "grad_norm": 1.4522650646617747,
      "learning_rate": 1.170947121184593e-06,
      "loss": 0.6163,
      "step": 8314
    },
    {
      "epoch": 10.177478580171359,
      "grad_norm": 1.2719371254014478,
      "learning_rate": 1.170404351060194e-06,
      "loss": 0.5912,
      "step": 8315
    },
    {
      "epoch": 10.178702570379437,
      "grad_norm": 3.0125555813835847,
      "learning_rate": 1.1698616683119032e-06,
      "loss": 0.91,
      "step": 8316
    },
    {
      "epoch": 10.179926560587516,
      "grad_norm": 1.3275171436275153,
      "learning_rate": 1.169319072975384e-06,
      "loss": 0.3945,
      "step": 8317
    },
    {
      "epoch": 10.181150550795593,
      "grad_norm": 2.695511269484224,
      "learning_rate": 1.1687765650862931e-06,
      "loss": 1.5551,
      "step": 8318
    },
    {
      "epoch": 10.182374541003671,
      "grad_norm": 1.6939889576153784,
      "learning_rate": 1.1682341446802817e-06,
      "loss": 1.4512,
      "step": 8319
    },
    {
      "epoch": 10.18359853121175,
      "grad_norm": 1.8149196040077349,
      "learning_rate": 1.1676918117929983e-06,
      "loss": 0.4596,
      "step": 8320
    },
    {
      "epoch": 10.184822521419829,
      "grad_norm": 1.5573051321121223,
      "learning_rate": 1.1671495664600804e-06,
      "loss": 0.3347,
      "step": 8321
    },
    {
      "epoch": 10.186046511627907,
      "grad_norm": 1.750645673531928,
      "learning_rate": 1.1666074087171628e-06,
      "loss": 0.9842,
      "step": 8322
    },
    {
      "epoch": 10.187270501835986,
      "grad_norm": 2.4020715131216086,
      "learning_rate": 1.1660653385998738e-06,
      "loss": 0.8615,
      "step": 8323
    },
    {
      "epoch": 10.188494492044065,
      "grad_norm": 1.4318909002906755,
      "learning_rate": 1.1655233561438375e-06,
      "loss": 0.7981,
      "step": 8324
    },
    {
      "epoch": 10.189718482252141,
      "grad_norm": 1.1306122936443008,
      "learning_rate": 1.1649814613846707e-06,
      "loss": 0.4167,
      "step": 8325
    },
    {
      "epoch": 10.19094247246022,
      "grad_norm": 1.1640537650362286,
      "learning_rate": 1.1644396543579842e-06,
      "loss": 0.5353,
      "step": 8326
    },
    {
      "epoch": 10.192166462668299,
      "grad_norm": 1.7140831606825007,
      "learning_rate": 1.1638979350993845e-06,
      "loss": 1.1736,
      "step": 8327
    },
    {
      "epoch": 10.193390452876377,
      "grad_norm": 2.2470977508041696,
      "learning_rate": 1.163356303644469e-06,
      "loss": 0.6721,
      "step": 8328
    },
    {
      "epoch": 10.194614443084456,
      "grad_norm": 1.5462304779107077,
      "learning_rate": 1.162814760028834e-06,
      "loss": 0.462,
      "step": 8329
    },
    {
      "epoch": 10.195838433292534,
      "grad_norm": 1.143920943810977,
      "learning_rate": 1.162273304288067e-06,
      "loss": 0.5965,
      "step": 8330
    },
    {
      "epoch": 10.197062423500611,
      "grad_norm": 2.0628194863949485,
      "learning_rate": 1.1617319364577508e-06,
      "loss": 0.8453,
      "step": 8331
    },
    {
      "epoch": 10.19828641370869,
      "grad_norm": 1.3388861742734397,
      "learning_rate": 1.1611906565734618e-06,
      "loss": 0.463,
      "step": 8332
    },
    {
      "epoch": 10.199510403916769,
      "grad_norm": 1.5654123475792425,
      "learning_rate": 1.1606494646707711e-06,
      "loss": 1.2579,
      "step": 8333
    },
    {
      "epoch": 10.200734394124847,
      "grad_norm": 1.8228851029165722,
      "learning_rate": 1.1601083607852433e-06,
      "loss": 1.8783,
      "step": 8334
    },
    {
      "epoch": 10.201958384332926,
      "grad_norm": 1.2500649201025675,
      "learning_rate": 1.1595673449524384e-06,
      "loss": 0.5001,
      "step": 8335
    },
    {
      "epoch": 10.203182374541004,
      "grad_norm": 3.646583771387821,
      "learning_rate": 1.1590264172079085e-06,
      "loss": 0.4199,
      "step": 8336
    },
    {
      "epoch": 10.204406364749081,
      "grad_norm": 2.190469362621497,
      "learning_rate": 1.1584855775872036e-06,
      "loss": 1.0131,
      "step": 8337
    },
    {
      "epoch": 10.20563035495716,
      "grad_norm": 2.3145338175383907,
      "learning_rate": 1.1579448261258654e-06,
      "loss": 0.9036,
      "step": 8338
    },
    {
      "epoch": 10.206854345165238,
      "grad_norm": 1.3557548223714138,
      "learning_rate": 1.1574041628594288e-06,
      "loss": 1.088,
      "step": 8339
    },
    {
      "epoch": 10.208078335373317,
      "grad_norm": 1.5903988371047444,
      "learning_rate": 1.1568635878234245e-06,
      "loss": 0.9209,
      "step": 8340
    },
    {
      "epoch": 10.209302325581396,
      "grad_norm": 1.210914049882735,
      "learning_rate": 1.1563231010533759e-06,
      "loss": 0.5301,
      "step": 8341
    },
    {
      "epoch": 10.210526315789474,
      "grad_norm": 1.0467182141933953,
      "learning_rate": 1.1557827025848048e-06,
      "loss": 0.7087,
      "step": 8342
    },
    {
      "epoch": 10.211750305997551,
      "grad_norm": 1.2897750019549414,
      "learning_rate": 1.1552423924532221e-06,
      "loss": 0.7182,
      "step": 8343
    },
    {
      "epoch": 10.21297429620563,
      "grad_norm": 1.617333055658577,
      "learning_rate": 1.1547021706941356e-06,
      "loss": 0.9815,
      "step": 8344
    },
    {
      "epoch": 10.214198286413708,
      "grad_norm": 1.8299443733612986,
      "learning_rate": 1.1541620373430466e-06,
      "loss": 0.6435,
      "step": 8345
    },
    {
      "epoch": 10.215422276621787,
      "grad_norm": 1.6015166042625728,
      "learning_rate": 1.1536219924354505e-06,
      "loss": 0.6386,
      "step": 8346
    },
    {
      "epoch": 10.216646266829866,
      "grad_norm": 1.699437146476779,
      "learning_rate": 1.1530820360068373e-06,
      "loss": 0.5719,
      "step": 8347
    },
    {
      "epoch": 10.217870257037944,
      "grad_norm": 1.6776732045011649,
      "learning_rate": 1.1525421680926904e-06,
      "loss": 0.7138,
      "step": 8348
    },
    {
      "epoch": 10.219094247246023,
      "grad_norm": 1.6357702570103423,
      "learning_rate": 1.1520023887284887e-06,
      "loss": 1.1736,
      "step": 8349
    },
    {
      "epoch": 10.2203182374541,
      "grad_norm": 1.8831999209550918,
      "learning_rate": 1.151462697949704e-06,
      "loss": 0.9931,
      "step": 8350
    },
    {
      "epoch": 10.221542227662178,
      "grad_norm": 1.912609688705624,
      "learning_rate": 1.1509230957918031e-06,
      "loss": 0.7452,
      "step": 8351
    },
    {
      "epoch": 10.222766217870257,
      "grad_norm": 1.4024398538551932,
      "learning_rate": 1.1503835822902462e-06,
      "loss": 0.455,
      "step": 8352
    },
    {
      "epoch": 10.223990208078336,
      "grad_norm": 0.918758601511904,
      "learning_rate": 1.1498441574804883e-06,
      "loss": 0.4639,
      "step": 8353
    },
    {
      "epoch": 10.225214198286414,
      "grad_norm": 2.5507953502152114,
      "learning_rate": 1.1493048213979778e-06,
      "loss": 0.6703,
      "step": 8354
    },
    {
      "epoch": 10.226438188494493,
      "grad_norm": 1.1059674081198785,
      "learning_rate": 1.1487655740781593e-06,
      "loss": 0.5653,
      "step": 8355
    },
    {
      "epoch": 10.22766217870257,
      "grad_norm": 3.4100362495835177,
      "learning_rate": 1.1482264155564704e-06,
      "loss": 0.413,
      "step": 8356
    },
    {
      "epoch": 10.228886168910648,
      "grad_norm": 3.0376916021455846,
      "learning_rate": 1.1476873458683407e-06,
      "loss": 0.5113,
      "step": 8357
    },
    {
      "epoch": 10.230110159118727,
      "grad_norm": 1.504266767393246,
      "learning_rate": 1.1471483650491957e-06,
      "loss": 0.5733,
      "step": 8358
    },
    {
      "epoch": 10.231334149326806,
      "grad_norm": 2.241040518246681,
      "learning_rate": 1.1466094731344577e-06,
      "loss": 0.4514,
      "step": 8359
    },
    {
      "epoch": 10.232558139534884,
      "grad_norm": 1.8791196078690673,
      "learning_rate": 1.1460706701595387e-06,
      "loss": 1.0033,
      "step": 8360
    },
    {
      "epoch": 10.233782129742963,
      "grad_norm": 1.3942239780713515,
      "learning_rate": 1.1455319561598478e-06,
      "loss": 0.4354,
      "step": 8361
    },
    {
      "epoch": 10.23500611995104,
      "grad_norm": 2.0887322855298183,
      "learning_rate": 1.1449933311707867e-06,
      "loss": 0.7139,
      "step": 8362
    },
    {
      "epoch": 10.236230110159118,
      "grad_norm": 1.1056140553296787,
      "learning_rate": 1.1444547952277522e-06,
      "loss": 0.6392,
      "step": 8363
    },
    {
      "epoch": 10.237454100367197,
      "grad_norm": 2.0290715332498626,
      "learning_rate": 1.1439163483661348e-06,
      "loss": 1.0704,
      "step": 8364
    },
    {
      "epoch": 10.238678090575275,
      "grad_norm": 1.4150767280861785,
      "learning_rate": 1.143377990621319e-06,
      "loss": 0.5298,
      "step": 8365
    },
    {
      "epoch": 10.239902080783354,
      "grad_norm": 1.4170582050575873,
      "learning_rate": 1.142839722028684e-06,
      "loss": 0.9999,
      "step": 8366
    },
    {
      "epoch": 10.241126070991433,
      "grad_norm": 1.4081944044271786,
      "learning_rate": 1.1423015426236023e-06,
      "loss": 1.2728,
      "step": 8367
    },
    {
      "epoch": 10.24235006119951,
      "grad_norm": 0.8783787897048889,
      "learning_rate": 1.1417634524414418e-06,
      "loss": 0.3488,
      "step": 8368
    },
    {
      "epoch": 10.243574051407588,
      "grad_norm": 1.5984788664020013,
      "learning_rate": 1.1412254515175633e-06,
      "loss": 0.4962,
      "step": 8369
    },
    {
      "epoch": 10.244798041615667,
      "grad_norm": 1.987829392514733,
      "learning_rate": 1.1406875398873224e-06,
      "loss": 0.5267,
      "step": 8370
    },
    {
      "epoch": 10.246022031823745,
      "grad_norm": 1.5639521959258251,
      "learning_rate": 1.140149717586068e-06,
      "loss": 0.5598,
      "step": 8371
    },
    {
      "epoch": 10.247246022031824,
      "grad_norm": 1.726152095353863,
      "learning_rate": 1.139611984649145e-06,
      "loss": 0.4489,
      "step": 8372
    },
    {
      "epoch": 10.248470012239903,
      "grad_norm": 2.0206152495520056,
      "learning_rate": 1.1390743411118907e-06,
      "loss": 0.419,
      "step": 8373
    },
    {
      "epoch": 10.249694002447981,
      "grad_norm": 3.3099355533192787,
      "learning_rate": 1.1385367870096379e-06,
      "loss": 0.4703,
      "step": 8374
    },
    {
      "epoch": 10.250917992656058,
      "grad_norm": 1.6844571898281888,
      "learning_rate": 1.13799932237771e-06,
      "loss": 0.6211,
      "step": 8375
    },
    {
      "epoch": 10.252141982864137,
      "grad_norm": 1.1269982840406254,
      "learning_rate": 1.1374619472514297e-06,
      "loss": 0.57,
      "step": 8376
    },
    {
      "epoch": 10.253365973072215,
      "grad_norm": 2.2250008274800535,
      "learning_rate": 1.1369246616661106e-06,
      "loss": 0.4441,
      "step": 8377
    },
    {
      "epoch": 10.254589963280294,
      "grad_norm": 1.596049240652772,
      "learning_rate": 1.1363874656570612e-06,
      "loss": 0.4982,
      "step": 8378
    },
    {
      "epoch": 10.255813953488373,
      "grad_norm": 1.2332760231446624,
      "learning_rate": 1.1358503592595837e-06,
      "loss": 1.0409,
      "step": 8379
    },
    {
      "epoch": 10.257037943696451,
      "grad_norm": 1.6205617388928162,
      "learning_rate": 1.1353133425089752e-06,
      "loss": 0.4733,
      "step": 8380
    },
    {
      "epoch": 10.258261933904528,
      "grad_norm": 1.7478631750592384,
      "learning_rate": 1.1347764154405262e-06,
      "loss": 0.9302,
      "step": 8381
    },
    {
      "epoch": 10.259485924112607,
      "grad_norm": 2.2085936672137376,
      "learning_rate": 1.1342395780895213e-06,
      "loss": 0.4659,
      "step": 8382
    },
    {
      "epoch": 10.260709914320685,
      "grad_norm": 2.3382153726279946,
      "learning_rate": 1.1337028304912399e-06,
      "loss": 0.4189,
      "step": 8383
    },
    {
      "epoch": 10.261933904528764,
      "grad_norm": 1.6922734960258807,
      "learning_rate": 1.133166172680954e-06,
      "loss": 1.265,
      "step": 8384
    },
    {
      "epoch": 10.263157894736842,
      "grad_norm": 1.768841475308521,
      "learning_rate": 1.1326296046939334e-06,
      "loss": 1.455,
      "step": 8385
    },
    {
      "epoch": 10.264381884944921,
      "grad_norm": 2.5550899911098384,
      "learning_rate": 1.1320931265654366e-06,
      "loss": 0.3588,
      "step": 8386
    },
    {
      "epoch": 10.265605875152998,
      "grad_norm": 1.5547804683775273,
      "learning_rate": 1.1315567383307203e-06,
      "loss": 0.7176,
      "step": 8387
    },
    {
      "epoch": 10.266829865361077,
      "grad_norm": 2.3686984188846814,
      "learning_rate": 1.1310204400250325e-06,
      "loss": 0.7347,
      "step": 8388
    },
    {
      "epoch": 10.268053855569155,
      "grad_norm": 2.5896477312817576,
      "learning_rate": 1.130484231683619e-06,
      "loss": 0.3931,
      "step": 8389
    },
    {
      "epoch": 10.269277845777234,
      "grad_norm": 1.4954587944254585,
      "learning_rate": 1.129948113341716e-06,
      "loss": 0.3217,
      "step": 8390
    },
    {
      "epoch": 10.270501835985312,
      "grad_norm": 1.796555392577347,
      "learning_rate": 1.1294120850345558e-06,
      "loss": 1.1709,
      "step": 8391
    },
    {
      "epoch": 10.271725826193391,
      "grad_norm": 1.585830372942829,
      "learning_rate": 1.1288761467973646e-06,
      "loss": 0.2952,
      "step": 8392
    },
    {
      "epoch": 10.27294981640147,
      "grad_norm": 1.610995756386133,
      "learning_rate": 1.12834029866536e-06,
      "loss": 0.6241,
      "step": 8393
    },
    {
      "epoch": 10.274173806609546,
      "grad_norm": 1.819408847372374,
      "learning_rate": 1.1278045406737584e-06,
      "loss": 0.2934,
      "step": 8394
    },
    {
      "epoch": 10.275397796817625,
      "grad_norm": 1.1400419023140156,
      "learning_rate": 1.127268872857767e-06,
      "loss": 0.4477,
      "step": 8395
    },
    {
      "epoch": 10.276621787025704,
      "grad_norm": 1.9444292724840881,
      "learning_rate": 1.1267332952525875e-06,
      "loss": 0.4805,
      "step": 8396
    },
    {
      "epoch": 10.277845777233782,
      "grad_norm": 2.0755990426405178,
      "learning_rate": 1.1261978078934168e-06,
      "loss": 0.6689,
      "step": 8397
    },
    {
      "epoch": 10.279069767441861,
      "grad_norm": 2.942265045004354,
      "learning_rate": 1.1256624108154449e-06,
      "loss": 0.8223,
      "step": 8398
    },
    {
      "epoch": 10.28029375764994,
      "grad_norm": 2.6342641796035897,
      "learning_rate": 1.1251271040538561e-06,
      "loss": 0.4195,
      "step": 8399
    },
    {
      "epoch": 10.281517747858016,
      "grad_norm": 2.805976035031549,
      "learning_rate": 1.1245918876438286e-06,
      "loss": 0.4467,
      "step": 8400
    },
    {
      "epoch": 10.282741738066095,
      "grad_norm": 1.6038462502907047,
      "learning_rate": 1.124056761620534e-06,
      "loss": 0.9341,
      "step": 8401
    },
    {
      "epoch": 10.283965728274174,
      "grad_norm": 2.2813991185786553,
      "learning_rate": 1.1235217260191406e-06,
      "loss": 0.5794,
      "step": 8402
    },
    {
      "epoch": 10.285189718482252,
      "grad_norm": 1.491523221012118,
      "learning_rate": 1.1229867808748093e-06,
      "loss": 0.9834,
      "step": 8403
    },
    {
      "epoch": 10.286413708690331,
      "grad_norm": 1.3781998760823395,
      "learning_rate": 1.1224519262226923e-06,
      "loss": 0.6246,
      "step": 8404
    },
    {
      "epoch": 10.28763769889841,
      "grad_norm": 1.607176538950427,
      "learning_rate": 1.1219171620979397e-06,
      "loss": 0.4661,
      "step": 8405
    },
    {
      "epoch": 10.288861689106486,
      "grad_norm": 1.8489478085080937,
      "learning_rate": 1.121382488535693e-06,
      "loss": 0.7074,
      "step": 8406
    },
    {
      "epoch": 10.290085679314565,
      "grad_norm": 2.1683448190993975,
      "learning_rate": 1.1208479055710912e-06,
      "loss": 0.5662,
      "step": 8407
    },
    {
      "epoch": 10.291309669522644,
      "grad_norm": 1.7655396358419815,
      "learning_rate": 1.1203134132392637e-06,
      "loss": 0.4061,
      "step": 8408
    },
    {
      "epoch": 10.292533659730722,
      "grad_norm": 2.1641590194239546,
      "learning_rate": 1.1197790115753366e-06,
      "loss": 0.6096,
      "step": 8409
    },
    {
      "epoch": 10.2937576499388,
      "grad_norm": 1.2520680820935073,
      "learning_rate": 1.1192447006144258e-06,
      "loss": 0.4763,
      "step": 8410
    },
    {
      "epoch": 10.29498164014688,
      "grad_norm": 1.490846415144508,
      "learning_rate": 1.1187104803916472e-06,
      "loss": 0.5454,
      "step": 8411
    },
    {
      "epoch": 10.296205630354958,
      "grad_norm": 1.7348293392828131,
      "learning_rate": 1.1181763509421068e-06,
      "loss": 0.516,
      "step": 8412
    },
    {
      "epoch": 10.297429620563035,
      "grad_norm": 1.1225321094959664,
      "learning_rate": 1.1176423123009056e-06,
      "loss": 0.369,
      "step": 8413
    },
    {
      "epoch": 10.298653610771114,
      "grad_norm": 2.2556798983018203,
      "learning_rate": 1.1171083645031385e-06,
      "loss": 0.4151,
      "step": 8414
    },
    {
      "epoch": 10.299877600979192,
      "grad_norm": 1.1510259088435155,
      "learning_rate": 1.116574507583895e-06,
      "loss": 1.1982,
      "step": 8415
    },
    {
      "epoch": 10.30110159118727,
      "grad_norm": 1.521151077397079,
      "learning_rate": 1.1160407415782578e-06,
      "loss": 0.6505,
      "step": 8416
    },
    {
      "epoch": 10.30232558139535,
      "grad_norm": 2.565945636579382,
      "learning_rate": 1.115507066521304e-06,
      "loss": 0.8568,
      "step": 8417
    },
    {
      "epoch": 10.303549571603428,
      "grad_norm": 1.8589512097746903,
      "learning_rate": 1.114973482448105e-06,
      "loss": 1.0799,
      "step": 8418
    },
    {
      "epoch": 10.304773561811505,
      "grad_norm": 1.469297712784982,
      "learning_rate": 1.1144399893937252e-06,
      "loss": 0.7116,
      "step": 8419
    },
    {
      "epoch": 10.305997552019583,
      "grad_norm": 1.765756101578622,
      "learning_rate": 1.1139065873932254e-06,
      "loss": 0.6298,
      "step": 8420
    },
    {
      "epoch": 10.307221542227662,
      "grad_norm": 2.0770371119312903,
      "learning_rate": 1.1133732764816586e-06,
      "loss": 0.8526,
      "step": 8421
    },
    {
      "epoch": 10.30844553243574,
      "grad_norm": 1.0403877838731264,
      "learning_rate": 1.11284005669407e-06,
      "loss": 0.5178,
      "step": 8422
    },
    {
      "epoch": 10.30966952264382,
      "grad_norm": 1.5937366936227606,
      "learning_rate": 1.1123069280655015e-06,
      "loss": 0.4744,
      "step": 8423
    },
    {
      "epoch": 10.310893512851898,
      "grad_norm": 1.6450343405730907,
      "learning_rate": 1.1117738906309902e-06,
      "loss": 0.5469,
      "step": 8424
    },
    {
      "epoch": 10.312117503059975,
      "grad_norm": 2.3251226468348043,
      "learning_rate": 1.1112409444255637e-06,
      "loss": 0.5172,
      "step": 8425
    },
    {
      "epoch": 10.313341493268053,
      "grad_norm": 1.726898309581724,
      "learning_rate": 1.110708089484246e-06,
      "loss": 0.5295,
      "step": 8426
    },
    {
      "epoch": 10.314565483476132,
      "grad_norm": 1.3796869583438913,
      "learning_rate": 1.1101753258420538e-06,
      "loss": 0.5917,
      "step": 8427
    },
    {
      "epoch": 10.31578947368421,
      "grad_norm": 2.0245731876288677,
      "learning_rate": 1.1096426535339985e-06,
      "loss": 0.4934,
      "step": 8428
    },
    {
      "epoch": 10.31701346389229,
      "grad_norm": 1.9540321646141319,
      "learning_rate": 1.1091100725950858e-06,
      "loss": 0.626,
      "step": 8429
    },
    {
      "epoch": 10.318237454100368,
      "grad_norm": 1.5745682173155513,
      "learning_rate": 1.1085775830603144e-06,
      "loss": 0.5905,
      "step": 8430
    },
    {
      "epoch": 10.319461444308445,
      "grad_norm": 1.7529256387205177,
      "learning_rate": 1.1080451849646776e-06,
      "loss": 0.6553,
      "step": 8431
    },
    {
      "epoch": 10.320685434516523,
      "grad_norm": 2.293978703904437,
      "learning_rate": 1.1075128783431626e-06,
      "loss": 0.4895,
      "step": 8432
    },
    {
      "epoch": 10.321909424724602,
      "grad_norm": 2.649267730402568,
      "learning_rate": 1.1069806632307513e-06,
      "loss": 0.4274,
      "step": 8433
    },
    {
      "epoch": 10.32313341493268,
      "grad_norm": 1.4920536372724003,
      "learning_rate": 1.106448539662418e-06,
      "loss": 0.5254,
      "step": 8434
    },
    {
      "epoch": 10.32435740514076,
      "grad_norm": 2.5704900251728153,
      "learning_rate": 1.1059165076731323e-06,
      "loss": 1.0254,
      "step": 8435
    },
    {
      "epoch": 10.325581395348838,
      "grad_norm": 1.773703460408595,
      "learning_rate": 1.1053845672978567e-06,
      "loss": 0.485,
      "step": 8436
    },
    {
      "epoch": 10.326805385556916,
      "grad_norm": 3.0486645792149627,
      "learning_rate": 1.1048527185715497e-06,
      "loss": 0.4595,
      "step": 8437
    },
    {
      "epoch": 10.328029375764993,
      "grad_norm": 2.6183222170709097,
      "learning_rate": 1.1043209615291625e-06,
      "loss": 0.7839,
      "step": 8438
    },
    {
      "epoch": 10.329253365973072,
      "grad_norm": 2.180740439115568,
      "learning_rate": 1.1037892962056387e-06,
      "loss": 0.483,
      "step": 8439
    },
    {
      "epoch": 10.33047735618115,
      "grad_norm": 2.0156348519167966,
      "learning_rate": 1.1032577226359172e-06,
      "loss": 0.2935,
      "step": 8440
    },
    {
      "epoch": 10.331701346389229,
      "grad_norm": 2.33427300877486,
      "learning_rate": 1.1027262408549334e-06,
      "loss": 0.4488,
      "step": 8441
    },
    {
      "epoch": 10.332925336597308,
      "grad_norm": 1.649204463122737,
      "learning_rate": 1.1021948508976122e-06,
      "loss": 0.4681,
      "step": 8442
    },
    {
      "epoch": 10.334149326805386,
      "grad_norm": 1.5667401243166743,
      "learning_rate": 1.1016635527988756e-06,
      "loss": 0.8094,
      "step": 8443
    },
    {
      "epoch": 10.335373317013463,
      "grad_norm": 2.5622843850130788,
      "learning_rate": 1.101132346593638e-06,
      "loss": 0.5512,
      "step": 8444
    },
    {
      "epoch": 10.336597307221542,
      "grad_norm": 2.9636747303880218,
      "learning_rate": 1.1006012323168089e-06,
      "loss": 0.4133,
      "step": 8445
    },
    {
      "epoch": 10.33782129742962,
      "grad_norm": 2.40250129964111,
      "learning_rate": 1.1000702100032903e-06,
      "loss": 0.537,
      "step": 8446
    },
    {
      "epoch": 10.339045287637699,
      "grad_norm": 2.090512835030234,
      "learning_rate": 1.09953927968798e-06,
      "loss": 0.834,
      "step": 8447
    },
    {
      "epoch": 10.340269277845778,
      "grad_norm": 2.524454845181693,
      "learning_rate": 1.0990084414057682e-06,
      "loss": 0.4771,
      "step": 8448
    },
    {
      "epoch": 10.341493268053856,
      "grad_norm": 3.0360937480957384,
      "learning_rate": 1.0984776951915385e-06,
      "loss": 0.3964,
      "step": 8449
    },
    {
      "epoch": 10.342717258261933,
      "grad_norm": 1.6879999746982848,
      "learning_rate": 1.097947041080173e-06,
      "loss": 0.8505,
      "step": 8450
    },
    {
      "epoch": 10.343941248470012,
      "grad_norm": 1.1691815199697158,
      "learning_rate": 1.0974164791065409e-06,
      "loss": 0.5359,
      "step": 8451
    },
    {
      "epoch": 10.34516523867809,
      "grad_norm": 2.1462666840432463,
      "learning_rate": 1.09688600930551e-06,
      "loss": 0.5126,
      "step": 8452
    },
    {
      "epoch": 10.346389228886169,
      "grad_norm": 1.8356308652447657,
      "learning_rate": 1.09635563171194e-06,
      "loss": 1.3052,
      "step": 8453
    },
    {
      "epoch": 10.347613219094248,
      "grad_norm": 1.7282599447279108,
      "learning_rate": 1.0958253463606872e-06,
      "loss": 1.5167,
      "step": 8454
    },
    {
      "epoch": 10.348837209302326,
      "grad_norm": 1.668961442470613,
      "learning_rate": 1.095295153286599e-06,
      "loss": 1.0814,
      "step": 8455
    },
    {
      "epoch": 10.350061199510403,
      "grad_norm": 2.595967695496556,
      "learning_rate": 1.0947650525245181e-06,
      "loss": 0.5352,
      "step": 8456
    },
    {
      "epoch": 10.351285189718482,
      "grad_norm": 1.7952438960163517,
      "learning_rate": 1.09423504410928e-06,
      "loss": 0.4463,
      "step": 8457
    },
    {
      "epoch": 10.35250917992656,
      "grad_norm": 2.309562721830781,
      "learning_rate": 1.0937051280757141e-06,
      "loss": 0.529,
      "step": 8458
    },
    {
      "epoch": 10.353733170134639,
      "grad_norm": 1.1468932196982267,
      "learning_rate": 1.0931753044586465e-06,
      "loss": 0.6099,
      "step": 8459
    },
    {
      "epoch": 10.354957160342718,
      "grad_norm": 1.170673848359504,
      "learning_rate": 1.0926455732928944e-06,
      "loss": 0.6046,
      "step": 8460
    },
    {
      "epoch": 10.356181150550796,
      "grad_norm": 1.7780906369407454,
      "learning_rate": 1.0921159346132697e-06,
      "loss": 1.5003,
      "step": 8461
    },
    {
      "epoch": 10.357405140758875,
      "grad_norm": 2.45774667451582,
      "learning_rate": 1.0915863884545789e-06,
      "loss": 0.7125,
      "step": 8462
    },
    {
      "epoch": 10.358629130966952,
      "grad_norm": 1.410064564754691,
      "learning_rate": 1.0910569348516206e-06,
      "loss": 0.3992,
      "step": 8463
    },
    {
      "epoch": 10.35985312117503,
      "grad_norm": 2.3153947208510592,
      "learning_rate": 1.0905275738391896e-06,
      "loss": 0.4272,
      "step": 8464
    },
    {
      "epoch": 10.361077111383109,
      "grad_norm": 2.6092059197116115,
      "learning_rate": 1.0899983054520732e-06,
      "loss": 0.3942,
      "step": 8465
    },
    {
      "epoch": 10.362301101591187,
      "grad_norm": 1.821736934644811,
      "learning_rate": 1.0894691297250528e-06,
      "loss": 0.8815,
      "step": 8466
    },
    {
      "epoch": 10.363525091799266,
      "grad_norm": 1.6484302717646422,
      "learning_rate": 1.088940046692904e-06,
      "loss": 0.5622,
      "step": 8467
    },
    {
      "epoch": 10.364749082007345,
      "grad_norm": 1.6179953938735485,
      "learning_rate": 1.0884110563903963e-06,
      "loss": 1.4848,
      "step": 8468
    },
    {
      "epoch": 10.365973072215422,
      "grad_norm": 1.57620223990001,
      "learning_rate": 1.0878821588522929e-06,
      "loss": 0.3544,
      "step": 8469
    },
    {
      "epoch": 10.3671970624235,
      "grad_norm": 0.5759291107983122,
      "learning_rate": 1.0873533541133507e-06,
      "loss": 0.1297,
      "step": 8470
    },
    {
      "epoch": 10.368421052631579,
      "grad_norm": 3.0648153432349017,
      "learning_rate": 1.0868246422083203e-06,
      "loss": 0.4475,
      "step": 8471
    },
    {
      "epoch": 10.369645042839657,
      "grad_norm": 2.616499137367613,
      "learning_rate": 1.0862960231719483e-06,
      "loss": 0.277,
      "step": 8472
    },
    {
      "epoch": 10.370869033047736,
      "grad_norm": 1.5288367613506255,
      "learning_rate": 1.0857674970389728e-06,
      "loss": 0.8205,
      "step": 8473
    },
    {
      "epoch": 10.372093023255815,
      "grad_norm": 2.0238795409573798,
      "learning_rate": 1.0852390638441273e-06,
      "loss": 0.7629,
      "step": 8474
    },
    {
      "epoch": 10.373317013463891,
      "grad_norm": 1.8461031123382783,
      "learning_rate": 1.084710723622136e-06,
      "loss": 0.4531,
      "step": 8475
    },
    {
      "epoch": 10.37454100367197,
      "grad_norm": 2.2634288277438324,
      "learning_rate": 1.084182476407722e-06,
      "loss": 0.8163,
      "step": 8476
    },
    {
      "epoch": 10.375764993880049,
      "grad_norm": 2.1175401077470717,
      "learning_rate": 1.0836543222355988e-06,
      "loss": 0.3873,
      "step": 8477
    },
    {
      "epoch": 10.376988984088127,
      "grad_norm": 1.9046717863016955,
      "learning_rate": 1.083126261140475e-06,
      "loss": 0.475,
      "step": 8478
    },
    {
      "epoch": 10.378212974296206,
      "grad_norm": 2.381146173054407,
      "learning_rate": 1.0825982931570527e-06,
      "loss": 1.1994,
      "step": 8479
    },
    {
      "epoch": 10.379436964504285,
      "grad_norm": 3.6868159414924055,
      "learning_rate": 1.082070418320028e-06,
      "loss": 0.2738,
      "step": 8480
    },
    {
      "epoch": 10.380660954712361,
      "grad_norm": 2.6036121991170247,
      "learning_rate": 1.0815426366640908e-06,
      "loss": 0.6751,
      "step": 8481
    },
    {
      "epoch": 10.38188494492044,
      "grad_norm": 2.4147349772096485,
      "learning_rate": 1.081014948223925e-06,
      "loss": 0.4479,
      "step": 8482
    },
    {
      "epoch": 10.383108935128519,
      "grad_norm": 1.687735603140283,
      "learning_rate": 1.080487353034208e-06,
      "loss": 0.44,
      "step": 8483
    },
    {
      "epoch": 10.384332925336597,
      "grad_norm": 1.9444074753120237,
      "learning_rate": 1.079959851129611e-06,
      "loss": 0.5245,
      "step": 8484
    },
    {
      "epoch": 10.385556915544676,
      "grad_norm": 2.455571156908325,
      "learning_rate": 1.0794324425448024e-06,
      "loss": 0.5175,
      "step": 8485
    },
    {
      "epoch": 10.386780905752754,
      "grad_norm": 2.2339148445775168,
      "learning_rate": 1.0789051273144383e-06,
      "loss": 0.6168,
      "step": 8486
    },
    {
      "epoch": 10.388004895960833,
      "grad_norm": 2.0290828295761054,
      "learning_rate": 1.0783779054731726e-06,
      "loss": 0.7334,
      "step": 8487
    },
    {
      "epoch": 10.38922888616891,
      "grad_norm": 1.6903125305731184,
      "learning_rate": 1.0778507770556523e-06,
      "loss": 0.9273,
      "step": 8488
    },
    {
      "epoch": 10.390452876376989,
      "grad_norm": 1.9205737346984166,
      "learning_rate": 1.0773237420965194e-06,
      "loss": 0.585,
      "step": 8489
    },
    {
      "epoch": 10.391676866585067,
      "grad_norm": 3.7927525297635896,
      "learning_rate": 1.0767968006304081e-06,
      "loss": 0.3788,
      "step": 8490
    },
    {
      "epoch": 10.392900856793146,
      "grad_norm": 2.240378435208915,
      "learning_rate": 1.076269952691947e-06,
      "loss": 0.3499,
      "step": 8491
    },
    {
      "epoch": 10.394124847001224,
      "grad_norm": 1.4477748483363693,
      "learning_rate": 1.0757431983157596e-06,
      "loss": 0.5325,
      "step": 8492
    },
    {
      "epoch": 10.395348837209303,
      "grad_norm": 1.3874320259018462,
      "learning_rate": 1.0752165375364593e-06,
      "loss": 1.5656,
      "step": 8493
    },
    {
      "epoch": 10.39657282741738,
      "grad_norm": 2.8363639458880465,
      "learning_rate": 1.0746899703886593e-06,
      "loss": 0.3526,
      "step": 8494
    },
    {
      "epoch": 10.397796817625458,
      "grad_norm": 2.138374078096518,
      "learning_rate": 1.0741634969069626e-06,
      "loss": 0.5121,
      "step": 8495
    },
    {
      "epoch": 10.399020807833537,
      "grad_norm": 1.1356686120448567,
      "learning_rate": 1.0736371171259668e-06,
      "loss": 0.4703,
      "step": 8496
    },
    {
      "epoch": 10.400244798041616,
      "grad_norm": 2.751851324266045,
      "learning_rate": 1.0731108310802644e-06,
      "loss": 0.5136,
      "step": 8497
    },
    {
      "epoch": 10.401468788249694,
      "grad_norm": 3.2914142929519596,
      "learning_rate": 1.07258463880444e-06,
      "loss": 0.3439,
      "step": 8498
    },
    {
      "epoch": 10.402692778457773,
      "grad_norm": 1.3125925852514093,
      "learning_rate": 1.0720585403330735e-06,
      "loss": 0.5825,
      "step": 8499
    },
    {
      "epoch": 10.403916768665852,
      "grad_norm": 2.0807189420410306,
      "learning_rate": 1.0715325357007383e-06,
      "loss": 0.6547,
      "step": 8500
    },
    {
      "epoch": 10.405140758873928,
      "grad_norm": 2.1604054101934342,
      "learning_rate": 1.0710066249420007e-06,
      "loss": 0.5354,
      "step": 8501
    },
    {
      "epoch": 10.406364749082007,
      "grad_norm": 2.6356944431209066,
      "learning_rate": 1.0704808080914228e-06,
      "loss": 0.3399,
      "step": 8502
    },
    {
      "epoch": 10.407588739290086,
      "grad_norm": 2.8971470984120393,
      "learning_rate": 1.0699550851835596e-06,
      "loss": 0.4209,
      "step": 8503
    },
    {
      "epoch": 10.408812729498164,
      "grad_norm": 1.0893131853232543,
      "learning_rate": 1.0694294562529581e-06,
      "loss": 0.5379,
      "step": 8504
    },
    {
      "epoch": 10.410036719706243,
      "grad_norm": 2.368927870768551,
      "learning_rate": 1.0689039213341615e-06,
      "loss": 0.6693,
      "step": 8505
    },
    {
      "epoch": 10.411260709914322,
      "grad_norm": 1.618998325175203,
      "learning_rate": 1.0683784804617048e-06,
      "loss": 0.5115,
      "step": 8506
    },
    {
      "epoch": 10.412484700122398,
      "grad_norm": 1.332734202616304,
      "learning_rate": 1.06785313367012e-06,
      "loss": 0.4026,
      "step": 8507
    },
    {
      "epoch": 10.413708690330477,
      "grad_norm": 2.4700827786989406,
      "learning_rate": 1.0673278809939305e-06,
      "loss": 1.2786,
      "step": 8508
    },
    {
      "epoch": 10.414932680538556,
      "grad_norm": 2.163851268908277,
      "learning_rate": 1.0668027224676541e-06,
      "loss": 1.0699,
      "step": 8509
    },
    {
      "epoch": 10.416156670746634,
      "grad_norm": 1.2365144974427893,
      "learning_rate": 1.0662776581258002e-06,
      "loss": 0.4284,
      "step": 8510
    },
    {
      "epoch": 10.417380660954713,
      "grad_norm": 1.853006161828569,
      "learning_rate": 1.0657526880028765e-06,
      "loss": 0.4219,
      "step": 8511
    },
    {
      "epoch": 10.418604651162791,
      "grad_norm": 2.716244962000596,
      "learning_rate": 1.065227812133381e-06,
      "loss": 0.9512,
      "step": 8512
    },
    {
      "epoch": 10.419828641370868,
      "grad_norm": 1.289329501372888,
      "learning_rate": 1.0647030305518072e-06,
      "loss": 0.5567,
      "step": 8513
    },
    {
      "epoch": 10.421052631578947,
      "grad_norm": 1.5825851498524606,
      "learning_rate": 1.0641783432926412e-06,
      "loss": 0.4049,
      "step": 8514
    },
    {
      "epoch": 10.422276621787026,
      "grad_norm": 3.243227755107791,
      "learning_rate": 1.0636537503903638e-06,
      "loss": 0.5923,
      "step": 8515
    },
    {
      "epoch": 10.423500611995104,
      "grad_norm": 2.0180419634127578,
      "learning_rate": 1.0631292518794493e-06,
      "loss": 0.4882,
      "step": 8516
    },
    {
      "epoch": 10.424724602203183,
      "grad_norm": 1.4154607401356551,
      "learning_rate": 1.0626048477943656e-06,
      "loss": 0.844,
      "step": 8517
    },
    {
      "epoch": 10.425948592411261,
      "grad_norm": 2.1397387088960964,
      "learning_rate": 1.0620805381695748e-06,
      "loss": 0.5346,
      "step": 8518
    },
    {
      "epoch": 10.427172582619338,
      "grad_norm": 2.0709679274138977,
      "learning_rate": 1.0615563230395319e-06,
      "loss": 0.4063,
      "step": 8519
    },
    {
      "epoch": 10.428396572827417,
      "grad_norm": 1.827648468172567,
      "learning_rate": 1.0610322024386873e-06,
      "loss": 0.5294,
      "step": 8520
    },
    {
      "epoch": 10.429620563035495,
      "grad_norm": 2.653414524870944,
      "learning_rate": 1.0605081764014854e-06,
      "loss": 0.4732,
      "step": 8521
    },
    {
      "epoch": 10.430844553243574,
      "grad_norm": 1.5313502745050567,
      "learning_rate": 1.0599842449623608e-06,
      "loss": 0.6231,
      "step": 8522
    },
    {
      "epoch": 10.432068543451653,
      "grad_norm": 1.9336528878368633,
      "learning_rate": 1.0594604081557445e-06,
      "loss": 0.611,
      "step": 8523
    },
    {
      "epoch": 10.433292533659731,
      "grad_norm": 2.0299623802001245,
      "learning_rate": 1.0589366660160627e-06,
      "loss": 0.5454,
      "step": 8524
    },
    {
      "epoch": 10.43451652386781,
      "grad_norm": 1.3729119294793166,
      "learning_rate": 1.0584130185777333e-06,
      "loss": 0.5235,
      "step": 8525
    },
    {
      "epoch": 10.435740514075887,
      "grad_norm": 1.2884984921244165,
      "learning_rate": 1.057889465875168e-06,
      "loss": 0.8946,
      "step": 8526
    },
    {
      "epoch": 10.436964504283965,
      "grad_norm": 1.325850595438034,
      "learning_rate": 1.0573660079427736e-06,
      "loss": 0.4913,
      "step": 8527
    },
    {
      "epoch": 10.438188494492044,
      "grad_norm": 1.61263251394551,
      "learning_rate": 1.0568426448149487e-06,
      "loss": 1.2441,
      "step": 8528
    },
    {
      "epoch": 10.439412484700123,
      "grad_norm": 2.090217960503598,
      "learning_rate": 1.0563193765260874e-06,
      "loss": 0.7318,
      "step": 8529
    },
    {
      "epoch": 10.440636474908201,
      "grad_norm": 1.294862815418946,
      "learning_rate": 1.055796203110577e-06,
      "loss": 0.7614,
      "step": 8530
    },
    {
      "epoch": 10.44186046511628,
      "grad_norm": 1.9132575257790458,
      "learning_rate": 1.0552731246027984e-06,
      "loss": 0.457,
      "step": 8531
    },
    {
      "epoch": 10.443084455324357,
      "grad_norm": 1.2814714036606616,
      "learning_rate": 1.054750141037126e-06,
      "loss": 0.9842,
      "step": 8532
    },
    {
      "epoch": 10.444308445532435,
      "grad_norm": 3.4945895559816735,
      "learning_rate": 1.0542272524479292e-06,
      "loss": 0.3477,
      "step": 8533
    },
    {
      "epoch": 10.445532435740514,
      "grad_norm": 1.6280409833084672,
      "learning_rate": 1.0537044588695698e-06,
      "loss": 1.0423,
      "step": 8534
    },
    {
      "epoch": 10.446756425948593,
      "grad_norm": 1.1946593891705124,
      "learning_rate": 1.0531817603364034e-06,
      "loss": 0.5404,
      "step": 8535
    },
    {
      "epoch": 10.447980416156671,
      "grad_norm": 1.5516899306316263,
      "learning_rate": 1.0526591568827798e-06,
      "loss": 0.4719,
      "step": 8536
    },
    {
      "epoch": 10.44920440636475,
      "grad_norm": 1.4791036006385672,
      "learning_rate": 1.052136648543044e-06,
      "loss": 0.2714,
      "step": 8537
    },
    {
      "epoch": 10.450428396572827,
      "grad_norm": 1.6057125036166295,
      "learning_rate": 1.051614235351533e-06,
      "loss": 0.5525,
      "step": 8538
    },
    {
      "epoch": 10.451652386780905,
      "grad_norm": 1.4560796067969188,
      "learning_rate": 1.0510919173425765e-06,
      "loss": 1.4085,
      "step": 8539
    },
    {
      "epoch": 10.452876376988984,
      "grad_norm": 1.3466907811331736,
      "learning_rate": 1.0505696945504993e-06,
      "loss": 0.5767,
      "step": 8540
    },
    {
      "epoch": 10.454100367197062,
      "grad_norm": 1.53148944518406,
      "learning_rate": 1.0500475670096214e-06,
      "loss": 0.7133,
      "step": 8541
    },
    {
      "epoch": 10.455324357405141,
      "grad_norm": 2.408949249335815,
      "learning_rate": 1.0495255347542544e-06,
      "loss": 0.3978,
      "step": 8542
    },
    {
      "epoch": 10.45654834761322,
      "grad_norm": 2.1987864312516665,
      "learning_rate": 1.0490035978187044e-06,
      "loss": 0.6024,
      "step": 8543
    },
    {
      "epoch": 10.457772337821297,
      "grad_norm": 1.073326998015257,
      "learning_rate": 1.048481756237271e-06,
      "loss": 0.6294,
      "step": 8544
    },
    {
      "epoch": 10.458996328029375,
      "grad_norm": 1.1647517581643725,
      "learning_rate": 1.0479600100442477e-06,
      "loss": 0.6928,
      "step": 8545
    },
    {
      "epoch": 10.460220318237454,
      "grad_norm": 1.84873925643271,
      "learning_rate": 1.047438359273922e-06,
      "loss": 0.3523,
      "step": 8546
    },
    {
      "epoch": 10.461444308445532,
      "grad_norm": 1.788973653532664,
      "learning_rate": 1.046916803960575e-06,
      "loss": 1.1933,
      "step": 8547
    },
    {
      "epoch": 10.462668298653611,
      "grad_norm": 1.730655429893381,
      "learning_rate": 1.0463953441384805e-06,
      "loss": 0.7487,
      "step": 8548
    },
    {
      "epoch": 10.46389228886169,
      "grad_norm": 2.448790105052832,
      "learning_rate": 1.0458739798419069e-06,
      "loss": 0.5002,
      "step": 8549
    },
    {
      "epoch": 10.465116279069768,
      "grad_norm": 1.8229118114534826,
      "learning_rate": 1.0453527111051183e-06,
      "loss": 0.4708,
      "step": 8550
    },
    {
      "epoch": 10.466340269277845,
      "grad_norm": 1.255274029461244,
      "learning_rate": 1.0448315379623686e-06,
      "loss": 0.6464,
      "step": 8551
    },
    {
      "epoch": 10.467564259485924,
      "grad_norm": 1.5478496195234754,
      "learning_rate": 1.044310460447908e-06,
      "loss": 0.2234,
      "step": 8552
    },
    {
      "epoch": 10.468788249694002,
      "grad_norm": 2.7132745979968043,
      "learning_rate": 1.0437894785959785e-06,
      "loss": 0.3596,
      "step": 8553
    },
    {
      "epoch": 10.470012239902081,
      "grad_norm": 1.1046243536726776,
      "learning_rate": 1.0432685924408195e-06,
      "loss": 0.5939,
      "step": 8554
    },
    {
      "epoch": 10.47123623011016,
      "grad_norm": 2.0125708444732764,
      "learning_rate": 1.0427478020166605e-06,
      "loss": 0.5471,
      "step": 8555
    },
    {
      "epoch": 10.472460220318238,
      "grad_norm": 1.689604537258621,
      "learning_rate": 1.0422271073577267e-06,
      "loss": 0.9057,
      "step": 8556
    },
    {
      "epoch": 10.473684210526315,
      "grad_norm": 1.8073316984831012,
      "learning_rate": 1.0417065084982347e-06,
      "loss": 1.0181,
      "step": 8557
    },
    {
      "epoch": 10.474908200734394,
      "grad_norm": 2.216652425391418,
      "learning_rate": 1.041186005472396e-06,
      "loss": 0.5383,
      "step": 8558
    },
    {
      "epoch": 10.476132190942472,
      "grad_norm": 1.929161702659491,
      "learning_rate": 1.0406655983144184e-06,
      "loss": 0.5542,
      "step": 8559
    },
    {
      "epoch": 10.477356181150551,
      "grad_norm": 2.380855596928087,
      "learning_rate": 1.0401452870585e-06,
      "loss": 0.5901,
      "step": 8560
    },
    {
      "epoch": 10.47858017135863,
      "grad_norm": 1.9690194840584812,
      "learning_rate": 1.0396250717388333e-06,
      "loss": 0.9588,
      "step": 8561
    },
    {
      "epoch": 10.479804161566708,
      "grad_norm": 2.5635183027017203,
      "learning_rate": 1.0391049523896058e-06,
      "loss": 0.6202,
      "step": 8562
    },
    {
      "epoch": 10.481028151774785,
      "grad_norm": 2.336644706463696,
      "learning_rate": 1.038584929044997e-06,
      "loss": 0.8969,
      "step": 8563
    },
    {
      "epoch": 10.482252141982864,
      "grad_norm": 1.393168657940036,
      "learning_rate": 1.0380650017391816e-06,
      "loss": 0.6936,
      "step": 8564
    },
    {
      "epoch": 10.483476132190942,
      "grad_norm": 2.3988903205596914,
      "learning_rate": 1.0375451705063266e-06,
      "loss": 0.4716,
      "step": 8565
    },
    {
      "epoch": 10.48470012239902,
      "grad_norm": 1.4249118757758763,
      "learning_rate": 1.037025435380593e-06,
      "loss": 0.9673,
      "step": 8566
    },
    {
      "epoch": 10.4859241126071,
      "grad_norm": 2.5041768966981386,
      "learning_rate": 1.0365057963961378e-06,
      "loss": 0.4478,
      "step": 8567
    },
    {
      "epoch": 10.487148102815178,
      "grad_norm": 3.046502042862484,
      "learning_rate": 1.035986253587109e-06,
      "loss": 0.3418,
      "step": 8568
    },
    {
      "epoch": 10.488372093023255,
      "grad_norm": 1.3838253755059224,
      "learning_rate": 1.035466806987648e-06,
      "loss": 0.7765,
      "step": 8569
    },
    {
      "epoch": 10.489596083231334,
      "grad_norm": 1.5085520372116943,
      "learning_rate": 1.0349474566318915e-06,
      "loss": 1.1161,
      "step": 8570
    },
    {
      "epoch": 10.490820073439412,
      "grad_norm": 2.905626955034027,
      "learning_rate": 1.0344282025539685e-06,
      "loss": 0.4615,
      "step": 8571
    },
    {
      "epoch": 10.49204406364749,
      "grad_norm": 1.4505173855674478,
      "learning_rate": 1.0339090447880038e-06,
      "loss": 0.6872,
      "step": 8572
    },
    {
      "epoch": 10.49326805385557,
      "grad_norm": 1.1811144475911761,
      "learning_rate": 1.033389983368114e-06,
      "loss": 0.5616,
      "step": 8573
    },
    {
      "epoch": 10.494492044063648,
      "grad_norm": 1.9184591243342848,
      "learning_rate": 1.032871018328411e-06,
      "loss": 1.4877,
      "step": 8574
    },
    {
      "epoch": 10.495716034271727,
      "grad_norm": 0.9759503101234503,
      "learning_rate": 1.0323521497029965e-06,
      "loss": 0.4276,
      "step": 8575
    },
    {
      "epoch": 10.496940024479803,
      "grad_norm": 1.4413719769243931,
      "learning_rate": 1.031833377525971e-06,
      "loss": 1.0234,
      "step": 8576
    },
    {
      "epoch": 10.498164014687882,
      "grad_norm": 2.404705329617172,
      "learning_rate": 1.0313147018314254e-06,
      "loss": 0.3636,
      "step": 8577
    },
    {
      "epoch": 10.49938800489596,
      "grad_norm": 1.9465073573921237,
      "learning_rate": 1.0307961226534455e-06,
      "loss": 0.4249,
      "step": 8578
    },
    {
      "epoch": 10.50061199510404,
      "grad_norm": 2.7092592426378204,
      "learning_rate": 1.03027764002611e-06,
      "loss": 0.3979,
      "step": 8579
    },
    {
      "epoch": 10.501835985312118,
      "grad_norm": 1.5498379859647287,
      "learning_rate": 1.0297592539834918e-06,
      "loss": 0.6045,
      "step": 8580
    },
    {
      "epoch": 10.503059975520197,
      "grad_norm": 2.033663399477973,
      "learning_rate": 1.0292409645596577e-06,
      "loss": 0.5741,
      "step": 8581
    },
    {
      "epoch": 10.504283965728273,
      "grad_norm": 1.2396010903048145,
      "learning_rate": 1.0287227717886674e-06,
      "loss": 0.4689,
      "step": 8582
    },
    {
      "epoch": 10.505507955936352,
      "grad_norm": 1.4061781072966644,
      "learning_rate": 1.0282046757045746e-06,
      "loss": 0.4861,
      "step": 8583
    },
    {
      "epoch": 10.50673194614443,
      "grad_norm": 1.411176176323051,
      "learning_rate": 1.027686676341426e-06,
      "loss": 0.6787,
      "step": 8584
    },
    {
      "epoch": 10.50795593635251,
      "grad_norm": 2.6524496951075696,
      "learning_rate": 1.0271687737332648e-06,
      "loss": 0.4892,
      "step": 8585
    },
    {
      "epoch": 10.509179926560588,
      "grad_norm": 1.9336231716134678,
      "learning_rate": 1.0266509679141234e-06,
      "loss": 0.6238,
      "step": 8586
    },
    {
      "epoch": 10.510403916768666,
      "grad_norm": 1.908748841258517,
      "learning_rate": 1.026133258918031e-06,
      "loss": 0.3384,
      "step": 8587
    },
    {
      "epoch": 10.511627906976745,
      "grad_norm": 1.0647839552556937,
      "learning_rate": 1.0256156467790081e-06,
      "loss": 0.5301,
      "step": 8588
    },
    {
      "epoch": 10.512851897184822,
      "grad_norm": 1.4737699117872558,
      "learning_rate": 1.0250981315310725e-06,
      "loss": 0.4304,
      "step": 8589
    },
    {
      "epoch": 10.5140758873929,
      "grad_norm": 1.586853684451767,
      "learning_rate": 1.0245807132082328e-06,
      "loss": 0.8166,
      "step": 8590
    },
    {
      "epoch": 10.51529987760098,
      "grad_norm": 2.473245525293707,
      "learning_rate": 1.0240633918444912e-06,
      "loss": 0.5376,
      "step": 8591
    },
    {
      "epoch": 10.516523867809058,
      "grad_norm": 2.3529658151318222,
      "learning_rate": 1.0235461674738443e-06,
      "loss": 1.078,
      "step": 8592
    },
    {
      "epoch": 10.517747858017136,
      "grad_norm": 1.8401405385427956,
      "learning_rate": 1.0230290401302822e-06,
      "loss": 1.3448,
      "step": 8593
    },
    {
      "epoch": 10.518971848225215,
      "grad_norm": 2.472234970532229,
      "learning_rate": 1.0225120098477887e-06,
      "loss": 0.4987,
      "step": 8594
    },
    {
      "epoch": 10.520195838433292,
      "grad_norm": 1.9533341669875603,
      "learning_rate": 1.0219950766603412e-06,
      "loss": 1.185,
      "step": 8595
    },
    {
      "epoch": 10.52141982864137,
      "grad_norm": 1.7165167903007745,
      "learning_rate": 1.0214782406019102e-06,
      "loss": 0.731,
      "step": 8596
    },
    {
      "epoch": 10.522643818849449,
      "grad_norm": 1.6499254116457798,
      "learning_rate": 1.0209615017064609e-06,
      "loss": 0.6991,
      "step": 8597
    },
    {
      "epoch": 10.523867809057528,
      "grad_norm": 1.534988972509923,
      "learning_rate": 1.0204448600079512e-06,
      "loss": 0.5498,
      "step": 8598
    },
    {
      "epoch": 10.525091799265606,
      "grad_norm": 2.7676899603373344,
      "learning_rate": 1.0199283155403326e-06,
      "loss": 0.9129,
      "step": 8599
    },
    {
      "epoch": 10.526315789473685,
      "grad_norm": 2.283960403297424,
      "learning_rate": 1.0194118683375502e-06,
      "loss": 0.639,
      "step": 8600
    },
    {
      "epoch": 10.527539779681762,
      "grad_norm": 1.7311773487278797,
      "learning_rate": 1.0188955184335434e-06,
      "loss": 1.0945,
      "step": 8601
    },
    {
      "epoch": 10.52876376988984,
      "grad_norm": 2.818609803500443,
      "learning_rate": 1.0183792658622454e-06,
      "loss": 0.3539,
      "step": 8602
    },
    {
      "epoch": 10.529987760097919,
      "grad_norm": 1.813462498281762,
      "learning_rate": 1.0178631106575828e-06,
      "loss": 1.1873,
      "step": 8603
    },
    {
      "epoch": 10.531211750305998,
      "grad_norm": 2.384475440937434,
      "learning_rate": 1.0173470528534738e-06,
      "loss": 0.5919,
      "step": 8604
    },
    {
      "epoch": 10.532435740514076,
      "grad_norm": 1.4996834917596134,
      "learning_rate": 1.0168310924838314e-06,
      "loss": 0.7503,
      "step": 8605
    },
    {
      "epoch": 10.533659730722155,
      "grad_norm": 1.4990267717439854,
      "learning_rate": 1.0163152295825651e-06,
      "loss": 0.6147,
      "step": 8606
    },
    {
      "epoch": 10.534883720930232,
      "grad_norm": 1.5818109512131402,
      "learning_rate": 1.0157994641835737e-06,
      "loss": 0.5704,
      "step": 8607
    },
    {
      "epoch": 10.53610771113831,
      "grad_norm": 1.7614122556750675,
      "learning_rate": 1.015283796320752e-06,
      "loss": 0.4147,
      "step": 8608
    },
    {
      "epoch": 10.537331701346389,
      "grad_norm": 2.068939008970077,
      "learning_rate": 1.0147682260279876e-06,
      "loss": 0.5738,
      "step": 8609
    },
    {
      "epoch": 10.538555691554468,
      "grad_norm": 1.5759608332509436,
      "learning_rate": 1.014252753339162e-06,
      "loss": 0.7462,
      "step": 8610
    },
    {
      "epoch": 10.539779681762546,
      "grad_norm": 2.250882510473957,
      "learning_rate": 1.0137373782881498e-06,
      "loss": 0.4675,
      "step": 8611
    },
    {
      "epoch": 10.541003671970625,
      "grad_norm": 1.6639344199171533,
      "learning_rate": 1.01322210090882e-06,
      "loss": 0.526,
      "step": 8612
    },
    {
      "epoch": 10.542227662178703,
      "grad_norm": 2.220784679362921,
      "learning_rate": 1.0127069212350344e-06,
      "loss": 0.9196,
      "step": 8613
    },
    {
      "epoch": 10.54345165238678,
      "grad_norm": 1.300612855856146,
      "learning_rate": 1.0121918393006488e-06,
      "loss": 0.7301,
      "step": 8614
    },
    {
      "epoch": 10.544675642594859,
      "grad_norm": 1.855938439511972,
      "learning_rate": 1.0116768551395127e-06,
      "loss": 0.4165,
      "step": 8615
    },
    {
      "epoch": 10.545899632802938,
      "grad_norm": 1.4739060358793459,
      "learning_rate": 1.0111619687854685e-06,
      "loss": 0.7944,
      "step": 8616
    },
    {
      "epoch": 10.547123623011016,
      "grad_norm": 1.0683053462846368,
      "learning_rate": 1.010647180272353e-06,
      "loss": 0.3939,
      "step": 8617
    },
    {
      "epoch": 10.548347613219095,
      "grad_norm": 1.8663184412326632,
      "learning_rate": 1.0101324896339962e-06,
      "loss": 0.486,
      "step": 8618
    },
    {
      "epoch": 10.549571603427173,
      "grad_norm": 1.3697879279837641,
      "learning_rate": 1.0096178969042203e-06,
      "loss": 0.4419,
      "step": 8619
    },
    {
      "epoch": 10.55079559363525,
      "grad_norm": 1.1819307295203072,
      "learning_rate": 1.009103402116845e-06,
      "loss": 0.5986,
      "step": 8620
    },
    {
      "epoch": 10.552019583843329,
      "grad_norm": 2.8455107984204524,
      "learning_rate": 1.0085890053056802e-06,
      "loss": 0.797,
      "step": 8621
    },
    {
      "epoch": 10.553243574051407,
      "grad_norm": 1.4189818917342532,
      "learning_rate": 1.0080747065045291e-06,
      "loss": 0.3661,
      "step": 8622
    },
    {
      "epoch": 10.554467564259486,
      "grad_norm": 1.4870154723582019,
      "learning_rate": 1.0075605057471888e-06,
      "loss": 0.5598,
      "step": 8623
    },
    {
      "epoch": 10.555691554467565,
      "grad_norm": 3.3156629681534624,
      "learning_rate": 1.007046403067453e-06,
      "loss": 0.4379,
      "step": 8624
    },
    {
      "epoch": 10.556915544675643,
      "grad_norm": 1.618991881944571,
      "learning_rate": 1.0065323984991055e-06,
      "loss": 0.7005,
      "step": 8625
    },
    {
      "epoch": 10.55813953488372,
      "grad_norm": 1.8464091349514562,
      "learning_rate": 1.0060184920759248e-06,
      "loss": 0.7448,
      "step": 8626
    },
    {
      "epoch": 10.559363525091799,
      "grad_norm": 2.70657871212265,
      "learning_rate": 1.005504683831683e-06,
      "loss": 0.5006,
      "step": 8627
    },
    {
      "epoch": 10.560587515299877,
      "grad_norm": 1.3999753808369797,
      "learning_rate": 1.0049909738001453e-06,
      "loss": 1.5577,
      "step": 8628
    },
    {
      "epoch": 10.561811505507956,
      "grad_norm": 2.0426703245408437,
      "learning_rate": 1.0044773620150714e-06,
      "loss": 1.5418,
      "step": 8629
    },
    {
      "epoch": 10.563035495716035,
      "grad_norm": 1.2065476412762617,
      "learning_rate": 1.0039638485102139e-06,
      "loss": 0.5026,
      "step": 8630
    },
    {
      "epoch": 10.564259485924113,
      "grad_norm": 2.0240925844000093,
      "learning_rate": 1.0034504333193184e-06,
      "loss": 0.7639,
      "step": 8631
    },
    {
      "epoch": 10.56548347613219,
      "grad_norm": 1.468703272721381,
      "learning_rate": 1.0029371164761254e-06,
      "loss": 1.1587,
      "step": 8632
    },
    {
      "epoch": 10.566707466340269,
      "grad_norm": 2.7534501882219993,
      "learning_rate": 1.0024238980143678e-06,
      "loss": 0.4463,
      "step": 8633
    },
    {
      "epoch": 10.567931456548347,
      "grad_norm": 2.388153447638572,
      "learning_rate": 1.0019107779677725e-06,
      "loss": 0.4443,
      "step": 8634
    },
    {
      "epoch": 10.569155446756426,
      "grad_norm": 1.3928934982659522,
      "learning_rate": 1.0013977563700598e-06,
      "loss": 0.6943,
      "step": 8635
    },
    {
      "epoch": 10.570379436964505,
      "grad_norm": 1.2741654113455056,
      "learning_rate": 1.0008848332549425e-06,
      "loss": 0.3862,
      "step": 8636
    },
    {
      "epoch": 10.571603427172583,
      "grad_norm": 1.2099311082184236,
      "learning_rate": 1.0003720086561302e-06,
      "loss": 0.5721,
      "step": 8637
    },
    {
      "epoch": 10.572827417380662,
      "grad_norm": 1.3736161263264792,
      "learning_rate": 9.998592826073229e-07,
      "loss": 1.0449,
      "step": 8638
    },
    {
      "epoch": 10.574051407588739,
      "grad_norm": 1.8240067095426356,
      "learning_rate": 9.993466551422154e-07,
      "loss": 0.5825,
      "step": 8639
    },
    {
      "epoch": 10.575275397796817,
      "grad_norm": 3.217571274147327,
      "learning_rate": 9.988341262944937e-07,
      "loss": 0.3487,
      "step": 8640
    },
    {
      "epoch": 10.576499388004896,
      "grad_norm": 1.657034530896635,
      "learning_rate": 9.983216960978418e-07,
      "loss": 0.3856,
      "step": 8641
    },
    {
      "epoch": 10.577723378212974,
      "grad_norm": 1.3708064243403908,
      "learning_rate": 9.978093645859332e-07,
      "loss": 1.3304,
      "step": 8642
    },
    {
      "epoch": 10.578947368421053,
      "grad_norm": 1.580619669799115,
      "learning_rate": 9.972971317924373e-07,
      "loss": 0.3792,
      "step": 8643
    },
    {
      "epoch": 10.580171358629132,
      "grad_norm": 1.3813460874033585,
      "learning_rate": 9.967849977510159e-07,
      "loss": 0.5153,
      "step": 8644
    },
    {
      "epoch": 10.581395348837209,
      "grad_norm": 1.3533727501677353,
      "learning_rate": 9.962729624953244e-07,
      "loss": 0.5263,
      "step": 8645
    },
    {
      "epoch": 10.582619339045287,
      "grad_norm": 1.761467395670106,
      "learning_rate": 9.957610260590117e-07,
      "loss": 1.6148,
      "step": 8646
    },
    {
      "epoch": 10.583843329253366,
      "grad_norm": 0.9766463317212936,
      "learning_rate": 9.952491884757207e-07,
      "loss": 0.4469,
      "step": 8647
    },
    {
      "epoch": 10.585067319461444,
      "grad_norm": 1.8151568760805872,
      "learning_rate": 9.947374497790873e-07,
      "loss": 0.5449,
      "step": 8648
    },
    {
      "epoch": 10.586291309669523,
      "grad_norm": 1.7224527971452135,
      "learning_rate": 9.942258100027401e-07,
      "loss": 0.6765,
      "step": 8649
    },
    {
      "epoch": 10.587515299877602,
      "grad_norm": 1.4977210939235652,
      "learning_rate": 9.937142691803053e-07,
      "loss": 0.71,
      "step": 8650
    },
    {
      "epoch": 10.588739290085678,
      "grad_norm": 1.6944018727479195,
      "learning_rate": 9.932028273453962e-07,
      "loss": 0.5209,
      "step": 8651
    },
    {
      "epoch": 10.589963280293757,
      "grad_norm": 2.2183766839352166,
      "learning_rate": 9.92691484531624e-07,
      "loss": 0.5768,
      "step": 8652
    },
    {
      "epoch": 10.591187270501836,
      "grad_norm": 1.0571109068385465,
      "learning_rate": 9.921802407725918e-07,
      "loss": 0.5947,
      "step": 8653
    },
    {
      "epoch": 10.592411260709914,
      "grad_norm": 2.582221738760998,
      "learning_rate": 9.916690961018977e-07,
      "loss": 0.5776,
      "step": 8654
    },
    {
      "epoch": 10.593635250917993,
      "grad_norm": 2.022265129630802,
      "learning_rate": 9.911580505531316e-07,
      "loss": 0.4807,
      "step": 8655
    },
    {
      "epoch": 10.594859241126072,
      "grad_norm": 1.231615140316117,
      "learning_rate": 9.906471041598786e-07,
      "loss": 0.5547,
      "step": 8656
    },
    {
      "epoch": 10.596083231334148,
      "grad_norm": 2.439320294952316,
      "learning_rate": 9.901362569557143e-07,
      "loss": 0.4984,
      "step": 8657
    },
    {
      "epoch": 10.597307221542227,
      "grad_norm": 1.4848563387223659,
      "learning_rate": 9.896255089742099e-07,
      "loss": 1.0723,
      "step": 8658
    },
    {
      "epoch": 10.598531211750306,
      "grad_norm": 1.5271439562341695,
      "learning_rate": 9.891148602489312e-07,
      "loss": 0.8646,
      "step": 8659
    },
    {
      "epoch": 10.599755201958384,
      "grad_norm": 1.9578350709093246,
      "learning_rate": 9.886043108134355e-07,
      "loss": 0.3934,
      "step": 8660
    },
    {
      "epoch": 10.600979192166463,
      "grad_norm": 1.8049429407361863,
      "learning_rate": 9.880938607012743e-07,
      "loss": 1.5604,
      "step": 8661
    },
    {
      "epoch": 10.602203182374542,
      "grad_norm": 1.366231074971778,
      "learning_rate": 9.875835099459923e-07,
      "loss": 0.3659,
      "step": 8662
    },
    {
      "epoch": 10.60342717258262,
      "grad_norm": 1.440098167803313,
      "learning_rate": 9.870732585811282e-07,
      "loss": 0.6654,
      "step": 8663
    },
    {
      "epoch": 10.604651162790697,
      "grad_norm": 1.9388807147106646,
      "learning_rate": 9.865631066402138e-07,
      "loss": 0.4815,
      "step": 8664
    },
    {
      "epoch": 10.605875152998776,
      "grad_norm": 2.903842649024455,
      "learning_rate": 9.860530541567739e-07,
      "loss": 0.7396,
      "step": 8665
    },
    {
      "epoch": 10.607099143206854,
      "grad_norm": 1.879653336823943,
      "learning_rate": 9.855431011643266e-07,
      "loss": 0.5056,
      "step": 8666
    },
    {
      "epoch": 10.608323133414933,
      "grad_norm": 1.35562382849663,
      "learning_rate": 9.850332476963861e-07,
      "loss": 0.5063,
      "step": 8667
    },
    {
      "epoch": 10.609547123623011,
      "grad_norm": 1.902253485951595,
      "learning_rate": 9.84523493786458e-07,
      "loss": 1.3884,
      "step": 8668
    },
    {
      "epoch": 10.61077111383109,
      "grad_norm": 2.3158541726232773,
      "learning_rate": 9.840138394680398e-07,
      "loss": 1.015,
      "step": 8669
    },
    {
      "epoch": 10.611995104039167,
      "grad_norm": 1.673576376310381,
      "learning_rate": 9.835042847746246e-07,
      "loss": 0.5081,
      "step": 8670
    },
    {
      "epoch": 10.613219094247246,
      "grad_norm": 2.3590375320053547,
      "learning_rate": 9.82994829739698e-07,
      "loss": 0.4699,
      "step": 8671
    },
    {
      "epoch": 10.614443084455324,
      "grad_norm": 1.2295717546046838,
      "learning_rate": 9.824854743967406e-07,
      "loss": 0.6836,
      "step": 8672
    },
    {
      "epoch": 10.615667074663403,
      "grad_norm": 3.2995308503302945,
      "learning_rate": 9.819762187792256e-07,
      "loss": 0.3792,
      "step": 8673
    },
    {
      "epoch": 10.616891064871481,
      "grad_norm": 1.1383697737551428,
      "learning_rate": 9.814670629206191e-07,
      "loss": 0.583,
      "step": 8674
    },
    {
      "epoch": 10.61811505507956,
      "grad_norm": 3.219568404590481,
      "learning_rate": 9.80958006854379e-07,
      "loss": 0.6838,
      "step": 8675
    },
    {
      "epoch": 10.619339045287639,
      "grad_norm": 1.6591554905641195,
      "learning_rate": 9.80449050613961e-07,
      "loss": 0.4724,
      "step": 8676
    },
    {
      "epoch": 10.620563035495715,
      "grad_norm": 1.9098929563326383,
      "learning_rate": 9.79940194232811e-07,
      "loss": 0.6928,
      "step": 8677
    },
    {
      "epoch": 10.621787025703794,
      "grad_norm": 2.2245099002831523,
      "learning_rate": 9.794314377443688e-07,
      "loss": 1.1699,
      "step": 8678
    },
    {
      "epoch": 10.623011015911873,
      "grad_norm": 1.3922823090475078,
      "learning_rate": 9.789227811820687e-07,
      "loss": 0.6065,
      "step": 8679
    },
    {
      "epoch": 10.624235006119951,
      "grad_norm": 1.9929958957257277,
      "learning_rate": 9.784142245793375e-07,
      "loss": 0.5309,
      "step": 8680
    },
    {
      "epoch": 10.62545899632803,
      "grad_norm": 1.74819941451009,
      "learning_rate": 9.77905767969595e-07,
      "loss": 0.4952,
      "step": 8681
    },
    {
      "epoch": 10.626682986536107,
      "grad_norm": 1.3671376813192477,
      "learning_rate": 9.77397411386256e-07,
      "loss": 0.6778,
      "step": 8682
    },
    {
      "epoch": 10.627906976744185,
      "grad_norm": 0.9317743979678326,
      "learning_rate": 9.768891548627272e-07,
      "loss": 0.5219,
      "step": 8683
    },
    {
      "epoch": 10.629130966952264,
      "grad_norm": 1.5656435340587507,
      "learning_rate": 9.763809984324086e-07,
      "loss": 0.535,
      "step": 8684
    },
    {
      "epoch": 10.630354957160343,
      "grad_norm": 2.965781577483978,
      "learning_rate": 9.758729421286973e-07,
      "loss": 0.4034,
      "step": 8685
    },
    {
      "epoch": 10.631578947368421,
      "grad_norm": 2.6088831859974113,
      "learning_rate": 9.753649859849775e-07,
      "loss": 1.178,
      "step": 8686
    },
    {
      "epoch": 10.6328029375765,
      "grad_norm": 1.1440959748705175,
      "learning_rate": 9.748571300346321e-07,
      "loss": 0.5074,
      "step": 8687
    },
    {
      "epoch": 10.634026927784578,
      "grad_norm": 1.3580094608487006,
      "learning_rate": 9.743493743110342e-07,
      "loss": 0.3884,
      "step": 8688
    },
    {
      "epoch": 10.635250917992655,
      "grad_norm": 1.7655060992618419,
      "learning_rate": 9.73841718847553e-07,
      "loss": 0.546,
      "step": 8689
    },
    {
      "epoch": 10.636474908200734,
      "grad_norm": 2.8686640765102758,
      "learning_rate": 9.733341636775493e-07,
      "loss": 0.4677,
      "step": 8690
    },
    {
      "epoch": 10.637698898408813,
      "grad_norm": 1.7772257892961434,
      "learning_rate": 9.728267088343774e-07,
      "loss": 0.565,
      "step": 8691
    },
    {
      "epoch": 10.638922888616891,
      "grad_norm": 0.9469959704166653,
      "learning_rate": 9.723193543513856e-07,
      "loss": 0.3448,
      "step": 8692
    },
    {
      "epoch": 10.64014687882497,
      "grad_norm": 1.2378826468750017,
      "learning_rate": 9.718121002619157e-07,
      "loss": 0.8448,
      "step": 8693
    },
    {
      "epoch": 10.641370869033048,
      "grad_norm": 2.6448299000974993,
      "learning_rate": 9.71304946599302e-07,
      "loss": 0.4008,
      "step": 8694
    },
    {
      "epoch": 10.642594859241125,
      "grad_norm": 2.318332669822744,
      "learning_rate": 9.707978933968728e-07,
      "loss": 1.0312,
      "step": 8695
    },
    {
      "epoch": 10.643818849449204,
      "grad_norm": 2.776212019999017,
      "learning_rate": 9.702909406879502e-07,
      "loss": 0.5121,
      "step": 8696
    },
    {
      "epoch": 10.645042839657282,
      "grad_norm": 1.8698647609886485,
      "learning_rate": 9.697840885058488e-07,
      "loss": 1.0071,
      "step": 8697
    },
    {
      "epoch": 10.646266829865361,
      "grad_norm": 1.2958359610298558,
      "learning_rate": 9.692773368838771e-07,
      "loss": 0.6018,
      "step": 8698
    },
    {
      "epoch": 10.64749082007344,
      "grad_norm": 2.6661637862725946,
      "learning_rate": 9.687706858553373e-07,
      "loss": 0.7899,
      "step": 8699
    },
    {
      "epoch": 10.648714810281518,
      "grad_norm": 1.5283666102224267,
      "learning_rate": 9.682641354535244e-07,
      "loss": 0.6181,
      "step": 8700
    },
    {
      "epoch": 10.649938800489597,
      "grad_norm": 1.5805063943634743,
      "learning_rate": 9.67757685711726e-07,
      "loss": 0.4633,
      "step": 8701
    },
    {
      "epoch": 10.651162790697674,
      "grad_norm": 1.5391531593192282,
      "learning_rate": 9.67251336663226e-07,
      "loss": 0.4031,
      "step": 8702
    },
    {
      "epoch": 10.652386780905752,
      "grad_norm": 1.612859561142017,
      "learning_rate": 9.667450883413e-07,
      "loss": 1.1694,
      "step": 8703
    },
    {
      "epoch": 10.653610771113831,
      "grad_norm": 1.3764455918373912,
      "learning_rate": 9.662389407792144e-07,
      "loss": 0.8637,
      "step": 8704
    },
    {
      "epoch": 10.65483476132191,
      "grad_norm": 1.452150281339315,
      "learning_rate": 9.65732894010232e-07,
      "loss": 0.8133,
      "step": 8705
    },
    {
      "epoch": 10.656058751529988,
      "grad_norm": 1.5110945313541537,
      "learning_rate": 9.6522694806761e-07,
      "loss": 1.1624,
      "step": 8706
    },
    {
      "epoch": 10.657282741738067,
      "grad_norm": 2.0113920849889753,
      "learning_rate": 9.64721102984596e-07,
      "loss": 0.458,
      "step": 8707
    },
    {
      "epoch": 10.658506731946144,
      "grad_norm": 2.4339823139710033,
      "learning_rate": 9.642153587944324e-07,
      "loss": 0.5764,
      "step": 8708
    },
    {
      "epoch": 10.659730722154222,
      "grad_norm": 1.7438866063595608,
      "learning_rate": 9.637097155303552e-07,
      "loss": 0.5396,
      "step": 8709
    },
    {
      "epoch": 10.660954712362301,
      "grad_norm": 1.6362915933231317,
      "learning_rate": 9.632041732255932e-07,
      "loss": 1.2278,
      "step": 8710
    },
    {
      "epoch": 10.66217870257038,
      "grad_norm": 1.888453850848573,
      "learning_rate": 9.626987319133684e-07,
      "loss": 1.105,
      "step": 8711
    },
    {
      "epoch": 10.663402692778458,
      "grad_norm": 1.3515672725581926,
      "learning_rate": 9.621933916268973e-07,
      "loss": 0.7372,
      "step": 8712
    },
    {
      "epoch": 10.664626682986537,
      "grad_norm": 1.3560359014460965,
      "learning_rate": 9.616881523993882e-07,
      "loss": 0.6713,
      "step": 8713
    },
    {
      "epoch": 10.665850673194614,
      "grad_norm": 1.5960417068658548,
      "learning_rate": 9.61183014264044e-07,
      "loss": 0.4691,
      "step": 8714
    },
    {
      "epoch": 10.667074663402692,
      "grad_norm": 1.370564965462065,
      "learning_rate": 9.606779772540603e-07,
      "loss": 0.5446,
      "step": 8715
    },
    {
      "epoch": 10.668298653610771,
      "grad_norm": 1.6866036025622224,
      "learning_rate": 9.601730414026264e-07,
      "loss": 0.6175,
      "step": 8716
    },
    {
      "epoch": 10.66952264381885,
      "grad_norm": 1.6359636711028158,
      "learning_rate": 9.59668206742925e-07,
      "loss": 1.6362,
      "step": 8717
    },
    {
      "epoch": 10.670746634026928,
      "grad_norm": 1.1278711875563878,
      "learning_rate": 9.591634733081307e-07,
      "loss": 0.5577,
      "step": 8718
    },
    {
      "epoch": 10.671970624235007,
      "grad_norm": 1.6534614343301133,
      "learning_rate": 9.586588411314145e-07,
      "loss": 0.489,
      "step": 8719
    },
    {
      "epoch": 10.673194614443084,
      "grad_norm": 3.172574123442707,
      "learning_rate": 9.581543102459386e-07,
      "loss": 0.4094,
      "step": 8720
    },
    {
      "epoch": 10.674418604651162,
      "grad_norm": 2.3492284913908716,
      "learning_rate": 9.576498806848592e-07,
      "loss": 0.6058,
      "step": 8721
    },
    {
      "epoch": 10.67564259485924,
      "grad_norm": 2.54798326735156,
      "learning_rate": 9.571455524813241e-07,
      "loss": 0.8169,
      "step": 8722
    },
    {
      "epoch": 10.67686658506732,
      "grad_norm": 2.9656438906393596,
      "learning_rate": 9.566413256684756e-07,
      "loss": 1.0203,
      "step": 8723
    },
    {
      "epoch": 10.678090575275398,
      "grad_norm": 1.6969211698690392,
      "learning_rate": 9.56137200279452e-07,
      "loss": 0.304,
      "step": 8724
    },
    {
      "epoch": 10.679314565483477,
      "grad_norm": 2.0417853280984164,
      "learning_rate": 9.556331763473808e-07,
      "loss": 1.0336,
      "step": 8725
    },
    {
      "epoch": 10.680538555691555,
      "grad_norm": 1.5624515352290547,
      "learning_rate": 9.551292539053851e-07,
      "loss": 0.6112,
      "step": 8726
    },
    {
      "epoch": 10.681762545899632,
      "grad_norm": 2.223770373062648,
      "learning_rate": 9.546254329865808e-07,
      "loss": 0.4748,
      "step": 8727
    },
    {
      "epoch": 10.68298653610771,
      "grad_norm": 2.184711110756065,
      "learning_rate": 9.541217136240773e-07,
      "loss": 0.6746,
      "step": 8728
    },
    {
      "epoch": 10.68421052631579,
      "grad_norm": 1.5391959924246308,
      "learning_rate": 9.536180958509769e-07,
      "loss": 2.2782,
      "step": 8729
    },
    {
      "epoch": 10.685434516523868,
      "grad_norm": 1.224400020218761,
      "learning_rate": 9.531145797003754e-07,
      "loss": 0.8429,
      "step": 8730
    },
    {
      "epoch": 10.686658506731947,
      "grad_norm": 2.179263562608608,
      "learning_rate": 9.526111652053613e-07,
      "loss": 0.3948,
      "step": 8731
    },
    {
      "epoch": 10.687882496940025,
      "grad_norm": 1.6256546591022707,
      "learning_rate": 9.521078523990199e-07,
      "loss": 1.0133,
      "step": 8732
    },
    {
      "epoch": 10.689106487148102,
      "grad_norm": 1.3739845897566165,
      "learning_rate": 9.516046413144242e-07,
      "loss": 0.5003,
      "step": 8733
    },
    {
      "epoch": 10.69033047735618,
      "grad_norm": 1.6591511978993947,
      "learning_rate": 9.511015319846445e-07,
      "loss": 0.4535,
      "step": 8734
    },
    {
      "epoch": 10.69155446756426,
      "grad_norm": 1.0952900532277625,
      "learning_rate": 9.505985244427432e-07,
      "loss": 0.5631,
      "step": 8735
    },
    {
      "epoch": 10.692778457772338,
      "grad_norm": 1.1330731040306519,
      "learning_rate": 9.50095618721775e-07,
      "loss": 0.8472,
      "step": 8736
    },
    {
      "epoch": 10.694002447980417,
      "grad_norm": 2.467991523325283,
      "learning_rate": 9.495928148547909e-07,
      "loss": 0.3958,
      "step": 8737
    },
    {
      "epoch": 10.695226438188495,
      "grad_norm": 3.135349117273822,
      "learning_rate": 9.490901128748325e-07,
      "loss": 0.3747,
      "step": 8738
    },
    {
      "epoch": 10.696450428396572,
      "grad_norm": 1.5127759286309204,
      "learning_rate": 9.485875128149364e-07,
      "loss": 0.6041,
      "step": 8739
    },
    {
      "epoch": 10.69767441860465,
      "grad_norm": 1.2177962501401907,
      "learning_rate": 9.480850147081289e-07,
      "loss": 0.66,
      "step": 8740
    },
    {
      "epoch": 10.69889840881273,
      "grad_norm": 1.8812683319909893,
      "learning_rate": 9.475826185874348e-07,
      "loss": 1.0097,
      "step": 8741
    },
    {
      "epoch": 10.700122399020808,
      "grad_norm": 1.6790218184462655,
      "learning_rate": 9.470803244858692e-07,
      "loss": 0.3713,
      "step": 8742
    },
    {
      "epoch": 10.701346389228886,
      "grad_norm": 2.038076377758818,
      "learning_rate": 9.465781324364409e-07,
      "loss": 0.6149,
      "step": 8743
    },
    {
      "epoch": 10.702570379436965,
      "grad_norm": 1.9828769609315426,
      "learning_rate": 9.460760424721519e-07,
      "loss": 0.727,
      "step": 8744
    },
    {
      "epoch": 10.703794369645042,
      "grad_norm": 1.6657334342773669,
      "learning_rate": 9.455740546259978e-07,
      "loss": 0.552,
      "step": 8745
    },
    {
      "epoch": 10.70501835985312,
      "grad_norm": 1.1585240425362378,
      "learning_rate": 9.450721689309675e-07,
      "loss": 0.6193,
      "step": 8746
    },
    {
      "epoch": 10.7062423500612,
      "grad_norm": 2.5833912737448155,
      "learning_rate": 9.445703854200428e-07,
      "loss": 0.4306,
      "step": 8747
    },
    {
      "epoch": 10.707466340269278,
      "grad_norm": 1.595504238547596,
      "learning_rate": 9.440687041261992e-07,
      "loss": 0.6001,
      "step": 8748
    },
    {
      "epoch": 10.708690330477356,
      "grad_norm": 1.1162940326982573,
      "learning_rate": 9.435671250824047e-07,
      "loss": 0.6395,
      "step": 8749
    },
    {
      "epoch": 10.709914320685435,
      "grad_norm": 0.9404529085916344,
      "learning_rate": 9.430656483216235e-07,
      "loss": 0.4394,
      "step": 8750
    },
    {
      "epoch": 10.711138310893514,
      "grad_norm": 2.6530022625131493,
      "learning_rate": 9.425642738768085e-07,
      "loss": 0.463,
      "step": 8751
    },
    {
      "epoch": 10.71236230110159,
      "grad_norm": 2.076440950055686,
      "learning_rate": 9.420630017809088e-07,
      "loss": 0.3832,
      "step": 8752
    },
    {
      "epoch": 10.713586291309669,
      "grad_norm": 1.4711210512883865,
      "learning_rate": 9.415618320668654e-07,
      "loss": 0.6273,
      "step": 8753
    },
    {
      "epoch": 10.714810281517748,
      "grad_norm": 2.9160325414065373,
      "learning_rate": 9.410607647676148e-07,
      "loss": 0.9598,
      "step": 8754
    },
    {
      "epoch": 10.716034271725826,
      "grad_norm": 2.1946812462566205,
      "learning_rate": 9.405597999160848e-07,
      "loss": 0.5048,
      "step": 8755
    },
    {
      "epoch": 10.717258261933905,
      "grad_norm": 2.3031194982500556,
      "learning_rate": 9.400589375451966e-07,
      "loss": 0.349,
      "step": 8756
    },
    {
      "epoch": 10.718482252141984,
      "grad_norm": 2.299966434251559,
      "learning_rate": 9.395581776878654e-07,
      "loss": 0.3265,
      "step": 8757
    },
    {
      "epoch": 10.71970624235006,
      "grad_norm": 1.8615364536448726,
      "learning_rate": 9.390575203769989e-07,
      "loss": 0.3894,
      "step": 8758
    },
    {
      "epoch": 10.720930232558139,
      "grad_norm": 1.1178188590796985,
      "learning_rate": 9.385569656454988e-07,
      "loss": 0.6617,
      "step": 8759
    },
    {
      "epoch": 10.722154222766218,
      "grad_norm": 3.5308817712691423,
      "learning_rate": 9.380565135262595e-07,
      "loss": 0.4249,
      "step": 8760
    },
    {
      "epoch": 10.723378212974296,
      "grad_norm": 1.5098062180095784,
      "learning_rate": 9.375561640521688e-07,
      "loss": 0.8715,
      "step": 8761
    },
    {
      "epoch": 10.724602203182375,
      "grad_norm": 1.090429577935605,
      "learning_rate": 9.370559172561081e-07,
      "loss": 0.5364,
      "step": 8762
    },
    {
      "epoch": 10.725826193390454,
      "grad_norm": 1.882517841787687,
      "learning_rate": 9.365557731709515e-07,
      "loss": 1.2128,
      "step": 8763
    },
    {
      "epoch": 10.727050183598532,
      "grad_norm": 2.415780578243912,
      "learning_rate": 9.360557318295666e-07,
      "loss": 0.4227,
      "step": 8764
    },
    {
      "epoch": 10.728274173806609,
      "grad_norm": 1.4104449616197556,
      "learning_rate": 9.355557932648143e-07,
      "loss": 1.1487,
      "step": 8765
    },
    {
      "epoch": 10.729498164014688,
      "grad_norm": 2.0027611719237717,
      "learning_rate": 9.350559575095483e-07,
      "loss": 0.7228,
      "step": 8766
    },
    {
      "epoch": 10.730722154222766,
      "grad_norm": 2.2556242479587594,
      "learning_rate": 9.345562245966169e-07,
      "loss": 0.4852,
      "step": 8767
    },
    {
      "epoch": 10.731946144430845,
      "grad_norm": 1.6321913971734976,
      "learning_rate": 9.340565945588614e-07,
      "loss": 0.4883,
      "step": 8768
    },
    {
      "epoch": 10.733170134638923,
      "grad_norm": 1.4217494546735152,
      "learning_rate": 9.335570674291136e-07,
      "loss": 0.4334,
      "step": 8769
    },
    {
      "epoch": 10.734394124847,
      "grad_norm": 1.933582147510335,
      "learning_rate": 9.330576432402014e-07,
      "loss": 0.4601,
      "step": 8770
    },
    {
      "epoch": 10.735618115055079,
      "grad_norm": 3.2266533748973805,
      "learning_rate": 9.325583220249443e-07,
      "loss": 0.4592,
      "step": 8771
    },
    {
      "epoch": 10.736842105263158,
      "grad_norm": 2.0951110288330064,
      "learning_rate": 9.320591038161575e-07,
      "loss": 1.119,
      "step": 8772
    },
    {
      "epoch": 10.738066095471236,
      "grad_norm": 1.7199314392050755,
      "learning_rate": 9.31559988646647e-07,
      "loss": 1.3985,
      "step": 8773
    },
    {
      "epoch": 10.739290085679315,
      "grad_norm": 1.470400044241514,
      "learning_rate": 9.310609765492137e-07,
      "loss": 0.5014,
      "step": 8774
    },
    {
      "epoch": 10.740514075887393,
      "grad_norm": 2.123458907503255,
      "learning_rate": 9.305620675566482e-07,
      "loss": 1.1858,
      "step": 8775
    },
    {
      "epoch": 10.741738066095472,
      "grad_norm": 2.572402597969765,
      "learning_rate": 9.300632617017396e-07,
      "loss": 0.5211,
      "step": 8776
    },
    {
      "epoch": 10.742962056303549,
      "grad_norm": 2.291279385010221,
      "learning_rate": 9.295645590172664e-07,
      "loss": 0.5232,
      "step": 8777
    },
    {
      "epoch": 10.744186046511627,
      "grad_norm": 1.9766594689003147,
      "learning_rate": 9.290659595360019e-07,
      "loss": 0.359,
      "step": 8778
    },
    {
      "epoch": 10.745410036719706,
      "grad_norm": 1.0969312860822165,
      "learning_rate": 9.285674632907121e-07,
      "loss": 0.6795,
      "step": 8779
    },
    {
      "epoch": 10.746634026927785,
      "grad_norm": 3.041037146917913,
      "learning_rate": 9.280690703141562e-07,
      "loss": 0.574,
      "step": 8780
    },
    {
      "epoch": 10.747858017135863,
      "grad_norm": 2.6418426451660504,
      "learning_rate": 9.275707806390871e-07,
      "loss": 0.7262,
      "step": 8781
    },
    {
      "epoch": 10.749082007343942,
      "grad_norm": 1.3024918877985507,
      "learning_rate": 9.2707259429825e-07,
      "loss": 0.9748,
      "step": 8782
    },
    {
      "epoch": 10.750305997552019,
      "grad_norm": 1.518329070600029,
      "learning_rate": 9.26574511324384e-07,
      "loss": 0.9156,
      "step": 8783
    },
    {
      "epoch": 10.751529987760097,
      "grad_norm": 1.1663899523561658,
      "learning_rate": 9.26076531750221e-07,
      "loss": 0.6702,
      "step": 8784
    },
    {
      "epoch": 10.752753977968176,
      "grad_norm": 1.3815846376700105,
      "learning_rate": 9.255786556084872e-07,
      "loss": 0.4513,
      "step": 8785
    },
    {
      "epoch": 10.753977968176255,
      "grad_norm": 3.7988457668781908,
      "learning_rate": 9.250808829319019e-07,
      "loss": 0.3555,
      "step": 8786
    },
    {
      "epoch": 10.755201958384333,
      "grad_norm": 1.8765315770441213,
      "learning_rate": 9.245832137531749e-07,
      "loss": 0.524,
      "step": 8787
    },
    {
      "epoch": 10.756425948592412,
      "grad_norm": 1.1778036011934452,
      "learning_rate": 9.240856481050111e-07,
      "loss": 0.5702,
      "step": 8788
    },
    {
      "epoch": 10.75764993880049,
      "grad_norm": 1.2883975459490684,
      "learning_rate": 9.235881860201104e-07,
      "loss": 0.5189,
      "step": 8789
    },
    {
      "epoch": 10.758873929008567,
      "grad_norm": 2.406356224106745,
      "learning_rate": 9.230908275311636e-07,
      "loss": 0.5474,
      "step": 8790
    },
    {
      "epoch": 10.760097919216646,
      "grad_norm": 1.4574476652097625,
      "learning_rate": 9.225935726708549e-07,
      "loss": 0.544,
      "step": 8791
    },
    {
      "epoch": 10.761321909424725,
      "grad_norm": 1.9744534664717306,
      "learning_rate": 9.220964214718622e-07,
      "loss": 0.3978,
      "step": 8792
    },
    {
      "epoch": 10.762545899632803,
      "grad_norm": 2.9176588876532046,
      "learning_rate": 9.215993739668563e-07,
      "loss": 0.3957,
      "step": 8793
    },
    {
      "epoch": 10.763769889840882,
      "grad_norm": 2.6633189263727477,
      "learning_rate": 9.211024301885016e-07,
      "loss": 0.6574,
      "step": 8794
    },
    {
      "epoch": 10.76499388004896,
      "grad_norm": 2.146300720611982,
      "learning_rate": 9.206055901694553e-07,
      "loss": 1.1045,
      "step": 8795
    },
    {
      "epoch": 10.766217870257037,
      "grad_norm": 1.581904086848936,
      "learning_rate": 9.201088539423675e-07,
      "loss": 0.5691,
      "step": 8796
    },
    {
      "epoch": 10.767441860465116,
      "grad_norm": 1.3893280587446335,
      "learning_rate": 9.196122215398825e-07,
      "loss": 0.8733,
      "step": 8797
    },
    {
      "epoch": 10.768665850673194,
      "grad_norm": 2.167982067488335,
      "learning_rate": 9.191156929946368e-07,
      "loss": 0.8206,
      "step": 8798
    },
    {
      "epoch": 10.769889840881273,
      "grad_norm": 3.277981078219065,
      "learning_rate": 9.186192683392603e-07,
      "loss": 0.5216,
      "step": 8799
    },
    {
      "epoch": 10.771113831089352,
      "grad_norm": 2.014471068521743,
      "learning_rate": 9.181229476063766e-07,
      "loss": 0.7658,
      "step": 8800
    },
    {
      "epoch": 10.77233782129743,
      "grad_norm": 2.5278238818525827,
      "learning_rate": 9.176267308286008e-07,
      "loss": 0.5576,
      "step": 8801
    },
    {
      "epoch": 10.773561811505507,
      "grad_norm": 1.43693889345621,
      "learning_rate": 9.171306180385445e-07,
      "loss": 0.7228,
      "step": 8802
    },
    {
      "epoch": 10.774785801713586,
      "grad_norm": 2.9054407027925486,
      "learning_rate": 9.166346092688099e-07,
      "loss": 0.6864,
      "step": 8803
    },
    {
      "epoch": 10.776009791921664,
      "grad_norm": 1.1486155007271546,
      "learning_rate": 9.161387045519918e-07,
      "loss": 0.6146,
      "step": 8804
    },
    {
      "epoch": 10.777233782129743,
      "grad_norm": 2.0138202757613364,
      "learning_rate": 9.156429039206785e-07,
      "loss": 0.5953,
      "step": 8805
    },
    {
      "epoch": 10.778457772337822,
      "grad_norm": 1.9856028274706146,
      "learning_rate": 9.151472074074547e-07,
      "loss": 1.2901,
      "step": 8806
    },
    {
      "epoch": 10.7796817625459,
      "grad_norm": 2.017828033716331,
      "learning_rate": 9.146516150448942e-07,
      "loss": 0.3913,
      "step": 8807
    },
    {
      "epoch": 10.780905752753977,
      "grad_norm": 1.0026918813187888,
      "learning_rate": 9.141561268655658e-07,
      "loss": 0.5101,
      "step": 8808
    },
    {
      "epoch": 10.782129742962056,
      "grad_norm": 1.8570817100163473,
      "learning_rate": 9.136607429020311e-07,
      "loss": 0.708,
      "step": 8809
    },
    {
      "epoch": 10.783353733170134,
      "grad_norm": 1.468635623858812,
      "learning_rate": 9.131654631868451e-07,
      "loss": 0.5578,
      "step": 8810
    },
    {
      "epoch": 10.784577723378213,
      "grad_norm": 1.750419702061781,
      "learning_rate": 9.126702877525554e-07,
      "loss": 0.6298,
      "step": 8811
    },
    {
      "epoch": 10.785801713586292,
      "grad_norm": 3.095617502427146,
      "learning_rate": 9.121752166317033e-07,
      "loss": 0.3049,
      "step": 8812
    },
    {
      "epoch": 10.78702570379437,
      "grad_norm": 2.48443147336221,
      "learning_rate": 9.116802498568231e-07,
      "loss": 1.0162,
      "step": 8813
    },
    {
      "epoch": 10.788249694002449,
      "grad_norm": 3.3719305723288735,
      "learning_rate": 9.11185387460441e-07,
      "loss": 0.354,
      "step": 8814
    },
    {
      "epoch": 10.789473684210526,
      "grad_norm": 1.4615253356058637,
      "learning_rate": 9.106906294750806e-07,
      "loss": 1.1294,
      "step": 8815
    },
    {
      "epoch": 10.790697674418604,
      "grad_norm": 2.0793834375042377,
      "learning_rate": 9.101959759332524e-07,
      "loss": 0.6516,
      "step": 8816
    },
    {
      "epoch": 10.791921664626683,
      "grad_norm": 1.6019213864908666,
      "learning_rate": 9.097014268674648e-07,
      "loss": 0.3562,
      "step": 8817
    },
    {
      "epoch": 10.793145654834762,
      "grad_norm": 2.1110277568050724,
      "learning_rate": 9.092069823102162e-07,
      "loss": 1.1941,
      "step": 8818
    },
    {
      "epoch": 10.79436964504284,
      "grad_norm": 2.4206268250881147,
      "learning_rate": 9.087126422940015e-07,
      "loss": 0.6073,
      "step": 8819
    },
    {
      "epoch": 10.795593635250919,
      "grad_norm": 2.0491564837202185,
      "learning_rate": 9.082184068513064e-07,
      "loss": 0.9624,
      "step": 8820
    },
    {
      "epoch": 10.796817625458996,
      "grad_norm": 2.73779763532413,
      "learning_rate": 9.077242760146108e-07,
      "loss": 0.439,
      "step": 8821
    },
    {
      "epoch": 10.798041615667074,
      "grad_norm": 1.28901050598713,
      "learning_rate": 9.072302498163857e-07,
      "loss": 0.7722,
      "step": 8822
    },
    {
      "epoch": 10.799265605875153,
      "grad_norm": 0.9961106599172523,
      "learning_rate": 9.067363282890962e-07,
      "loss": 0.4404,
      "step": 8823
    },
    {
      "epoch": 10.800489596083231,
      "grad_norm": 1.992592677874508,
      "learning_rate": 9.062425114652033e-07,
      "loss": 0.3881,
      "step": 8824
    },
    {
      "epoch": 10.80171358629131,
      "grad_norm": 1.3886421224080288,
      "learning_rate": 9.057487993771571e-07,
      "loss": 0.6446,
      "step": 8825
    },
    {
      "epoch": 10.802937576499389,
      "grad_norm": 1.9573628949458135,
      "learning_rate": 9.052551920574034e-07,
      "loss": 0.4887,
      "step": 8826
    },
    {
      "epoch": 10.804161566707466,
      "grad_norm": 1.6141869935493223,
      "learning_rate": 9.047616895383801e-07,
      "loss": 0.4652,
      "step": 8827
    },
    {
      "epoch": 10.805385556915544,
      "grad_norm": 2.0748215029726778,
      "learning_rate": 9.04268291852518e-07,
      "loss": 0.4078,
      "step": 8828
    },
    {
      "epoch": 10.806609547123623,
      "grad_norm": 1.1916634668897612,
      "learning_rate": 9.037749990322416e-07,
      "loss": 0.3834,
      "step": 8829
    },
    {
      "epoch": 10.807833537331701,
      "grad_norm": 1.5741422237682339,
      "learning_rate": 9.032818111099681e-07,
      "loss": 0.5224,
      "step": 8830
    },
    {
      "epoch": 10.80905752753978,
      "grad_norm": 1.0967594781715873,
      "learning_rate": 9.027887281181075e-07,
      "loss": 0.3619,
      "step": 8831
    },
    {
      "epoch": 10.810281517747859,
      "grad_norm": 1.7010932689977778,
      "learning_rate": 9.022957500890656e-07,
      "loss": 0.9959,
      "step": 8832
    },
    {
      "epoch": 10.811505507955935,
      "grad_norm": 2.1447784234308123,
      "learning_rate": 9.018028770552368e-07,
      "loss": 0.8109,
      "step": 8833
    },
    {
      "epoch": 10.812729498164014,
      "grad_norm": 1.695069812187889,
      "learning_rate": 9.013101090490114e-07,
      "loss": 1.5546,
      "step": 8834
    },
    {
      "epoch": 10.813953488372093,
      "grad_norm": 1.3802544845683025,
      "learning_rate": 9.008174461027724e-07,
      "loss": 0.4101,
      "step": 8835
    },
    {
      "epoch": 10.815177478580171,
      "grad_norm": 1.3397612469354259,
      "learning_rate": 9.003248882488952e-07,
      "loss": 0.7652,
      "step": 8836
    },
    {
      "epoch": 10.81640146878825,
      "grad_norm": 2.374832251367542,
      "learning_rate": 8.998324355197505e-07,
      "loss": 0.4148,
      "step": 8837
    },
    {
      "epoch": 10.817625458996329,
      "grad_norm": 1.9212712624656891,
      "learning_rate": 8.993400879476994e-07,
      "loss": 0.7362,
      "step": 8838
    },
    {
      "epoch": 10.818849449204407,
      "grad_norm": 1.8151163957596244,
      "learning_rate": 8.988478455650981e-07,
      "loss": 0.5339,
      "step": 8839
    },
    {
      "epoch": 10.820073439412484,
      "grad_norm": 1.7236462859569641,
      "learning_rate": 8.983557084042923e-07,
      "loss": 0.6586,
      "step": 8840
    },
    {
      "epoch": 10.821297429620563,
      "grad_norm": 1.8186051428188494,
      "learning_rate": 8.978636764976264e-07,
      "loss": 0.6106,
      "step": 8841
    },
    {
      "epoch": 10.822521419828641,
      "grad_norm": 2.6130220010581917,
      "learning_rate": 8.973717498774337e-07,
      "loss": 0.4753,
      "step": 8842
    },
    {
      "epoch": 10.82374541003672,
      "grad_norm": 1.1784443711672994,
      "learning_rate": 8.968799285760419e-07,
      "loss": 0.4435,
      "step": 8843
    },
    {
      "epoch": 10.824969400244798,
      "grad_norm": 1.480096826780721,
      "learning_rate": 8.963882126257717e-07,
      "loss": 0.6461,
      "step": 8844
    },
    {
      "epoch": 10.826193390452877,
      "grad_norm": 1.3355849944382563,
      "learning_rate": 8.958966020589369e-07,
      "loss": 0.5526,
      "step": 8845
    },
    {
      "epoch": 10.827417380660954,
      "grad_norm": 2.9573673857930625,
      "learning_rate": 8.95405096907844e-07,
      "loss": 0.3402,
      "step": 8846
    },
    {
      "epoch": 10.828641370869033,
      "grad_norm": 1.2175949492004883,
      "learning_rate": 8.949136972047931e-07,
      "loss": 0.4725,
      "step": 8847
    },
    {
      "epoch": 10.829865361077111,
      "grad_norm": 1.9324568925558754,
      "learning_rate": 8.944224029820775e-07,
      "loss": 0.5312,
      "step": 8848
    },
    {
      "epoch": 10.83108935128519,
      "grad_norm": 1.5117286796643794,
      "learning_rate": 8.939312142719819e-07,
      "loss": 1.4009,
      "step": 8849
    },
    {
      "epoch": 10.832313341493268,
      "grad_norm": 1.724660881008733,
      "learning_rate": 8.934401311067884e-07,
      "loss": 1.3589,
      "step": 8850
    },
    {
      "epoch": 10.833537331701347,
      "grad_norm": 2.855986319733748,
      "learning_rate": 8.929491535187659e-07,
      "loss": 0.3571,
      "step": 8851
    },
    {
      "epoch": 10.834761321909426,
      "grad_norm": 1.5480842598278306,
      "learning_rate": 8.924582815401814e-07,
      "loss": 0.9117,
      "step": 8852
    },
    {
      "epoch": 10.835985312117502,
      "grad_norm": 1.9782238554740605,
      "learning_rate": 8.919675152032917e-07,
      "loss": 0.3648,
      "step": 8853
    },
    {
      "epoch": 10.837209302325581,
      "grad_norm": 1.9931295510586355,
      "learning_rate": 8.914768545403501e-07,
      "loss": 0.906,
      "step": 8854
    },
    {
      "epoch": 10.83843329253366,
      "grad_norm": 1.5563984858716138,
      "learning_rate": 8.909862995836002e-07,
      "loss": 0.5227,
      "step": 8855
    },
    {
      "epoch": 10.839657282741738,
      "grad_norm": 1.4763442324442477,
      "learning_rate": 8.904958503652794e-07,
      "loss": 1.1161,
      "step": 8856
    },
    {
      "epoch": 10.840881272949817,
      "grad_norm": 2.5731467557224432,
      "learning_rate": 8.90005506917618e-07,
      "loss": 0.3308,
      "step": 8857
    },
    {
      "epoch": 10.842105263157894,
      "grad_norm": 2.1096327329981457,
      "learning_rate": 8.895152692728398e-07,
      "loss": 0.4174,
      "step": 8858
    },
    {
      "epoch": 10.843329253365972,
      "grad_norm": 1.6968615663372175,
      "learning_rate": 8.890251374631614e-07,
      "loss": 0.4296,
      "step": 8859
    },
    {
      "epoch": 10.844553243574051,
      "grad_norm": 1.6031819088825963,
      "learning_rate": 8.885351115207921e-07,
      "loss": 1.0664,
      "step": 8860
    },
    {
      "epoch": 10.84577723378213,
      "grad_norm": 1.9787644028906364,
      "learning_rate": 8.880451914779351e-07,
      "loss": 1.0485,
      "step": 8861
    },
    {
      "epoch": 10.847001223990208,
      "grad_norm": 1.1386704710765776,
      "learning_rate": 8.875553773667861e-07,
      "loss": 0.6304,
      "step": 8862
    },
    {
      "epoch": 10.848225214198287,
      "grad_norm": 2.4623961254858315,
      "learning_rate": 8.870656692195334e-07,
      "loss": 0.4867,
      "step": 8863
    },
    {
      "epoch": 10.849449204406366,
      "grad_norm": 2.033289342438134,
      "learning_rate": 8.865760670683591e-07,
      "loss": 0.502,
      "step": 8864
    },
    {
      "epoch": 10.850673194614442,
      "grad_norm": 1.503310129177509,
      "learning_rate": 8.86086570945438e-07,
      "loss": 0.3597,
      "step": 8865
    },
    {
      "epoch": 10.851897184822521,
      "grad_norm": 1.804326272697163,
      "learning_rate": 8.855971808829373e-07,
      "loss": 0.9527,
      "step": 8866
    },
    {
      "epoch": 10.8531211750306,
      "grad_norm": 1.8934647556548647,
      "learning_rate": 8.851078969130194e-07,
      "loss": 0.4726,
      "step": 8867
    },
    {
      "epoch": 10.854345165238678,
      "grad_norm": 1.2660418171643988,
      "learning_rate": 8.846187190678382e-07,
      "loss": 0.5716,
      "step": 8868
    },
    {
      "epoch": 10.855569155446757,
      "grad_norm": 2.128250943968023,
      "learning_rate": 8.841296473795391e-07,
      "loss": 1.204,
      "step": 8869
    },
    {
      "epoch": 10.856793145654835,
      "grad_norm": 2.6262913407430775,
      "learning_rate": 8.836406818802623e-07,
      "loss": 0.5077,
      "step": 8870
    },
    {
      "epoch": 10.858017135862912,
      "grad_norm": 1.3742961016978275,
      "learning_rate": 8.831518226021419e-07,
      "loss": 0.7373,
      "step": 8871
    },
    {
      "epoch": 10.859241126070991,
      "grad_norm": 3.0228403149208636,
      "learning_rate": 8.826630695773037e-07,
      "loss": 0.5288,
      "step": 8872
    },
    {
      "epoch": 10.86046511627907,
      "grad_norm": 2.425391869869488,
      "learning_rate": 8.821744228378662e-07,
      "loss": 0.5433,
      "step": 8873
    },
    {
      "epoch": 10.861689106487148,
      "grad_norm": 1.1007681533374445,
      "learning_rate": 8.816858824159424e-07,
      "loss": 0.5467,
      "step": 8874
    },
    {
      "epoch": 10.862913096695227,
      "grad_norm": 2.30995393899878,
      "learning_rate": 8.811974483436353e-07,
      "loss": 0.5483,
      "step": 8875
    },
    {
      "epoch": 10.864137086903305,
      "grad_norm": 2.4173971476024754,
      "learning_rate": 8.807091206530452e-07,
      "loss": 0.4602,
      "step": 8876
    },
    {
      "epoch": 10.865361077111384,
      "grad_norm": 1.2289146632248107,
      "learning_rate": 8.802208993762624e-07,
      "loss": 0.5375,
      "step": 8877
    },
    {
      "epoch": 10.86658506731946,
      "grad_norm": 2.61760851738504,
      "learning_rate": 8.797327845453707e-07,
      "loss": 0.9302,
      "step": 8878
    },
    {
      "epoch": 10.86780905752754,
      "grad_norm": 2.2179158246506674,
      "learning_rate": 8.792447761924475e-07,
      "loss": 0.9524,
      "step": 8879
    },
    {
      "epoch": 10.869033047735618,
      "grad_norm": 2.1980296200282776,
      "learning_rate": 8.787568743495628e-07,
      "loss": 0.3754,
      "step": 8880
    },
    {
      "epoch": 10.870257037943697,
      "grad_norm": 1.6277589894324616,
      "learning_rate": 8.7826907904878e-07,
      "loss": 0.5937,
      "step": 8881
    },
    {
      "epoch": 10.871481028151775,
      "grad_norm": 2.2051376806958802,
      "learning_rate": 8.777813903221549e-07,
      "loss": 0.5712,
      "step": 8882
    },
    {
      "epoch": 10.872705018359854,
      "grad_norm": 1.476588258781014,
      "learning_rate": 8.772938082017358e-07,
      "loss": 0.3546,
      "step": 8883
    },
    {
      "epoch": 10.87392900856793,
      "grad_norm": 3.4460999950284417,
      "learning_rate": 8.768063327195664e-07,
      "loss": 0.3369,
      "step": 8884
    },
    {
      "epoch": 10.87515299877601,
      "grad_norm": 1.4378715659017955,
      "learning_rate": 8.763189639076811e-07,
      "loss": 1.311,
      "step": 8885
    },
    {
      "epoch": 10.876376988984088,
      "grad_norm": 1.3537408903904238,
      "learning_rate": 8.758317017981086e-07,
      "loss": 0.6273,
      "step": 8886
    },
    {
      "epoch": 10.877600979192167,
      "grad_norm": 1.8066542490506188,
      "learning_rate": 8.753445464228685e-07,
      "loss": 0.4667,
      "step": 8887
    },
    {
      "epoch": 10.878824969400245,
      "grad_norm": 1.5869184308356272,
      "learning_rate": 8.748574978139748e-07,
      "loss": 1.571,
      "step": 8888
    },
    {
      "epoch": 10.880048959608324,
      "grad_norm": 2.1683892162459593,
      "learning_rate": 8.743705560034363e-07,
      "loss": 0.5993,
      "step": 8889
    },
    {
      "epoch": 10.8812729498164,
      "grad_norm": 2.0160370797276594,
      "learning_rate": 8.73883721023252e-07,
      "loss": 0.6309,
      "step": 8890
    },
    {
      "epoch": 10.88249694002448,
      "grad_norm": 2.0128745696234103,
      "learning_rate": 8.733969929054151e-07,
      "loss": 0.9254,
      "step": 8891
    },
    {
      "epoch": 10.883720930232558,
      "grad_norm": 1.4845349374718784,
      "learning_rate": 8.729103716819113e-07,
      "loss": 1.3185,
      "step": 8892
    },
    {
      "epoch": 10.884944920440637,
      "grad_norm": 1.2901472525128017,
      "learning_rate": 8.724238573847197e-07,
      "loss": 0.4491,
      "step": 8893
    },
    {
      "epoch": 10.886168910648715,
      "grad_norm": 1.5625853692253024,
      "learning_rate": 8.71937450045812e-07,
      "loss": 0.4264,
      "step": 8894
    },
    {
      "epoch": 10.887392900856794,
      "grad_norm": 1.6127149977192443,
      "learning_rate": 8.714511496971536e-07,
      "loss": 1.2939,
      "step": 8895
    },
    {
      "epoch": 10.88861689106487,
      "grad_norm": 1.1796151376489348,
      "learning_rate": 8.709649563707009e-07,
      "loss": 0.4761,
      "step": 8896
    },
    {
      "epoch": 10.88984088127295,
      "grad_norm": 1.3788548707825221,
      "learning_rate": 8.704788700984079e-07,
      "loss": 0.6402,
      "step": 8897
    },
    {
      "epoch": 10.891064871481028,
      "grad_norm": 1.4475268137371076,
      "learning_rate": 8.699928909122155e-07,
      "loss": 0.9475,
      "step": 8898
    },
    {
      "epoch": 10.892288861689106,
      "grad_norm": 2.799403731285428,
      "learning_rate": 8.695070188440613e-07,
      "loss": 0.5801,
      "step": 8899
    },
    {
      "epoch": 10.893512851897185,
      "grad_norm": 1.3275860635447232,
      "learning_rate": 8.69021253925875e-07,
      "loss": 0.5086,
      "step": 8900
    },
    {
      "epoch": 10.894736842105264,
      "grad_norm": 1.1760502646054831,
      "learning_rate": 8.685355961895783e-07,
      "loss": 0.5095,
      "step": 8901
    },
    {
      "epoch": 10.895960832313342,
      "grad_norm": 1.073271358261181,
      "learning_rate": 8.68050045667089e-07,
      "loss": 0.512,
      "step": 8902
    },
    {
      "epoch": 10.89718482252142,
      "grad_norm": 1.7787931892178646,
      "learning_rate": 8.67564602390315e-07,
      "loss": 0.6877,
      "step": 8903
    },
    {
      "epoch": 10.898408812729498,
      "grad_norm": 1.6730917363727054,
      "learning_rate": 8.67079266391157e-07,
      "loss": 0.4579,
      "step": 8904
    },
    {
      "epoch": 10.899632802937576,
      "grad_norm": 1.894003840891833,
      "learning_rate": 8.665940377015086e-07,
      "loss": 0.4594,
      "step": 8905
    },
    {
      "epoch": 10.900856793145655,
      "grad_norm": 1.4035659571897081,
      "learning_rate": 8.661089163532596e-07,
      "loss": 0.3591,
      "step": 8906
    },
    {
      "epoch": 10.902080783353734,
      "grad_norm": 2.3197191936502284,
      "learning_rate": 8.656239023782892e-07,
      "loss": 0.4804,
      "step": 8907
    },
    {
      "epoch": 10.903304773561812,
      "grad_norm": 2.1401141376686272,
      "learning_rate": 8.651389958084708e-07,
      "loss": 1.1199,
      "step": 8908
    },
    {
      "epoch": 10.904528763769889,
      "grad_norm": 1.9578680327250007,
      "learning_rate": 8.646541966756705e-07,
      "loss": 0.5221,
      "step": 8909
    },
    {
      "epoch": 10.905752753977968,
      "grad_norm": 1.477165036695425,
      "learning_rate": 8.64169505011748e-07,
      "loss": 0.5916,
      "step": 8910
    },
    {
      "epoch": 10.906976744186046,
      "grad_norm": 1.2970344280290176,
      "learning_rate": 8.636849208485548e-07,
      "loss": 0.6566,
      "step": 8911
    },
    {
      "epoch": 10.908200734394125,
      "grad_norm": 1.7399100675252694,
      "learning_rate": 8.632004442179365e-07,
      "loss": 1.8016,
      "step": 8912
    },
    {
      "epoch": 10.909424724602204,
      "grad_norm": 1.5198510557130038,
      "learning_rate": 8.627160751517308e-07,
      "loss": 1.2277,
      "step": 8913
    },
    {
      "epoch": 10.910648714810282,
      "grad_norm": 2.220286328304501,
      "learning_rate": 8.622318136817678e-07,
      "loss": 0.9632,
      "step": 8914
    },
    {
      "epoch": 10.911872705018359,
      "grad_norm": 1.147575674201209,
      "learning_rate": 8.617476598398742e-07,
      "loss": 0.7447,
      "step": 8915
    },
    {
      "epoch": 10.913096695226438,
      "grad_norm": 2.725471610613177,
      "learning_rate": 8.61263613657864e-07,
      "loss": 0.8274,
      "step": 8916
    },
    {
      "epoch": 10.914320685434516,
      "grad_norm": 0.8657904250779648,
      "learning_rate": 8.607796751675479e-07,
      "loss": 0.3509,
      "step": 8917
    },
    {
      "epoch": 10.915544675642595,
      "grad_norm": 0.7891734698344517,
      "learning_rate": 8.602958444007276e-07,
      "loss": 0.2181,
      "step": 8918
    },
    {
      "epoch": 10.916768665850674,
      "grad_norm": 2.823844167352141,
      "learning_rate": 8.598121213892002e-07,
      "loss": 0.5811,
      "step": 8919
    },
    {
      "epoch": 10.917992656058752,
      "grad_norm": 1.8467338048152613,
      "learning_rate": 8.593285061647539e-07,
      "loss": 0.4994,
      "step": 8920
    },
    {
      "epoch": 10.919216646266829,
      "grad_norm": 1.501730814717575,
      "learning_rate": 8.588449987591704e-07,
      "loss": 0.4768,
      "step": 8921
    },
    {
      "epoch": 10.920440636474908,
      "grad_norm": 1.6695639631283132,
      "learning_rate": 8.583615992042227e-07,
      "loss": 0.5084,
      "step": 8922
    },
    {
      "epoch": 10.921664626682986,
      "grad_norm": 1.111795791156414,
      "learning_rate": 8.578783075316777e-07,
      "loss": 0.6025,
      "step": 8923
    },
    {
      "epoch": 10.922888616891065,
      "grad_norm": 1.3371035246155394,
      "learning_rate": 8.573951237732972e-07,
      "loss": 1.0881,
      "step": 8924
    },
    {
      "epoch": 10.924112607099143,
      "grad_norm": 1.9300967908001243,
      "learning_rate": 8.569120479608339e-07,
      "loss": 0.6019,
      "step": 8925
    },
    {
      "epoch": 10.925336597307222,
      "grad_norm": 2.0573930980525854,
      "learning_rate": 8.564290801260333e-07,
      "loss": 1.1509,
      "step": 8926
    },
    {
      "epoch": 10.9265605875153,
      "grad_norm": 2.2991558761633013,
      "learning_rate": 8.559462203006344e-07,
      "loss": 1.4073,
      "step": 8927
    },
    {
      "epoch": 10.927784577723378,
      "grad_norm": 2.013762494315006,
      "learning_rate": 8.55463468516369e-07,
      "loss": 0.5142,
      "step": 8928
    },
    {
      "epoch": 10.929008567931456,
      "grad_norm": 2.3225612139624987,
      "learning_rate": 8.549808248049618e-07,
      "loss": 0.2862,
      "step": 8929
    },
    {
      "epoch": 10.930232558139535,
      "grad_norm": 1.5665924910995463,
      "learning_rate": 8.544982891981304e-07,
      "loss": 0.5678,
      "step": 8930
    },
    {
      "epoch": 10.931456548347613,
      "grad_norm": 1.3780835733691228,
      "learning_rate": 8.54015861727584e-07,
      "loss": 0.309,
      "step": 8931
    },
    {
      "epoch": 10.932680538555692,
      "grad_norm": 2.9374634730738003,
      "learning_rate": 8.535335424250282e-07,
      "loss": 0.2104,
      "step": 8932
    },
    {
      "epoch": 10.93390452876377,
      "grad_norm": 1.846529014987819,
      "learning_rate": 8.530513313221589e-07,
      "loss": 0.9246,
      "step": 8933
    },
    {
      "epoch": 10.935128518971847,
      "grad_norm": 1.1023103887743304,
      "learning_rate": 8.525692284506634e-07,
      "loss": 0.3871,
      "step": 8934
    },
    {
      "epoch": 10.936352509179926,
      "grad_norm": 1.054171522443247,
      "learning_rate": 8.520872338422251e-07,
      "loss": 0.3978,
      "step": 8935
    },
    {
      "epoch": 10.937576499388005,
      "grad_norm": 2.329122658947561,
      "learning_rate": 8.516053475285174e-07,
      "loss": 0.4964,
      "step": 8936
    },
    {
      "epoch": 10.938800489596083,
      "grad_norm": 1.2607869420555138,
      "learning_rate": 8.511235695412104e-07,
      "loss": 0.7486,
      "step": 8937
    },
    {
      "epoch": 10.940024479804162,
      "grad_norm": 1.6993531676981326,
      "learning_rate": 8.506418999119636e-07,
      "loss": 0.5714,
      "step": 8938
    },
    {
      "epoch": 10.94124847001224,
      "grad_norm": 1.9166522475525072,
      "learning_rate": 8.501603386724313e-07,
      "loss": 0.3608,
      "step": 8939
    },
    {
      "epoch": 10.94247246022032,
      "grad_norm": 1.0470781885175395,
      "learning_rate": 8.496788858542577e-07,
      "loss": 0.5906,
      "step": 8940
    },
    {
      "epoch": 10.943696450428396,
      "grad_norm": 2.0769101671348174,
      "learning_rate": 8.491975414890844e-07,
      "loss": 1.3167,
      "step": 8941
    },
    {
      "epoch": 10.944920440636475,
      "grad_norm": 2.042378957842182,
      "learning_rate": 8.487163056085427e-07,
      "loss": 0.5075,
      "step": 8942
    },
    {
      "epoch": 10.946144430844553,
      "grad_norm": 1.8228634964297847,
      "learning_rate": 8.482351782442577e-07,
      "loss": 0.9836,
      "step": 8943
    },
    {
      "epoch": 10.947368421052632,
      "grad_norm": 2.7149327295018866,
      "learning_rate": 8.477541594278474e-07,
      "loss": 0.9528,
      "step": 8944
    },
    {
      "epoch": 10.94859241126071,
      "grad_norm": 1.1930688404965586,
      "learning_rate": 8.472732491909225e-07,
      "loss": 0.8828,
      "step": 8945
    },
    {
      "epoch": 10.949816401468787,
      "grad_norm": 1.9291657922578964,
      "learning_rate": 8.467924475650865e-07,
      "loss": 0.5889,
      "step": 8946
    },
    {
      "epoch": 10.951040391676866,
      "grad_norm": 1.957743495582537,
      "learning_rate": 8.463117545819363e-07,
      "loss": 0.7267,
      "step": 8947
    },
    {
      "epoch": 10.952264381884945,
      "grad_norm": 1.751108947639815,
      "learning_rate": 8.458311702730607e-07,
      "loss": 0.5002,
      "step": 8948
    },
    {
      "epoch": 10.953488372093023,
      "grad_norm": 1.8483341817237053,
      "learning_rate": 8.453506946700419e-07,
      "loss": 0.464,
      "step": 8949
    },
    {
      "epoch": 10.954712362301102,
      "grad_norm": 2.0358231351121745,
      "learning_rate": 8.448703278044565e-07,
      "loss": 1.1752,
      "step": 8950
    },
    {
      "epoch": 10.95593635250918,
      "grad_norm": 2.2377133061067744,
      "learning_rate": 8.443900697078705e-07,
      "loss": 0.3958,
      "step": 8951
    },
    {
      "epoch": 10.957160342717259,
      "grad_norm": 1.9164186425605303,
      "learning_rate": 8.439099204118456e-07,
      "loss": 0.5828,
      "step": 8952
    },
    {
      "epoch": 10.958384332925336,
      "grad_norm": 1.5316622325990201,
      "learning_rate": 8.434298799479346e-07,
      "loss": 0.5665,
      "step": 8953
    },
    {
      "epoch": 10.959608323133414,
      "grad_norm": 1.661033787555597,
      "learning_rate": 8.429499483476853e-07,
      "loss": 1.1663,
      "step": 8954
    },
    {
      "epoch": 10.960832313341493,
      "grad_norm": 1.665625099147359,
      "learning_rate": 8.424701256426362e-07,
      "loss": 0.6374,
      "step": 8955
    },
    {
      "epoch": 10.962056303549572,
      "grad_norm": 1.975149715147597,
      "learning_rate": 8.419904118643199e-07,
      "loss": 0.683,
      "step": 8956
    },
    {
      "epoch": 10.96328029375765,
      "grad_norm": 1.6937528719568988,
      "learning_rate": 8.415108070442612e-07,
      "loss": 1.2298,
      "step": 8957
    },
    {
      "epoch": 10.964504283965729,
      "grad_norm": 1.754314461498832,
      "learning_rate": 8.410313112139778e-07,
      "loss": 0.3121,
      "step": 8958
    },
    {
      "epoch": 10.965728274173806,
      "grad_norm": 1.4272650585910678,
      "learning_rate": 8.405519244049806e-07,
      "loss": 1.2816,
      "step": 8959
    },
    {
      "epoch": 10.966952264381884,
      "grad_norm": 1.9662501427400545,
      "learning_rate": 8.400726466487732e-07,
      "loss": 0.5383,
      "step": 8960
    },
    {
      "epoch": 10.968176254589963,
      "grad_norm": 2.0489953724926746,
      "learning_rate": 8.395934779768517e-07,
      "loss": 0.5775,
      "step": 8961
    },
    {
      "epoch": 10.969400244798042,
      "grad_norm": 1.4849446446480306,
      "learning_rate": 8.391144184207053e-07,
      "loss": 0.5883,
      "step": 8962
    },
    {
      "epoch": 10.97062423500612,
      "grad_norm": 2.3237842079024786,
      "learning_rate": 8.386354680118161e-07,
      "loss": 0.9702,
      "step": 8963
    },
    {
      "epoch": 10.971848225214199,
      "grad_norm": 1.8979662160273187,
      "learning_rate": 8.381566267816593e-07,
      "loss": 0.6506,
      "step": 8964
    },
    {
      "epoch": 10.973072215422278,
      "grad_norm": 1.0105928205098453,
      "learning_rate": 8.37677894761702e-07,
      "loss": 0.4415,
      "step": 8965
    },
    {
      "epoch": 10.974296205630354,
      "grad_norm": 1.8026098561749249,
      "learning_rate": 8.371992719834041e-07,
      "loss": 1.4205,
      "step": 8966
    },
    {
      "epoch": 10.975520195838433,
      "grad_norm": 2.0120388489481553,
      "learning_rate": 8.367207584782205e-07,
      "loss": 0.4893,
      "step": 8967
    },
    {
      "epoch": 10.976744186046512,
      "grad_norm": 1.6500466456438387,
      "learning_rate": 8.362423542775977e-07,
      "loss": 0.6702,
      "step": 8968
    },
    {
      "epoch": 10.97796817625459,
      "grad_norm": 1.4490398125283181,
      "learning_rate": 8.357640594129726e-07,
      "loss": 0.6181,
      "step": 8969
    },
    {
      "epoch": 10.979192166462669,
      "grad_norm": 2.8983370979898364,
      "learning_rate": 8.352858739157774e-07,
      "loss": 0.4072,
      "step": 8970
    },
    {
      "epoch": 10.980416156670747,
      "grad_norm": 1.6274523799665237,
      "learning_rate": 8.348077978174377e-07,
      "loss": 0.4543,
      "step": 8971
    },
    {
      "epoch": 10.981640146878824,
      "grad_norm": 1.87577985792482,
      "learning_rate": 8.343298311493706e-07,
      "loss": 1.134,
      "step": 8972
    },
    {
      "epoch": 10.982864137086903,
      "grad_norm": 2.041703163954681,
      "learning_rate": 8.338519739429859e-07,
      "loss": 0.5284,
      "step": 8973
    },
    {
      "epoch": 10.984088127294982,
      "grad_norm": 1.7900977711684403,
      "learning_rate": 8.333742262296868e-07,
      "loss": 0.5777,
      "step": 8974
    },
    {
      "epoch": 10.98531211750306,
      "grad_norm": 2.159972071983754,
      "learning_rate": 8.328965880408694e-07,
      "loss": 0.97,
      "step": 8975
    },
    {
      "epoch": 10.986536107711139,
      "grad_norm": 2.310039857771576,
      "learning_rate": 8.324190594079218e-07,
      "loss": 0.3745,
      "step": 8976
    },
    {
      "epoch": 10.987760097919217,
      "grad_norm": 1.7053630985828936,
      "learning_rate": 8.319416403622258e-07,
      "loss": 0.9098,
      "step": 8977
    },
    {
      "epoch": 10.988984088127294,
      "grad_norm": 1.5961032425641881,
      "learning_rate": 8.314643309351553e-07,
      "loss": 0.9821,
      "step": 8978
    },
    {
      "epoch": 10.990208078335373,
      "grad_norm": 1.9702172662424797,
      "learning_rate": 8.309871311580778e-07,
      "loss": 0.4393,
      "step": 8979
    },
    {
      "epoch": 10.991432068543451,
      "grad_norm": 2.790925789912982,
      "learning_rate": 8.305100410623526e-07,
      "loss": 0.5051,
      "step": 8980
    },
    {
      "epoch": 10.99265605875153,
      "grad_norm": 2.9752884029539763,
      "learning_rate": 8.300330606793325e-07,
      "loss": 0.4907,
      "step": 8981
    },
    {
      "epoch": 10.993880048959609,
      "grad_norm": 2.325050124519071,
      "learning_rate": 8.295561900403629e-07,
      "loss": 1.3339,
      "step": 8982
    },
    {
      "epoch": 10.995104039167687,
      "grad_norm": 1.5812850622737917,
      "learning_rate": 8.290794291767809e-07,
      "loss": 0.69,
      "step": 8983
    },
    {
      "epoch": 10.996328029375764,
      "grad_norm": 1.9202572705944772,
      "learning_rate": 8.286027781199197e-07,
      "loss": 0.9627,
      "step": 8984
    },
    {
      "epoch": 10.997552019583843,
      "grad_norm": 3.5864533521274944,
      "learning_rate": 8.281262369011014e-07,
      "loss": 0.7625,
      "step": 8985
    },
    {
      "epoch": 10.998776009791921,
      "grad_norm": 1.3888240083779075,
      "learning_rate": 8.276498055516441e-07,
      "loss": 0.5563,
      "step": 8986
    },
    {
      "epoch": 11.0,
      "grad_norm": 0.9693536178833745,
      "learning_rate": 8.271734841028553e-07,
      "loss": 0.4933,
      "step": 8987
    },
    {
      "epoch": 11.001223990208079,
      "grad_norm": 1.1289925391657307,
      "learning_rate": 8.266972725860367e-07,
      "loss": 0.5689,
      "step": 8988
    },
    {
      "epoch": 11.002447980416157,
      "grad_norm": 1.8221020220221489,
      "learning_rate": 8.262211710324853e-07,
      "loss": 1.4173,
      "step": 8989
    },
    {
      "epoch": 11.003671970624236,
      "grad_norm": 1.1791998370160566,
      "learning_rate": 8.257451794734872e-07,
      "loss": 0.5592,
      "step": 8990
    },
    {
      "epoch": 11.004895960832313,
      "grad_norm": 1.9316723770530155,
      "learning_rate": 8.252692979403237e-07,
      "loss": 0.3354,
      "step": 8991
    },
    {
      "epoch": 11.006119951040391,
      "grad_norm": 1.3749000030580507,
      "learning_rate": 8.247935264642673e-07,
      "loss": 0.6034,
      "step": 8992
    },
    {
      "epoch": 11.00734394124847,
      "grad_norm": 2.0877353307772943,
      "learning_rate": 8.243178650765843e-07,
      "loss": 0.485,
      "step": 8993
    },
    {
      "epoch": 11.008567931456549,
      "grad_norm": 2.0878536127052523,
      "learning_rate": 8.238423138085333e-07,
      "loss": 0.4043,
      "step": 8994
    },
    {
      "epoch": 11.009791921664627,
      "grad_norm": 2.058080277792716,
      "learning_rate": 8.233668726913657e-07,
      "loss": 1.313,
      "step": 8995
    },
    {
      "epoch": 11.011015911872706,
      "grad_norm": 1.216947685758841,
      "learning_rate": 8.228915417563249e-07,
      "loss": 0.5002,
      "step": 8996
    },
    {
      "epoch": 11.012239902080783,
      "grad_norm": 1.9404533427866895,
      "learning_rate": 8.224163210346506e-07,
      "loss": 0.479,
      "step": 8997
    },
    {
      "epoch": 11.013463892288861,
      "grad_norm": 2.2693046245292816,
      "learning_rate": 8.219412105575697e-07,
      "loss": 0.5201,
      "step": 8998
    },
    {
      "epoch": 11.01468788249694,
      "grad_norm": 2.463331257307761,
      "learning_rate": 8.214662103563059e-07,
      "loss": 0.5386,
      "step": 8999
    },
    {
      "epoch": 11.015911872705018,
      "grad_norm": 2.959995702971444,
      "learning_rate": 8.209913204620743e-07,
      "loss": 0.398,
      "step": 9000
    },
    {
      "epoch": 11.017135862913097,
      "grad_norm": 0.8337580078870223,
      "learning_rate": 8.205165409060819e-07,
      "loss": 0.4318,
      "step": 9001
    },
    {
      "epoch": 11.018359853121176,
      "grad_norm": 3.0379771090488275,
      "learning_rate": 8.200418717195316e-07,
      "loss": 0.3805,
      "step": 9002
    },
    {
      "epoch": 11.019583843329253,
      "grad_norm": 1.6598730419960264,
      "learning_rate": 8.195673129336155e-07,
      "loss": 0.6753,
      "step": 9003
    },
    {
      "epoch": 11.020807833537331,
      "grad_norm": 3.38254614824278,
      "learning_rate": 8.19092864579521e-07,
      "loss": 0.6778,
      "step": 9004
    },
    {
      "epoch": 11.02203182374541,
      "grad_norm": 1.2449635255493556,
      "learning_rate": 8.186185266884245e-07,
      "loss": 0.7469,
      "step": 9005
    },
    {
      "epoch": 11.023255813953488,
      "grad_norm": 1.6497189206451062,
      "learning_rate": 8.181442992915001e-07,
      "loss": 0.4858,
      "step": 9006
    },
    {
      "epoch": 11.024479804161567,
      "grad_norm": 1.3058462142843823,
      "learning_rate": 8.176701824199118e-07,
      "loss": 0.7144,
      "step": 9007
    },
    {
      "epoch": 11.025703794369646,
      "grad_norm": 3.2527790581892657,
      "learning_rate": 8.171961761048167e-07,
      "loss": 0.5882,
      "step": 9008
    },
    {
      "epoch": 11.026927784577722,
      "grad_norm": 1.5922785584790589,
      "learning_rate": 8.167222803773645e-07,
      "loss": 0.8138,
      "step": 9009
    },
    {
      "epoch": 11.028151774785801,
      "grad_norm": 1.1924356260514575,
      "learning_rate": 8.162484952686981e-07,
      "loss": 0.4981,
      "step": 9010
    },
    {
      "epoch": 11.02937576499388,
      "grad_norm": 1.9626999528009017,
      "learning_rate": 8.157748208099525e-07,
      "loss": 0.7235,
      "step": 9011
    },
    {
      "epoch": 11.030599755201958,
      "grad_norm": 1.5058697414033677,
      "learning_rate": 8.153012570322566e-07,
      "loss": 0.4737,
      "step": 9012
    },
    {
      "epoch": 11.031823745410037,
      "grad_norm": 1.0028706580506175,
      "learning_rate": 8.148278039667306e-07,
      "loss": 0.438,
      "step": 9013
    },
    {
      "epoch": 11.033047735618116,
      "grad_norm": 1.9264411742542642,
      "learning_rate": 8.143544616444876e-07,
      "loss": 0.3575,
      "step": 9014
    },
    {
      "epoch": 11.034271725826194,
      "grad_norm": 1.6045848568118388,
      "learning_rate": 8.138812300966361e-07,
      "loss": 0.3528,
      "step": 9015
    },
    {
      "epoch": 11.035495716034271,
      "grad_norm": 2.0386715085310905,
      "learning_rate": 8.134081093542728e-07,
      "loss": 0.4772,
      "step": 9016
    },
    {
      "epoch": 11.03671970624235,
      "grad_norm": 2.4086092216733186,
      "learning_rate": 8.129350994484905e-07,
      "loss": 0.4433,
      "step": 9017
    },
    {
      "epoch": 11.037943696450428,
      "grad_norm": 1.7455660542088032,
      "learning_rate": 8.124622004103725e-07,
      "loss": 0.8219,
      "step": 9018
    },
    {
      "epoch": 11.039167686658507,
      "grad_norm": 1.4532366693555925,
      "learning_rate": 8.119894122709975e-07,
      "loss": 0.7217,
      "step": 9019
    },
    {
      "epoch": 11.040391676866586,
      "grad_norm": 2.117616021720788,
      "learning_rate": 8.11516735061435e-07,
      "loss": 0.6643,
      "step": 9020
    },
    {
      "epoch": 11.041615667074664,
      "grad_norm": 1.5605732319617553,
      "learning_rate": 8.110441688127485e-07,
      "loss": 0.5589,
      "step": 9021
    },
    {
      "epoch": 11.042839657282741,
      "grad_norm": 1.854794137892344,
      "learning_rate": 8.105717135559901e-07,
      "loss": 0.3489,
      "step": 9022
    },
    {
      "epoch": 11.04406364749082,
      "grad_norm": 1.1604213782803097,
      "learning_rate": 8.100993693222111e-07,
      "loss": 0.5567,
      "step": 9023
    },
    {
      "epoch": 11.045287637698898,
      "grad_norm": 2.428132684325884,
      "learning_rate": 8.09627136142451e-07,
      "loss": 0.8956,
      "step": 9024
    },
    {
      "epoch": 11.046511627906977,
      "grad_norm": 3.532493026301642,
      "learning_rate": 8.09155014047743e-07,
      "loss": 0.4088,
      "step": 9025
    },
    {
      "epoch": 11.047735618115055,
      "grad_norm": 2.6419304656065727,
      "learning_rate": 8.086830030691137e-07,
      "loss": 0.945,
      "step": 9026
    },
    {
      "epoch": 11.048959608323134,
      "grad_norm": 2.0147598842992127,
      "learning_rate": 8.082111032375817e-07,
      "loss": 0.4888,
      "step": 9027
    },
    {
      "epoch": 11.050183598531211,
      "grad_norm": 1.7221598166560412,
      "learning_rate": 8.077393145841583e-07,
      "loss": 0.8554,
      "step": 9028
    },
    {
      "epoch": 11.05140758873929,
      "grad_norm": 2.847893662916916,
      "learning_rate": 8.07267637139848e-07,
      "loss": 0.352,
      "step": 9029
    },
    {
      "epoch": 11.052631578947368,
      "grad_norm": 1.7539307739055112,
      "learning_rate": 8.067960709356479e-07,
      "loss": 0.7451,
      "step": 9030
    },
    {
      "epoch": 11.053855569155447,
      "grad_norm": 1.8805115400790082,
      "learning_rate": 8.063246160025462e-07,
      "loss": 0.483,
      "step": 9031
    },
    {
      "epoch": 11.055079559363525,
      "grad_norm": 1.4678512270655446,
      "learning_rate": 8.058532723715271e-07,
      "loss": 0.5549,
      "step": 9032
    },
    {
      "epoch": 11.056303549571604,
      "grad_norm": 1.5594408200453251,
      "learning_rate": 8.053820400735654e-07,
      "loss": 1.1142,
      "step": 9033
    },
    {
      "epoch": 11.057527539779683,
      "grad_norm": 1.5758230819702244,
      "learning_rate": 8.049109191396274e-07,
      "loss": 0.4715,
      "step": 9034
    },
    {
      "epoch": 11.05875152998776,
      "grad_norm": 2.4567145809287037,
      "learning_rate": 8.044399096006733e-07,
      "loss": 0.4498,
      "step": 9035
    },
    {
      "epoch": 11.059975520195838,
      "grad_norm": 1.4285953589742924,
      "learning_rate": 8.039690114876578e-07,
      "loss": 1.1269,
      "step": 9036
    },
    {
      "epoch": 11.061199510403917,
      "grad_norm": 1.2639135334186864,
      "learning_rate": 8.034982248315254e-07,
      "loss": 0.4477,
      "step": 9037
    },
    {
      "epoch": 11.062423500611995,
      "grad_norm": 2.0556390954012675,
      "learning_rate": 8.030275496632148e-07,
      "loss": 0.4154,
      "step": 9038
    },
    {
      "epoch": 11.063647490820074,
      "grad_norm": 1.3588686954988605,
      "learning_rate": 8.025569860136581e-07,
      "loss": 1.0416,
      "step": 9039
    },
    {
      "epoch": 11.064871481028153,
      "grad_norm": 1.5282918284009752,
      "learning_rate": 8.020865339137759e-07,
      "loss": 0.9821,
      "step": 9040
    },
    {
      "epoch": 11.06609547123623,
      "grad_norm": 1.534433316137476,
      "learning_rate": 8.016161933944874e-07,
      "loss": 0.7076,
      "step": 9041
    },
    {
      "epoch": 11.067319461444308,
      "grad_norm": 2.2054124690883317,
      "learning_rate": 8.011459644867009e-07,
      "loss": 0.5352,
      "step": 9042
    },
    {
      "epoch": 11.068543451652387,
      "grad_norm": 1.4714474783642406,
      "learning_rate": 8.00675847221318e-07,
      "loss": 0.5572,
      "step": 9043
    },
    {
      "epoch": 11.069767441860465,
      "grad_norm": 1.3320788204444765,
      "learning_rate": 8.00205841629233e-07,
      "loss": 0.574,
      "step": 9044
    },
    {
      "epoch": 11.070991432068544,
      "grad_norm": 1.1670758620919202,
      "learning_rate": 7.99735947741333e-07,
      "loss": 0.4736,
      "step": 9045
    },
    {
      "epoch": 11.072215422276622,
      "grad_norm": 1.8480131502968522,
      "learning_rate": 7.992661655884975e-07,
      "loss": 0.5313,
      "step": 9046
    },
    {
      "epoch": 11.0734394124847,
      "grad_norm": 2.8361725815278214,
      "learning_rate": 7.987964952015994e-07,
      "loss": 0.3577,
      "step": 9047
    },
    {
      "epoch": 11.074663402692778,
      "grad_norm": 2.027972525331852,
      "learning_rate": 7.983269366115023e-07,
      "loss": 0.7601,
      "step": 9048
    },
    {
      "epoch": 11.075887392900857,
      "grad_norm": 2.026673609144047,
      "learning_rate": 7.978574898490657e-07,
      "loss": 0.7706,
      "step": 9049
    },
    {
      "epoch": 11.077111383108935,
      "grad_norm": 1.2065909096414953,
      "learning_rate": 7.973881549451398e-07,
      "loss": 0.4704,
      "step": 9050
    },
    {
      "epoch": 11.078335373317014,
      "grad_norm": 3.839933680884358,
      "learning_rate": 7.969189319305662e-07,
      "loss": 0.3739,
      "step": 9051
    },
    {
      "epoch": 11.079559363525092,
      "grad_norm": 1.7507623043046063,
      "learning_rate": 7.96449820836181e-07,
      "loss": 0.5359,
      "step": 9052
    },
    {
      "epoch": 11.080783353733171,
      "grad_norm": 2.072053109697878,
      "learning_rate": 7.959808216928116e-07,
      "loss": 0.4024,
      "step": 9053
    },
    {
      "epoch": 11.082007343941248,
      "grad_norm": 2.1449997890513215,
      "learning_rate": 7.955119345312809e-07,
      "loss": 0.4154,
      "step": 9054
    },
    {
      "epoch": 11.083231334149326,
      "grad_norm": 2.4011598293805614,
      "learning_rate": 7.950431593824015e-07,
      "loss": 0.5864,
      "step": 9055
    },
    {
      "epoch": 11.084455324357405,
      "grad_norm": 2.1241554889745897,
      "learning_rate": 7.945744962769791e-07,
      "loss": 0.7425,
      "step": 9056
    },
    {
      "epoch": 11.085679314565484,
      "grad_norm": 1.6662494654909628,
      "learning_rate": 7.941059452458133e-07,
      "loss": 0.5051,
      "step": 9057
    },
    {
      "epoch": 11.086903304773562,
      "grad_norm": 1.0550890742901429,
      "learning_rate": 7.936375063196949e-07,
      "loss": 0.3959,
      "step": 9058
    },
    {
      "epoch": 11.088127294981641,
      "grad_norm": 1.254876338661226,
      "learning_rate": 7.931691795294083e-07,
      "loss": 0.578,
      "step": 9059
    },
    {
      "epoch": 11.089351285189718,
      "grad_norm": 2.8877306921798107,
      "learning_rate": 7.9270096490573e-07,
      "loss": 0.3495,
      "step": 9060
    },
    {
      "epoch": 11.090575275397796,
      "grad_norm": 1.0833302202086688,
      "learning_rate": 7.922328624794285e-07,
      "loss": 0.5891,
      "step": 9061
    },
    {
      "epoch": 11.091799265605875,
      "grad_norm": 1.4817649774576878,
      "learning_rate": 7.917648722812684e-07,
      "loss": 0.5549,
      "step": 9062
    },
    {
      "epoch": 11.093023255813954,
      "grad_norm": 2.953505720386357,
      "learning_rate": 7.912969943420018e-07,
      "loss": 0.4077,
      "step": 9063
    },
    {
      "epoch": 11.094247246022032,
      "grad_norm": 1.89259016000055,
      "learning_rate": 7.908292286923766e-07,
      "loss": 0.3858,
      "step": 9064
    },
    {
      "epoch": 11.095471236230111,
      "grad_norm": 2.2833507750051956,
      "learning_rate": 7.903615753631327e-07,
      "loss": 0.4424,
      "step": 9065
    },
    {
      "epoch": 11.096695226438188,
      "grad_norm": 2.695021120179962,
      "learning_rate": 7.898940343850017e-07,
      "loss": 0.4557,
      "step": 9066
    },
    {
      "epoch": 11.097919216646266,
      "grad_norm": 2.274765322029223,
      "learning_rate": 7.894266057887107e-07,
      "loss": 0.961,
      "step": 9067
    },
    {
      "epoch": 11.099143206854345,
      "grad_norm": 1.306886182953371,
      "learning_rate": 7.889592896049767e-07,
      "loss": 0.9722,
      "step": 9068
    },
    {
      "epoch": 11.100367197062424,
      "grad_norm": 2.0656082370102236,
      "learning_rate": 7.884920858645088e-07,
      "loss": 0.4034,
      "step": 9069
    },
    {
      "epoch": 11.101591187270502,
      "grad_norm": 3.7382445197822136,
      "learning_rate": 7.880249945980095e-07,
      "loss": 0.4169,
      "step": 9070
    },
    {
      "epoch": 11.10281517747858,
      "grad_norm": 3.1049568336389965,
      "learning_rate": 7.875580158361765e-07,
      "loss": 0.3003,
      "step": 9071
    },
    {
      "epoch": 11.104039167686658,
      "grad_norm": 1.3420903798518247,
      "learning_rate": 7.870911496096967e-07,
      "loss": 0.6379,
      "step": 9072
    },
    {
      "epoch": 11.105263157894736,
      "grad_norm": 1.296158044267461,
      "learning_rate": 7.866243959492509e-07,
      "loss": 0.9826,
      "step": 9073
    },
    {
      "epoch": 11.106487148102815,
      "grad_norm": 1.2906180768021773,
      "learning_rate": 7.861577548855123e-07,
      "loss": 0.7591,
      "step": 9074
    },
    {
      "epoch": 11.107711138310894,
      "grad_norm": 1.2433176899451637,
      "learning_rate": 7.856912264491473e-07,
      "loss": 0.5279,
      "step": 9075
    },
    {
      "epoch": 11.108935128518972,
      "grad_norm": 2.918391339038749,
      "learning_rate": 7.852248106708141e-07,
      "loss": 0.4171,
      "step": 9076
    },
    {
      "epoch": 11.11015911872705,
      "grad_norm": 1.5352841303201348,
      "learning_rate": 7.847585075811636e-07,
      "loss": 0.8624,
      "step": 9077
    },
    {
      "epoch": 11.11138310893513,
      "grad_norm": 1.7867949910916066,
      "learning_rate": 7.8429231721084e-07,
      "loss": 0.5427,
      "step": 9078
    },
    {
      "epoch": 11.112607099143206,
      "grad_norm": 1.1824868865946672,
      "learning_rate": 7.838262395904792e-07,
      "loss": 0.691,
      "step": 9079
    },
    {
      "epoch": 11.113831089351285,
      "grad_norm": 1.3146322324244117,
      "learning_rate": 7.833602747507101e-07,
      "loss": 0.7286,
      "step": 9080
    },
    {
      "epoch": 11.115055079559363,
      "grad_norm": 1.6182449169451625,
      "learning_rate": 7.828944227221546e-07,
      "loss": 1.2426,
      "step": 9081
    },
    {
      "epoch": 11.116279069767442,
      "grad_norm": 1.5955642993106962,
      "learning_rate": 7.824286835354263e-07,
      "loss": 0.9277,
      "step": 9082
    },
    {
      "epoch": 11.11750305997552,
      "grad_norm": 0.7424132452994368,
      "learning_rate": 7.819630572211315e-07,
      "loss": 0.2555,
      "step": 9083
    },
    {
      "epoch": 11.1187270501836,
      "grad_norm": 1.484020607951024,
      "learning_rate": 7.814975438098704e-07,
      "loss": 0.3579,
      "step": 9084
    },
    {
      "epoch": 11.119951040391676,
      "grad_norm": 2.2102104942832668,
      "learning_rate": 7.810321433322348e-07,
      "loss": 0.4653,
      "step": 9085
    },
    {
      "epoch": 11.121175030599755,
      "grad_norm": 1.4141735938394249,
      "learning_rate": 7.805668558188095e-07,
      "loss": 0.4281,
      "step": 9086
    },
    {
      "epoch": 11.122399020807833,
      "grad_norm": 2.2722903076937175,
      "learning_rate": 7.801016813001696e-07,
      "loss": 0.5753,
      "step": 9087
    },
    {
      "epoch": 11.123623011015912,
      "grad_norm": 1.6896563783615277,
      "learning_rate": 7.796366198068853e-07,
      "loss": 0.7264,
      "step": 9088
    },
    {
      "epoch": 11.12484700122399,
      "grad_norm": 1.9011428103531902,
      "learning_rate": 7.791716713695196e-07,
      "loss": 0.5987,
      "step": 9089
    },
    {
      "epoch": 11.12607099143207,
      "grad_norm": 1.3223997013010873,
      "learning_rate": 7.78706836018627e-07,
      "loss": 0.7352,
      "step": 9090
    },
    {
      "epoch": 11.127294981640146,
      "grad_norm": 1.4186627612434166,
      "learning_rate": 7.782421137847545e-07,
      "loss": 0.4313,
      "step": 9091
    },
    {
      "epoch": 11.128518971848225,
      "grad_norm": 3.66746973237799,
      "learning_rate": 7.77777504698442e-07,
      "loss": 0.4725,
      "step": 9092
    },
    {
      "epoch": 11.129742962056303,
      "grad_norm": 2.1174106319585206,
      "learning_rate": 7.773130087902217e-07,
      "loss": 0.6332,
      "step": 9093
    },
    {
      "epoch": 11.130966952264382,
      "grad_norm": 1.2773208865439043,
      "learning_rate": 7.768486260906186e-07,
      "loss": 0.8927,
      "step": 9094
    },
    {
      "epoch": 11.13219094247246,
      "grad_norm": 2.4292649627633094,
      "learning_rate": 7.763843566301505e-07,
      "loss": 0.5545,
      "step": 9095
    },
    {
      "epoch": 11.13341493268054,
      "grad_norm": 1.3557435645291944,
      "learning_rate": 7.759202004393263e-07,
      "loss": 0.5312,
      "step": 9096
    },
    {
      "epoch": 11.134638922888616,
      "grad_norm": 3.447571551851755,
      "learning_rate": 7.754561575486513e-07,
      "loss": 0.3307,
      "step": 9097
    },
    {
      "epoch": 11.135862913096695,
      "grad_norm": 1.638558041786235,
      "learning_rate": 7.749922279886183e-07,
      "loss": 1.1662,
      "step": 9098
    },
    {
      "epoch": 11.137086903304773,
      "grad_norm": 1.499538569700982,
      "learning_rate": 7.745284117897156e-07,
      "loss": 0.5427,
      "step": 9099
    },
    {
      "epoch": 11.138310893512852,
      "grad_norm": 1.089358414460181,
      "learning_rate": 7.740647089824236e-07,
      "loss": 0.3922,
      "step": 9100
    },
    {
      "epoch": 11.13953488372093,
      "grad_norm": 1.9188916057645615,
      "learning_rate": 7.736011195972143e-07,
      "loss": 1.1016,
      "step": 9101
    },
    {
      "epoch": 11.140758873929009,
      "grad_norm": 1.5536141968734014,
      "learning_rate": 7.731376436645547e-07,
      "loss": 0.3327,
      "step": 9102
    },
    {
      "epoch": 11.141982864137088,
      "grad_norm": 1.6680839934894849,
      "learning_rate": 7.726742812149018e-07,
      "loss": 1.1716,
      "step": 9103
    },
    {
      "epoch": 11.143206854345165,
      "grad_norm": 1.9484252697281426,
      "learning_rate": 7.722110322787072e-07,
      "loss": 0.489,
      "step": 9104
    },
    {
      "epoch": 11.144430844553243,
      "grad_norm": 2.569829002386726,
      "learning_rate": 7.717478968864111e-07,
      "loss": 0.3272,
      "step": 9105
    },
    {
      "epoch": 11.145654834761322,
      "grad_norm": 1.1535842832355192,
      "learning_rate": 7.71284875068452e-07,
      "loss": 0.5643,
      "step": 9106
    },
    {
      "epoch": 11.1468788249694,
      "grad_norm": 2.8093778229001214,
      "learning_rate": 7.708219668552566e-07,
      "loss": 0.4425,
      "step": 9107
    },
    {
      "epoch": 11.148102815177479,
      "grad_norm": 2.330324877420494,
      "learning_rate": 7.703591722772457e-07,
      "loss": 0.6354,
      "step": 9108
    },
    {
      "epoch": 11.149326805385558,
      "grad_norm": 1.732857296987824,
      "learning_rate": 7.698964913648327e-07,
      "loss": 0.4924,
      "step": 9109
    },
    {
      "epoch": 11.150550795593634,
      "grad_norm": 2.395273694420052,
      "learning_rate": 7.694339241484231e-07,
      "loss": 0.4109,
      "step": 9110
    },
    {
      "epoch": 11.151774785801713,
      "grad_norm": 1.1402521487978383,
      "learning_rate": 7.68971470658415e-07,
      "loss": 0.367,
      "step": 9111
    },
    {
      "epoch": 11.152998776009792,
      "grad_norm": 1.2498823238802579,
      "learning_rate": 7.685091309251991e-07,
      "loss": 0.5531,
      "step": 9112
    },
    {
      "epoch": 11.15422276621787,
      "grad_norm": 2.3409235946449924,
      "learning_rate": 7.680469049791589e-07,
      "loss": 0.9805,
      "step": 9113
    },
    {
      "epoch": 11.155446756425949,
      "grad_norm": 1.8843083817100326,
      "learning_rate": 7.675847928506691e-07,
      "loss": 0.5227,
      "step": 9114
    },
    {
      "epoch": 11.156670746634028,
      "grad_norm": 1.2280877648771822,
      "learning_rate": 7.671227945701007e-07,
      "loss": 0.6573,
      "step": 9115
    },
    {
      "epoch": 11.157894736842104,
      "grad_norm": 1.9848982625545588,
      "learning_rate": 7.666609101678121e-07,
      "loss": 0.3564,
      "step": 9116
    },
    {
      "epoch": 11.159118727050183,
      "grad_norm": 1.7460817593611593,
      "learning_rate": 7.661991396741569e-07,
      "loss": 1.5531,
      "step": 9117
    },
    {
      "epoch": 11.160342717258262,
      "grad_norm": 1.5512508292757279,
      "learning_rate": 7.657374831194811e-07,
      "loss": 0.3319,
      "step": 9118
    },
    {
      "epoch": 11.16156670746634,
      "grad_norm": 1.168089979472258,
      "learning_rate": 7.652759405341237e-07,
      "loss": 0.6023,
      "step": 9119
    },
    {
      "epoch": 11.162790697674419,
      "grad_norm": 2.0344674463992787,
      "learning_rate": 7.648145119484152e-07,
      "loss": 0.7586,
      "step": 9120
    },
    {
      "epoch": 11.164014687882498,
      "grad_norm": 1.747980283228499,
      "learning_rate": 7.643531973926791e-07,
      "loss": 0.482,
      "step": 9121
    },
    {
      "epoch": 11.165238678090576,
      "grad_norm": 1.4712900649859744,
      "learning_rate": 7.638919968972311e-07,
      "loss": 1.3046,
      "step": 9122
    },
    {
      "epoch": 11.166462668298653,
      "grad_norm": 1.153693911464379,
      "learning_rate": 7.634309104923798e-07,
      "loss": 0.4453,
      "step": 9123
    },
    {
      "epoch": 11.167686658506732,
      "grad_norm": 1.1122289361877795,
      "learning_rate": 7.629699382084257e-07,
      "loss": 0.6785,
      "step": 9124
    },
    {
      "epoch": 11.16891064871481,
      "grad_norm": 1.6293839506096226,
      "learning_rate": 7.625090800756626e-07,
      "loss": 1.0644,
      "step": 9125
    },
    {
      "epoch": 11.170134638922889,
      "grad_norm": 3.142294094788809,
      "learning_rate": 7.620483361243761e-07,
      "loss": 0.3691,
      "step": 9126
    },
    {
      "epoch": 11.171358629130967,
      "grad_norm": 2.6980444388815017,
      "learning_rate": 7.615877063848448e-07,
      "loss": 0.6525,
      "step": 9127
    },
    {
      "epoch": 11.172582619339046,
      "grad_norm": 1.481898591842148,
      "learning_rate": 7.611271908873397e-07,
      "loss": 0.7476,
      "step": 9128
    },
    {
      "epoch": 11.173806609547123,
      "grad_norm": 1.8642462988957658,
      "learning_rate": 7.606667896621236e-07,
      "loss": 0.5438,
      "step": 9129
    },
    {
      "epoch": 11.175030599755202,
      "grad_norm": 1.5543913081424334,
      "learning_rate": 7.602065027394528e-07,
      "loss": 0.6206,
      "step": 9130
    },
    {
      "epoch": 11.17625458996328,
      "grad_norm": 1.9162629538149112,
      "learning_rate": 7.59746330149575e-07,
      "loss": 0.4541,
      "step": 9131
    },
    {
      "epoch": 11.177478580171359,
      "grad_norm": 1.6064608708152974,
      "learning_rate": 7.592862719227323e-07,
      "loss": 0.5754,
      "step": 9132
    },
    {
      "epoch": 11.178702570379437,
      "grad_norm": 1.2616563022953007,
      "learning_rate": 7.588263280891581e-07,
      "loss": 0.5883,
      "step": 9133
    },
    {
      "epoch": 11.179926560587516,
      "grad_norm": 1.9146964240642799,
      "learning_rate": 7.583664986790767e-07,
      "loss": 0.4771,
      "step": 9134
    },
    {
      "epoch": 11.181150550795593,
      "grad_norm": 1.0736450356230924,
      "learning_rate": 7.579067837227064e-07,
      "loss": 0.5357,
      "step": 9135
    },
    {
      "epoch": 11.182374541003671,
      "grad_norm": 2.618260070759814,
      "learning_rate": 7.574471832502595e-07,
      "loss": 0.387,
      "step": 9136
    },
    {
      "epoch": 11.18359853121175,
      "grad_norm": 2.207984561315172,
      "learning_rate": 7.569876972919384e-07,
      "loss": 0.5316,
      "step": 9137
    },
    {
      "epoch": 11.184822521419829,
      "grad_norm": 2.473155313859028,
      "learning_rate": 7.565283258779388e-07,
      "loss": 0.3911,
      "step": 9138
    },
    {
      "epoch": 11.186046511627907,
      "grad_norm": 2.435780528158061,
      "learning_rate": 7.560690690384501e-07,
      "loss": 0.441,
      "step": 9139
    },
    {
      "epoch": 11.187270501835986,
      "grad_norm": 2.9536713629994464,
      "learning_rate": 7.556099268036498e-07,
      "loss": 0.3597,
      "step": 9140
    },
    {
      "epoch": 11.188494492044065,
      "grad_norm": 1.5630554290864693,
      "learning_rate": 7.55150899203714e-07,
      "loss": 0.9144,
      "step": 9141
    },
    {
      "epoch": 11.189718482252141,
      "grad_norm": 2.131091107003431,
      "learning_rate": 7.546919862688076e-07,
      "loss": 0.543,
      "step": 9142
    },
    {
      "epoch": 11.19094247246022,
      "grad_norm": 1.710285488802874,
      "learning_rate": 7.542331880290882e-07,
      "loss": 0.4134,
      "step": 9143
    },
    {
      "epoch": 11.192166462668299,
      "grad_norm": 2.2787525268794533,
      "learning_rate": 7.537745045147066e-07,
      "loss": 0.3687,
      "step": 9144
    },
    {
      "epoch": 11.193390452876377,
      "grad_norm": 1.6808967914357382,
      "learning_rate": 7.533159357558056e-07,
      "loss": 0.4521,
      "step": 9145
    },
    {
      "epoch": 11.194614443084456,
      "grad_norm": 2.6093955952367422,
      "learning_rate": 7.52857481782521e-07,
      "loss": 0.8137,
      "step": 9146
    },
    {
      "epoch": 11.195838433292534,
      "grad_norm": 1.963454899072075,
      "learning_rate": 7.523991426249802e-07,
      "loss": 0.5188,
      "step": 9147
    },
    {
      "epoch": 11.197062423500611,
      "grad_norm": 1.5012452829516905,
      "learning_rate": 7.519409183133033e-07,
      "loss": 0.6006,
      "step": 9148
    },
    {
      "epoch": 11.19828641370869,
      "grad_norm": 1.9075276671449244,
      "learning_rate": 7.514828088776041e-07,
      "loss": 0.727,
      "step": 9149
    },
    {
      "epoch": 11.199510403916769,
      "grad_norm": 1.7047109922290788,
      "learning_rate": 7.510248143479876e-07,
      "loss": 0.4509,
      "step": 9150
    },
    {
      "epoch": 11.200734394124847,
      "grad_norm": 1.8531093354105534,
      "learning_rate": 7.505669347545522e-07,
      "loss": 0.4685,
      "step": 9151
    },
    {
      "epoch": 11.201958384332926,
      "grad_norm": 1.6176632290320638,
      "learning_rate": 7.501091701273861e-07,
      "loss": 0.472,
      "step": 9152
    },
    {
      "epoch": 11.203182374541004,
      "grad_norm": 1.1426095234925107,
      "learning_rate": 7.496515204965723e-07,
      "loss": 0.6381,
      "step": 9153
    },
    {
      "epoch": 11.204406364749081,
      "grad_norm": 1.9169721458303726,
      "learning_rate": 7.491939858921871e-07,
      "loss": 0.5799,
      "step": 9154
    },
    {
      "epoch": 11.20563035495716,
      "grad_norm": 1.4332003874143537,
      "learning_rate": 7.487365663442975e-07,
      "loss": 0.43,
      "step": 9155
    },
    {
      "epoch": 11.206854345165238,
      "grad_norm": 2.204336299885022,
      "learning_rate": 7.482792618829635e-07,
      "loss": 0.7124,
      "step": 9156
    },
    {
      "epoch": 11.208078335373317,
      "grad_norm": 3.552237915978077,
      "learning_rate": 7.478220725382368e-07,
      "loss": 0.4215,
      "step": 9157
    },
    {
      "epoch": 11.209302325581396,
      "grad_norm": 2.2082325054828913,
      "learning_rate": 7.473649983401629e-07,
      "loss": 0.5635,
      "step": 9158
    },
    {
      "epoch": 11.210526315789474,
      "grad_norm": 1.6393584012439335,
      "learning_rate": 7.469080393187786e-07,
      "loss": 0.4008,
      "step": 9159
    },
    {
      "epoch": 11.211750305997551,
      "grad_norm": 1.2567536383872553,
      "learning_rate": 7.464511955041139e-07,
      "loss": 0.682,
      "step": 9160
    },
    {
      "epoch": 11.21297429620563,
      "grad_norm": 2.7340917580872452,
      "learning_rate": 7.459944669261898e-07,
      "loss": 0.9466,
      "step": 9161
    },
    {
      "epoch": 11.214198286413708,
      "grad_norm": 1.4058596927258034,
      "learning_rate": 7.455378536150234e-07,
      "loss": 0.8703,
      "step": 9162
    },
    {
      "epoch": 11.215422276621787,
      "grad_norm": 1.8665857222648208,
      "learning_rate": 7.45081355600619e-07,
      "loss": 0.6411,
      "step": 9163
    },
    {
      "epoch": 11.216646266829866,
      "grad_norm": 1.6787495099736394,
      "learning_rate": 7.44624972912977e-07,
      "loss": 0.6343,
      "step": 9164
    },
    {
      "epoch": 11.217870257037944,
      "grad_norm": 1.7791783845458926,
      "learning_rate": 7.441687055820893e-07,
      "loss": 0.6262,
      "step": 9165
    },
    {
      "epoch": 11.219094247246023,
      "grad_norm": 3.5719544405103507,
      "learning_rate": 7.437125536379389e-07,
      "loss": 0.758,
      "step": 9166
    },
    {
      "epoch": 11.2203182374541,
      "grad_norm": 2.144505497733659,
      "learning_rate": 7.432565171105041e-07,
      "loss": 1.1171,
      "step": 9167
    },
    {
      "epoch": 11.221542227662178,
      "grad_norm": 2.3872502001618265,
      "learning_rate": 7.428005960297541e-07,
      "loss": 0.4658,
      "step": 9168
    },
    {
      "epoch": 11.222766217870257,
      "grad_norm": 1.847329083113318,
      "learning_rate": 7.423447904256489e-07,
      "loss": 1.0824,
      "step": 9169
    },
    {
      "epoch": 11.223990208078336,
      "grad_norm": 2.2244255178768264,
      "learning_rate": 7.418891003281419e-07,
      "loss": 0.6726,
      "step": 9170
    },
    {
      "epoch": 11.225214198286414,
      "grad_norm": 3.0257141022428806,
      "learning_rate": 7.414335257671814e-07,
      "loss": 0.3409,
      "step": 9171
    },
    {
      "epoch": 11.226438188494493,
      "grad_norm": 2.6617880999155075,
      "learning_rate": 7.409780667727051e-07,
      "loss": 0.398,
      "step": 9172
    },
    {
      "epoch": 11.22766217870257,
      "grad_norm": 1.4008937151922232,
      "learning_rate": 7.405227233746439e-07,
      "loss": 0.7739,
      "step": 9173
    },
    {
      "epoch": 11.228886168910648,
      "grad_norm": 2.174140888281186,
      "learning_rate": 7.400674956029216e-07,
      "loss": 0.4332,
      "step": 9174
    },
    {
      "epoch": 11.230110159118727,
      "grad_norm": 1.295026517459073,
      "learning_rate": 7.396123834874541e-07,
      "loss": 0.4748,
      "step": 9175
    },
    {
      "epoch": 11.231334149326806,
      "grad_norm": 1.6189994835044081,
      "learning_rate": 7.391573870581492e-07,
      "loss": 0.5668,
      "step": 9176
    },
    {
      "epoch": 11.232558139534884,
      "grad_norm": 1.6400974007676203,
      "learning_rate": 7.387025063449082e-07,
      "loss": 1.2916,
      "step": 9177
    },
    {
      "epoch": 11.233782129742963,
      "grad_norm": 1.4401681692308206,
      "learning_rate": 7.382477413776237e-07,
      "loss": 1.1474,
      "step": 9178
    },
    {
      "epoch": 11.23500611995104,
      "grad_norm": 1.123553867978941,
      "learning_rate": 7.377930921861806e-07,
      "loss": 0.494,
      "step": 9179
    },
    {
      "epoch": 11.236230110159118,
      "grad_norm": 2.1241558425948175,
      "learning_rate": 7.373385588004592e-07,
      "loss": 1.201,
      "step": 9180
    },
    {
      "epoch": 11.237454100367197,
      "grad_norm": 2.6032484453652613,
      "learning_rate": 7.368841412503275e-07,
      "loss": 0.3166,
      "step": 9181
    },
    {
      "epoch": 11.238678090575275,
      "grad_norm": 1.516883684419195,
      "learning_rate": 7.364298395656483e-07,
      "loss": 1.3987,
      "step": 9182
    },
    {
      "epoch": 11.239902080783354,
      "grad_norm": 1.1605717128897368,
      "learning_rate": 7.359756537762766e-07,
      "loss": 0.5334,
      "step": 9183
    },
    {
      "epoch": 11.241126070991433,
      "grad_norm": 1.3880649547828297,
      "learning_rate": 7.355215839120608e-07,
      "loss": 0.4982,
      "step": 9184
    },
    {
      "epoch": 11.24235006119951,
      "grad_norm": 1.9746009760310927,
      "learning_rate": 7.350676300028403e-07,
      "loss": 0.7242,
      "step": 9185
    },
    {
      "epoch": 11.243574051407588,
      "grad_norm": 1.964830344463645,
      "learning_rate": 7.346137920784477e-07,
      "loss": 0.4557,
      "step": 9186
    },
    {
      "epoch": 11.244798041615667,
      "grad_norm": 1.3535945993292509,
      "learning_rate": 7.341600701687058e-07,
      "loss": 0.6225,
      "step": 9187
    },
    {
      "epoch": 11.246022031823745,
      "grad_norm": 2.0156665478079185,
      "learning_rate": 7.33706464303433e-07,
      "loss": 0.9995,
      "step": 9188
    },
    {
      "epoch": 11.247246022031824,
      "grad_norm": 1.65450977795214,
      "learning_rate": 7.332529745124386e-07,
      "loss": 1.6339,
      "step": 9189
    },
    {
      "epoch": 11.248470012239903,
      "grad_norm": 1.853633702127774,
      "learning_rate": 7.32799600825524e-07,
      "loss": 0.612,
      "step": 9190
    },
    {
      "epoch": 11.249694002447981,
      "grad_norm": 1.7687767927752847,
      "learning_rate": 7.323463432724831e-07,
      "loss": 0.5203,
      "step": 9191
    },
    {
      "epoch": 11.250917992656058,
      "grad_norm": 2.164772063946902,
      "learning_rate": 7.318932018831024e-07,
      "loss": 0.391,
      "step": 9192
    },
    {
      "epoch": 11.252141982864137,
      "grad_norm": 1.2557300534939466,
      "learning_rate": 7.314401766871607e-07,
      "loss": 1.0389,
      "step": 9193
    },
    {
      "epoch": 11.253365973072215,
      "grad_norm": 2.189965988523399,
      "learning_rate": 7.309872677144291e-07,
      "loss": 0.5955,
      "step": 9194
    },
    {
      "epoch": 11.254589963280294,
      "grad_norm": 1.2826601182778723,
      "learning_rate": 7.305344749946714e-07,
      "loss": 0.6445,
      "step": 9195
    },
    {
      "epoch": 11.255813953488373,
      "grad_norm": 1.2707907329611157,
      "learning_rate": 7.300817985576422e-07,
      "loss": 0.8924,
      "step": 9196
    },
    {
      "epoch": 11.257037943696451,
      "grad_norm": 1.4735894090703174,
      "learning_rate": 7.296292384330924e-07,
      "loss": 0.6637,
      "step": 9197
    },
    {
      "epoch": 11.258261933904528,
      "grad_norm": 1.626665321068316,
      "learning_rate": 7.2917679465076e-07,
      "loss": 0.5494,
      "step": 9198
    },
    {
      "epoch": 11.259485924112607,
      "grad_norm": 1.9587523873809236,
      "learning_rate": 7.28724467240379e-07,
      "loss": 0.4567,
      "step": 9199
    },
    {
      "epoch": 11.260709914320685,
      "grad_norm": 1.395759840563724,
      "learning_rate": 7.282722562316735e-07,
      "loss": 1.5626,
      "step": 9200
    },
    {
      "epoch": 11.261933904528764,
      "grad_norm": 1.5934285031704438,
      "learning_rate": 7.278201616543631e-07,
      "loss": 0.6878,
      "step": 9201
    },
    {
      "epoch": 11.263157894736842,
      "grad_norm": 2.1316361555822154,
      "learning_rate": 7.273681835381569e-07,
      "loss": 1.0999,
      "step": 9202
    },
    {
      "epoch": 11.264381884944921,
      "grad_norm": 2.047506549009432,
      "learning_rate": 7.26916321912757e-07,
      "loss": 0.3168,
      "step": 9203
    },
    {
      "epoch": 11.265605875152998,
      "grad_norm": 2.3429887012076076,
      "learning_rate": 7.26464576807859e-07,
      "loss": 0.9683,
      "step": 9204
    },
    {
      "epoch": 11.266829865361077,
      "grad_norm": 1.8865114985989797,
      "learning_rate": 7.260129482531477e-07,
      "loss": 0.609,
      "step": 9205
    },
    {
      "epoch": 11.268053855569155,
      "grad_norm": 2.4163214580647634,
      "learning_rate": 7.255614362783048e-07,
      "loss": 0.543,
      "step": 9206
    },
    {
      "epoch": 11.269277845777234,
      "grad_norm": 2.3934640945231846,
      "learning_rate": 7.251100409130008e-07,
      "loss": 0.416,
      "step": 9207
    },
    {
      "epoch": 11.270501835985312,
      "grad_norm": 1.58977672184993,
      "learning_rate": 7.246587621869003e-07,
      "loss": 0.351,
      "step": 9208
    },
    {
      "epoch": 11.271725826193391,
      "grad_norm": 2.628137159952994,
      "learning_rate": 7.242076001296594e-07,
      "loss": 0.9276,
      "step": 9209
    },
    {
      "epoch": 11.27294981640147,
      "grad_norm": 1.0762399565429819,
      "learning_rate": 7.237565547709266e-07,
      "loss": 0.5452,
      "step": 9210
    },
    {
      "epoch": 11.274173806609546,
      "grad_norm": 3.046337802339018,
      "learning_rate": 7.233056261403434e-07,
      "loss": 0.3155,
      "step": 9211
    },
    {
      "epoch": 11.275397796817625,
      "grad_norm": 2.0005295084866375,
      "learning_rate": 7.228548142675429e-07,
      "loss": 0.5441,
      "step": 9212
    },
    {
      "epoch": 11.276621787025704,
      "grad_norm": 1.358163587024687,
      "learning_rate": 7.224041191821499e-07,
      "loss": 0.4078,
      "step": 9213
    },
    {
      "epoch": 11.277845777233782,
      "grad_norm": 1.0591034443143093,
      "learning_rate": 7.219535409137843e-07,
      "loss": 0.5924,
      "step": 9214
    },
    {
      "epoch": 11.279069767441861,
      "grad_norm": 3.130781399445709,
      "learning_rate": 7.215030794920561e-07,
      "loss": 0.4567,
      "step": 9215
    },
    {
      "epoch": 11.28029375764994,
      "grad_norm": 2.4149238868241656,
      "learning_rate": 7.210527349465662e-07,
      "loss": 0.5885,
      "step": 9216
    },
    {
      "epoch": 11.281517747858016,
      "grad_norm": 3.3263824217706914,
      "learning_rate": 7.206025073069112e-07,
      "loss": 0.5176,
      "step": 9217
    },
    {
      "epoch": 11.282741738066095,
      "grad_norm": 1.200507949084988,
      "learning_rate": 7.201523966026769e-07,
      "loss": 0.6228,
      "step": 9218
    },
    {
      "epoch": 11.283965728274174,
      "grad_norm": 2.3254970094874405,
      "learning_rate": 7.197024028634447e-07,
      "loss": 1.4033,
      "step": 9219
    },
    {
      "epoch": 11.285189718482252,
      "grad_norm": 1.3277088100389418,
      "learning_rate": 7.192525261187858e-07,
      "loss": 0.7712,
      "step": 9220
    },
    {
      "epoch": 11.286413708690331,
      "grad_norm": 1.4289192678138443,
      "learning_rate": 7.188027663982642e-07,
      "loss": 0.4835,
      "step": 9221
    },
    {
      "epoch": 11.28763769889841,
      "grad_norm": 2.0537694934047264,
      "learning_rate": 7.183531237314364e-07,
      "loss": 1.0321,
      "step": 9222
    },
    {
      "epoch": 11.288861689106486,
      "grad_norm": 1.7661501941189586,
      "learning_rate": 7.179035981478519e-07,
      "loss": 1.7992,
      "step": 9223
    },
    {
      "epoch": 11.290085679314565,
      "grad_norm": 2.2568519707800605,
      "learning_rate": 7.174541896770512e-07,
      "loss": 0.3,
      "step": 9224
    },
    {
      "epoch": 11.291309669522644,
      "grad_norm": 1.2173106595406769,
      "learning_rate": 7.170048983485681e-07,
      "loss": 0.5702,
      "step": 9225
    },
    {
      "epoch": 11.292533659730722,
      "grad_norm": 1.9271085523600158,
      "learning_rate": 7.16555724191928e-07,
      "loss": 0.4589,
      "step": 9226
    },
    {
      "epoch": 11.2937576499388,
      "grad_norm": 1.8319674704634554,
      "learning_rate": 7.161066672366493e-07,
      "loss": 1.4968,
      "step": 9227
    },
    {
      "epoch": 11.29498164014688,
      "grad_norm": 2.126143438838529,
      "learning_rate": 7.156577275122423e-07,
      "loss": 0.611,
      "step": 9228
    },
    {
      "epoch": 11.296205630354958,
      "grad_norm": 1.9567019535296075,
      "learning_rate": 7.152089050482097e-07,
      "loss": 0.3936,
      "step": 9229
    },
    {
      "epoch": 11.297429620563035,
      "grad_norm": 1.3403997662206386,
      "learning_rate": 7.147601998740464e-07,
      "loss": 0.5882,
      "step": 9230
    },
    {
      "epoch": 11.298653610771114,
      "grad_norm": 1.616725754870654,
      "learning_rate": 7.143116120192386e-07,
      "loss": 0.4932,
      "step": 9231
    },
    {
      "epoch": 11.299877600979192,
      "grad_norm": 1.1790510775569667,
      "learning_rate": 7.138631415132677e-07,
      "loss": 0.6086,
      "step": 9232
    },
    {
      "epoch": 11.30110159118727,
      "grad_norm": 1.479883386081326,
      "learning_rate": 7.134147883856055e-07,
      "loss": 1.1139,
      "step": 9233
    },
    {
      "epoch": 11.30232558139535,
      "grad_norm": 2.2427046828883177,
      "learning_rate": 7.129665526657145e-07,
      "loss": 0.4472,
      "step": 9234
    },
    {
      "epoch": 11.303549571603428,
      "grad_norm": 1.429985698959392,
      "learning_rate": 7.125184343830513e-07,
      "loss": 0.5136,
      "step": 9235
    },
    {
      "epoch": 11.304773561811505,
      "grad_norm": 3.2859486962040343,
      "learning_rate": 7.120704335670656e-07,
      "loss": 0.4471,
      "step": 9236
    },
    {
      "epoch": 11.305997552019583,
      "grad_norm": 1.584146874112754,
      "learning_rate": 7.116225502471983e-07,
      "loss": 0.4031,
      "step": 9237
    },
    {
      "epoch": 11.307221542227662,
      "grad_norm": 1.4774946297530975,
      "learning_rate": 7.111747844528818e-07,
      "loss": 0.4996,
      "step": 9238
    },
    {
      "epoch": 11.30844553243574,
      "grad_norm": 3.1950924773787652,
      "learning_rate": 7.107271362135431e-07,
      "loss": 0.404,
      "step": 9239
    },
    {
      "epoch": 11.30966952264382,
      "grad_norm": 1.1187894276931432,
      "learning_rate": 7.102796055585975e-07,
      "loss": 0.3856,
      "step": 9240
    },
    {
      "epoch": 11.310893512851898,
      "grad_norm": 2.236064202470563,
      "learning_rate": 7.09832192517457e-07,
      "loss": 0.5745,
      "step": 9241
    },
    {
      "epoch": 11.312117503059975,
      "grad_norm": 2.4889539003623096,
      "learning_rate": 7.093848971195235e-07,
      "loss": 0.4829,
      "step": 9242
    },
    {
      "epoch": 11.313341493268053,
      "grad_norm": 2.0113920492199044,
      "learning_rate": 7.089377193941918e-07,
      "loss": 0.4126,
      "step": 9243
    },
    {
      "epoch": 11.314565483476132,
      "grad_norm": 1.4575247010121124,
      "learning_rate": 7.084906593708483e-07,
      "loss": 0.5274,
      "step": 9244
    },
    {
      "epoch": 11.31578947368421,
      "grad_norm": 1.6967011959221514,
      "learning_rate": 7.080437170788723e-07,
      "loss": 1.4492,
      "step": 9245
    },
    {
      "epoch": 11.31701346389229,
      "grad_norm": 1.6556376058000686,
      "learning_rate": 7.075968925476353e-07,
      "loss": 0.6989,
      "step": 9246
    },
    {
      "epoch": 11.318237454100368,
      "grad_norm": 2.0108899132862255,
      "learning_rate": 7.071501858065008e-07,
      "loss": 0.5107,
      "step": 9247
    },
    {
      "epoch": 11.319461444308445,
      "grad_norm": 1.9845941203102704,
      "learning_rate": 7.06703596884824e-07,
      "loss": 0.4858,
      "step": 9248
    },
    {
      "epoch": 11.320685434516523,
      "grad_norm": 1.6660343767372723,
      "learning_rate": 7.062571258119547e-07,
      "loss": 0.5911,
      "step": 9249
    },
    {
      "epoch": 11.321909424724602,
      "grad_norm": 1.3579458820539614,
      "learning_rate": 7.058107726172325e-07,
      "loss": 0.6761,
      "step": 9250
    },
    {
      "epoch": 11.32313341493268,
      "grad_norm": 1.6401436283944666,
      "learning_rate": 7.053645373299909e-07,
      "loss": 1.5544,
      "step": 9251
    },
    {
      "epoch": 11.32435740514076,
      "grad_norm": 1.4415503443343662,
      "learning_rate": 7.049184199795531e-07,
      "loss": 0.3641,
      "step": 9252
    },
    {
      "epoch": 11.325581395348838,
      "grad_norm": 1.6748704312231806,
      "learning_rate": 7.044724205952366e-07,
      "loss": 0.6145,
      "step": 9253
    },
    {
      "epoch": 11.326805385556916,
      "grad_norm": 3.3291981179399035,
      "learning_rate": 7.040265392063522e-07,
      "loss": 0.3392,
      "step": 9254
    },
    {
      "epoch": 11.328029375764993,
      "grad_norm": 2.745298199409814,
      "learning_rate": 7.035807758422006e-07,
      "loss": 0.4001,
      "step": 9255
    },
    {
      "epoch": 11.329253365973072,
      "grad_norm": 1.6478941029105787,
      "learning_rate": 7.031351305320761e-07,
      "loss": 0.5972,
      "step": 9256
    },
    {
      "epoch": 11.33047735618115,
      "grad_norm": 1.8171200256913236,
      "learning_rate": 7.026896033052647e-07,
      "loss": 0.6729,
      "step": 9257
    },
    {
      "epoch": 11.331701346389229,
      "grad_norm": 1.539496196105262,
      "learning_rate": 7.022441941910446e-07,
      "loss": 0.5241,
      "step": 9258
    },
    {
      "epoch": 11.332925336597308,
      "grad_norm": 2.418844382352003,
      "learning_rate": 7.017989032186867e-07,
      "loss": 0.4781,
      "step": 9259
    },
    {
      "epoch": 11.334149326805386,
      "grad_norm": 2.6292267949732278,
      "learning_rate": 7.013537304174537e-07,
      "loss": 0.4685,
      "step": 9260
    },
    {
      "epoch": 11.335373317013463,
      "grad_norm": 1.3485552864058679,
      "learning_rate": 7.009086758166001e-07,
      "loss": 0.5812,
      "step": 9261
    },
    {
      "epoch": 11.336597307221542,
      "grad_norm": 1.1445512436033012,
      "learning_rate": 7.004637394453756e-07,
      "loss": 0.6606,
      "step": 9262
    },
    {
      "epoch": 11.33782129742962,
      "grad_norm": 1.7408587670652917,
      "learning_rate": 7.000189213330171e-07,
      "loss": 0.994,
      "step": 9263
    },
    {
      "epoch": 11.339045287637699,
      "grad_norm": 2.438142777804475,
      "learning_rate": 6.995742215087578e-07,
      "loss": 0.3708,
      "step": 9264
    },
    {
      "epoch": 11.340269277845778,
      "grad_norm": 2.504701754848156,
      "learning_rate": 6.991296400018211e-07,
      "loss": 0.5173,
      "step": 9265
    },
    {
      "epoch": 11.341493268053856,
      "grad_norm": 1.5316765298500248,
      "learning_rate": 6.986851768414227e-07,
      "loss": 0.4893,
      "step": 9266
    },
    {
      "epoch": 11.342717258261933,
      "grad_norm": 1.9461252815952585,
      "learning_rate": 6.982408320567726e-07,
      "loss": 1.2106,
      "step": 9267
    },
    {
      "epoch": 11.343941248470012,
      "grad_norm": 2.61650354750077,
      "learning_rate": 6.977966056770719e-07,
      "loss": 0.3413,
      "step": 9268
    },
    {
      "epoch": 11.34516523867809,
      "grad_norm": 1.9296089579297702,
      "learning_rate": 6.973524977315114e-07,
      "loss": 0.4785,
      "step": 9269
    },
    {
      "epoch": 11.346389228886169,
      "grad_norm": 2.563334358725834,
      "learning_rate": 6.969085082492763e-07,
      "loss": 1.2763,
      "step": 9270
    },
    {
      "epoch": 11.347613219094248,
      "grad_norm": 1.5490257736661406,
      "learning_rate": 6.964646372595457e-07,
      "loss": 1.5693,
      "step": 9271
    },
    {
      "epoch": 11.348837209302326,
      "grad_norm": 1.0504999307705611,
      "learning_rate": 6.960208847914885e-07,
      "loss": 0.5832,
      "step": 9272
    },
    {
      "epoch": 11.350061199510403,
      "grad_norm": 1.5001011555471966,
      "learning_rate": 6.955772508742661e-07,
      "loss": 0.6149,
      "step": 9273
    },
    {
      "epoch": 11.351285189718482,
      "grad_norm": 2.082757985889313,
      "learning_rate": 6.951337355370328e-07,
      "loss": 0.4982,
      "step": 9274
    },
    {
      "epoch": 11.35250917992656,
      "grad_norm": 1.397001870090096,
      "learning_rate": 6.946903388089343e-07,
      "loss": 0.5416,
      "step": 9275
    },
    {
      "epoch": 11.353733170134639,
      "grad_norm": 1.6740243079566808,
      "learning_rate": 6.942470607191099e-07,
      "loss": 0.6964,
      "step": 9276
    },
    {
      "epoch": 11.354957160342718,
      "grad_norm": 1.51939938291957,
      "learning_rate": 6.938039012966894e-07,
      "loss": 1.3093,
      "step": 9277
    },
    {
      "epoch": 11.356181150550796,
      "grad_norm": 2.1098612718995247,
      "learning_rate": 6.933608605707959e-07,
      "loss": 0.5306,
      "step": 9278
    },
    {
      "epoch": 11.357405140758875,
      "grad_norm": 2.324104354015338,
      "learning_rate": 6.929179385705434e-07,
      "loss": 0.3697,
      "step": 9279
    },
    {
      "epoch": 11.358629130966952,
      "grad_norm": 1.0628003048229993,
      "learning_rate": 6.924751353250416e-07,
      "loss": 0.5238,
      "step": 9280
    },
    {
      "epoch": 11.35985312117503,
      "grad_norm": 1.6323572505380874,
      "learning_rate": 6.920324508633878e-07,
      "loss": 0.4637,
      "step": 9281
    },
    {
      "epoch": 11.361077111383109,
      "grad_norm": 1.5893445870669265,
      "learning_rate": 6.915898852146741e-07,
      "loss": 0.6266,
      "step": 9282
    },
    {
      "epoch": 11.362301101591187,
      "grad_norm": 3.4202677545557543,
      "learning_rate": 6.911474384079833e-07,
      "loss": 0.3468,
      "step": 9283
    },
    {
      "epoch": 11.363525091799266,
      "grad_norm": 1.4429774597057228,
      "learning_rate": 6.907051104723933e-07,
      "loss": 0.5362,
      "step": 9284
    },
    {
      "epoch": 11.364749082007345,
      "grad_norm": 1.813095750432886,
      "learning_rate": 6.902629014369714e-07,
      "loss": 1.5402,
      "step": 9285
    },
    {
      "epoch": 11.365973072215422,
      "grad_norm": 1.6739840417757066,
      "learning_rate": 6.898208113307784e-07,
      "loss": 0.3829,
      "step": 9286
    },
    {
      "epoch": 11.3671970624235,
      "grad_norm": 2.3730018276317026,
      "learning_rate": 6.893788401828649e-07,
      "loss": 0.4587,
      "step": 9287
    },
    {
      "epoch": 11.368421052631579,
      "grad_norm": 2.0577917038363496,
      "learning_rate": 6.889369880222776e-07,
      "loss": 0.5244,
      "step": 9288
    },
    {
      "epoch": 11.369645042839657,
      "grad_norm": 1.8321011967334995,
      "learning_rate": 6.88495254878053e-07,
      "loss": 0.5637,
      "step": 9289
    },
    {
      "epoch": 11.370869033047736,
      "grad_norm": 1.9997078362009568,
      "learning_rate": 6.880536407792199e-07,
      "loss": 0.3856,
      "step": 9290
    },
    {
      "epoch": 11.372093023255815,
      "grad_norm": 1.5212789368544173,
      "learning_rate": 6.876121457547996e-07,
      "loss": 0.7921,
      "step": 9291
    },
    {
      "epoch": 11.373317013463891,
      "grad_norm": 1.8689514556691684,
      "learning_rate": 6.871707698338057e-07,
      "loss": 0.4493,
      "step": 9292
    },
    {
      "epoch": 11.37454100367197,
      "grad_norm": 0.5860894790672225,
      "learning_rate": 6.867295130452434e-07,
      "loss": 0.1288,
      "step": 9293
    },
    {
      "epoch": 11.375764993880049,
      "grad_norm": 2.5729602225669703,
      "learning_rate": 6.862883754181107e-07,
      "loss": 0.4554,
      "step": 9294
    },
    {
      "epoch": 11.376988984088127,
      "grad_norm": 1.5944769672047863,
      "learning_rate": 6.858473569813976e-07,
      "loss": 0.3766,
      "step": 9295
    },
    {
      "epoch": 11.378212974296206,
      "grad_norm": 1.8328470671983572,
      "learning_rate": 6.854064577640851e-07,
      "loss": 0.4567,
      "step": 9296
    },
    {
      "epoch": 11.379436964504285,
      "grad_norm": 3.366510416089552,
      "learning_rate": 6.849656777951493e-07,
      "loss": 0.3765,
      "step": 9297
    },
    {
      "epoch": 11.380660954712361,
      "grad_norm": 1.36816397449437,
      "learning_rate": 6.845250171035566e-07,
      "loss": 0.4406,
      "step": 9298
    },
    {
      "epoch": 11.38188494492044,
      "grad_norm": 3.315066917680868,
      "learning_rate": 6.840844757182644e-07,
      "loss": 0.403,
      "step": 9299
    },
    {
      "epoch": 11.383108935128519,
      "grad_norm": 1.8403379772981971,
      "learning_rate": 6.836440536682226e-07,
      "loss": 0.4609,
      "step": 9300
    },
    {
      "epoch": 11.384332925336597,
      "grad_norm": 1.4349016803106063,
      "learning_rate": 6.832037509823763e-07,
      "loss": 0.5417,
      "step": 9301
    },
    {
      "epoch": 11.385556915544676,
      "grad_norm": 1.981789330852303,
      "learning_rate": 6.827635676896597e-07,
      "loss": 0.3908,
      "step": 9302
    },
    {
      "epoch": 11.386780905752754,
      "grad_norm": 1.4544851213683228,
      "learning_rate": 6.823235038189999e-07,
      "loss": 0.5288,
      "step": 9303
    },
    {
      "epoch": 11.388004895960833,
      "grad_norm": 2.365766202134165,
      "learning_rate": 6.81883559399317e-07,
      "loss": 0.494,
      "step": 9304
    },
    {
      "epoch": 11.38922888616891,
      "grad_norm": 1.8490652727922594,
      "learning_rate": 6.814437344595204e-07,
      "loss": 1.2295,
      "step": 9305
    },
    {
      "epoch": 11.390452876376989,
      "grad_norm": 1.8432405822810964,
      "learning_rate": 6.810040290285161e-07,
      "loss": 1.3025,
      "step": 9306
    },
    {
      "epoch": 11.391676866585067,
      "grad_norm": 2.78969850879817,
      "learning_rate": 6.805644431351987e-07,
      "loss": 1.1758,
      "step": 9307
    },
    {
      "epoch": 11.392900856793146,
      "grad_norm": 2.489499023623702,
      "learning_rate": 6.801249768084567e-07,
      "loss": 0.7093,
      "step": 9308
    },
    {
      "epoch": 11.394124847001224,
      "grad_norm": 1.2427983658380841,
      "learning_rate": 6.796856300771698e-07,
      "loss": 0.5366,
      "step": 9309
    },
    {
      "epoch": 11.395348837209303,
      "grad_norm": 1.5839588848534054,
      "learning_rate": 6.792464029702104e-07,
      "loss": 0.5653,
      "step": 9310
    },
    {
      "epoch": 11.39657282741738,
      "grad_norm": 2.183935233561735,
      "learning_rate": 6.788072955164429e-07,
      "loss": 1.1275,
      "step": 9311
    },
    {
      "epoch": 11.397796817625458,
      "grad_norm": 2.6073809612114474,
      "learning_rate": 6.783683077447239e-07,
      "loss": 0.5466,
      "step": 9312
    },
    {
      "epoch": 11.399020807833537,
      "grad_norm": 1.396541699922828,
      "learning_rate": 6.779294396839015e-07,
      "loss": 0.4484,
      "step": 9313
    },
    {
      "epoch": 11.400244798041616,
      "grad_norm": 2.8044411711207724,
      "learning_rate": 6.774906913628173e-07,
      "loss": 0.442,
      "step": 9314
    },
    {
      "epoch": 11.401468788249694,
      "grad_norm": 2.0390283123857063,
      "learning_rate": 6.770520628103048e-07,
      "loss": 0.7287,
      "step": 9315
    },
    {
      "epoch": 11.402692778457773,
      "grad_norm": 1.7418741331760392,
      "learning_rate": 6.766135540551871e-07,
      "loss": 0.4287,
      "step": 9316
    },
    {
      "epoch": 11.403916768665852,
      "grad_norm": 0.9766012577225778,
      "learning_rate": 6.761751651262827e-07,
      "loss": 0.4259,
      "step": 9317
    },
    {
      "epoch": 11.405140758873928,
      "grad_norm": 1.3795023581344423,
      "learning_rate": 6.757368960523996e-07,
      "loss": 0.67,
      "step": 9318
    },
    {
      "epoch": 11.406364749082007,
      "grad_norm": 2.9815500578584464,
      "learning_rate": 6.752987468623412e-07,
      "loss": 0.7354,
      "step": 9319
    },
    {
      "epoch": 11.407588739290086,
      "grad_norm": 2.0960857193191114,
      "learning_rate": 6.748607175848998e-07,
      "loss": 1.173,
      "step": 9320
    },
    {
      "epoch": 11.408812729498164,
      "grad_norm": 2.8223928325938297,
      "learning_rate": 6.744228082488616e-07,
      "loss": 0.5604,
      "step": 9321
    },
    {
      "epoch": 11.410036719706243,
      "grad_norm": 1.7528478969582482,
      "learning_rate": 6.739850188830038e-07,
      "loss": 0.5262,
      "step": 9322
    },
    {
      "epoch": 11.411260709914322,
      "grad_norm": 1.8434779777186812,
      "learning_rate": 6.735473495160969e-07,
      "loss": 1.0168,
      "step": 9323
    },
    {
      "epoch": 11.412484700122398,
      "grad_norm": 1.2691214575206564,
      "learning_rate": 6.731098001769023e-07,
      "loss": 0.7658,
      "step": 9324
    },
    {
      "epoch": 11.413708690330477,
      "grad_norm": 1.1871344284104646,
      "learning_rate": 6.726723708941745e-07,
      "loss": 0.7437,
      "step": 9325
    },
    {
      "epoch": 11.414932680538556,
      "grad_norm": 1.4489109913176736,
      "learning_rate": 6.722350616966591e-07,
      "loss": 0.9457,
      "step": 9326
    },
    {
      "epoch": 11.416156670746634,
      "grad_norm": 2.5199606476496235,
      "learning_rate": 6.717978726130966e-07,
      "loss": 0.5986,
      "step": 9327
    },
    {
      "epoch": 11.417380660954713,
      "grad_norm": 2.3108250341800822,
      "learning_rate": 6.713608036722149e-07,
      "loss": 0.5447,
      "step": 9328
    },
    {
      "epoch": 11.418604651162791,
      "grad_norm": 2.0685717782707655,
      "learning_rate": 6.709238549027375e-07,
      "loss": 0.5742,
      "step": 9329
    },
    {
      "epoch": 11.419828641370868,
      "grad_norm": 2.664088766293819,
      "learning_rate": 6.704870263333791e-07,
      "loss": 0.4217,
      "step": 9330
    },
    {
      "epoch": 11.421052631578947,
      "grad_norm": 2.537077886017719,
      "learning_rate": 6.700503179928458e-07,
      "loss": 1.0211,
      "step": 9331
    },
    {
      "epoch": 11.422276621787026,
      "grad_norm": 2.259080738592429,
      "learning_rate": 6.696137299098376e-07,
      "loss": 0.463,
      "step": 9332
    },
    {
      "epoch": 11.423500611995104,
      "grad_norm": 1.3577656594067957,
      "learning_rate": 6.691772621130457e-07,
      "loss": 0.5528,
      "step": 9333
    },
    {
      "epoch": 11.424724602203183,
      "grad_norm": 2.0495467986219253,
      "learning_rate": 6.687409146311519e-07,
      "loss": 0.5223,
      "step": 9334
    },
    {
      "epoch": 11.425948592411261,
      "grad_norm": 1.019803359664632,
      "learning_rate": 6.683046874928309e-07,
      "loss": 0.4455,
      "step": 9335
    },
    {
      "epoch": 11.427172582619338,
      "grad_norm": 1.5276721896186174,
      "learning_rate": 6.678685807267518e-07,
      "loss": 1.0796,
      "step": 9336
    },
    {
      "epoch": 11.428396572827417,
      "grad_norm": 2.6436119130361404,
      "learning_rate": 6.674325943615728e-07,
      "loss": 0.9097,
      "step": 9337
    },
    {
      "epoch": 11.429620563035495,
      "grad_norm": 1.5486684725355158,
      "learning_rate": 6.669967284259455e-07,
      "loss": 0.7146,
      "step": 9338
    },
    {
      "epoch": 11.430844553243574,
      "grad_norm": 1.5736753520620796,
      "learning_rate": 6.665609829485134e-07,
      "loss": 1.2112,
      "step": 9339
    },
    {
      "epoch": 11.432068543451653,
      "grad_norm": 2.184064177381339,
      "learning_rate": 6.661253579579121e-07,
      "loss": 0.9172,
      "step": 9340
    },
    {
      "epoch": 11.433292533659731,
      "grad_norm": 2.733677004674629,
      "learning_rate": 6.656898534827694e-07,
      "loss": 0.4599,
      "step": 9341
    },
    {
      "epoch": 11.43451652386781,
      "grad_norm": 1.5137001092022975,
      "learning_rate": 6.652544695517047e-07,
      "loss": 0.6122,
      "step": 9342
    },
    {
      "epoch": 11.435740514075887,
      "grad_norm": 1.8437874592768346,
      "learning_rate": 6.648192061933301e-07,
      "loss": 0.5806,
      "step": 9343
    },
    {
      "epoch": 11.436964504283965,
      "grad_norm": 1.4771095207972567,
      "learning_rate": 6.643840634362492e-07,
      "loss": 0.5904,
      "step": 9344
    },
    {
      "epoch": 11.438188494492044,
      "grad_norm": 1.141457399416088,
      "learning_rate": 6.639490413090585e-07,
      "loss": 0.4145,
      "step": 9345
    },
    {
      "epoch": 11.439412484700123,
      "grad_norm": 2.6422063932865876,
      "learning_rate": 6.635141398403455e-07,
      "loss": 0.3912,
      "step": 9346
    },
    {
      "epoch": 11.440636474908201,
      "grad_norm": 2.6793064550060053,
      "learning_rate": 6.630793590586906e-07,
      "loss": 0.7855,
      "step": 9347
    },
    {
      "epoch": 11.44186046511628,
      "grad_norm": 3.0440009967159574,
      "learning_rate": 6.626446989926652e-07,
      "loss": 0.6914,
      "step": 9348
    },
    {
      "epoch": 11.443084455324357,
      "grad_norm": 1.9529672859514824,
      "learning_rate": 6.622101596708349e-07,
      "loss": 0.6203,
      "step": 9349
    },
    {
      "epoch": 11.444308445532435,
      "grad_norm": 1.7416631964775362,
      "learning_rate": 6.617757411217554e-07,
      "loss": 0.4376,
      "step": 9350
    },
    {
      "epoch": 11.445532435740514,
      "grad_norm": 1.2058856693433617,
      "learning_rate": 6.61341443373976e-07,
      "loss": 0.8404,
      "step": 9351
    },
    {
      "epoch": 11.446756425948593,
      "grad_norm": 2.613322319471002,
      "learning_rate": 6.609072664560348e-07,
      "loss": 0.4262,
      "step": 9352
    },
    {
      "epoch": 11.447980416156671,
      "grad_norm": 1.131205544125502,
      "learning_rate": 6.604732103964661e-07,
      "loss": 0.4682,
      "step": 9353
    },
    {
      "epoch": 11.44920440636475,
      "grad_norm": 2.5730355731953307,
      "learning_rate": 6.600392752237945e-07,
      "loss": 0.2286,
      "step": 9354
    },
    {
      "epoch": 11.450428396572827,
      "grad_norm": 1.8761472128530874,
      "learning_rate": 6.596054609665359e-07,
      "loss": 0.9226,
      "step": 9355
    },
    {
      "epoch": 11.451652386780905,
      "grad_norm": 2.4582676566269335,
      "learning_rate": 6.591717676531991e-07,
      "loss": 0.4208,
      "step": 9356
    },
    {
      "epoch": 11.452876376988984,
      "grad_norm": 1.7868558556866694,
      "learning_rate": 6.587381953122854e-07,
      "loss": 0.4703,
      "step": 9357
    },
    {
      "epoch": 11.454100367197062,
      "grad_norm": 2.702666971179652,
      "learning_rate": 6.583047439722867e-07,
      "loss": 0.7219,
      "step": 9358
    },
    {
      "epoch": 11.455324357405141,
      "grad_norm": 1.8576364512876349,
      "learning_rate": 6.578714136616887e-07,
      "loss": 0.5408,
      "step": 9359
    },
    {
      "epoch": 11.45654834761322,
      "grad_norm": 1.1406870415609436,
      "learning_rate": 6.574382044089675e-07,
      "loss": 0.595,
      "step": 9360
    },
    {
      "epoch": 11.457772337821297,
      "grad_norm": 1.9511794120514112,
      "learning_rate": 6.57005116242592e-07,
      "loss": 0.5805,
      "step": 9361
    },
    {
      "epoch": 11.458996328029375,
      "grad_norm": 1.4294897263028947,
      "learning_rate": 6.565721491910245e-07,
      "loss": 1.2793,
      "step": 9362
    },
    {
      "epoch": 11.460220318237454,
      "grad_norm": 1.4461134639400834,
      "learning_rate": 6.561393032827165e-07,
      "loss": 0.7954,
      "step": 9363
    },
    {
      "epoch": 11.461444308445532,
      "grad_norm": 1.7847648027413354,
      "learning_rate": 6.557065785461137e-07,
      "loss": 0.5139,
      "step": 9364
    },
    {
      "epoch": 11.462668298653611,
      "grad_norm": 1.642418468034481,
      "learning_rate": 6.552739750096518e-07,
      "loss": 0.2935,
      "step": 9365
    },
    {
      "epoch": 11.46389228886169,
      "grad_norm": 1.7144186977569338,
      "learning_rate": 6.548414927017622e-07,
      "loss": 0.3684,
      "step": 9366
    },
    {
      "epoch": 11.465116279069768,
      "grad_norm": 2.566570151228328,
      "learning_rate": 6.544091316508647e-07,
      "loss": 0.6024,
      "step": 9367
    },
    {
      "epoch": 11.466340269277845,
      "grad_norm": 2.6190794824893993,
      "learning_rate": 6.539768918853729e-07,
      "loss": 0.8518,
      "step": 9368
    },
    {
      "epoch": 11.467564259485924,
      "grad_norm": 1.4362923647664918,
      "learning_rate": 6.535447734336922e-07,
      "loss": 0.6914,
      "step": 9369
    },
    {
      "epoch": 11.468788249694002,
      "grad_norm": 2.5567207056089805,
      "learning_rate": 6.531127763242182e-07,
      "loss": 0.4741,
      "step": 9370
    },
    {
      "epoch": 11.470012239902081,
      "grad_norm": 0.9368338150378875,
      "learning_rate": 6.526809005853421e-07,
      "loss": 0.463,
      "step": 9371
    },
    {
      "epoch": 11.47123623011016,
      "grad_norm": 1.357493197167332,
      "learning_rate": 6.522491462454445e-07,
      "loss": 0.5502,
      "step": 9372
    },
    {
      "epoch": 11.472460220318238,
      "grad_norm": 1.697535879445398,
      "learning_rate": 6.518175133328985e-07,
      "loss": 0.4551,
      "step": 9373
    },
    {
      "epoch": 11.473684210526315,
      "grad_norm": 2.076418045952692,
      "learning_rate": 6.513860018760698e-07,
      "loss": 0.6527,
      "step": 9374
    },
    {
      "epoch": 11.474908200734394,
      "grad_norm": 2.9402604494363027,
      "learning_rate": 6.509546119033153e-07,
      "loss": 0.6829,
      "step": 9375
    },
    {
      "epoch": 11.476132190942472,
      "grad_norm": 1.5866154278013265,
      "learning_rate": 6.505233434429847e-07,
      "loss": 0.5667,
      "step": 9376
    },
    {
      "epoch": 11.477356181150551,
      "grad_norm": 1.9411699637088407,
      "learning_rate": 6.500921965234191e-07,
      "loss": 0.5876,
      "step": 9377
    },
    {
      "epoch": 11.47858017135863,
      "grad_norm": 1.6885148853326255,
      "learning_rate": 6.496611711729514e-07,
      "loss": 1.2276,
      "step": 9378
    },
    {
      "epoch": 11.479804161566708,
      "grad_norm": 2.508540547839427,
      "learning_rate": 6.492302674199083e-07,
      "loss": 0.5117,
      "step": 9379
    },
    {
      "epoch": 11.481028151774785,
      "grad_norm": 1.5704231515551814,
      "learning_rate": 6.487994852926072e-07,
      "loss": 1.1615,
      "step": 9380
    },
    {
      "epoch": 11.482252141982864,
      "grad_norm": 1.816472301899404,
      "learning_rate": 6.483688248193562e-07,
      "loss": 0.6264,
      "step": 9381
    },
    {
      "epoch": 11.483476132190942,
      "grad_norm": 2.0342517102019415,
      "learning_rate": 6.479382860284572e-07,
      "loss": 0.9581,
      "step": 9382
    },
    {
      "epoch": 11.48470012239902,
      "grad_norm": 1.9488865219991744,
      "learning_rate": 6.475078689482029e-07,
      "loss": 0.6482,
      "step": 9383
    },
    {
      "epoch": 11.4859241126071,
      "grad_norm": 1.5686318557573427,
      "learning_rate": 6.470775736068805e-07,
      "loss": 0.4593,
      "step": 9384
    },
    {
      "epoch": 11.487148102815178,
      "grad_norm": 1.863604772546068,
      "learning_rate": 6.466474000327663e-07,
      "loss": 0.2897,
      "step": 9385
    },
    {
      "epoch": 11.488372093023255,
      "grad_norm": 2.671725793882436,
      "learning_rate": 6.462173482541306e-07,
      "loss": 0.4854,
      "step": 9386
    },
    {
      "epoch": 11.489596083231334,
      "grad_norm": 1.3471363828147445,
      "learning_rate": 6.457874182992327e-07,
      "loss": 0.3924,
      "step": 9387
    },
    {
      "epoch": 11.490820073439412,
      "grad_norm": 1.1374167441263758,
      "learning_rate": 6.453576101963282e-07,
      "loss": 0.5357,
      "step": 9388
    },
    {
      "epoch": 11.49204406364749,
      "grad_norm": 1.228177491981677,
      "learning_rate": 6.44927923973662e-07,
      "loss": 0.5079,
      "step": 9389
    },
    {
      "epoch": 11.49326805385557,
      "grad_norm": 2.646953232340129,
      "learning_rate": 6.444983596594706e-07,
      "loss": 0.3555,
      "step": 9390
    },
    {
      "epoch": 11.494492044063648,
      "grad_norm": 2.4750944415248246,
      "learning_rate": 6.440689172819845e-07,
      "loss": 0.5739,
      "step": 9391
    },
    {
      "epoch": 11.495716034271727,
      "grad_norm": 1.6597516084644672,
      "learning_rate": 6.436395968694242e-07,
      "loss": 0.4946,
      "step": 9392
    },
    {
      "epoch": 11.496940024479803,
      "grad_norm": 1.4711426170695057,
      "learning_rate": 6.432103984500035e-07,
      "loss": 0.8416,
      "step": 9393
    },
    {
      "epoch": 11.498164014687882,
      "grad_norm": 1.4167379271349267,
      "learning_rate": 6.427813220519277e-07,
      "loss": 0.6435,
      "step": 9394
    },
    {
      "epoch": 11.49938800489596,
      "grad_norm": 2.862736222518574,
      "learning_rate": 6.423523677033941e-07,
      "loss": 1.0555,
      "step": 9395
    },
    {
      "epoch": 11.50061199510404,
      "grad_norm": 2.0874055097795963,
      "learning_rate": 6.419235354325909e-07,
      "loss": 0.4371,
      "step": 9396
    },
    {
      "epoch": 11.501835985312118,
      "grad_norm": 1.7346817029848305,
      "learning_rate": 6.414948252677014e-07,
      "loss": 1.1708,
      "step": 9397
    },
    {
      "epoch": 11.503059975520197,
      "grad_norm": 1.8036958900179971,
      "learning_rate": 6.410662372368984e-07,
      "loss": 0.9831,
      "step": 9398
    },
    {
      "epoch": 11.504283965728273,
      "grad_norm": 1.5429064125064982,
      "learning_rate": 6.406377713683459e-07,
      "loss": 0.6677,
      "step": 9399
    },
    {
      "epoch": 11.505507955936352,
      "grad_norm": 1.6997807116334063,
      "learning_rate": 6.40209427690201e-07,
      "loss": 0.4261,
      "step": 9400
    },
    {
      "epoch": 11.50673194614443,
      "grad_norm": 1.6401340156260826,
      "learning_rate": 6.397812062306144e-07,
      "loss": 0.6673,
      "step": 9401
    },
    {
      "epoch": 11.50795593635251,
      "grad_norm": 3.0972981050120687,
      "learning_rate": 6.393531070177267e-07,
      "loss": 0.525,
      "step": 9402
    },
    {
      "epoch": 11.509179926560588,
      "grad_norm": 2.1329677927517623,
      "learning_rate": 6.389251300796704e-07,
      "loss": 0.3082,
      "step": 9403
    },
    {
      "epoch": 11.510403916768666,
      "grad_norm": 0.9145916739906653,
      "learning_rate": 6.384972754445718e-07,
      "loss": 0.3469,
      "step": 9404
    },
    {
      "epoch": 11.511627906976745,
      "grad_norm": 3.2843586112023253,
      "learning_rate": 6.380695431405453e-07,
      "loss": 0.3448,
      "step": 9405
    },
    {
      "epoch": 11.512851897184822,
      "grad_norm": 1.8362014002992468,
      "learning_rate": 6.376419331957028e-07,
      "loss": 1.8758,
      "step": 9406
    },
    {
      "epoch": 11.5140758873929,
      "grad_norm": 2.0648461342158537,
      "learning_rate": 6.372144456381438e-07,
      "loss": 0.7203,
      "step": 9407
    },
    {
      "epoch": 11.51529987760098,
      "grad_norm": 2.2416023371834384,
      "learning_rate": 6.367870804959616e-07,
      "loss": 0.4815,
      "step": 9408
    },
    {
      "epoch": 11.516523867809058,
      "grad_norm": 1.6502714203530948,
      "learning_rate": 6.363598377972407e-07,
      "loss": 0.4517,
      "step": 9409
    },
    {
      "epoch": 11.517747858017136,
      "grad_norm": 1.5710552233058677,
      "learning_rate": 6.359327175700583e-07,
      "loss": 0.5718,
      "step": 9410
    },
    {
      "epoch": 11.518971848225215,
      "grad_norm": 1.6922258138894823,
      "learning_rate": 6.35505719842483e-07,
      "loss": 1.6211,
      "step": 9411
    },
    {
      "epoch": 11.520195838433292,
      "grad_norm": 1.1530322472599936,
      "learning_rate": 6.350788446425754e-07,
      "loss": 0.5055,
      "step": 9412
    },
    {
      "epoch": 11.52141982864137,
      "grad_norm": 1.810848603551753,
      "learning_rate": 6.346520919983876e-07,
      "loss": 0.951,
      "step": 9413
    },
    {
      "epoch": 11.522643818849449,
      "grad_norm": 1.0803274652226387,
      "learning_rate": 6.342254619379657e-07,
      "loss": 0.5102,
      "step": 9414
    },
    {
      "epoch": 11.523867809057528,
      "grad_norm": 1.6834970943589604,
      "learning_rate": 6.337989544893461e-07,
      "loss": 0.5443,
      "step": 9415
    },
    {
      "epoch": 11.525091799265606,
      "grad_norm": 1.550240642881175,
      "learning_rate": 6.333725696805557e-07,
      "loss": 0.2205,
      "step": 9416
    },
    {
      "epoch": 11.526315789473685,
      "grad_norm": 1.7550914022539013,
      "learning_rate": 6.329463075396161e-07,
      "loss": 0.5688,
      "step": 9417
    },
    {
      "epoch": 11.527539779681762,
      "grad_norm": 2.1468247433281813,
      "learning_rate": 6.325201680945384e-07,
      "loss": 0.6483,
      "step": 9418
    },
    {
      "epoch": 11.52876376988984,
      "grad_norm": 2.1983005171120307,
      "learning_rate": 6.320941513733289e-07,
      "loss": 0.5952,
      "step": 9419
    },
    {
      "epoch": 11.529987760097919,
      "grad_norm": 1.2053465087849038,
      "learning_rate": 6.316682574039828e-07,
      "loss": 0.6691,
      "step": 9420
    },
    {
      "epoch": 11.531211750305998,
      "grad_norm": 2.849578731485106,
      "learning_rate": 6.312424862144884e-07,
      "loss": 0.463,
      "step": 9421
    },
    {
      "epoch": 11.532435740514076,
      "grad_norm": 2.8121690419913494,
      "learning_rate": 6.308168378328255e-07,
      "loss": 0.9492,
      "step": 9422
    },
    {
      "epoch": 11.533659730722155,
      "grad_norm": 2.181656274077179,
      "learning_rate": 6.303913122869668e-07,
      "loss": 0.6068,
      "step": 9423
    },
    {
      "epoch": 11.534883720930232,
      "grad_norm": 1.532927647137027,
      "learning_rate": 6.299659096048754e-07,
      "loss": 0.5583,
      "step": 9424
    },
    {
      "epoch": 11.53610771113831,
      "grad_norm": 1.8450190753702351,
      "learning_rate": 6.295406298145079e-07,
      "loss": 1.359,
      "step": 9425
    },
    {
      "epoch": 11.537331701346389,
      "grad_norm": 1.853350699501881,
      "learning_rate": 6.291154729438112e-07,
      "loss": 1.3998,
      "step": 9426
    },
    {
      "epoch": 11.538555691554468,
      "grad_norm": 2.1151419411436905,
      "learning_rate": 6.286904390207268e-07,
      "loss": 0.8428,
      "step": 9427
    },
    {
      "epoch": 11.539779681762546,
      "grad_norm": 1.757763529562232,
      "learning_rate": 6.28265528073185e-07,
      "loss": 0.5696,
      "step": 9428
    },
    {
      "epoch": 11.541003671970625,
      "grad_norm": 2.3846000553087774,
      "learning_rate": 6.278407401291092e-07,
      "loss": 0.4465,
      "step": 9429
    },
    {
      "epoch": 11.542227662178703,
      "grad_norm": 2.2684426215800335,
      "learning_rate": 6.274160752164154e-07,
      "loss": 1.012,
      "step": 9430
    },
    {
      "epoch": 11.54345165238678,
      "grad_norm": 1.3539459496677917,
      "learning_rate": 6.269915333630106e-07,
      "loss": 0.7641,
      "step": 9431
    },
    {
      "epoch": 11.544675642594859,
      "grad_norm": 1.4217083874877705,
      "learning_rate": 6.265671145967947e-07,
      "loss": 0.5006,
      "step": 9432
    },
    {
      "epoch": 11.545899632802938,
      "grad_norm": 3.6682815649729883,
      "learning_rate": 6.261428189456598e-07,
      "loss": 0.2685,
      "step": 9433
    },
    {
      "epoch": 11.547123623011016,
      "grad_norm": 0.9713195854723147,
      "learning_rate": 6.257186464374868e-07,
      "loss": 0.4117,
      "step": 9434
    },
    {
      "epoch": 11.548347613219095,
      "grad_norm": 2.290869652631053,
      "learning_rate": 6.252945971001514e-07,
      "loss": 0.4813,
      "step": 9435
    },
    {
      "epoch": 11.549571603427173,
      "grad_norm": 1.8975000771074098,
      "learning_rate": 6.248706709615216e-07,
      "loss": 0.4686,
      "step": 9436
    },
    {
      "epoch": 11.55079559363525,
      "grad_norm": 1.5233555045160905,
      "learning_rate": 6.244468680494559e-07,
      "loss": 1.0702,
      "step": 9437
    },
    {
      "epoch": 11.552019583843329,
      "grad_norm": 2.7898146098075514,
      "learning_rate": 6.240231883918046e-07,
      "loss": 0.577,
      "step": 9438
    },
    {
      "epoch": 11.553243574051407,
      "grad_norm": 1.7495884372160342,
      "learning_rate": 6.235996320164106e-07,
      "loss": 1.611,
      "step": 9439
    },
    {
      "epoch": 11.554467564259486,
      "grad_norm": 1.6200547462551997,
      "learning_rate": 6.231761989511087e-07,
      "loss": 0.7444,
      "step": 9440
    },
    {
      "epoch": 11.555691554467565,
      "grad_norm": 2.0260390656487703,
      "learning_rate": 6.22752889223725e-07,
      "loss": 1.1356,
      "step": 9441
    },
    {
      "epoch": 11.556915544675643,
      "grad_norm": 2.082440265072196,
      "learning_rate": 6.223297028620779e-07,
      "loss": 1.1483,
      "step": 9442
    },
    {
      "epoch": 11.55813953488372,
      "grad_norm": 1.4749237126467665,
      "learning_rate": 6.219066398939777e-07,
      "loss": 0.3574,
      "step": 9443
    },
    {
      "epoch": 11.559363525091799,
      "grad_norm": 1.761999904525041,
      "learning_rate": 6.214837003472265e-07,
      "loss": 0.929,
      "step": 9444
    },
    {
      "epoch": 11.560587515299877,
      "grad_norm": 2.77295256341615,
      "learning_rate": 6.210608842496186e-07,
      "loss": 0.4959,
      "step": 9445
    },
    {
      "epoch": 11.561811505507956,
      "grad_norm": 2.149433350932017,
      "learning_rate": 6.206381916289394e-07,
      "loss": 0.5084,
      "step": 9446
    },
    {
      "epoch": 11.563035495716035,
      "grad_norm": 1.1695936472661572,
      "learning_rate": 6.202156225129671e-07,
      "loss": 0.6288,
      "step": 9447
    },
    {
      "epoch": 11.564259485924113,
      "grad_norm": 1.8714435745461542,
      "learning_rate": 6.197931769294707e-07,
      "loss": 0.7052,
      "step": 9448
    },
    {
      "epoch": 11.56548347613219,
      "grad_norm": 1.4157664770669391,
      "learning_rate": 6.193708549062127e-07,
      "loss": 0.3643,
      "step": 9449
    },
    {
      "epoch": 11.566707466340269,
      "grad_norm": 1.722963581819155,
      "learning_rate": 6.189486564709463e-07,
      "loss": 0.5053,
      "step": 9450
    },
    {
      "epoch": 11.567931456548347,
      "grad_norm": 1.7248667980968144,
      "learning_rate": 6.185265816514174e-07,
      "loss": 0.4652,
      "step": 9451
    },
    {
      "epoch": 11.569155446756426,
      "grad_norm": 2.1274036119049824,
      "learning_rate": 6.181046304753611e-07,
      "loss": 0.8085,
      "step": 9452
    },
    {
      "epoch": 11.570379436964505,
      "grad_norm": 1.0947370160279923,
      "learning_rate": 6.176828029705084e-07,
      "loss": 0.7068,
      "step": 9453
    },
    {
      "epoch": 11.571603427172583,
      "grad_norm": 1.1292977135952094,
      "learning_rate": 6.172610991645797e-07,
      "loss": 0.5063,
      "step": 9454
    },
    {
      "epoch": 11.572827417380662,
      "grad_norm": 1.6530603780957207,
      "learning_rate": 6.16839519085288e-07,
      "loss": 0.98,
      "step": 9455
    },
    {
      "epoch": 11.574051407588739,
      "grad_norm": 2.4215853500892095,
      "learning_rate": 6.164180627603375e-07,
      "loss": 1.1965,
      "step": 9456
    },
    {
      "epoch": 11.575275397796817,
      "grad_norm": 2.6556120802929737,
      "learning_rate": 6.159967302174252e-07,
      "loss": 0.3949,
      "step": 9457
    },
    {
      "epoch": 11.576499388004896,
      "grad_norm": 2.739246997380678,
      "learning_rate": 6.155755214842393e-07,
      "loss": 0.8217,
      "step": 9458
    },
    {
      "epoch": 11.577723378212974,
      "grad_norm": 0.9752810173512048,
      "learning_rate": 6.151544365884601e-07,
      "loss": 0.4922,
      "step": 9459
    },
    {
      "epoch": 11.578947368421053,
      "grad_norm": 1.8472663396139954,
      "learning_rate": 6.147334755577597e-07,
      "loss": 1.4536,
      "step": 9460
    },
    {
      "epoch": 11.580171358629132,
      "grad_norm": 2.0649290709130694,
      "learning_rate": 6.143126384198015e-07,
      "loss": 1.067,
      "step": 9461
    },
    {
      "epoch": 11.581395348837209,
      "grad_norm": 1.1783126247932845,
      "learning_rate": 6.138919252022435e-07,
      "loss": 0.4426,
      "step": 9462
    },
    {
      "epoch": 11.582619339045287,
      "grad_norm": 1.8403710362480752,
      "learning_rate": 6.134713359327313e-07,
      "loss": 0.4435,
      "step": 9463
    },
    {
      "epoch": 11.583843329253366,
      "grad_norm": 1.2876206364339005,
      "learning_rate": 6.130508706389049e-07,
      "loss": 0.3846,
      "step": 9464
    },
    {
      "epoch": 11.585067319461444,
      "grad_norm": 1.9883212504611005,
      "learning_rate": 6.126305293483955e-07,
      "loss": 1.046,
      "step": 9465
    },
    {
      "epoch": 11.586291309669523,
      "grad_norm": 2.0579301321725287,
      "learning_rate": 6.122103120888274e-07,
      "loss": 0.4081,
      "step": 9466
    },
    {
      "epoch": 11.587515299877602,
      "grad_norm": 1.9039768957820018,
      "learning_rate": 6.117902188878149e-07,
      "loss": 0.503,
      "step": 9467
    },
    {
      "epoch": 11.588739290085678,
      "grad_norm": 2.922871407513824,
      "learning_rate": 6.113702497729657e-07,
      "loss": 0.4044,
      "step": 9468
    },
    {
      "epoch": 11.589963280293757,
      "grad_norm": 2.655156423371508,
      "learning_rate": 6.109504047718784e-07,
      "loss": 0.3875,
      "step": 9469
    },
    {
      "epoch": 11.591187270501836,
      "grad_norm": 2.147228957227427,
      "learning_rate": 6.105306839121422e-07,
      "loss": 0.3449,
      "step": 9470
    },
    {
      "epoch": 11.592411260709914,
      "grad_norm": 1.8738959464910643,
      "learning_rate": 6.101110872213414e-07,
      "loss": 1.5882,
      "step": 9471
    },
    {
      "epoch": 11.593635250917993,
      "grad_norm": 1.3890029956696825,
      "learning_rate": 6.096916147270498e-07,
      "loss": 1.329,
      "step": 9472
    },
    {
      "epoch": 11.594859241126072,
      "grad_norm": 2.581151621586061,
      "learning_rate": 6.092722664568335e-07,
      "loss": 0.4603,
      "step": 9473
    },
    {
      "epoch": 11.596083231334148,
      "grad_norm": 1.7458444187393871,
      "learning_rate": 6.088530424382505e-07,
      "loss": 1.2626,
      "step": 9474
    },
    {
      "epoch": 11.597307221542227,
      "grad_norm": 2.4913860679633095,
      "learning_rate": 6.084339426988508e-07,
      "loss": 0.6048,
      "step": 9475
    },
    {
      "epoch": 11.598531211750306,
      "grad_norm": 1.3654420454040583,
      "learning_rate": 6.080149672661759e-07,
      "loss": 0.5232,
      "step": 9476
    },
    {
      "epoch": 11.599755201958384,
      "grad_norm": 1.1377069029551743,
      "learning_rate": 6.075961161677593e-07,
      "loss": 0.5926,
      "step": 9477
    },
    {
      "epoch": 11.600979192166463,
      "grad_norm": 2.0680513647202674,
      "learning_rate": 6.071773894311256e-07,
      "loss": 0.592,
      "step": 9478
    },
    {
      "epoch": 11.602203182374542,
      "grad_norm": 1.8833169021707632,
      "learning_rate": 6.067587870837937e-07,
      "loss": 0.5276,
      "step": 9479
    },
    {
      "epoch": 11.60342717258262,
      "grad_norm": 1.236447288569973,
      "learning_rate": 6.063403091532724e-07,
      "loss": 0.8153,
      "step": 9480
    },
    {
      "epoch": 11.604651162790697,
      "grad_norm": 1.9767530223068288,
      "learning_rate": 6.059219556670612e-07,
      "loss": 0.5142,
      "step": 9481
    },
    {
      "epoch": 11.605875152998776,
      "grad_norm": 2.4242420091064703,
      "learning_rate": 6.055037266526534e-07,
      "loss": 0.4111,
      "step": 9482
    },
    {
      "epoch": 11.607099143206854,
      "grad_norm": 1.0854337592879344,
      "learning_rate": 6.050856221375323e-07,
      "loss": 0.5455,
      "step": 9483
    },
    {
      "epoch": 11.608323133414933,
      "grad_norm": 1.4044124752527214,
      "learning_rate": 6.046676421491765e-07,
      "loss": 0.3065,
      "step": 9484
    },
    {
      "epoch": 11.609547123623011,
      "grad_norm": 1.1900528172932887,
      "learning_rate": 6.042497867150527e-07,
      "loss": 0.479,
      "step": 9485
    },
    {
      "epoch": 11.61077111383109,
      "grad_norm": 1.1558885736567517,
      "learning_rate": 6.03832055862621e-07,
      "loss": 0.6122,
      "step": 9486
    },
    {
      "epoch": 11.611995104039167,
      "grad_norm": 1.1350885562527167,
      "learning_rate": 6.034144496193331e-07,
      "loss": 0.3067,
      "step": 9487
    },
    {
      "epoch": 11.613219094247246,
      "grad_norm": 2.4384332191331803,
      "learning_rate": 6.029969680126326e-07,
      "loss": 1.0754,
      "step": 9488
    },
    {
      "epoch": 11.614443084455324,
      "grad_norm": 1.7748893006656772,
      "learning_rate": 6.02579611069955e-07,
      "loss": 0.4986,
      "step": 9489
    },
    {
      "epoch": 11.615667074663403,
      "grad_norm": 3.107767185736674,
      "learning_rate": 6.021623788187273e-07,
      "loss": 0.4424,
      "step": 9490
    },
    {
      "epoch": 11.616891064871481,
      "grad_norm": 2.284322654872905,
      "learning_rate": 6.017452712863684e-07,
      "loss": 0.9507,
      "step": 9491
    },
    {
      "epoch": 11.61811505507956,
      "grad_norm": 2.0001695079378905,
      "learning_rate": 6.013282885002889e-07,
      "loss": 0.5364,
      "step": 9492
    },
    {
      "epoch": 11.619339045287639,
      "grad_norm": 2.391370226821474,
      "learning_rate": 6.009114304878916e-07,
      "loss": 0.6028,
      "step": 9493
    },
    {
      "epoch": 11.620563035495715,
      "grad_norm": 1.0988512118208684,
      "learning_rate": 6.00494697276571e-07,
      "loss": 0.506,
      "step": 9494
    },
    {
      "epoch": 11.621787025703794,
      "grad_norm": 2.8549939380085387,
      "learning_rate": 6.000780888937133e-07,
      "loss": 0.5747,
      "step": 9495
    },
    {
      "epoch": 11.623011015911873,
      "grad_norm": 1.8255568636601338,
      "learning_rate": 5.996616053666951e-07,
      "loss": 0.9275,
      "step": 9496
    },
    {
      "epoch": 11.624235006119951,
      "grad_norm": 2.686001985120119,
      "learning_rate": 5.992452467228884e-07,
      "loss": 0.5098,
      "step": 9497
    },
    {
      "epoch": 11.62545899632803,
      "grad_norm": 2.2384048670298244,
      "learning_rate": 5.988290129896543e-07,
      "loss": 0.5062,
      "step": 9498
    },
    {
      "epoch": 11.626682986536107,
      "grad_norm": 1.7862959659658524,
      "learning_rate": 5.984129041943448e-07,
      "loss": 0.6569,
      "step": 9499
    },
    {
      "epoch": 11.627906976744185,
      "grad_norm": 2.5197831258493535,
      "learning_rate": 5.979969203643049e-07,
      "loss": 0.4562,
      "step": 9500
    },
    {
      "epoch": 11.629130966952264,
      "grad_norm": 2.3411191503359503,
      "learning_rate": 5.975810615268731e-07,
      "loss": 0.5453,
      "step": 9501
    },
    {
      "epoch": 11.630354957160343,
      "grad_norm": 2.1608442336168947,
      "learning_rate": 5.971653277093778e-07,
      "loss": 0.3846,
      "step": 9502
    },
    {
      "epoch": 11.631578947368421,
      "grad_norm": 2.2959339098494604,
      "learning_rate": 5.967497189391385e-07,
      "loss": 0.615,
      "step": 9503
    },
    {
      "epoch": 11.6328029375765,
      "grad_norm": 2.6585807592232187,
      "learning_rate": 5.963342352434684e-07,
      "loss": 0.3341,
      "step": 9504
    },
    {
      "epoch": 11.634026927784578,
      "grad_norm": 3.47811285495677,
      "learning_rate": 5.959188766496713e-07,
      "loss": 0.467,
      "step": 9505
    },
    {
      "epoch": 11.635250917992655,
      "grad_norm": 1.1703020743265737,
      "learning_rate": 5.955036431850425e-07,
      "loss": 1.1966,
      "step": 9506
    },
    {
      "epoch": 11.636474908200734,
      "grad_norm": 1.8603818613953647,
      "learning_rate": 5.950885348768704e-07,
      "loss": 0.9084,
      "step": 9507
    },
    {
      "epoch": 11.637698898408813,
      "grad_norm": 3.589560565253967,
      "learning_rate": 5.946735517524337e-07,
      "loss": 0.3445,
      "step": 9508
    },
    {
      "epoch": 11.638922888616891,
      "grad_norm": 2.6745580152636625,
      "learning_rate": 5.942586938390041e-07,
      "loss": 0.8505,
      "step": 9509
    },
    {
      "epoch": 11.64014687882497,
      "grad_norm": 2.3572813334675113,
      "learning_rate": 5.938439611638442e-07,
      "loss": 0.814,
      "step": 9510
    },
    {
      "epoch": 11.641370869033048,
      "grad_norm": 2.3652719622979563,
      "learning_rate": 5.934293537542085e-07,
      "loss": 0.4239,
      "step": 9511
    },
    {
      "epoch": 11.642594859241125,
      "grad_norm": 2.0078424270760995,
      "learning_rate": 5.93014871637344e-07,
      "loss": 0.9549,
      "step": 9512
    },
    {
      "epoch": 11.643818849449204,
      "grad_norm": 1.6909670289061323,
      "learning_rate": 5.926005148404881e-07,
      "loss": 0.5153,
      "step": 9513
    },
    {
      "epoch": 11.645042839657282,
      "grad_norm": 1.8099705798867596,
      "learning_rate": 5.921862833908721e-07,
      "loss": 0.4736,
      "step": 9514
    },
    {
      "epoch": 11.646266829865361,
      "grad_norm": 2.0170978190786197,
      "learning_rate": 5.91772177315717e-07,
      "loss": 0.3611,
      "step": 9515
    },
    {
      "epoch": 11.64749082007344,
      "grad_norm": 1.4939581817327328,
      "learning_rate": 5.913581966422368e-07,
      "loss": 1.4075,
      "step": 9516
    },
    {
      "epoch": 11.648714810281518,
      "grad_norm": 1.1667402584822508,
      "learning_rate": 5.909443413976351e-07,
      "loss": 0.6179,
      "step": 9517
    },
    {
      "epoch": 11.649938800489597,
      "grad_norm": 1.7502141504277084,
      "learning_rate": 5.905306116091111e-07,
      "loss": 0.3021,
      "step": 9518
    },
    {
      "epoch": 11.651162790697674,
      "grad_norm": 1.1512375490074636,
      "learning_rate": 5.901170073038523e-07,
      "loss": 0.6378,
      "step": 9519
    },
    {
      "epoch": 11.652386780905752,
      "grad_norm": 1.8336913782993471,
      "learning_rate": 5.897035285090399e-07,
      "loss": 0.6854,
      "step": 9520
    },
    {
      "epoch": 11.653610771113831,
      "grad_norm": 1.87926866171637,
      "learning_rate": 5.892901752518457e-07,
      "loss": 1.3791,
      "step": 9521
    },
    {
      "epoch": 11.65483476132191,
      "grad_norm": 1.429720210691919,
      "learning_rate": 5.888769475594341e-07,
      "loss": 1.5557,
      "step": 9522
    },
    {
      "epoch": 11.656058751529988,
      "grad_norm": 3.093397373363434,
      "learning_rate": 5.884638454589609e-07,
      "loss": 0.5076,
      "step": 9523
    },
    {
      "epoch": 11.657282741738067,
      "grad_norm": 1.557990771263695,
      "learning_rate": 5.880508689775732e-07,
      "loss": 0.6164,
      "step": 9524
    },
    {
      "epoch": 11.658506731946144,
      "grad_norm": 2.079403310944342,
      "learning_rate": 5.876380181424107e-07,
      "loss": 0.5289,
      "step": 9525
    },
    {
      "epoch": 11.659730722154222,
      "grad_norm": 2.923485543074061,
      "learning_rate": 5.872252929806035e-07,
      "loss": 0.8196,
      "step": 9526
    },
    {
      "epoch": 11.660954712362301,
      "grad_norm": 2.1328601495085557,
      "learning_rate": 5.868126935192766e-07,
      "loss": 0.3816,
      "step": 9527
    },
    {
      "epoch": 11.66217870257038,
      "grad_norm": 2.525772039425355,
      "learning_rate": 5.864002197855425e-07,
      "loss": 0.5339,
      "step": 9528
    },
    {
      "epoch": 11.663402692778458,
      "grad_norm": 2.236291198492012,
      "learning_rate": 5.859878718065082e-07,
      "loss": 0.3934,
      "step": 9529
    },
    {
      "epoch": 11.664626682986537,
      "grad_norm": 1.5515494548596904,
      "learning_rate": 5.855756496092704e-07,
      "loss": 0.7119,
      "step": 9530
    },
    {
      "epoch": 11.665850673194614,
      "grad_norm": 2.104289627859134,
      "learning_rate": 5.85163553220921e-07,
      "loss": 0.7284,
      "step": 9531
    },
    {
      "epoch": 11.667074663402692,
      "grad_norm": 1.5653892279137194,
      "learning_rate": 5.847515826685401e-07,
      "loss": 0.6088,
      "step": 9532
    },
    {
      "epoch": 11.668298653610771,
      "grad_norm": 1.8814160373624913,
      "learning_rate": 5.843397379792021e-07,
      "loss": 1.5593,
      "step": 9533
    },
    {
      "epoch": 11.66952264381885,
      "grad_norm": 2.1035337091764,
      "learning_rate": 5.839280191799704e-07,
      "loss": 0.5051,
      "step": 9534
    },
    {
      "epoch": 11.670746634026928,
      "grad_norm": 1.3795937285578812,
      "learning_rate": 5.835164262979013e-07,
      "loss": 0.3865,
      "step": 9535
    },
    {
      "epoch": 11.671970624235007,
      "grad_norm": 2.4860021038401476,
      "learning_rate": 5.831049593600447e-07,
      "loss": 0.4827,
      "step": 9536
    },
    {
      "epoch": 11.673194614443084,
      "grad_norm": 1.7179572497580777,
      "learning_rate": 5.826936183934403e-07,
      "loss": 0.5199,
      "step": 9537
    },
    {
      "epoch": 11.674418604651162,
      "grad_norm": 2.191317693430263,
      "learning_rate": 5.822824034251195e-07,
      "loss": 0.5101,
      "step": 9538
    },
    {
      "epoch": 11.67564259485924,
      "grad_norm": 2.3327127561215573,
      "learning_rate": 5.818713144821062e-07,
      "loss": 1.0274,
      "step": 9539
    },
    {
      "epoch": 11.67686658506732,
      "grad_norm": 1.6289944897805386,
      "learning_rate": 5.814603515914153e-07,
      "loss": 0.4738,
      "step": 9540
    },
    {
      "epoch": 11.678090575275398,
      "grad_norm": 2.0401840764728836,
      "learning_rate": 5.810495147800538e-07,
      "loss": 1.0025,
      "step": 9541
    },
    {
      "epoch": 11.679314565483477,
      "grad_norm": 0.9675589717544911,
      "learning_rate": 5.806388040750205e-07,
      "loss": 0.4387,
      "step": 9542
    },
    {
      "epoch": 11.680538555691555,
      "grad_norm": 2.903803409098663,
      "learning_rate": 5.80228219503306e-07,
      "loss": 0.3527,
      "step": 9543
    },
    {
      "epoch": 11.681762545899632,
      "grad_norm": 1.3156137160977062,
      "learning_rate": 5.798177610918912e-07,
      "loss": 0.6545,
      "step": 9544
    },
    {
      "epoch": 11.68298653610771,
      "grad_norm": 2.786977548137845,
      "learning_rate": 5.794074288677526e-07,
      "loss": 0.5107,
      "step": 9545
    },
    {
      "epoch": 11.68421052631579,
      "grad_norm": 2.3185822128760805,
      "learning_rate": 5.78997222857853e-07,
      "loss": 0.5251,
      "step": 9546
    },
    {
      "epoch": 11.685434516523868,
      "grad_norm": 2.649237750016963,
      "learning_rate": 5.785871430891508e-07,
      "loss": 0.531,
      "step": 9547
    },
    {
      "epoch": 11.686658506731947,
      "grad_norm": 1.9812315258378146,
      "learning_rate": 5.781771895885937e-07,
      "loss": 1.3855,
      "step": 9548
    },
    {
      "epoch": 11.687882496940025,
      "grad_norm": 1.727502148485613,
      "learning_rate": 5.777673623831245e-07,
      "loss": 0.5599,
      "step": 9549
    },
    {
      "epoch": 11.689106487148102,
      "grad_norm": 1.7332818529486536,
      "learning_rate": 5.773576614996743e-07,
      "loss": 1.3305,
      "step": 9550
    },
    {
      "epoch": 11.69033047735618,
      "grad_norm": 2.225294020139932,
      "learning_rate": 5.769480869651678e-07,
      "loss": 0.4306,
      "step": 9551
    },
    {
      "epoch": 11.69155446756426,
      "grad_norm": 2.4005172831474253,
      "learning_rate": 5.765386388065189e-07,
      "loss": 0.6671,
      "step": 9552
    },
    {
      "epoch": 11.692778457772338,
      "grad_norm": 1.029971418326937,
      "learning_rate": 5.761293170506368e-07,
      "loss": 0.4401,
      "step": 9553
    },
    {
      "epoch": 11.694002447980417,
      "grad_norm": 1.4590314649251621,
      "learning_rate": 5.757201217244201e-07,
      "loss": 0.9655,
      "step": 9554
    },
    {
      "epoch": 11.695226438188495,
      "grad_norm": 1.710504564042804,
      "learning_rate": 5.753110528547596e-07,
      "loss": 1.0798,
      "step": 9555
    },
    {
      "epoch": 11.696450428396572,
      "grad_norm": 1.1015609029008078,
      "learning_rate": 5.749021104685379e-07,
      "loss": 0.5277,
      "step": 9556
    },
    {
      "epoch": 11.69767441860465,
      "grad_norm": 2.7549988942225436,
      "learning_rate": 5.74493294592629e-07,
      "loss": 0.4693,
      "step": 9557
    },
    {
      "epoch": 11.69889840881273,
      "grad_norm": 2.722589493264937,
      "learning_rate": 5.740846052538986e-07,
      "loss": 0.5858,
      "step": 9558
    },
    {
      "epoch": 11.700122399020808,
      "grad_norm": 0.9665818710005806,
      "learning_rate": 5.736760424792048e-07,
      "loss": 0.3887,
      "step": 9559
    },
    {
      "epoch": 11.701346389228886,
      "grad_norm": 1.6589136810326994,
      "learning_rate": 5.732676062953963e-07,
      "loss": 1.0112,
      "step": 9560
    },
    {
      "epoch": 11.702570379436965,
      "grad_norm": 1.7588763839172836,
      "learning_rate": 5.728592967293137e-07,
      "loss": 0.6186,
      "step": 9561
    },
    {
      "epoch": 11.703794369645042,
      "grad_norm": 2.0581047604530016,
      "learning_rate": 5.724511138077912e-07,
      "loss": 0.5881,
      "step": 9562
    },
    {
      "epoch": 11.70501835985312,
      "grad_norm": 1.7002094710676772,
      "learning_rate": 5.720430575576516e-07,
      "loss": 0.5492,
      "step": 9563
    },
    {
      "epoch": 11.7062423500612,
      "grad_norm": 1.6530125895485392,
      "learning_rate": 5.716351280057111e-07,
      "loss": 0.9325,
      "step": 9564
    },
    {
      "epoch": 11.707466340269278,
      "grad_norm": 2.2769363773842866,
      "learning_rate": 5.712273251787769e-07,
      "loss": 0.551,
      "step": 9565
    },
    {
      "epoch": 11.708690330477356,
      "grad_norm": 1.977321946021986,
      "learning_rate": 5.708196491036494e-07,
      "loss": 0.6089,
      "step": 9566
    },
    {
      "epoch": 11.709914320685435,
      "grad_norm": 2.3933127931980858,
      "learning_rate": 5.704120998071192e-07,
      "loss": 0.5144,
      "step": 9567
    },
    {
      "epoch": 11.711138310893514,
      "grad_norm": 1.6914215646448734,
      "learning_rate": 5.700046773159685e-07,
      "loss": 0.8486,
      "step": 9568
    },
    {
      "epoch": 11.71236230110159,
      "grad_norm": 1.5643157705745718,
      "learning_rate": 5.695973816569728e-07,
      "loss": 0.9091,
      "step": 9569
    },
    {
      "epoch": 11.713586291309669,
      "grad_norm": 1.7978107033483646,
      "learning_rate": 5.691902128568954e-07,
      "loss": 0.4454,
      "step": 9570
    },
    {
      "epoch": 11.714810281517748,
      "grad_norm": 1.8838832532215717,
      "learning_rate": 5.687831709424965e-07,
      "loss": 1.1693,
      "step": 9571
    },
    {
      "epoch": 11.716034271725826,
      "grad_norm": 1.1343877244493665,
      "learning_rate": 5.683762559405245e-07,
      "loss": 0.3605,
      "step": 9572
    },
    {
      "epoch": 11.717258261933905,
      "grad_norm": 1.9333511989666385,
      "learning_rate": 5.679694678777206e-07,
      "loss": 0.9604,
      "step": 9573
    },
    {
      "epoch": 11.718482252141984,
      "grad_norm": 1.8553587692472726,
      "learning_rate": 5.675628067808169e-07,
      "loss": 1.1862,
      "step": 9574
    },
    {
      "epoch": 11.71970624235006,
      "grad_norm": 3.0291618423154607,
      "learning_rate": 5.671562726765381e-07,
      "loss": 0.486,
      "step": 9575
    },
    {
      "epoch": 11.720930232558139,
      "grad_norm": 2.1280383493704855,
      "learning_rate": 5.667498655916002e-07,
      "loss": 1.1908,
      "step": 9576
    },
    {
      "epoch": 11.722154222766218,
      "grad_norm": 1.5382589574316063,
      "learning_rate": 5.6634358555271e-07,
      "loss": 0.811,
      "step": 9577
    },
    {
      "epoch": 11.723378212974296,
      "grad_norm": 2.326124098616378,
      "learning_rate": 5.659374325865668e-07,
      "loss": 1.3312,
      "step": 9578
    },
    {
      "epoch": 11.724602203182375,
      "grad_norm": 2.9264766995794607,
      "learning_rate": 5.655314067198628e-07,
      "loss": 0.206,
      "step": 9579
    },
    {
      "epoch": 11.725826193390454,
      "grad_norm": 1.5824709748210595,
      "learning_rate": 5.6512550797928e-07,
      "loss": 0.9187,
      "step": 9580
    },
    {
      "epoch": 11.727050183598532,
      "grad_norm": 1.9987175533476114,
      "learning_rate": 5.647197363914919e-07,
      "loss": 0.9907,
      "step": 9581
    },
    {
      "epoch": 11.728274173806609,
      "grad_norm": 1.3964943902835223,
      "learning_rate": 5.643140919831644e-07,
      "loss": 0.7094,
      "step": 9582
    },
    {
      "epoch": 11.729498164014688,
      "grad_norm": 1.5759832362872677,
      "learning_rate": 5.639085747809544e-07,
      "loss": 0.424,
      "step": 9583
    },
    {
      "epoch": 11.730722154222766,
      "grad_norm": 1.8150244741887251,
      "learning_rate": 5.635031848115124e-07,
      "loss": 0.4057,
      "step": 9584
    },
    {
      "epoch": 11.731946144430845,
      "grad_norm": 1.8934606554631424,
      "learning_rate": 5.630979221014787e-07,
      "loss": 0.4852,
      "step": 9585
    },
    {
      "epoch": 11.733170134638923,
      "grad_norm": 2.1657169690417035,
      "learning_rate": 5.626927866774857e-07,
      "loss": 1.1832,
      "step": 9586
    },
    {
      "epoch": 11.734394124847,
      "grad_norm": 2.3388524037771856,
      "learning_rate": 5.622877785661565e-07,
      "loss": 0.4854,
      "step": 9587
    },
    {
      "epoch": 11.735618115055079,
      "grad_norm": 2.726010922472211,
      "learning_rate": 5.61882897794108e-07,
      "loss": 0.2732,
      "step": 9588
    },
    {
      "epoch": 11.736842105263158,
      "grad_norm": 1.5541067439079779,
      "learning_rate": 5.614781443879463e-07,
      "loss": 0.5477,
      "step": 9589
    },
    {
      "epoch": 11.738066095471236,
      "grad_norm": 1.7496871356654915,
      "learning_rate": 5.610735183742713e-07,
      "loss": 0.9033,
      "step": 9590
    },
    {
      "epoch": 11.739290085679315,
      "grad_norm": 2.838162018131953,
      "learning_rate": 5.60669019779673e-07,
      "loss": 0.4372,
      "step": 9591
    },
    {
      "epoch": 11.740514075887393,
      "grad_norm": 2.533352684473776,
      "learning_rate": 5.602646486307334e-07,
      "loss": 0.4156,
      "step": 9592
    },
    {
      "epoch": 11.741738066095472,
      "grad_norm": 1.4886163409163444,
      "learning_rate": 5.598604049540268e-07,
      "loss": 0.6859,
      "step": 9593
    },
    {
      "epoch": 11.742962056303549,
      "grad_norm": 2.272022435309857,
      "learning_rate": 5.594562887761182e-07,
      "loss": 0.4707,
      "step": 9594
    },
    {
      "epoch": 11.744186046511627,
      "grad_norm": 1.439255800322469,
      "learning_rate": 5.590523001235646e-07,
      "loss": 1.2692,
      "step": 9595
    },
    {
      "epoch": 11.745410036719706,
      "grad_norm": 2.74586383663098,
      "learning_rate": 5.586484390229141e-07,
      "loss": 0.8912,
      "step": 9596
    },
    {
      "epoch": 11.746634026927785,
      "grad_norm": 1.4566049036829696,
      "learning_rate": 5.582447055007085e-07,
      "loss": 0.4534,
      "step": 9597
    },
    {
      "epoch": 11.747858017135863,
      "grad_norm": 2.068558175716681,
      "learning_rate": 5.578410995834793e-07,
      "loss": 0.3883,
      "step": 9598
    },
    {
      "epoch": 11.749082007343942,
      "grad_norm": 1.7657748824746549,
      "learning_rate": 5.57437621297749e-07,
      "loss": 0.7278,
      "step": 9599
    },
    {
      "epoch": 11.750305997552019,
      "grad_norm": 3.245117604734685,
      "learning_rate": 5.570342706700324e-07,
      "loss": 0.3063,
      "step": 9600
    },
    {
      "epoch": 11.751529987760097,
      "grad_norm": 1.9554461881287157,
      "learning_rate": 5.566310477268378e-07,
      "loss": 0.5278,
      "step": 9601
    },
    {
      "epoch": 11.752753977968176,
      "grad_norm": 1.996805271592485,
      "learning_rate": 5.562279524946623e-07,
      "loss": 0.9023,
      "step": 9602
    },
    {
      "epoch": 11.753977968176255,
      "grad_norm": 1.1559807910758335,
      "learning_rate": 5.558249849999967e-07,
      "loss": 0.602,
      "step": 9603
    },
    {
      "epoch": 11.755201958384333,
      "grad_norm": 1.1923324390588945,
      "learning_rate": 5.554221452693217e-07,
      "loss": 0.5679,
      "step": 9604
    },
    {
      "epoch": 11.756425948592412,
      "grad_norm": 1.403489346857492,
      "learning_rate": 5.550194333291112e-07,
      "loss": 0.7361,
      "step": 9605
    },
    {
      "epoch": 11.75764993880049,
      "grad_norm": 2.0561954092293897,
      "learning_rate": 5.546168492058293e-07,
      "loss": 0.9224,
      "step": 9606
    },
    {
      "epoch": 11.758873929008567,
      "grad_norm": 1.4788114971130026,
      "learning_rate": 5.542143929259327e-07,
      "loss": 0.6173,
      "step": 9607
    },
    {
      "epoch": 11.760097919216646,
      "grad_norm": 1.3344460188646743,
      "learning_rate": 5.538120645158692e-07,
      "loss": 0.5695,
      "step": 9608
    },
    {
      "epoch": 11.761321909424725,
      "grad_norm": 2.169798093171633,
      "learning_rate": 5.534098640020785e-07,
      "loss": 0.5026,
      "step": 9609
    },
    {
      "epoch": 11.762545899632803,
      "grad_norm": 2.8968636319003047,
      "learning_rate": 5.530077914109916e-07,
      "loss": 1.0183,
      "step": 9610
    },
    {
      "epoch": 11.763769889840882,
      "grad_norm": 1.5914225178824548,
      "learning_rate": 5.52605846769031e-07,
      "loss": 0.5329,
      "step": 9611
    },
    {
      "epoch": 11.76499388004896,
      "grad_norm": 2.757851758052305,
      "learning_rate": 5.522040301026114e-07,
      "loss": 0.4293,
      "step": 9612
    },
    {
      "epoch": 11.766217870257037,
      "grad_norm": 1.833039035028096,
      "learning_rate": 5.518023414381376e-07,
      "loss": 0.4645,
      "step": 9613
    },
    {
      "epoch": 11.767441860465116,
      "grad_norm": 2.2455754329775863,
      "learning_rate": 5.514007808020091e-07,
      "loss": 0.5719,
      "step": 9614
    },
    {
      "epoch": 11.768665850673194,
      "grad_norm": 2.534834805176596,
      "learning_rate": 5.509993482206136e-07,
      "loss": 0.5334,
      "step": 9615
    },
    {
      "epoch": 11.769889840881273,
      "grad_norm": 2.0622556236026863,
      "learning_rate": 5.505980437203329e-07,
      "loss": 0.5421,
      "step": 9616
    },
    {
      "epoch": 11.771113831089352,
      "grad_norm": 1.530941806701124,
      "learning_rate": 5.501968673275371e-07,
      "loss": 0.8066,
      "step": 9617
    },
    {
      "epoch": 11.77233782129743,
      "grad_norm": 2.1057014525580606,
      "learning_rate": 5.497958190685921e-07,
      "loss": 0.8304,
      "step": 9618
    },
    {
      "epoch": 11.773561811505507,
      "grad_norm": 2.8724926569680633,
      "learning_rate": 5.493948989698525e-07,
      "loss": 0.7937,
      "step": 9619
    },
    {
      "epoch": 11.774785801713586,
      "grad_norm": 1.6279512007918664,
      "learning_rate": 5.489941070576651e-07,
      "loss": 1.0397,
      "step": 9620
    },
    {
      "epoch": 11.776009791921664,
      "grad_norm": 1.7321569042192375,
      "learning_rate": 5.485934433583689e-07,
      "loss": 0.4275,
      "step": 9621
    },
    {
      "epoch": 11.777233782129743,
      "grad_norm": 2.1659002004250443,
      "learning_rate": 5.481929078982939e-07,
      "loss": 0.571,
      "step": 9622
    },
    {
      "epoch": 11.778457772337822,
      "grad_norm": 1.6172109566542565,
      "learning_rate": 5.477925007037619e-07,
      "loss": 0.4953,
      "step": 9623
    },
    {
      "epoch": 11.7796817625459,
      "grad_norm": 1.3366547369943436,
      "learning_rate": 5.473922218010857e-07,
      "loss": 1.087,
      "step": 9624
    },
    {
      "epoch": 11.780905752753977,
      "grad_norm": 1.508268997408351,
      "learning_rate": 5.469920712165708e-07,
      "loss": 0.5649,
      "step": 9625
    },
    {
      "epoch": 11.782129742962056,
      "grad_norm": 2.017968565615283,
      "learning_rate": 5.465920489765126e-07,
      "loss": 0.7628,
      "step": 9626
    },
    {
      "epoch": 11.783353733170134,
      "grad_norm": 1.7975004105553078,
      "learning_rate": 5.461921551072011e-07,
      "loss": 1.3571,
      "step": 9627
    },
    {
      "epoch": 11.784577723378213,
      "grad_norm": 1.9185394893974115,
      "learning_rate": 5.457923896349137e-07,
      "loss": 1.1039,
      "step": 9628
    },
    {
      "epoch": 11.785801713586292,
      "grad_norm": 1.976862238328463,
      "learning_rate": 5.453927525859223e-07,
      "loss": 1.0057,
      "step": 9629
    },
    {
      "epoch": 11.78702570379437,
      "grad_norm": 1.855032791202365,
      "learning_rate": 5.449932439864891e-07,
      "loss": 0.4195,
      "step": 9630
    },
    {
      "epoch": 11.788249694002449,
      "grad_norm": 2.625899833500155,
      "learning_rate": 5.445938638628695e-07,
      "loss": 0.6653,
      "step": 9631
    },
    {
      "epoch": 11.789473684210526,
      "grad_norm": 1.5105860511753944,
      "learning_rate": 5.441946122413086e-07,
      "loss": 0.433,
      "step": 9632
    },
    {
      "epoch": 11.790697674418604,
      "grad_norm": 1.530683490723502,
      "learning_rate": 5.437954891480443e-07,
      "loss": 0.6782,
      "step": 9633
    },
    {
      "epoch": 11.791921664626683,
      "grad_norm": 0.8629613158319624,
      "learning_rate": 5.433964946093045e-07,
      "loss": 0.3498,
      "step": 9634
    },
    {
      "epoch": 11.793145654834762,
      "grad_norm": 1.293572005345208,
      "learning_rate": 5.429976286513095e-07,
      "loss": 0.5548,
      "step": 9635
    },
    {
      "epoch": 11.79436964504284,
      "grad_norm": 3.850537247821223,
      "learning_rate": 5.425988913002722e-07,
      "loss": 0.3482,
      "step": 9636
    },
    {
      "epoch": 11.795593635250919,
      "grad_norm": 1.7943969858624227,
      "learning_rate": 5.422002825823961e-07,
      "loss": 1.1571,
      "step": 9637
    },
    {
      "epoch": 11.796817625458996,
      "grad_norm": 1.2768930838219346,
      "learning_rate": 5.418018025238758e-07,
      "loss": 0.4378,
      "step": 9638
    },
    {
      "epoch": 11.798041615667074,
      "grad_norm": 1.6957228805528795,
      "learning_rate": 5.414034511508983e-07,
      "loss": 0.7116,
      "step": 9639
    },
    {
      "epoch": 11.799265605875153,
      "grad_norm": 2.4637895774969207,
      "learning_rate": 5.410052284896419e-07,
      "loss": 0.498,
      "step": 9640
    },
    {
      "epoch": 11.800489596083231,
      "grad_norm": 1.7045370227986845,
      "learning_rate": 5.406071345662758e-07,
      "loss": 0.9076,
      "step": 9641
    },
    {
      "epoch": 11.80171358629131,
      "grad_norm": 2.922652563543034,
      "learning_rate": 5.402091694069614e-07,
      "loss": 0.9556,
      "step": 9642
    },
    {
      "epoch": 11.802937576499389,
      "grad_norm": 1.6447772945225934,
      "learning_rate": 5.398113330378513e-07,
      "loss": 0.5092,
      "step": 9643
    },
    {
      "epoch": 11.804161566707466,
      "grad_norm": 2.0881910115327713,
      "learning_rate": 5.394136254850906e-07,
      "loss": 1.1834,
      "step": 9644
    },
    {
      "epoch": 11.805385556915544,
      "grad_norm": 1.3886443180614014,
      "learning_rate": 5.390160467748154e-07,
      "loss": 0.504,
      "step": 9645
    },
    {
      "epoch": 11.806609547123623,
      "grad_norm": 1.722356172312867,
      "learning_rate": 5.386185969331517e-07,
      "loss": 1.3957,
      "step": 9646
    },
    {
      "epoch": 11.807833537331701,
      "grad_norm": 1.55599222316431,
      "learning_rate": 5.382212759862193e-07,
      "loss": 0.8694,
      "step": 9647
    },
    {
      "epoch": 11.80905752753978,
      "grad_norm": 1.3862346624743738,
      "learning_rate": 5.378240839601279e-07,
      "loss": 0.4612,
      "step": 9648
    },
    {
      "epoch": 11.810281517747859,
      "grad_norm": 2.362886969129296,
      "learning_rate": 5.374270208809807e-07,
      "loss": 0.3235,
      "step": 9649
    },
    {
      "epoch": 11.811505507955935,
      "grad_norm": 1.946137995379407,
      "learning_rate": 5.370300867748707e-07,
      "loss": 0.4211,
      "step": 9650
    },
    {
      "epoch": 11.812729498164014,
      "grad_norm": 1.902486642100752,
      "learning_rate": 5.366332816678835e-07,
      "loss": 0.4147,
      "step": 9651
    },
    {
      "epoch": 11.813953488372093,
      "grad_norm": 1.5762733339610606,
      "learning_rate": 5.362366055860935e-07,
      "loss": 0.3203,
      "step": 9652
    },
    {
      "epoch": 11.815177478580171,
      "grad_norm": 1.7960480211161267,
      "learning_rate": 5.358400585555712e-07,
      "loss": 1.0983,
      "step": 9653
    },
    {
      "epoch": 11.81640146878825,
      "grad_norm": 1.6861139069739322,
      "learning_rate": 5.354436406023752e-07,
      "loss": 0.4637,
      "step": 9654
    },
    {
      "epoch": 11.817625458996329,
      "grad_norm": 1.4899835558924546,
      "learning_rate": 5.35047351752557e-07,
      "loss": 1.3164,
      "step": 9655
    },
    {
      "epoch": 11.818849449204407,
      "grad_norm": 1.1140949937771847,
      "learning_rate": 5.346511920321587e-07,
      "loss": 0.5926,
      "step": 9656
    },
    {
      "epoch": 11.820073439412484,
      "grad_norm": 1.9757958411602672,
      "learning_rate": 5.342551614672148e-07,
      "loss": 0.6217,
      "step": 9657
    },
    {
      "epoch": 11.821297429620563,
      "grad_norm": 1.5729933232992068,
      "learning_rate": 5.338592600837511e-07,
      "loss": 0.6487,
      "step": 9658
    },
    {
      "epoch": 11.822521419828641,
      "grad_norm": 1.495183544550604,
      "learning_rate": 5.334634879077843e-07,
      "loss": 0.3529,
      "step": 9659
    },
    {
      "epoch": 11.82374541003672,
      "grad_norm": 2.085559635477468,
      "learning_rate": 5.330678449653234e-07,
      "loss": 0.8509,
      "step": 9660
    },
    {
      "epoch": 11.824969400244798,
      "grad_norm": 2.6324113940106697,
      "learning_rate": 5.326723312823681e-07,
      "loss": 0.573,
      "step": 9661
    },
    {
      "epoch": 11.826193390452877,
      "grad_norm": 1.3086173675091548,
      "learning_rate": 5.322769468849112e-07,
      "loss": 0.5175,
      "step": 9662
    },
    {
      "epoch": 11.827417380660954,
      "grad_norm": 1.3529903911702288,
      "learning_rate": 5.318816917989359e-07,
      "loss": 0.7168,
      "step": 9663
    },
    {
      "epoch": 11.828641370869033,
      "grad_norm": 1.2232475260676607,
      "learning_rate": 5.314865660504154e-07,
      "loss": 0.5815,
      "step": 9664
    },
    {
      "epoch": 11.829865361077111,
      "grad_norm": 3.0262573449101153,
      "learning_rate": 5.31091569665316e-07,
      "loss": 0.3941,
      "step": 9665
    },
    {
      "epoch": 11.83108935128519,
      "grad_norm": 1.857187413573764,
      "learning_rate": 5.30696702669597e-07,
      "loss": 1.3419,
      "step": 9666
    },
    {
      "epoch": 11.832313341493268,
      "grad_norm": 2.329777507072392,
      "learning_rate": 5.303019650892069e-07,
      "loss": 1.0322,
      "step": 9667
    },
    {
      "epoch": 11.833537331701347,
      "grad_norm": 1.8241093899538658,
      "learning_rate": 5.299073569500859e-07,
      "loss": 0.2915,
      "step": 9668
    },
    {
      "epoch": 11.834761321909426,
      "grad_norm": 2.0119448533452253,
      "learning_rate": 5.29512878278167e-07,
      "loss": 1.3067,
      "step": 9669
    },
    {
      "epoch": 11.835985312117502,
      "grad_norm": 2.3127739638229197,
      "learning_rate": 5.291185290993731e-07,
      "loss": 0.348,
      "step": 9670
    },
    {
      "epoch": 11.837209302325581,
      "grad_norm": 1.064128634279048,
      "learning_rate": 5.287243094396196e-07,
      "loss": 0.6264,
      "step": 9671
    },
    {
      "epoch": 11.83843329253366,
      "grad_norm": 2.17380824465741,
      "learning_rate": 5.283302193248132e-07,
      "loss": 0.9667,
      "step": 9672
    },
    {
      "epoch": 11.839657282741738,
      "grad_norm": 1.2234325371074593,
      "learning_rate": 5.279362587808523e-07,
      "loss": 0.539,
      "step": 9673
    },
    {
      "epoch": 11.840881272949817,
      "grad_norm": 1.1207950026267357,
      "learning_rate": 5.275424278336263e-07,
      "loss": 0.5618,
      "step": 9674
    },
    {
      "epoch": 11.842105263157894,
      "grad_norm": 2.6916425047957713,
      "learning_rate": 5.271487265090163e-07,
      "loss": 0.5045,
      "step": 9675
    },
    {
      "epoch": 11.843329253365972,
      "grad_norm": 2.7844439896510824,
      "learning_rate": 5.26755154832895e-07,
      "loss": 0.3558,
      "step": 9676
    },
    {
      "epoch": 11.844553243574051,
      "grad_norm": 2.2344096504865987,
      "learning_rate": 5.263617128311263e-07,
      "loss": 0.5686,
      "step": 9677
    },
    {
      "epoch": 11.84577723378213,
      "grad_norm": 1.1707255814781243,
      "learning_rate": 5.259684005295651e-07,
      "loss": 0.8456,
      "step": 9678
    },
    {
      "epoch": 11.847001223990208,
      "grad_norm": 1.8119866222418062,
      "learning_rate": 5.255752179540599e-07,
      "loss": 1.1896,
      "step": 9679
    },
    {
      "epoch": 11.848225214198287,
      "grad_norm": 1.7650960015505934,
      "learning_rate": 5.251821651304493e-07,
      "loss": 0.4979,
      "step": 9680
    },
    {
      "epoch": 11.849449204406366,
      "grad_norm": 2.0587448765329195,
      "learning_rate": 5.247892420845616e-07,
      "loss": 1.2884,
      "step": 9681
    },
    {
      "epoch": 11.850673194614442,
      "grad_norm": 2.320058181991726,
      "learning_rate": 5.243964488422185e-07,
      "loss": 0.6699,
      "step": 9682
    },
    {
      "epoch": 11.851897184822521,
      "grad_norm": 2.1584854181575603,
      "learning_rate": 5.240037854292343e-07,
      "loss": 1.1178,
      "step": 9683
    },
    {
      "epoch": 11.8531211750306,
      "grad_norm": 2.440839453620434,
      "learning_rate": 5.236112518714126e-07,
      "loss": 0.7301,
      "step": 9684
    },
    {
      "epoch": 11.854345165238678,
      "grad_norm": 1.4017775228277878,
      "learning_rate": 5.232188481945491e-07,
      "loss": 0.8612,
      "step": 9685
    },
    {
      "epoch": 11.855569155446757,
      "grad_norm": 1.4109160011250932,
      "learning_rate": 5.228265744244315e-07,
      "loss": 0.6916,
      "step": 9686
    },
    {
      "epoch": 11.856793145654835,
      "grad_norm": 3.206276308205612,
      "learning_rate": 5.224344305868381e-07,
      "loss": 0.9057,
      "step": 9687
    },
    {
      "epoch": 11.858017135862912,
      "grad_norm": 3.124955060085674,
      "learning_rate": 5.220424167075397e-07,
      "loss": 0.3936,
      "step": 9688
    },
    {
      "epoch": 11.859241126070991,
      "grad_norm": 2.2900567465888475,
      "learning_rate": 5.216505328122975e-07,
      "loss": 0.5994,
      "step": 9689
    },
    {
      "epoch": 11.86046511627907,
      "grad_norm": 1.3050467030486939,
      "learning_rate": 5.21258778926865e-07,
      "loss": 0.4002,
      "step": 9690
    },
    {
      "epoch": 11.861689106487148,
      "grad_norm": 1.5494740754288179,
      "learning_rate": 5.208671550769859e-07,
      "loss": 1.2261,
      "step": 9691
    },
    {
      "epoch": 11.862913096695227,
      "grad_norm": 1.136125403916061,
      "learning_rate": 5.204756612883985e-07,
      "loss": 0.577,
      "step": 9692
    },
    {
      "epoch": 11.864137086903305,
      "grad_norm": 1.8089213280564602,
      "learning_rate": 5.200842975868283e-07,
      "loss": 0.4124,
      "step": 9693
    },
    {
      "epoch": 11.865361077111384,
      "grad_norm": 2.0656860208362864,
      "learning_rate": 5.196930639979947e-07,
      "loss": 0.4255,
      "step": 9694
    },
    {
      "epoch": 11.86658506731946,
      "grad_norm": 1.504090737986394,
      "learning_rate": 5.193019605476077e-07,
      "loss": 0.2684,
      "step": 9695
    },
    {
      "epoch": 11.86780905752754,
      "grad_norm": 1.915281171046154,
      "learning_rate": 5.189109872613704e-07,
      "loss": 0.7417,
      "step": 9696
    },
    {
      "epoch": 11.869033047735618,
      "grad_norm": 1.1946191683253597,
      "learning_rate": 5.185201441649757e-07,
      "loss": 0.8805,
      "step": 9697
    },
    {
      "epoch": 11.870257037943697,
      "grad_norm": 1.1753436650387616,
      "learning_rate": 5.181294312841084e-07,
      "loss": 0.3811,
      "step": 9698
    },
    {
      "epoch": 11.871481028151775,
      "grad_norm": 1.5965487529600186,
      "learning_rate": 5.17738848644444e-07,
      "loss": 0.5212,
      "step": 9699
    },
    {
      "epoch": 11.872705018359854,
      "grad_norm": 3.131803704473436,
      "learning_rate": 5.173483962716499e-07,
      "loss": 0.5694,
      "step": 9700
    },
    {
      "epoch": 11.87392900856793,
      "grad_norm": 1.6189207930527607,
      "learning_rate": 5.169580741913865e-07,
      "loss": 2.2776,
      "step": 9701
    },
    {
      "epoch": 11.87515299877601,
      "grad_norm": 1.469925519128763,
      "learning_rate": 5.165678824293038e-07,
      "loss": 0.9973,
      "step": 9702
    },
    {
      "epoch": 11.876376988984088,
      "grad_norm": 1.6199140722888992,
      "learning_rate": 5.161778210110436e-07,
      "loss": 0.467,
      "step": 9703
    },
    {
      "epoch": 11.877600979192167,
      "grad_norm": 1.1576113713205967,
      "learning_rate": 5.157878899622393e-07,
      "loss": 0.5816,
      "step": 9704
    },
    {
      "epoch": 11.878824969400245,
      "grad_norm": 0.935159554091864,
      "learning_rate": 5.153980893085159e-07,
      "loss": 0.5203,
      "step": 9705
    },
    {
      "epoch": 11.880048959608324,
      "grad_norm": 2.537927233611365,
      "learning_rate": 5.150084190754895e-07,
      "loss": 0.3751,
      "step": 9706
    },
    {
      "epoch": 11.8812729498164,
      "grad_norm": 1.68789373503456,
      "learning_rate": 5.146188792887677e-07,
      "loss": 1.1672,
      "step": 9707
    },
    {
      "epoch": 11.88249694002448,
      "grad_norm": 1.6904049965287256,
      "learning_rate": 5.1422946997395e-07,
      "loss": 1.1648,
      "step": 9708
    },
    {
      "epoch": 11.883720930232558,
      "grad_norm": 1.3395828442593432,
      "learning_rate": 5.138401911566265e-07,
      "loss": 0.4891,
      "step": 9709
    },
    {
      "epoch": 11.884944920440637,
      "grad_norm": 1.784389486001225,
      "learning_rate": 5.134510428623793e-07,
      "loss": 1.0915,
      "step": 9710
    },
    {
      "epoch": 11.886168910648715,
      "grad_norm": 2.0459106624934917,
      "learning_rate": 5.130620251167817e-07,
      "loss": 0.5231,
      "step": 9711
    },
    {
      "epoch": 11.887392900856794,
      "grad_norm": 1.8766319960086206,
      "learning_rate": 5.126731379453989e-07,
      "loss": 0.5418,
      "step": 9712
    },
    {
      "epoch": 11.88861689106487,
      "grad_norm": 1.5803806809553844,
      "learning_rate": 5.122843813737863e-07,
      "loss": 0.4682,
      "step": 9713
    },
    {
      "epoch": 11.88984088127295,
      "grad_norm": 1.7017072148577388,
      "learning_rate": 5.118957554274925e-07,
      "loss": 0.4864,
      "step": 9714
    },
    {
      "epoch": 11.891064871481028,
      "grad_norm": 1.3854031300408698,
      "learning_rate": 5.115072601320561e-07,
      "loss": 0.522,
      "step": 9715
    },
    {
      "epoch": 11.892288861689106,
      "grad_norm": 2.883032235571502,
      "learning_rate": 5.111188955130087e-07,
      "loss": 0.4966,
      "step": 9716
    },
    {
      "epoch": 11.893512851897185,
      "grad_norm": 2.3695324441161896,
      "learning_rate": 5.107306615958693e-07,
      "loss": 0.2827,
      "step": 9717
    },
    {
      "epoch": 11.894736842105264,
      "grad_norm": 2.654315444297391,
      "learning_rate": 5.103425584061539e-07,
      "loss": 0.6726,
      "step": 9718
    },
    {
      "epoch": 11.895960832313342,
      "grad_norm": 3.421252877498952,
      "learning_rate": 5.099545859693664e-07,
      "loss": 0.4333,
      "step": 9719
    },
    {
      "epoch": 11.89718482252142,
      "grad_norm": 1.5473007777313559,
      "learning_rate": 5.095667443110028e-07,
      "loss": 1.2563,
      "step": 9720
    },
    {
      "epoch": 11.898408812729498,
      "grad_norm": 1.9100711860955442,
      "learning_rate": 5.091790334565508e-07,
      "loss": 0.7052,
      "step": 9721
    },
    {
      "epoch": 11.899632802937576,
      "grad_norm": 1.8940775234630107,
      "learning_rate": 5.087914534314889e-07,
      "loss": 0.4362,
      "step": 9722
    },
    {
      "epoch": 11.900856793145655,
      "grad_norm": 1.952782023059173,
      "learning_rate": 5.084040042612878e-07,
      "loss": 0.6901,
      "step": 9723
    },
    {
      "epoch": 11.902080783353734,
      "grad_norm": 1.1448534616713462,
      "learning_rate": 5.080166859714092e-07,
      "loss": 0.5347,
      "step": 9724
    },
    {
      "epoch": 11.903304773561812,
      "grad_norm": 2.5433503020435313,
      "learning_rate": 5.076294985873062e-07,
      "loss": 1.013,
      "step": 9725
    },
    {
      "epoch": 11.904528763769889,
      "grad_norm": 2.428415254490329,
      "learning_rate": 5.072424421344224e-07,
      "loss": 0.4676,
      "step": 9726
    },
    {
      "epoch": 11.905752753977968,
      "grad_norm": 2.447362214768637,
      "learning_rate": 5.068555166381961e-07,
      "loss": 0.3948,
      "step": 9727
    },
    {
      "epoch": 11.906976744186046,
      "grad_norm": 1.8078675432097615,
      "learning_rate": 5.064687221240525e-07,
      "loss": 1.132,
      "step": 9728
    },
    {
      "epoch": 11.908200734394125,
      "grad_norm": 1.3396673725894388,
      "learning_rate": 5.060820586174109e-07,
      "loss": 0.5065,
      "step": 9729
    },
    {
      "epoch": 11.909424724602204,
      "grad_norm": 1.0802932201395359,
      "learning_rate": 5.056955261436808e-07,
      "loss": 0.5159,
      "step": 9730
    },
    {
      "epoch": 11.910648714810282,
      "grad_norm": 0.7957081907104413,
      "learning_rate": 5.053091247282649e-07,
      "loss": 0.2172,
      "step": 9731
    },
    {
      "epoch": 11.911872705018359,
      "grad_norm": 1.5561072459775909,
      "learning_rate": 5.049228543965553e-07,
      "loss": 0.5226,
      "step": 9732
    },
    {
      "epoch": 11.913096695226438,
      "grad_norm": 1.8352110696108188,
      "learning_rate": 5.045367151739366e-07,
      "loss": 1.0056,
      "step": 9733
    },
    {
      "epoch": 11.914320685434516,
      "grad_norm": 2.100933243137448,
      "learning_rate": 5.041507070857852e-07,
      "loss": 0.4367,
      "step": 9734
    },
    {
      "epoch": 11.915544675642595,
      "grad_norm": 3.056708505341668,
      "learning_rate": 5.037648301574655e-07,
      "loss": 0.3375,
      "step": 9735
    },
    {
      "epoch": 11.916768665850674,
      "grad_norm": 1.5175747888218345,
      "learning_rate": 5.033790844143383e-07,
      "loss": 0.4513,
      "step": 9736
    },
    {
      "epoch": 11.917992656058752,
      "grad_norm": 1.9036691472886569,
      "learning_rate": 5.029934698817526e-07,
      "loss": 0.4517,
      "step": 9737
    },
    {
      "epoch": 11.919216646266829,
      "grad_norm": 1.7101949465399282,
      "learning_rate": 5.026079865850497e-07,
      "loss": 0.9251,
      "step": 9738
    },
    {
      "epoch": 11.920440636474908,
      "grad_norm": 2.48098954052835,
      "learning_rate": 5.022226345495624e-07,
      "loss": 0.3609,
      "step": 9739
    },
    {
      "epoch": 11.921664626682986,
      "grad_norm": 1.8301958282344546,
      "learning_rate": 5.01837413800614e-07,
      "loss": 0.5236,
      "step": 9740
    },
    {
      "epoch": 11.922888616891065,
      "grad_norm": 2.0491555028504864,
      "learning_rate": 5.0145232436352e-07,
      "loss": 0.6293,
      "step": 9741
    },
    {
      "epoch": 11.924112607099143,
      "grad_norm": 1.6395358698209082,
      "learning_rate": 5.010673662635873e-07,
      "loss": 0.6209,
      "step": 9742
    },
    {
      "epoch": 11.925336597307222,
      "grad_norm": 1.4037162715236198,
      "learning_rate": 5.00682539526113e-07,
      "loss": 1.0867,
      "step": 9743
    },
    {
      "epoch": 11.9265605875153,
      "grad_norm": 1.3999126230809755,
      "learning_rate": 5.002978441763881e-07,
      "loss": 0.6253,
      "step": 9744
    },
    {
      "epoch": 11.927784577723378,
      "grad_norm": 1.7653018523398902,
      "learning_rate": 4.999132802396927e-07,
      "loss": 0.3088,
      "step": 9745
    },
    {
      "epoch": 11.929008567931456,
      "grad_norm": 1.8258976037985457,
      "learning_rate": 4.995288477412985e-07,
      "loss": 0.6524,
      "step": 9746
    },
    {
      "epoch": 11.930232558139535,
      "grad_norm": 1.8157786506421503,
      "learning_rate": 4.99144546706469e-07,
      "loss": 0.5753,
      "step": 9747
    },
    {
      "epoch": 11.931456548347613,
      "grad_norm": 1.0224478759969862,
      "learning_rate": 4.987603771604588e-07,
      "loss": 0.5085,
      "step": 9748
    },
    {
      "epoch": 11.932680538555692,
      "grad_norm": 2.169745791481339,
      "learning_rate": 4.983763391285148e-07,
      "loss": 0.957,
      "step": 9749
    },
    {
      "epoch": 11.93390452876377,
      "grad_norm": 1.2438674301624664,
      "learning_rate": 4.979924326358745e-07,
      "loss": 0.426,
      "step": 9750
    },
    {
      "epoch": 11.935128518971847,
      "grad_norm": 1.4921247591356994,
      "learning_rate": 4.976086577077674e-07,
      "loss": 1.0218,
      "step": 9751
    },
    {
      "epoch": 11.936352509179926,
      "grad_norm": 1.0480078655922382,
      "learning_rate": 4.972250143694116e-07,
      "loss": 0.3435,
      "step": 9752
    },
    {
      "epoch": 11.937576499388005,
      "grad_norm": 1.9765446088836802,
      "learning_rate": 4.968415026460207e-07,
      "loss": 0.6809,
      "step": 9753
    },
    {
      "epoch": 11.938800489596083,
      "grad_norm": 1.6189816817936633,
      "learning_rate": 4.964581225627971e-07,
      "loss": 0.4623,
      "step": 9754
    },
    {
      "epoch": 11.940024479804162,
      "grad_norm": 1.8027419077705586,
      "learning_rate": 4.960748741449351e-07,
      "loss": 0.8804,
      "step": 9755
    },
    {
      "epoch": 11.94124847001224,
      "grad_norm": 1.8541038103766476,
      "learning_rate": 4.956917574176204e-07,
      "loss": 0.4968,
      "step": 9756
    },
    {
      "epoch": 11.94247246022032,
      "grad_norm": 2.5449358366994974,
      "learning_rate": 4.953087724060299e-07,
      "loss": 0.9668,
      "step": 9757
    },
    {
      "epoch": 11.943696450428396,
      "grad_norm": 2.3273895373275293,
      "learning_rate": 4.94925919135332e-07,
      "loss": 0.6276,
      "step": 9758
    },
    {
      "epoch": 11.944920440636475,
      "grad_norm": 2.442466034286734,
      "learning_rate": 4.945431976306863e-07,
      "loss": 0.4958,
      "step": 9759
    },
    {
      "epoch": 11.946144430844553,
      "grad_norm": 1.4990254208663694,
      "learning_rate": 4.941606079172443e-07,
      "loss": 0.6447,
      "step": 9760
    },
    {
      "epoch": 11.947368421052632,
      "grad_norm": 2.3219606290541654,
      "learning_rate": 4.937781500201475e-07,
      "loss": 0.3458,
      "step": 9761
    },
    {
      "epoch": 11.94859241126071,
      "grad_norm": 2.3844456749417966,
      "learning_rate": 4.933958239645304e-07,
      "loss": 0.9008,
      "step": 9762
    },
    {
      "epoch": 11.949816401468787,
      "grad_norm": 1.5364613348038414,
      "learning_rate": 4.930136297755189e-07,
      "loss": 0.8875,
      "step": 9763
    },
    {
      "epoch": 11.951040391676866,
      "grad_norm": 2.5643217208850495,
      "learning_rate": 4.926315674782273e-07,
      "loss": 0.4964,
      "step": 9764
    },
    {
      "epoch": 11.952264381884945,
      "grad_norm": 1.9294251259921487,
      "learning_rate": 4.922496370977639e-07,
      "loss": 0.7353,
      "step": 9765
    },
    {
      "epoch": 11.953488372093023,
      "grad_norm": 1.2829427997574674,
      "learning_rate": 4.918678386592288e-07,
      "loss": 0.4677,
      "step": 9766
    },
    {
      "epoch": 11.954712362301102,
      "grad_norm": 1.455683230986563,
      "learning_rate": 4.914861721877118e-07,
      "loss": 0.3964,
      "step": 9767
    },
    {
      "epoch": 11.95593635250918,
      "grad_norm": 1.672975215369213,
      "learning_rate": 4.911046377082945e-07,
      "loss": 0.6372,
      "step": 9768
    },
    {
      "epoch": 11.957160342717259,
      "grad_norm": 1.8520041810593246,
      "learning_rate": 4.907232352460503e-07,
      "loss": 0.9809,
      "step": 9769
    },
    {
      "epoch": 11.958384332925336,
      "grad_norm": 1.68820672878603,
      "learning_rate": 4.90341964826043e-07,
      "loss": 0.5565,
      "step": 9770
    },
    {
      "epoch": 11.959608323133414,
      "grad_norm": 2.7011217773256733,
      "learning_rate": 4.899608264733288e-07,
      "loss": 1.0109,
      "step": 9771
    },
    {
      "epoch": 11.960832313341493,
      "grad_norm": 1.901002300265054,
      "learning_rate": 4.895798202129543e-07,
      "loss": 1.4837,
      "step": 9772
    },
    {
      "epoch": 11.962056303549572,
      "grad_norm": 1.6346174957740531,
      "learning_rate": 4.891989460699578e-07,
      "loss": 1.482,
      "step": 9773
    },
    {
      "epoch": 11.96328029375765,
      "grad_norm": 1.5028344920080638,
      "learning_rate": 4.888182040693695e-07,
      "loss": 0.616,
      "step": 9774
    },
    {
      "epoch": 11.964504283965729,
      "grad_norm": 1.6352758187086325,
      "learning_rate": 4.884375942362096e-07,
      "loss": 0.9808,
      "step": 9775
    },
    {
      "epoch": 11.965728274173806,
      "grad_norm": 1.5788750109075158,
      "learning_rate": 4.880571165954912e-07,
      "loss": 0.5888,
      "step": 9776
    },
    {
      "epoch": 11.966952264381884,
      "grad_norm": 2.618036414249568,
      "learning_rate": 4.876767711722172e-07,
      "loss": 0.6162,
      "step": 9777
    },
    {
      "epoch": 11.968176254589963,
      "grad_norm": 2.6713428448670817,
      "learning_rate": 4.872965579913818e-07,
      "loss": 0.7802,
      "step": 9778
    },
    {
      "epoch": 11.969400244798042,
      "grad_norm": 2.929651143516116,
      "learning_rate": 4.869164770779733e-07,
      "loss": 0.4162,
      "step": 9779
    },
    {
      "epoch": 11.97062423500612,
      "grad_norm": 1.8890342145958732,
      "learning_rate": 4.865365284569684e-07,
      "loss": 1.0786,
      "step": 9780
    },
    {
      "epoch": 11.971848225214199,
      "grad_norm": 1.4854633095506888,
      "learning_rate": 4.861567121533351e-07,
      "loss": 1.1563,
      "step": 9781
    },
    {
      "epoch": 11.973072215422278,
      "grad_norm": 2.2381693687775344,
      "learning_rate": 4.857770281920329e-07,
      "loss": 0.8184,
      "step": 9782
    },
    {
      "epoch": 11.974296205630354,
      "grad_norm": 1.5934141898976182,
      "learning_rate": 4.853974765980154e-07,
      "loss": 0.689,
      "step": 9783
    },
    {
      "epoch": 11.975520195838433,
      "grad_norm": 1.5936599310354573,
      "learning_rate": 4.85018057396224e-07,
      "loss": 0.8192,
      "step": 9784
    },
    {
      "epoch": 11.976744186046512,
      "grad_norm": 2.182123113014971,
      "learning_rate": 4.846387706115932e-07,
      "loss": 1.0654,
      "step": 9785
    },
    {
      "epoch": 11.97796817625459,
      "grad_norm": 1.530252467702324,
      "learning_rate": 4.84259616269048e-07,
      "loss": 0.5055,
      "step": 9786
    },
    {
      "epoch": 11.979192166462669,
      "grad_norm": 2.0005235936531105,
      "learning_rate": 4.838805943935049e-07,
      "loss": 0.5513,
      "step": 9787
    },
    {
      "epoch": 11.980416156670747,
      "grad_norm": 1.3220546376351436,
      "learning_rate": 4.835017050098722e-07,
      "loss": 0.6003,
      "step": 9788
    },
    {
      "epoch": 11.981640146878824,
      "grad_norm": 2.9388842573965133,
      "learning_rate": 4.831229481430491e-07,
      "loss": 0.5266,
      "step": 9789
    },
    {
      "epoch": 11.982864137086903,
      "grad_norm": 1.2484547917169084,
      "learning_rate": 4.827443238179258e-07,
      "loss": 0.843,
      "step": 9790
    },
    {
      "epoch": 11.984088127294982,
      "grad_norm": 3.330047648081599,
      "learning_rate": 4.823658320593835e-07,
      "loss": 0.4561,
      "step": 9791
    },
    {
      "epoch": 11.98531211750306,
      "grad_norm": 2.982033096852002,
      "learning_rate": 4.819874728922974e-07,
      "loss": 0.4573,
      "step": 9792
    },
    {
      "epoch": 11.986536107711139,
      "grad_norm": 1.2345595751857097,
      "learning_rate": 4.816092463415298e-07,
      "loss": 0.5979,
      "step": 9793
    },
    {
      "epoch": 11.987760097919217,
      "grad_norm": 1.5479198301944783,
      "learning_rate": 4.812311524319369e-07,
      "loss": 0.7829,
      "step": 9794
    },
    {
      "epoch": 11.988984088127294,
      "grad_norm": 1.6277053177986331,
      "learning_rate": 4.808531911883651e-07,
      "loss": 0.5206,
      "step": 9795
    },
    {
      "epoch": 11.990208078335373,
      "grad_norm": 2.487736751957783,
      "learning_rate": 4.80475362635654e-07,
      "loss": 0.8581,
      "step": 9796
    },
    {
      "epoch": 11.991432068543451,
      "grad_norm": 1.5145936690931638,
      "learning_rate": 4.800976667986324e-07,
      "loss": 0.5853,
      "step": 9797
    },
    {
      "epoch": 11.99265605875153,
      "grad_norm": 2.071004183405466,
      "learning_rate": 4.797201037021218e-07,
      "loss": 0.2904,
      "step": 9798
    },
    {
      "epoch": 11.993880048959609,
      "grad_norm": 2.813941153099257,
      "learning_rate": 4.793426733709325e-07,
      "loss": 0.5011,
      "step": 9799
    },
    {
      "epoch": 11.995104039167687,
      "grad_norm": 2.553918810004041,
      "learning_rate": 4.789653758298684e-07,
      "loss": 0.4038,
      "step": 9800
    },
    {
      "epoch": 11.996328029375764,
      "grad_norm": 1.5772928371842687,
      "learning_rate": 4.785882111037252e-07,
      "loss": 0.6026,
      "step": 9801
    },
    {
      "epoch": 11.997552019583843,
      "grad_norm": 3.1949451206746886,
      "learning_rate": 4.78211179217288e-07,
      "loss": 0.5376,
      "step": 9802
    },
    {
      "epoch": 11.998776009791921,
      "grad_norm": 2.568425502389638,
      "learning_rate": 4.77834280195334e-07,
      "loss": 0.4461,
      "step": 9803
    },
    {
      "epoch": 12.0,
      "grad_norm": 1.763552642288535,
      "learning_rate": 4.774575140626317e-07,
      "loss": 1.5151,
      "step": 9804
    },
    {
      "epoch": 12.001223990208079,
      "grad_norm": 1.06937955991284,
      "learning_rate": 4.770808808439406e-07,
      "loss": 0.3953,
      "step": 9805
    },
    {
      "epoch": 12.002447980416157,
      "grad_norm": 1.3995286713720143,
      "learning_rate": 4.7670438056401174e-07,
      "loss": 0.447,
      "step": 9806
    },
    {
      "epoch": 12.003671970624236,
      "grad_norm": 1.5357033098348605,
      "learning_rate": 4.7632801324758754e-07,
      "loss": 1.3972,
      "step": 9807
    },
    {
      "epoch": 12.004895960832313,
      "grad_norm": 2.2988835833570582,
      "learning_rate": 4.7595177891940044e-07,
      "loss": 0.4785,
      "step": 9808
    },
    {
      "epoch": 12.006119951040391,
      "grad_norm": 1.621073675979508,
      "learning_rate": 4.755756776041773e-07,
      "loss": 0.4709,
      "step": 9809
    },
    {
      "epoch": 12.00734394124847,
      "grad_norm": 3.7232724328294435,
      "learning_rate": 4.7519970932663223e-07,
      "loss": 0.2661,
      "step": 9810
    },
    {
      "epoch": 12.008567931456549,
      "grad_norm": 1.3208411991142264,
      "learning_rate": 4.74823874111473e-07,
      "loss": 0.5988,
      "step": 9811
    },
    {
      "epoch": 12.009791921664627,
      "grad_norm": 1.9331367855600958,
      "learning_rate": 4.744481719833982e-07,
      "loss": 0.4807,
      "step": 9812
    },
    {
      "epoch": 12.011015911872706,
      "grad_norm": 1.787282415937334,
      "learning_rate": 4.7407260296709666e-07,
      "loss": 0.9278,
      "step": 9813
    },
    {
      "epoch": 12.012239902080783,
      "grad_norm": 1.3485463933977873,
      "learning_rate": 4.736971670872512e-07,
      "loss": 0.5799,
      "step": 9814
    },
    {
      "epoch": 12.013463892288861,
      "grad_norm": 2.2013901081983223,
      "learning_rate": 4.7332186436853307e-07,
      "loss": 1.1246,
      "step": 9815
    },
    {
      "epoch": 12.01468788249694,
      "grad_norm": 2.7074331637650175,
      "learning_rate": 4.729466948356065e-07,
      "loss": 0.7786,
      "step": 9816
    },
    {
      "epoch": 12.015911872705018,
      "grad_norm": 1.4606019344408274,
      "learning_rate": 4.7257165851312434e-07,
      "loss": 1.2787,
      "step": 9817
    },
    {
      "epoch": 12.017135862913097,
      "grad_norm": 2.052112431251952,
      "learning_rate": 4.7219675542573426e-07,
      "loss": 0.6275,
      "step": 9818
    },
    {
      "epoch": 12.018359853121176,
      "grad_norm": 2.7036255660517328,
      "learning_rate": 4.7182198559807323e-07,
      "loss": 0.7182,
      "step": 9819
    },
    {
      "epoch": 12.019583843329253,
      "grad_norm": 2.297677543451564,
      "learning_rate": 4.714473490547697e-07,
      "loss": 0.3449,
      "step": 9820
    },
    {
      "epoch": 12.020807833537331,
      "grad_norm": 2.457431787984592,
      "learning_rate": 4.710728458204433e-07,
      "loss": 0.4436,
      "step": 9821
    },
    {
      "epoch": 12.02203182374541,
      "grad_norm": 2.181894738407353,
      "learning_rate": 4.7069847591970475e-07,
      "loss": 0.6328,
      "step": 9822
    },
    {
      "epoch": 12.023255813953488,
      "grad_norm": 2.5880225741496283,
      "learning_rate": 4.703242393771568e-07,
      "loss": 0.6,
      "step": 9823
    },
    {
      "epoch": 12.024479804161567,
      "grad_norm": 1.2046080301760016,
      "learning_rate": 4.699501362173925e-07,
      "loss": 0.5673,
      "step": 9824
    },
    {
      "epoch": 12.025703794369646,
      "grad_norm": 1.1043697253791462,
      "learning_rate": 4.6957616646499643e-07,
      "loss": 0.5914,
      "step": 9825
    },
    {
      "epoch": 12.026927784577722,
      "grad_norm": 3.2463734644032005,
      "learning_rate": 4.692023301445442e-07,
      "loss": 0.3993,
      "step": 9826
    },
    {
      "epoch": 12.028151774785801,
      "grad_norm": 1.889592040310895,
      "learning_rate": 4.6882862728060475e-07,
      "loss": 0.4565,
      "step": 9827
    },
    {
      "epoch": 12.02937576499388,
      "grad_norm": 1.252174922983469,
      "learning_rate": 4.684550578977343e-07,
      "loss": 0.8143,
      "step": 9828
    },
    {
      "epoch": 12.030599755201958,
      "grad_norm": 1.8263571837891874,
      "learning_rate": 4.6808162202048343e-07,
      "loss": 0.9491,
      "step": 9829
    },
    {
      "epoch": 12.031823745410037,
      "grad_norm": 1.6679685461209077,
      "learning_rate": 4.6770831967339224e-07,
      "loss": 0.5487,
      "step": 9830
    },
    {
      "epoch": 12.033047735618116,
      "grad_norm": 2.448271005093146,
      "learning_rate": 4.6733515088099366e-07,
      "loss": 0.4164,
      "step": 9831
    },
    {
      "epoch": 12.034271725826194,
      "grad_norm": 1.377955968704842,
      "learning_rate": 4.6696211566781114e-07,
      "loss": 0.3909,
      "step": 9832
    },
    {
      "epoch": 12.035495716034271,
      "grad_norm": 2.2689595276360435,
      "learning_rate": 4.6658921405835835e-07,
      "loss": 0.6694,
      "step": 9833
    },
    {
      "epoch": 12.03671970624235,
      "grad_norm": 3.0943201450568636,
      "learning_rate": 4.662164460771415e-07,
      "loss": 0.9016,
      "step": 9834
    },
    {
      "epoch": 12.037943696450428,
      "grad_norm": 1.8728311781104834,
      "learning_rate": 4.6584381174865717e-07,
      "loss": 0.4176,
      "step": 9835
    },
    {
      "epoch": 12.039167686658507,
      "grad_norm": 1.4691314105199358,
      "learning_rate": 4.654713110973938e-07,
      "loss": 0.9435,
      "step": 9836
    },
    {
      "epoch": 12.040391676866586,
      "grad_norm": 2.1515442702737526,
      "learning_rate": 4.6509894414783095e-07,
      "loss": 1.1805,
      "step": 9837
    },
    {
      "epoch": 12.041615667074664,
      "grad_norm": 2.3327982597598984,
      "learning_rate": 4.6472671092443856e-07,
      "loss": 0.5425,
      "step": 9838
    },
    {
      "epoch": 12.042839657282741,
      "grad_norm": 1.589967871035463,
      "learning_rate": 4.6435461145167894e-07,
      "loss": 0.817,
      "step": 9839
    },
    {
      "epoch": 12.04406364749082,
      "grad_norm": 3.9085738900765192,
      "learning_rate": 4.6398264575400476e-07,
      "loss": 0.3701,
      "step": 9840
    },
    {
      "epoch": 12.045287637698898,
      "grad_norm": 2.293361326179226,
      "learning_rate": 4.6361081385586066e-07,
      "loss": 0.6126,
      "step": 9841
    },
    {
      "epoch": 12.046511627906977,
      "grad_norm": 1.146308521680064,
      "learning_rate": 4.632391157816815e-07,
      "loss": 0.3596,
      "step": 9842
    },
    {
      "epoch": 12.047735618115055,
      "grad_norm": 2.0531558712071036,
      "learning_rate": 4.628675515558936e-07,
      "loss": 0.9567,
      "step": 9843
    },
    {
      "epoch": 12.048959608323134,
      "grad_norm": 1.720940632293993,
      "learning_rate": 4.6249612120291635e-07,
      "loss": 0.9062,
      "step": 9844
    },
    {
      "epoch": 12.050183598531211,
      "grad_norm": 3.162836671029267,
      "learning_rate": 4.621248247471583e-07,
      "loss": 0.2977,
      "step": 9845
    },
    {
      "epoch": 12.05140758873929,
      "grad_norm": 1.9974275505551535,
      "learning_rate": 4.6175366221301837e-07,
      "loss": 0.4759,
      "step": 9846
    },
    {
      "epoch": 12.052631578947368,
      "grad_norm": 2.972693030628046,
      "learning_rate": 4.6138263362488806e-07,
      "loss": 0.4536,
      "step": 9847
    },
    {
      "epoch": 12.053855569155447,
      "grad_norm": 2.4616670578156983,
      "learning_rate": 4.610117390071517e-07,
      "loss": 0.465,
      "step": 9848
    },
    {
      "epoch": 12.055079559363525,
      "grad_norm": 1.422730272753212,
      "learning_rate": 4.60640978384182e-07,
      "loss": 0.7727,
      "step": 9849
    },
    {
      "epoch": 12.056303549571604,
      "grad_norm": 2.754119810554422,
      "learning_rate": 4.6027035178034417e-07,
      "loss": 0.8888,
      "step": 9850
    },
    {
      "epoch": 12.057527539779683,
      "grad_norm": 2.3891237683044912,
      "learning_rate": 4.5989985921999436e-07,
      "loss": 0.4916,
      "step": 9851
    },
    {
      "epoch": 12.05875152998776,
      "grad_norm": 1.1243558004549665,
      "learning_rate": 4.5952950072748e-07,
      "loss": 0.384,
      "step": 9852
    },
    {
      "epoch": 12.059975520195838,
      "grad_norm": 2.493761485173966,
      "learning_rate": 4.591592763271399e-07,
      "loss": 1.0089,
      "step": 9853
    },
    {
      "epoch": 12.061199510403917,
      "grad_norm": 1.1549428173005303,
      "learning_rate": 4.587891860433033e-07,
      "loss": 0.563,
      "step": 9854
    },
    {
      "epoch": 12.062423500611995,
      "grad_norm": 2.2935070150009675,
      "learning_rate": 4.5841922990029153e-07,
      "loss": 0.4616,
      "step": 9855
    },
    {
      "epoch": 12.063647490820074,
      "grad_norm": 1.1457588230874467,
      "learning_rate": 4.5804940792241684e-07,
      "loss": 0.4928,
      "step": 9856
    },
    {
      "epoch": 12.064871481028153,
      "grad_norm": 1.3943106284741955,
      "learning_rate": 4.576797201339822e-07,
      "loss": 0.6211,
      "step": 9857
    },
    {
      "epoch": 12.06609547123623,
      "grad_norm": 3.8900900268856082,
      "learning_rate": 4.5731016655928244e-07,
      "loss": 0.3466,
      "step": 9858
    },
    {
      "epoch": 12.067319461444308,
      "grad_norm": 1.6664150449720165,
      "learning_rate": 4.5694074722260314e-07,
      "loss": 0.462,
      "step": 9859
    },
    {
      "epoch": 12.068543451652387,
      "grad_norm": 2.340235700044478,
      "learning_rate": 4.5657146214822116e-07,
      "loss": 0.9666,
      "step": 9860
    },
    {
      "epoch": 12.069767441860465,
      "grad_norm": 1.0119321829613022,
      "learning_rate": 4.5620231136040414e-07,
      "loss": 0.4445,
      "step": 9861
    },
    {
      "epoch": 12.070991432068544,
      "grad_norm": 1.8829654314919344,
      "learning_rate": 4.558332948834121e-07,
      "loss": 0.4669,
      "step": 9862
    },
    {
      "epoch": 12.072215422276622,
      "grad_norm": 2.4532484608435126,
      "learning_rate": 4.5546441274149587e-07,
      "loss": 0.5828,
      "step": 9863
    },
    {
      "epoch": 12.0734394124847,
      "grad_norm": 1.5266845747389997,
      "learning_rate": 4.5509566495889555e-07,
      "loss": 0.7464,
      "step": 9864
    },
    {
      "epoch": 12.074663402692778,
      "grad_norm": 1.9942204621335993,
      "learning_rate": 4.547270515598437e-07,
      "loss": 0.4101,
      "step": 9865
    },
    {
      "epoch": 12.075887392900857,
      "grad_norm": 1.7291123741491843,
      "learning_rate": 4.543585725685659e-07,
      "loss": 1.4469,
      "step": 9866
    },
    {
      "epoch": 12.077111383108935,
      "grad_norm": 2.2852536769183307,
      "learning_rate": 4.539902280092764e-07,
      "loss": 0.5983,
      "step": 9867
    },
    {
      "epoch": 12.078335373317014,
      "grad_norm": 1.137836332383699,
      "learning_rate": 4.536220179061815e-07,
      "loss": 0.576,
      "step": 9868
    },
    {
      "epoch": 12.079559363525092,
      "grad_norm": 1.6148989892829824,
      "learning_rate": 4.5325394228347865e-07,
      "loss": 0.4599,
      "step": 9869
    },
    {
      "epoch": 12.080783353733171,
      "grad_norm": 1.1661929163130917,
      "learning_rate": 4.5288600116535605e-07,
      "loss": 0.4442,
      "step": 9870
    },
    {
      "epoch": 12.082007343941248,
      "grad_norm": 2.4521939122646987,
      "learning_rate": 4.525181945759938e-07,
      "loss": 0.7272,
      "step": 9871
    },
    {
      "epoch": 12.083231334149326,
      "grad_norm": 1.9861315474217156,
      "learning_rate": 4.521505225395628e-07,
      "loss": 0.5265,
      "step": 9872
    },
    {
      "epoch": 12.084455324357405,
      "grad_norm": 1.431905026346671,
      "learning_rate": 4.5178298508022494e-07,
      "loss": 0.4976,
      "step": 9873
    },
    {
      "epoch": 12.085679314565484,
      "grad_norm": 2.3893437288797177,
      "learning_rate": 4.5141558222213335e-07,
      "loss": 0.281,
      "step": 9874
    },
    {
      "epoch": 12.086903304773562,
      "grad_norm": 2.034914385477065,
      "learning_rate": 4.5104831398943255e-07,
      "loss": 0.7412,
      "step": 9875
    },
    {
      "epoch": 12.088127294981641,
      "grad_norm": 3.5822319490159376,
      "learning_rate": 4.50681180406258e-07,
      "loss": 0.3393,
      "step": 9876
    },
    {
      "epoch": 12.089351285189718,
      "grad_norm": 1.2244974744568673,
      "learning_rate": 4.503141814967366e-07,
      "loss": 0.5808,
      "step": 9877
    },
    {
      "epoch": 12.090575275397796,
      "grad_norm": 1.9876717054998867,
      "learning_rate": 4.4994731728498496e-07,
      "loss": 0.5092,
      "step": 9878
    },
    {
      "epoch": 12.091799265605875,
      "grad_norm": 2.649614995869319,
      "learning_rate": 4.49580587795114e-07,
      "loss": 1.0097,
      "step": 9879
    },
    {
      "epoch": 12.093023255813954,
      "grad_norm": 1.7790289257921463,
      "learning_rate": 4.4921399305122264e-07,
      "loss": 1.7968,
      "step": 9880
    },
    {
      "epoch": 12.094247246022032,
      "grad_norm": 1.8469362614924132,
      "learning_rate": 4.488475330774028e-07,
      "loss": 1.188,
      "step": 9881
    },
    {
      "epoch": 12.095471236230111,
      "grad_norm": 1.7208819606883992,
      "learning_rate": 4.484812078977352e-07,
      "loss": 0.9788,
      "step": 9882
    },
    {
      "epoch": 12.096695226438188,
      "grad_norm": 3.325942111886066,
      "learning_rate": 4.4811501753629557e-07,
      "loss": 0.3713,
      "step": 9883
    },
    {
      "epoch": 12.097919216646266,
      "grad_norm": 1.2198282354821652,
      "learning_rate": 4.4774896201714714e-07,
      "loss": 0.5966,
      "step": 9884
    },
    {
      "epoch": 12.099143206854345,
      "grad_norm": 1.488033900934936,
      "learning_rate": 4.4738304136434624e-07,
      "loss": 0.3515,
      "step": 9885
    },
    {
      "epoch": 12.100367197062424,
      "grad_norm": 1.5387776287805777,
      "learning_rate": 4.4701725560194003e-07,
      "loss": 0.5042,
      "step": 9886
    },
    {
      "epoch": 12.101591187270502,
      "grad_norm": 1.5998503885101045,
      "learning_rate": 4.4665160475396625e-07,
      "loss": 1.5388,
      "step": 9887
    },
    {
      "epoch": 12.10281517747858,
      "grad_norm": 1.4489724960266488,
      "learning_rate": 4.4628608884445395e-07,
      "loss": 0.6896,
      "step": 9888
    },
    {
      "epoch": 12.104039167686658,
      "grad_norm": 2.1859678800892475,
      "learning_rate": 4.45920707897424e-07,
      "loss": 0.6038,
      "step": 9889
    },
    {
      "epoch": 12.105263157894736,
      "grad_norm": 1.8365406759265028,
      "learning_rate": 4.455554619368874e-07,
      "loss": 0.5221,
      "step": 9890
    },
    {
      "epoch": 12.106487148102815,
      "grad_norm": 2.404010668777895,
      "learning_rate": 4.451903509868463e-07,
      "loss": 0.4573,
      "step": 9891
    },
    {
      "epoch": 12.107711138310894,
      "grad_norm": 2.0314762404045648,
      "learning_rate": 4.448253750712966e-07,
      "loss": 0.571,
      "step": 9892
    },
    {
      "epoch": 12.108935128518972,
      "grad_norm": 2.103575760472522,
      "learning_rate": 4.4446053421422107e-07,
      "loss": 0.4952,
      "step": 9893
    },
    {
      "epoch": 12.11015911872705,
      "grad_norm": 0.8003613923814068,
      "learning_rate": 4.4409582843959604e-07,
      "loss": 0.2165,
      "step": 9894
    },
    {
      "epoch": 12.11138310893513,
      "grad_norm": 2.5936684983742615,
      "learning_rate": 4.437312577713887e-07,
      "loss": 0.515,
      "step": 9895
    },
    {
      "epoch": 12.112607099143206,
      "grad_norm": 1.5208508362513038,
      "learning_rate": 4.433668222335577e-07,
      "loss": 0.7901,
      "step": 9896
    },
    {
      "epoch": 12.113831089351285,
      "grad_norm": 1.3546266302253611,
      "learning_rate": 4.430025218500525e-07,
      "loss": 0.6374,
      "step": 9897
    },
    {
      "epoch": 12.115055079559363,
      "grad_norm": 2.5274244906538925,
      "learning_rate": 4.4263835664481364e-07,
      "loss": 0.5329,
      "step": 9898
    },
    {
      "epoch": 12.116279069767442,
      "grad_norm": 1.898962987957822,
      "learning_rate": 4.4227432664177174e-07,
      "loss": 0.5007,
      "step": 9899
    },
    {
      "epoch": 12.11750305997552,
      "grad_norm": 1.6225671791530905,
      "learning_rate": 4.4191043186484954e-07,
      "loss": 0.7425,
      "step": 9900
    },
    {
      "epoch": 12.1187270501836,
      "grad_norm": 1.796820810908527,
      "learning_rate": 4.4154667233796215e-07,
      "loss": 0.5121,
      "step": 9901
    },
    {
      "epoch": 12.119951040391676,
      "grad_norm": 1.494615173288428,
      "learning_rate": 4.411830480850135e-07,
      "loss": 0.5538,
      "step": 9902
    },
    {
      "epoch": 12.121175030599755,
      "grad_norm": 1.6488432393290298,
      "learning_rate": 4.408195591299e-07,
      "loss": 0.5733,
      "step": 9903
    },
    {
      "epoch": 12.122399020807833,
      "grad_norm": 1.1461736505441344,
      "learning_rate": 4.404562054965089e-07,
      "loss": 0.3658,
      "step": 9904
    },
    {
      "epoch": 12.123623011015912,
      "grad_norm": 1.0758920038287227,
      "learning_rate": 4.4009298720871787e-07,
      "loss": 0.5876,
      "step": 9905
    },
    {
      "epoch": 12.12484700122399,
      "grad_norm": 2.435880849707312,
      "learning_rate": 4.397299042903969e-07,
      "loss": 0.4376,
      "step": 9906
    },
    {
      "epoch": 12.12607099143207,
      "grad_norm": 2.8805629482823836,
      "learning_rate": 4.393669567654063e-07,
      "loss": 0.5575,
      "step": 9907
    },
    {
      "epoch": 12.127294981640146,
      "grad_norm": 2.126941497861444,
      "learning_rate": 4.3900414465759657e-07,
      "loss": 0.3783,
      "step": 9908
    },
    {
      "epoch": 12.128518971848225,
      "grad_norm": 1.580872832876345,
      "learning_rate": 4.3864146799081244e-07,
      "loss": 0.6779,
      "step": 9909
    },
    {
      "epoch": 12.129742962056303,
      "grad_norm": 1.2147103196048037,
      "learning_rate": 4.382789267888868e-07,
      "loss": 0.4409,
      "step": 9910
    },
    {
      "epoch": 12.130966952264382,
      "grad_norm": 1.9571349064248427,
      "learning_rate": 4.3791652107564393e-07,
      "loss": 0.5972,
      "step": 9911
    },
    {
      "epoch": 12.13219094247246,
      "grad_norm": 1.6867800487866882,
      "learning_rate": 4.375542508749001e-07,
      "loss": 0.5576,
      "step": 9912
    },
    {
      "epoch": 12.13341493268054,
      "grad_norm": 1.4019699319878516,
      "learning_rate": 4.371921162104617e-07,
      "loss": 1.0408,
      "step": 9913
    },
    {
      "epoch": 12.134638922888616,
      "grad_norm": 2.214537381654303,
      "learning_rate": 4.3683011710612844e-07,
      "loss": 0.5014,
      "step": 9914
    },
    {
      "epoch": 12.135862913096695,
      "grad_norm": 1.8141860048296625,
      "learning_rate": 4.364682535856887e-07,
      "loss": 0.5727,
      "step": 9915
    },
    {
      "epoch": 12.137086903304773,
      "grad_norm": 2.3518674391936374,
      "learning_rate": 4.361065256729236e-07,
      "loss": 0.3672,
      "step": 9916
    },
    {
      "epoch": 12.138310893512852,
      "grad_norm": 1.7520402836416917,
      "learning_rate": 4.3574493339160224e-07,
      "loss": 0.6169,
      "step": 9917
    },
    {
      "epoch": 12.13953488372093,
      "grad_norm": 1.1273110087370142,
      "learning_rate": 4.3538347676548965e-07,
      "loss": 0.6769,
      "step": 9918
    },
    {
      "epoch": 12.140758873929009,
      "grad_norm": 2.4557490320529634,
      "learning_rate": 4.3502215581833846e-07,
      "loss": 0.511,
      "step": 9919
    },
    {
      "epoch": 12.141982864137088,
      "grad_norm": 1.1527724959225012,
      "learning_rate": 4.34660970573893e-07,
      "loss": 0.6591,
      "step": 9920
    },
    {
      "epoch": 12.143206854345165,
      "grad_norm": 2.180329050129752,
      "learning_rate": 4.342999210558899e-07,
      "loss": 1.115,
      "step": 9921
    },
    {
      "epoch": 12.144430844553243,
      "grad_norm": 2.7513851335840895,
      "learning_rate": 4.33939007288055e-07,
      "loss": 0.4276,
      "step": 9922
    },
    {
      "epoch": 12.145654834761322,
      "grad_norm": 1.9186742962310244,
      "learning_rate": 4.335782292941071e-07,
      "loss": 1.3762,
      "step": 9923
    },
    {
      "epoch": 12.1468788249694,
      "grad_norm": 1.9456019512713338,
      "learning_rate": 4.332175870977548e-07,
      "loss": 0.6794,
      "step": 9924
    },
    {
      "epoch": 12.148102815177479,
      "grad_norm": 1.5483407121093726,
      "learning_rate": 4.3285708072269785e-07,
      "loss": 0.6472,
      "step": 9925
    },
    {
      "epoch": 12.149326805385558,
      "grad_norm": 2.22433313110445,
      "learning_rate": 4.324967101926272e-07,
      "loss": 0.8157,
      "step": 9926
    },
    {
      "epoch": 12.150550795593634,
      "grad_norm": 1.3835871257512595,
      "learning_rate": 4.321364755312271e-07,
      "loss": 0.7076,
      "step": 9927
    },
    {
      "epoch": 12.151774785801713,
      "grad_norm": 2.1274055432143437,
      "learning_rate": 4.3177637676216856e-07,
      "loss": 0.849,
      "step": 9928
    },
    {
      "epoch": 12.152998776009792,
      "grad_norm": 1.4834720329633477,
      "learning_rate": 4.3141641390911675e-07,
      "loss": 1.3031,
      "step": 9929
    },
    {
      "epoch": 12.15422276621787,
      "grad_norm": 2.3829031925716606,
      "learning_rate": 4.310565869957262e-07,
      "loss": 0.4745,
      "step": 9930
    },
    {
      "epoch": 12.155446756425949,
      "grad_norm": 1.775159001372628,
      "learning_rate": 4.306968960456451e-07,
      "loss": 0.4964,
      "step": 9931
    },
    {
      "epoch": 12.156670746634028,
      "grad_norm": 1.7019094221349393,
      "learning_rate": 4.3033734108251013e-07,
      "loss": 0.8537,
      "step": 9932
    },
    {
      "epoch": 12.157894736842104,
      "grad_norm": 2.3290088843286036,
      "learning_rate": 4.299779221299499e-07,
      "loss": 0.5173,
      "step": 9933
    },
    {
      "epoch": 12.159118727050183,
      "grad_norm": 1.8607224235749455,
      "learning_rate": 4.296186392115842e-07,
      "loss": 0.7041,
      "step": 9934
    },
    {
      "epoch": 12.160342717258262,
      "grad_norm": 2.079523877477626,
      "learning_rate": 4.2925949235102377e-07,
      "loss": 0.2887,
      "step": 9935
    },
    {
      "epoch": 12.16156670746634,
      "grad_norm": 1.5694188618569103,
      "learning_rate": 4.289004815718703e-07,
      "loss": 0.912,
      "step": 9936
    },
    {
      "epoch": 12.162790697674419,
      "grad_norm": 1.6331903089491029,
      "learning_rate": 4.2854160689771665e-07,
      "loss": 1.1128,
      "step": 9937
    },
    {
      "epoch": 12.164014687882498,
      "grad_norm": 1.292644305368941,
      "learning_rate": 4.281828683521469e-07,
      "loss": 0.8919,
      "step": 9938
    },
    {
      "epoch": 12.165238678090576,
      "grad_norm": 2.352174852138731,
      "learning_rate": 4.278242659587359e-07,
      "loss": 0.4433,
      "step": 9939
    },
    {
      "epoch": 12.166462668298653,
      "grad_norm": 1.5745901328115093,
      "learning_rate": 4.2746579974104976e-07,
      "loss": 0.6078,
      "step": 9940
    },
    {
      "epoch": 12.167686658506732,
      "grad_norm": 1.5518559801000849,
      "learning_rate": 4.2710746972264547e-07,
      "loss": 1.0694,
      "step": 9941
    },
    {
      "epoch": 12.16891064871481,
      "grad_norm": 1.8807369220365866,
      "learning_rate": 4.2674927592707083e-07,
      "loss": 0.5398,
      "step": 9942
    },
    {
      "epoch": 12.170134638922889,
      "grad_norm": 1.2832099513160966,
      "learning_rate": 4.263912183778651e-07,
      "loss": 0.474,
      "step": 9943
    },
    {
      "epoch": 12.171358629130967,
      "grad_norm": 1.4316272011652191,
      "learning_rate": 4.260332970985592e-07,
      "loss": 0.675,
      "step": 9944
    },
    {
      "epoch": 12.172582619339046,
      "grad_norm": 1.6388463887116504,
      "learning_rate": 4.2567551211267487e-07,
      "loss": 0.5941,
      "step": 9945
    },
    {
      "epoch": 12.173806609547123,
      "grad_norm": 2.622173142198984,
      "learning_rate": 4.253178634437224e-07,
      "loss": 0.6634,
      "step": 9946
    },
    {
      "epoch": 12.175030599755202,
      "grad_norm": 1.5513127763662728,
      "learning_rate": 4.2496035111520587e-07,
      "loss": 0.8675,
      "step": 9947
    },
    {
      "epoch": 12.17625458996328,
      "grad_norm": 2.0108700065424903,
      "learning_rate": 4.246029751506206e-07,
      "loss": 0.4841,
      "step": 9948
    },
    {
      "epoch": 12.177478580171359,
      "grad_norm": 1.6024169841184293,
      "learning_rate": 4.2424573557345167e-07,
      "loss": 0.9785,
      "step": 9949
    },
    {
      "epoch": 12.178702570379437,
      "grad_norm": 2.4820886547416374,
      "learning_rate": 4.23888632407175e-07,
      "loss": 0.5381,
      "step": 9950
    },
    {
      "epoch": 12.179926560587516,
      "grad_norm": 1.6954237683852358,
      "learning_rate": 4.235316656752586e-07,
      "loss": 0.449,
      "step": 9951
    },
    {
      "epoch": 12.181150550795593,
      "grad_norm": 1.9971082501659663,
      "learning_rate": 4.2317483540116053e-07,
      "loss": 0.5131,
      "step": 9952
    },
    {
      "epoch": 12.182374541003671,
      "grad_norm": 2.0228969980278353,
      "learning_rate": 4.22818141608331e-07,
      "loss": 0.9541,
      "step": 9953
    },
    {
      "epoch": 12.18359853121175,
      "grad_norm": 3.1675237629912316,
      "learning_rate": 4.224615843202101e-07,
      "loss": 0.4531,
      "step": 9954
    },
    {
      "epoch": 12.184822521419829,
      "grad_norm": 1.994616569733034,
      "learning_rate": 4.221051635602294e-07,
      "loss": 0.5348,
      "step": 9955
    },
    {
      "epoch": 12.186046511627907,
      "grad_norm": 2.8997049103132784,
      "learning_rate": 4.217488793518121e-07,
      "loss": 0.4413,
      "step": 9956
    },
    {
      "epoch": 12.187270501835986,
      "grad_norm": 1.406741195723224,
      "learning_rate": 4.2139273171837134e-07,
      "loss": 0.8691,
      "step": 9957
    },
    {
      "epoch": 12.188494492044065,
      "grad_norm": 1.8371999451431766,
      "learning_rate": 4.210367206833119e-07,
      "loss": 0.684,
      "step": 9958
    },
    {
      "epoch": 12.189718482252141,
      "grad_norm": 3.084155337624326,
      "learning_rate": 4.2068084627002985e-07,
      "loss": 0.4442,
      "step": 9959
    },
    {
      "epoch": 12.19094247246022,
      "grad_norm": 2.142789729763367,
      "learning_rate": 4.2032510850191085e-07,
      "loss": 0.8283,
      "step": 9960
    },
    {
      "epoch": 12.192166462668299,
      "grad_norm": 2.2191790454478513,
      "learning_rate": 4.199695074023344e-07,
      "loss": 1.098,
      "step": 9961
    },
    {
      "epoch": 12.193390452876377,
      "grad_norm": 2.7477338368596724,
      "learning_rate": 4.1961404299466823e-07,
      "loss": 0.467,
      "step": 9962
    },
    {
      "epoch": 12.194614443084456,
      "grad_norm": 1.9050595841989424,
      "learning_rate": 4.1925871530227314e-07,
      "loss": 0.4124,
      "step": 9963
    },
    {
      "epoch": 12.195838433292534,
      "grad_norm": 2.220837347203319,
      "learning_rate": 4.189035243484982e-07,
      "loss": 1.1708,
      "step": 9964
    },
    {
      "epoch": 12.197062423500611,
      "grad_norm": 2.3434813200185336,
      "learning_rate": 4.185484701566858e-07,
      "loss": 1.3293,
      "step": 9965
    },
    {
      "epoch": 12.19828641370869,
      "grad_norm": 1.8929994169967128,
      "learning_rate": 4.181935527501699e-07,
      "loss": 0.4477,
      "step": 9966
    },
    {
      "epoch": 12.199510403916769,
      "grad_norm": 2.47015602636036,
      "learning_rate": 4.1783877215227363e-07,
      "loss": 0.5404,
      "step": 9967
    },
    {
      "epoch": 12.200734394124847,
      "grad_norm": 1.9153343766803266,
      "learning_rate": 4.174841283863118e-07,
      "loss": 0.3847,
      "step": 9968
    },
    {
      "epoch": 12.201958384332926,
      "grad_norm": 1.7100869861670718,
      "learning_rate": 4.1712962147559045e-07,
      "loss": 1.3291,
      "step": 9969
    },
    {
      "epoch": 12.203182374541004,
      "grad_norm": 1.1849610023040595,
      "learning_rate": 4.167752514434062e-07,
      "loss": 0.668,
      "step": 9970
    },
    {
      "epoch": 12.204406364749081,
      "grad_norm": 2.1366878715968047,
      "learning_rate": 4.164210183130471e-07,
      "loss": 0.6623,
      "step": 9971
    },
    {
      "epoch": 12.20563035495716,
      "grad_norm": 2.286220997230615,
      "learning_rate": 4.1606692210779203e-07,
      "loss": 0.4686,
      "step": 9972
    },
    {
      "epoch": 12.206854345165238,
      "grad_norm": 1.4050380501273065,
      "learning_rate": 4.1571296285091046e-07,
      "loss": 0.5532,
      "step": 9973
    },
    {
      "epoch": 12.208078335373317,
      "grad_norm": 2.0542360339015735,
      "learning_rate": 4.1535914056566484e-07,
      "loss": 1.1811,
      "step": 9974
    },
    {
      "epoch": 12.209302325581396,
      "grad_norm": 1.527193322654163,
      "learning_rate": 4.150054552753055e-07,
      "loss": 0.611,
      "step": 9975
    },
    {
      "epoch": 12.210526315789474,
      "grad_norm": 1.771356214462141,
      "learning_rate": 4.146519070030758e-07,
      "loss": 0.4362,
      "step": 9976
    },
    {
      "epoch": 12.211750305997551,
      "grad_norm": 1.2592842422992234,
      "learning_rate": 4.1429849577220933e-07,
      "loss": 0.5758,
      "step": 9977
    },
    {
      "epoch": 12.21297429620563,
      "grad_norm": 2.3607252046550555,
      "learning_rate": 4.139452216059309e-07,
      "loss": 1.4017,
      "step": 9978
    },
    {
      "epoch": 12.214198286413708,
      "grad_norm": 1.5527299601845577,
      "learning_rate": 4.1359208452745713e-07,
      "loss": 0.6659,
      "step": 9979
    },
    {
      "epoch": 12.215422276621787,
      "grad_norm": 1.1831708547825637,
      "learning_rate": 4.132390845599946e-07,
      "loss": 0.7424,
      "step": 9980
    },
    {
      "epoch": 12.216646266829866,
      "grad_norm": 2.333177974027238,
      "learning_rate": 4.1288622172674155e-07,
      "loss": 0.5909,
      "step": 9981
    },
    {
      "epoch": 12.217870257037944,
      "grad_norm": 2.0052352550097132,
      "learning_rate": 4.1253349605088506e-07,
      "loss": 0.7029,
      "step": 9982
    },
    {
      "epoch": 12.219094247246023,
      "grad_norm": 1.7383829323130149,
      "learning_rate": 4.1218090755560685e-07,
      "loss": 0.5044,
      "step": 9983
    },
    {
      "epoch": 12.2203182374541,
      "grad_norm": 2.5088337749745255,
      "learning_rate": 4.118284562640773e-07,
      "loss": 0.48,
      "step": 9984
    },
    {
      "epoch": 12.221542227662178,
      "grad_norm": 1.1694476124081807,
      "learning_rate": 4.1147614219945785e-07,
      "loss": 0.6165,
      "step": 9985
    },
    {
      "epoch": 12.222766217870257,
      "grad_norm": 1.8218305335472436,
      "learning_rate": 4.111239653849014e-07,
      "loss": 0.4028,
      "step": 9986
    },
    {
      "epoch": 12.223990208078336,
      "grad_norm": 0.8524906928475731,
      "learning_rate": 4.1077192584355165e-07,
      "loss": 0.4316,
      "step": 9987
    },
    {
      "epoch": 12.225214198286414,
      "grad_norm": 1.6003722853086957,
      "learning_rate": 4.1042002359854336e-07,
      "loss": 0.5875,
      "step": 9988
    },
    {
      "epoch": 12.226438188494493,
      "grad_norm": 1.2728991174416264,
      "learning_rate": 4.100682586730026e-07,
      "loss": 0.7642,
      "step": 9989
    },
    {
      "epoch": 12.22766217870257,
      "grad_norm": 1.9070716802546566,
      "learning_rate": 4.097166310900455e-07,
      "loss": 0.4348,
      "step": 9990
    },
    {
      "epoch": 12.228886168910648,
      "grad_norm": 1.3733714915735937,
      "learning_rate": 4.0936514087277943e-07,
      "loss": 0.488,
      "step": 9991
    },
    {
      "epoch": 12.230110159118727,
      "grad_norm": 1.3757391297952897,
      "learning_rate": 4.0901378804430486e-07,
      "loss": 0.4597,
      "step": 9992
    },
    {
      "epoch": 12.231334149326806,
      "grad_norm": 1.5502071189758553,
      "learning_rate": 4.0866257262770935e-07,
      "loss": 0.6243,
      "step": 9993
    },
    {
      "epoch": 12.232558139534884,
      "grad_norm": 2.4335702767304084,
      "learning_rate": 4.083114946460745e-07,
      "loss": 0.4095,
      "step": 9994
    },
    {
      "epoch": 12.233782129742963,
      "grad_norm": 1.9912566742233184,
      "learning_rate": 4.079605541224707e-07,
      "loss": 0.6199,
      "step": 9995
    },
    {
      "epoch": 12.23500611995104,
      "grad_norm": 1.2020196701503614,
      "learning_rate": 4.076097510799623e-07,
      "loss": 0.879,
      "step": 9996
    },
    {
      "epoch": 12.236230110159118,
      "grad_norm": 2.0340832703288165,
      "learning_rate": 4.0725908554160173e-07,
      "loss": 0.722,
      "step": 9997
    },
    {
      "epoch": 12.237454100367197,
      "grad_norm": 1.493228722510594,
      "learning_rate": 4.0690855753043404e-07,
      "loss": 0.6628,
      "step": 9998
    },
    {
      "epoch": 12.238678090575275,
      "grad_norm": 2.8264583789260915,
      "learning_rate": 4.065581670694929e-07,
      "loss": 0.5739,
      "step": 9999
    },
    {
      "epoch": 12.239902080783354,
      "grad_norm": 1.5619494379292729,
      "learning_rate": 4.0620791418180677e-07,
      "loss": 0.9795,
      "step": 10000
    },
    {
      "epoch": 12.241126070991433,
      "grad_norm": 2.755488288692661,
      "learning_rate": 4.0585779889039194e-07,
      "loss": 0.4146,
      "step": 10001
    },
    {
      "epoch": 12.24235006119951,
      "grad_norm": 2.586528908266464,
      "learning_rate": 4.055078212182567e-07,
      "loss": 0.2242,
      "step": 10002
    },
    {
      "epoch": 12.243574051407588,
      "grad_norm": 2.7427314668334923,
      "learning_rate": 4.051579811884007e-07,
      "loss": 0.7832,
      "step": 10003
    },
    {
      "epoch": 12.244798041615667,
      "grad_norm": 1.918886883613435,
      "learning_rate": 4.048082788238139e-07,
      "loss": 0.9207,
      "step": 10004
    },
    {
      "epoch": 12.246022031823745,
      "grad_norm": 1.1061764334775783,
      "learning_rate": 4.0445871414747733e-07,
      "loss": 0.5047,
      "step": 10005
    },
    {
      "epoch": 12.247246022031824,
      "grad_norm": 2.0847005800472758,
      "learning_rate": 4.041092871823632e-07,
      "loss": 1.1348,
      "step": 10006
    },
    {
      "epoch": 12.248470012239903,
      "grad_norm": 2.806116138859567,
      "learning_rate": 4.037599979514345e-07,
      "loss": 0.5075,
      "step": 10007
    },
    {
      "epoch": 12.249694002447981,
      "grad_norm": 1.8668924833122575,
      "learning_rate": 4.034108464776448e-07,
      "loss": 1.3398,
      "step": 10008
    },
    {
      "epoch": 12.250917992656058,
      "grad_norm": 2.5171763428542104,
      "learning_rate": 4.030618327839403e-07,
      "loss": 0.4803,
      "step": 10009
    },
    {
      "epoch": 12.252141982864137,
      "grad_norm": 3.742339060151395,
      "learning_rate": 4.027129568932564e-07,
      "loss": 0.4671,
      "step": 10010
    },
    {
      "epoch": 12.253365973072215,
      "grad_norm": 1.7408260289125281,
      "learning_rate": 4.02364218828519e-07,
      "loss": 0.9015,
      "step": 10011
    },
    {
      "epoch": 12.254589963280294,
      "grad_norm": 1.49256320180312,
      "learning_rate": 4.020156186126467e-07,
      "loss": 0.7947,
      "step": 10012
    },
    {
      "epoch": 12.255813953488373,
      "grad_norm": 2.8831431496059428,
      "learning_rate": 4.016671562685476e-07,
      "loss": 0.2715,
      "step": 10013
    },
    {
      "epoch": 12.257037943696451,
      "grad_norm": 3.3185030395573585,
      "learning_rate": 4.013188318191222e-07,
      "loss": 0.677,
      "step": 10014
    },
    {
      "epoch": 12.258261933904528,
      "grad_norm": 1.9840215712208988,
      "learning_rate": 4.0097064528726056e-07,
      "loss": 0.5866,
      "step": 10015
    },
    {
      "epoch": 12.259485924112607,
      "grad_norm": 1.9107566388762782,
      "learning_rate": 4.006225966958452e-07,
      "loss": 0.4599,
      "step": 10016
    },
    {
      "epoch": 12.260709914320685,
      "grad_norm": 2.0038572057368507,
      "learning_rate": 4.0027468606774615e-07,
      "loss": 0.3535,
      "step": 10017
    },
    {
      "epoch": 12.261933904528764,
      "grad_norm": 1.3988257899836924,
      "learning_rate": 3.9992691342582933e-07,
      "loss": 1.086,
      "step": 10018
    },
    {
      "epoch": 12.263157894736842,
      "grad_norm": 1.6438355637458306,
      "learning_rate": 3.995792787929481e-07,
      "loss": 0.5651,
      "step": 10019
    },
    {
      "epoch": 12.264381884944921,
      "grad_norm": 1.868734178806193,
      "learning_rate": 3.9923178219194753e-07,
      "loss": 1.1838,
      "step": 10020
    },
    {
      "epoch": 12.265605875152998,
      "grad_norm": 3.398866594945571,
      "learning_rate": 3.988844236456643e-07,
      "loss": 0.5147,
      "step": 10021
    },
    {
      "epoch": 12.266829865361077,
      "grad_norm": 2.000663868568739,
      "learning_rate": 3.985372031769252e-07,
      "loss": 0.9884,
      "step": 10022
    },
    {
      "epoch": 12.268053855569155,
      "grad_norm": 1.3799089394145148,
      "learning_rate": 3.981901208085481e-07,
      "loss": 0.5221,
      "step": 10023
    },
    {
      "epoch": 12.269277845777234,
      "grad_norm": 2.6092420017637155,
      "learning_rate": 3.9784317656334256e-07,
      "loss": 1.02,
      "step": 10024
    },
    {
      "epoch": 12.270501835985312,
      "grad_norm": 2.1413183405983784,
      "learning_rate": 3.9749637046410774e-07,
      "loss": 0.4281,
      "step": 10025
    },
    {
      "epoch": 12.271725826193391,
      "grad_norm": 1.5400396788837625,
      "learning_rate": 3.9714970253363463e-07,
      "loss": 1.2244,
      "step": 10026
    },
    {
      "epoch": 12.27294981640147,
      "grad_norm": 1.149487698949321,
      "learning_rate": 3.9680317279470555e-07,
      "loss": 0.6113,
      "step": 10027
    },
    {
      "epoch": 12.274173806609546,
      "grad_norm": 1.9384348488266716,
      "learning_rate": 3.964567812700934e-07,
      "loss": 0.9577,
      "step": 10028
    },
    {
      "epoch": 12.275397796817625,
      "grad_norm": 2.047353002800829,
      "learning_rate": 3.9611052798256025e-07,
      "loss": 0.5269,
      "step": 10029
    },
    {
      "epoch": 12.276621787025704,
      "grad_norm": 2.1860460927732888,
      "learning_rate": 3.957644129548613e-07,
      "loss": 0.9645,
      "step": 10030
    },
    {
      "epoch": 12.277845777233782,
      "grad_norm": 0.9468071453126047,
      "learning_rate": 3.954184362097424e-07,
      "loss": 0.4619,
      "step": 10031
    },
    {
      "epoch": 12.279069767441861,
      "grad_norm": 1.1102950184187865,
      "learning_rate": 3.950725977699396e-07,
      "loss": 0.5347,
      "step": 10032
    },
    {
      "epoch": 12.28029375764994,
      "grad_norm": 2.1567358049812326,
      "learning_rate": 3.947268976581803e-07,
      "loss": 1.188,
      "step": 10033
    },
    {
      "epoch": 12.281517747858016,
      "grad_norm": 1.2457808252523448,
      "learning_rate": 3.943813358971821e-07,
      "loss": 0.8422,
      "step": 10034
    },
    {
      "epoch": 12.282741738066095,
      "grad_norm": 1.3278506445009297,
      "learning_rate": 3.9403591250965464e-07,
      "loss": 0.5874,
      "step": 10035
    },
    {
      "epoch": 12.283965728274174,
      "grad_norm": 2.0246986452607434,
      "learning_rate": 3.936906275182975e-07,
      "loss": 0.7229,
      "step": 10036
    },
    {
      "epoch": 12.285189718482252,
      "grad_norm": 2.6648738550025857,
      "learning_rate": 3.933454809458015e-07,
      "loss": 0.6685,
      "step": 10037
    },
    {
      "epoch": 12.286413708690331,
      "grad_norm": 2.3367123683774307,
      "learning_rate": 3.930004728148487e-07,
      "loss": 0.668,
      "step": 10038
    },
    {
      "epoch": 12.28763769889841,
      "grad_norm": 1.5690994989515628,
      "learning_rate": 3.926556031481113e-07,
      "loss": 0.9257,
      "step": 10039
    },
    {
      "epoch": 12.288861689106486,
      "grad_norm": 1.7680102562467166,
      "learning_rate": 3.9231087196825335e-07,
      "loss": 0.6557,
      "step": 10040
    },
    {
      "epoch": 12.290085679314565,
      "grad_norm": 2.1810152576741904,
      "learning_rate": 3.9196627929792926e-07,
      "loss": 0.298,
      "step": 10041
    },
    {
      "epoch": 12.291309669522644,
      "grad_norm": 1.6861214474210215,
      "learning_rate": 3.9162182515978394e-07,
      "loss": 0.6972,
      "step": 10042
    },
    {
      "epoch": 12.292533659730722,
      "grad_norm": 1.421759382476858,
      "learning_rate": 3.912775095764534e-07,
      "loss": 0.3562,
      "step": 10043
    },
    {
      "epoch": 12.2937576499388,
      "grad_norm": 1.2701974788846186,
      "learning_rate": 3.909333325705658e-07,
      "loss": 0.4694,
      "step": 10044
    },
    {
      "epoch": 12.29498164014688,
      "grad_norm": 1.1425560510057682,
      "learning_rate": 3.905892941647393e-07,
      "loss": 0.5332,
      "step": 10045
    },
    {
      "epoch": 12.296205630354958,
      "grad_norm": 2.767562164639848,
      "learning_rate": 3.902453943815818e-07,
      "loss": 0.4946,
      "step": 10046
    },
    {
      "epoch": 12.297429620563035,
      "grad_norm": 1.173317110730678,
      "learning_rate": 3.8990163324369277e-07,
      "loss": 0.8449,
      "step": 10047
    },
    {
      "epoch": 12.298653610771114,
      "grad_norm": 1.6572332160276138,
      "learning_rate": 3.8955801077366463e-07,
      "loss": 0.4937,
      "step": 10048
    },
    {
      "epoch": 12.299877600979192,
      "grad_norm": 1.8582740884911884,
      "learning_rate": 3.8921452699407753e-07,
      "loss": 0.4629,
      "step": 10049
    },
    {
      "epoch": 12.30110159118727,
      "grad_norm": 2.10424507030358,
      "learning_rate": 3.8887118192750485e-07,
      "loss": 0.4348,
      "step": 10050
    },
    {
      "epoch": 12.30232558139535,
      "grad_norm": 2.5629995728757953,
      "learning_rate": 3.8852797559650936e-07,
      "loss": 0.4017,
      "step": 10051
    },
    {
      "epoch": 12.303549571603428,
      "grad_norm": 1.4077115386922814,
      "learning_rate": 3.881849080236455e-07,
      "loss": 0.6755,
      "step": 10052
    },
    {
      "epoch": 12.304773561811505,
      "grad_norm": 2.2576209123480995,
      "learning_rate": 3.878419792314586e-07,
      "loss": 0.5676,
      "step": 10053
    },
    {
      "epoch": 12.305997552019583,
      "grad_norm": 1.2980430776017775,
      "learning_rate": 3.8749918924248453e-07,
      "loss": 0.6431,
      "step": 10054
    },
    {
      "epoch": 12.307221542227662,
      "grad_norm": 2.48994872691508,
      "learning_rate": 3.8715653807924996e-07,
      "loss": 0.8554,
      "step": 10055
    },
    {
      "epoch": 12.30844553243574,
      "grad_norm": 1.6072094839106092,
      "learning_rate": 3.8681402576427216e-07,
      "loss": 0.5316,
      "step": 10056
    },
    {
      "epoch": 12.30966952264382,
      "grad_norm": 2.9813305660343516,
      "learning_rate": 3.864716523200618e-07,
      "loss": 0.4144,
      "step": 10057
    },
    {
      "epoch": 12.310893512851898,
      "grad_norm": 1.2717284139284284,
      "learning_rate": 3.861294177691163e-07,
      "loss": 0.6809,
      "step": 10058
    },
    {
      "epoch": 12.312117503059975,
      "grad_norm": 1.1745400635565966,
      "learning_rate": 3.85787322133927e-07,
      "loss": 0.6074,
      "step": 10059
    },
    {
      "epoch": 12.313341493268053,
      "grad_norm": 2.153387970241339,
      "learning_rate": 3.854453654369744e-07,
      "loss": 0.6466,
      "step": 10060
    },
    {
      "epoch": 12.314565483476132,
      "grad_norm": 2.177572509072106,
      "learning_rate": 3.851035477007317e-07,
      "loss": 0.5688,
      "step": 10061
    },
    {
      "epoch": 12.31578947368421,
      "grad_norm": 1.2524280924900757,
      "learning_rate": 3.847618689476612e-07,
      "loss": 0.5693,
      "step": 10062
    },
    {
      "epoch": 12.31701346389229,
      "grad_norm": 1.5581670551342393,
      "learning_rate": 3.844203292002177e-07,
      "loss": 0.5693,
      "step": 10063
    },
    {
      "epoch": 12.318237454100368,
      "grad_norm": 2.187174698495828,
      "learning_rate": 3.840789284808447e-07,
      "loss": 0.5605,
      "step": 10064
    },
    {
      "epoch": 12.319461444308445,
      "grad_norm": 1.5820517206423277,
      "learning_rate": 3.8373766681197745e-07,
      "loss": 0.2198,
      "step": 10065
    },
    {
      "epoch": 12.320685434516523,
      "grad_norm": 2.505797087811427,
      "learning_rate": 3.833965442160439e-07,
      "loss": 0.4487,
      "step": 10066
    },
    {
      "epoch": 12.321909424724602,
      "grad_norm": 1.1053389283515012,
      "learning_rate": 3.8305556071546097e-07,
      "loss": 0.5269,
      "step": 10067
    },
    {
      "epoch": 12.32313341493268,
      "grad_norm": 2.878333556506974,
      "learning_rate": 3.827147163326361e-07,
      "loss": 0.5727,
      "step": 10068
    },
    {
      "epoch": 12.32435740514076,
      "grad_norm": 1.5763898240411935,
      "learning_rate": 3.8237401108996903e-07,
      "loss": 1.0784,
      "step": 10069
    },
    {
      "epoch": 12.325581395348838,
      "grad_norm": 1.6527215375752544,
      "learning_rate": 3.8203344500984924e-07,
      "loss": 1.1636,
      "step": 10070
    },
    {
      "epoch": 12.326805385556916,
      "grad_norm": 3.636656996534286,
      "learning_rate": 3.8169301811465773e-07,
      "loss": 0.7551,
      "step": 10071
    },
    {
      "epoch": 12.328029375764993,
      "grad_norm": 1.6569435088492923,
      "learning_rate": 3.8135273042676613e-07,
      "loss": 0.2917,
      "step": 10072
    },
    {
      "epoch": 12.329253365973072,
      "grad_norm": 2.524449712059774,
      "learning_rate": 3.810125819685359e-07,
      "loss": 0.4955,
      "step": 10073
    },
    {
      "epoch": 12.33047735618115,
      "grad_norm": 1.9888043601032568,
      "learning_rate": 3.8067257276232255e-07,
      "loss": 0.549,
      "step": 10074
    },
    {
      "epoch": 12.331701346389229,
      "grad_norm": 1.2367683666664937,
      "learning_rate": 3.803327028304682e-07,
      "loss": 0.5383,
      "step": 10075
    },
    {
      "epoch": 12.332925336597308,
      "grad_norm": 1.5620771403302534,
      "learning_rate": 3.799929721953086e-07,
      "loss": 0.4872,
      "step": 10076
    },
    {
      "epoch": 12.334149326805386,
      "grad_norm": 1.5615057197115185,
      "learning_rate": 3.796533808791694e-07,
      "loss": 0.8598,
      "step": 10077
    },
    {
      "epoch": 12.335373317013463,
      "grad_norm": 1.6426163632338924,
      "learning_rate": 3.793139289043668e-07,
      "loss": 0.3521,
      "step": 10078
    },
    {
      "epoch": 12.336597307221542,
      "grad_norm": 1.147273181998868,
      "learning_rate": 3.789746162932098e-07,
      "loss": 0.5607,
      "step": 10079
    },
    {
      "epoch": 12.33782129742962,
      "grad_norm": 1.6076230362467414,
      "learning_rate": 3.786354430679956e-07,
      "loss": 0.3995,
      "step": 10080
    },
    {
      "epoch": 12.339045287637699,
      "grad_norm": 1.1851554721726174,
      "learning_rate": 3.7829640925101413e-07,
      "loss": 0.3806,
      "step": 10081
    },
    {
      "epoch": 12.340269277845778,
      "grad_norm": 2.1309015447436046,
      "learning_rate": 3.7795751486454425e-07,
      "loss": 0.4354,
      "step": 10082
    },
    {
      "epoch": 12.341493268053856,
      "grad_norm": 2.5154015374187098,
      "learning_rate": 3.7761875993085783e-07,
      "loss": 0.7047,
      "step": 10083
    },
    {
      "epoch": 12.342717258261933,
      "grad_norm": 3.162608994604238,
      "learning_rate": 3.772801444722166e-07,
      "loss": 0.3989,
      "step": 10084
    },
    {
      "epoch": 12.343941248470012,
      "grad_norm": 1.5813263898851395,
      "learning_rate": 3.769416685108726e-07,
      "loss": 0.887,
      "step": 10085
    },
    {
      "epoch": 12.34516523867809,
      "grad_norm": 2.6767825295849503,
      "learning_rate": 3.7660333206906986e-07,
      "loss": 0.3531,
      "step": 10086
    },
    {
      "epoch": 12.346389228886169,
      "grad_norm": 1.7641467857458457,
      "learning_rate": 3.762651351690419e-07,
      "loss": 0.9931,
      "step": 10087
    },
    {
      "epoch": 12.347613219094248,
      "grad_norm": 2.683230107274646,
      "learning_rate": 3.759270778330143e-07,
      "loss": 0.3919,
      "step": 10088
    },
    {
      "epoch": 12.348837209302326,
      "grad_norm": 2.704333470274218,
      "learning_rate": 3.7558916008320263e-07,
      "loss": 0.5296,
      "step": 10089
    },
    {
      "epoch": 12.350061199510403,
      "grad_norm": 1.4592034155164992,
      "learning_rate": 3.752513819418138e-07,
      "loss": 0.4301,
      "step": 10090
    },
    {
      "epoch": 12.351285189718482,
      "grad_norm": 1.9298902624247505,
      "learning_rate": 3.7491374343104457e-07,
      "loss": 0.3473,
      "step": 10091
    },
    {
      "epoch": 12.35250917992656,
      "grad_norm": 1.6783151816565625,
      "learning_rate": 3.745762445730852e-07,
      "loss": 0.6852,
      "step": 10092
    },
    {
      "epoch": 12.353733170134639,
      "grad_norm": 2.4251828365747192,
      "learning_rate": 3.742388853901133e-07,
      "loss": 0.9,
      "step": 10093
    },
    {
      "epoch": 12.354957160342718,
      "grad_norm": 1.5110005712408319,
      "learning_rate": 3.7390166590429876e-07,
      "loss": 0.997,
      "step": 10094
    },
    {
      "epoch": 12.356181150550796,
      "grad_norm": 2.0893612489991757,
      "learning_rate": 3.735645861378026e-07,
      "loss": 0.7291,
      "step": 10095
    },
    {
      "epoch": 12.357405140758875,
      "grad_norm": 1.7467157352317273,
      "learning_rate": 3.732276461127771e-07,
      "loss": 1.2255,
      "step": 10096
    },
    {
      "epoch": 12.358629130966952,
      "grad_norm": 3.330642419214167,
      "learning_rate": 3.728908458513644e-07,
      "loss": 0.3419,
      "step": 10097
    },
    {
      "epoch": 12.35985312117503,
      "grad_norm": 1.7491526154124082,
      "learning_rate": 3.725541853756975e-07,
      "loss": 0.9244,
      "step": 10098
    },
    {
      "epoch": 12.361077111383109,
      "grad_norm": 0.8664559294467663,
      "learning_rate": 3.7221766470790093e-07,
      "loss": 0.3488,
      "step": 10099
    },
    {
      "epoch": 12.362301101591187,
      "grad_norm": 1.069984346174424,
      "learning_rate": 3.7188128387008915e-07,
      "loss": 0.5082,
      "step": 10100
    },
    {
      "epoch": 12.363525091799266,
      "grad_norm": 1.3703224266725897,
      "learning_rate": 3.715450428843681e-07,
      "loss": 0.5301,
      "step": 10101
    },
    {
      "epoch": 12.364749082007345,
      "grad_norm": 1.514110692644841,
      "learning_rate": 3.71208941772834e-07,
      "loss": 0.4258,
      "step": 10102
    },
    {
      "epoch": 12.365973072215422,
      "grad_norm": 2.2832271872744956,
      "learning_rate": 3.7087298055757473e-07,
      "loss": 1.0094,
      "step": 10103
    },
    {
      "epoch": 12.3671970624235,
      "grad_norm": 1.3743471999808747,
      "learning_rate": 3.70537159260668e-07,
      "loss": 0.6679,
      "step": 10104
    },
    {
      "epoch": 12.368421052631579,
      "grad_norm": 1.4145066924085268,
      "learning_rate": 3.7020147790418266e-07,
      "loss": 1.3274,
      "step": 10105
    },
    {
      "epoch": 12.369645042839657,
      "grad_norm": 1.7766840737330394,
      "learning_rate": 3.698659365101787e-07,
      "loss": 0.7265,
      "step": 10106
    },
    {
      "epoch": 12.370869033047736,
      "grad_norm": 2.675778544477697,
      "learning_rate": 3.6953053510070643e-07,
      "loss": 0.9252,
      "step": 10107
    },
    {
      "epoch": 12.372093023255815,
      "grad_norm": 1.2062622739389488,
      "learning_rate": 3.6919527369780687e-07,
      "loss": 0.4964,
      "step": 10108
    },
    {
      "epoch": 12.373317013463891,
      "grad_norm": 1.6745607747112512,
      "learning_rate": 3.688601523235133e-07,
      "loss": 1.5521,
      "step": 10109
    },
    {
      "epoch": 12.37454100367197,
      "grad_norm": 2.1514967477965796,
      "learning_rate": 3.685251709998486e-07,
      "loss": 0.3061,
      "step": 10110
    },
    {
      "epoch": 12.375764993880049,
      "grad_norm": 1.1910871345374336,
      "learning_rate": 3.6819032974882513e-07,
      "loss": 0.4731,
      "step": 10111
    },
    {
      "epoch": 12.376988984088127,
      "grad_norm": 2.8872679051790873,
      "learning_rate": 3.6785562859244754e-07,
      "loss": 0.4385,
      "step": 10112
    },
    {
      "epoch": 12.378212974296206,
      "grad_norm": 2.446572264942344,
      "learning_rate": 3.6752106755271245e-07,
      "loss": 1.1939,
      "step": 10113
    },
    {
      "epoch": 12.379436964504285,
      "grad_norm": 1.8775611224207334,
      "learning_rate": 3.671866466516055e-07,
      "loss": 0.639,
      "step": 10114
    },
    {
      "epoch": 12.380660954712361,
      "grad_norm": 1.952213120907043,
      "learning_rate": 3.6685236591110314e-07,
      "loss": 1.0037,
      "step": 10115
    },
    {
      "epoch": 12.38188494492044,
      "grad_norm": 1.2054140475732251,
      "learning_rate": 3.6651822535317404e-07,
      "loss": 0.4778,
      "step": 10116
    },
    {
      "epoch": 12.383108935128519,
      "grad_norm": 1.642683215168807,
      "learning_rate": 3.661842249997746e-07,
      "loss": 0.4719,
      "step": 10117
    },
    {
      "epoch": 12.384332925336597,
      "grad_norm": 2.0414246015244153,
      "learning_rate": 3.658503648728562e-07,
      "loss": 0.7191,
      "step": 10118
    },
    {
      "epoch": 12.385556915544676,
      "grad_norm": 1.5978917418614638,
      "learning_rate": 3.6551664499435814e-07,
      "loss": 1.5683,
      "step": 10119
    },
    {
      "epoch": 12.386780905752754,
      "grad_norm": 1.1532625286040887,
      "learning_rate": 3.651830653862115e-07,
      "loss": 0.6375,
      "step": 10120
    },
    {
      "epoch": 12.388004895960833,
      "grad_norm": 1.4591648550459952,
      "learning_rate": 3.648496260703374e-07,
      "loss": 1.1451,
      "step": 10121
    },
    {
      "epoch": 12.38922888616891,
      "grad_norm": 3.5826061051209557,
      "learning_rate": 3.645163270686483e-07,
      "loss": 0.4048,
      "step": 10122
    },
    {
      "epoch": 12.390452876376989,
      "grad_norm": 1.0436087263399831,
      "learning_rate": 3.6418316840304805e-07,
      "loss": 0.4375,
      "step": 10123
    },
    {
      "epoch": 12.391676866585067,
      "grad_norm": 1.5465518830442484,
      "learning_rate": 3.6385015009542973e-07,
      "loss": 0.6433,
      "step": 10124
    },
    {
      "epoch": 12.392900856793146,
      "grad_norm": 2.314928098606368,
      "learning_rate": 3.63517272167678e-07,
      "loss": 0.6253,
      "step": 10125
    },
    {
      "epoch": 12.394124847001224,
      "grad_norm": 2.639917217968421,
      "learning_rate": 3.631845346416696e-07,
      "loss": 0.5712,
      "step": 10126
    },
    {
      "epoch": 12.395348837209303,
      "grad_norm": 1.9566367336943487,
      "learning_rate": 3.6285193753927e-07,
      "loss": 1.1005,
      "step": 10127
    },
    {
      "epoch": 12.39657282741738,
      "grad_norm": 1.4547435787002216,
      "learning_rate": 3.625194808823368e-07,
      "loss": 0.3622,
      "step": 10128
    },
    {
      "epoch": 12.397796817625458,
      "grad_norm": 2.235789977544165,
      "learning_rate": 3.6218716469271665e-07,
      "loss": 0.4787,
      "step": 10129
    },
    {
      "epoch": 12.399020807833537,
      "grad_norm": 2.0083394342922314,
      "learning_rate": 3.618549889922482e-07,
      "loss": 0.3923,
      "step": 10130
    },
    {
      "epoch": 12.400244798041616,
      "grad_norm": 1.4614367779830644,
      "learning_rate": 3.6152295380276234e-07,
      "loss": 0.5277,
      "step": 10131
    },
    {
      "epoch": 12.401468788249694,
      "grad_norm": 2.1602435008547083,
      "learning_rate": 3.611910591460779e-07,
      "loss": 0.3836,
      "step": 10132
    },
    {
      "epoch": 12.402692778457773,
      "grad_norm": 1.618286820905577,
      "learning_rate": 3.608593050440065e-07,
      "loss": 0.9078,
      "step": 10133
    },
    {
      "epoch": 12.403916768665852,
      "grad_norm": 2.157167795412002,
      "learning_rate": 3.605276915183492e-07,
      "loss": 1.1983,
      "step": 10134
    },
    {
      "epoch": 12.405140758873928,
      "grad_norm": 1.7091084156789649,
      "learning_rate": 3.601962185908986e-07,
      "loss": 1.5503,
      "step": 10135
    },
    {
      "epoch": 12.406364749082007,
      "grad_norm": 1.8212733841991549,
      "learning_rate": 3.5986488628343807e-07,
      "loss": 0.4107,
      "step": 10136
    },
    {
      "epoch": 12.407588739290086,
      "grad_norm": 1.4801638111471427,
      "learning_rate": 3.5953369461774134e-07,
      "loss": 1.1552,
      "step": 10137
    },
    {
      "epoch": 12.408812729498164,
      "grad_norm": 2.72914681221252,
      "learning_rate": 3.5920264361557266e-07,
      "loss": 0.9435,
      "step": 10138
    },
    {
      "epoch": 12.410036719706243,
      "grad_norm": 2.253379564449145,
      "learning_rate": 3.588717332986891e-07,
      "loss": 0.9155,
      "step": 10139
    },
    {
      "epoch": 12.411260709914322,
      "grad_norm": 1.685581444068458,
      "learning_rate": 3.585409636888351e-07,
      "loss": 0.3814,
      "step": 10140
    },
    {
      "epoch": 12.412484700122398,
      "grad_norm": 2.3636302590825076,
      "learning_rate": 3.5821033480774843e-07,
      "loss": 0.3429,
      "step": 10141
    },
    {
      "epoch": 12.413708690330477,
      "grad_norm": 2.657079168393158,
      "learning_rate": 3.578798466771563e-07,
      "loss": 0.6153,
      "step": 10142
    },
    {
      "epoch": 12.414932680538556,
      "grad_norm": 1.201482044576642,
      "learning_rate": 3.5754949931877725e-07,
      "loss": 0.5578,
      "step": 10143
    },
    {
      "epoch": 12.416156670746634,
      "grad_norm": 1.8890586575070734,
      "learning_rate": 3.572192927543211e-07,
      "loss": 1.5857,
      "step": 10144
    },
    {
      "epoch": 12.417380660954713,
      "grad_norm": 1.76857961487169,
      "learning_rate": 3.5688922700548826e-07,
      "loss": 0.3007,
      "step": 10145
    },
    {
      "epoch": 12.418604651162791,
      "grad_norm": 1.791252915491405,
      "learning_rate": 3.5655930209396784e-07,
      "loss": 1.6101,
      "step": 10146
    },
    {
      "epoch": 12.419828641370868,
      "grad_norm": 1.4936640102579652,
      "learning_rate": 3.562295180414413e-07,
      "loss": 1.4062,
      "step": 10147
    },
    {
      "epoch": 12.421052631578947,
      "grad_norm": 1.626191404677855,
      "learning_rate": 3.558998748695824e-07,
      "loss": 0.5077,
      "step": 10148
    },
    {
      "epoch": 12.422276621787026,
      "grad_norm": 1.783319789473257,
      "learning_rate": 3.555703726000531e-07,
      "loss": 1.0902,
      "step": 10149
    },
    {
      "epoch": 12.423500611995104,
      "grad_norm": 1.9335713097662122,
      "learning_rate": 3.5524101125450737e-07,
      "loss": 1.2087,
      "step": 10150
    },
    {
      "epoch": 12.424724602203183,
      "grad_norm": 0.5943678039609892,
      "learning_rate": 3.549117908545893e-07,
      "loss": 0.1286,
      "step": 10151
    },
    {
      "epoch": 12.425948592411261,
      "grad_norm": 2.141183373080161,
      "learning_rate": 3.545827114219344e-07,
      "loss": 0.954,
      "step": 10152
    },
    {
      "epoch": 12.427172582619338,
      "grad_norm": 1.315501832948651,
      "learning_rate": 3.542537729781684e-07,
      "loss": 0.5161,
      "step": 10153
    },
    {
      "epoch": 12.428396572827417,
      "grad_norm": 2.0791392529035737,
      "learning_rate": 3.5392497554490755e-07,
      "loss": 0.4233,
      "step": 10154
    },
    {
      "epoch": 12.429620563035495,
      "grad_norm": 1.6473902084908767,
      "learning_rate": 3.535963191437597e-07,
      "loss": 0.4615,
      "step": 10155
    },
    {
      "epoch": 12.430844553243574,
      "grad_norm": 2.099305150470503,
      "learning_rate": 3.532678037963222e-07,
      "loss": 0.3158,
      "step": 10156
    },
    {
      "epoch": 12.432068543451653,
      "grad_norm": 0.9864552343235745,
      "learning_rate": 3.529394295241856e-07,
      "loss": 0.4251,
      "step": 10157
    },
    {
      "epoch": 12.433292533659731,
      "grad_norm": 1.9209945833242439,
      "learning_rate": 3.526111963489276e-07,
      "loss": 1.4827,
      "step": 10158
    },
    {
      "epoch": 12.43451652386781,
      "grad_norm": 1.6608347914606691,
      "learning_rate": 3.5228310429211934e-07,
      "loss": 0.4503,
      "step": 10159
    },
    {
      "epoch": 12.435740514075887,
      "grad_norm": 1.6888052055073246,
      "learning_rate": 3.519551533753207e-07,
      "loss": 1.0106,
      "step": 10160
    },
    {
      "epoch": 12.436964504283965,
      "grad_norm": 1.7838017273957312,
      "learning_rate": 3.516273436200851e-07,
      "loss": 0.672,
      "step": 10161
    },
    {
      "epoch": 12.438188494492044,
      "grad_norm": 1.3720950434270993,
      "learning_rate": 3.5129967504795415e-07,
      "loss": 0.5733,
      "step": 10162
    },
    {
      "epoch": 12.439412484700123,
      "grad_norm": 2.692792322607686,
      "learning_rate": 3.509721476804617e-07,
      "loss": 0.9063,
      "step": 10163
    },
    {
      "epoch": 12.440636474908201,
      "grad_norm": 2.3178450368740435,
      "learning_rate": 3.5064476153913036e-07,
      "loss": 1.029,
      "step": 10164
    },
    {
      "epoch": 12.44186046511628,
      "grad_norm": 2.0120546205846064,
      "learning_rate": 3.5031751664547475e-07,
      "loss": 1.0431,
      "step": 10165
    },
    {
      "epoch": 12.443084455324357,
      "grad_norm": 2.9497064953831584,
      "learning_rate": 3.4999041302100164e-07,
      "loss": 0.7306,
      "step": 10166
    },
    {
      "epoch": 12.444308445532435,
      "grad_norm": 1.300586266279099,
      "learning_rate": 3.4966345068720597e-07,
      "loss": 0.9712,
      "step": 10167
    },
    {
      "epoch": 12.445532435740514,
      "grad_norm": 1.5306742093767491,
      "learning_rate": 3.493366296655748e-07,
      "loss": 0.7814,
      "step": 10168
    },
    {
      "epoch": 12.446756425948593,
      "grad_norm": 1.7925765286209163,
      "learning_rate": 3.4900994997758566e-07,
      "loss": 0.7439,
      "step": 10169
    },
    {
      "epoch": 12.447980416156671,
      "grad_norm": 1.8804417699946472,
      "learning_rate": 3.4868341164470685e-07,
      "loss": 0.5619,
      "step": 10170
    },
    {
      "epoch": 12.44920440636475,
      "grad_norm": 3.022492781697122,
      "learning_rate": 3.483570146883969e-07,
      "loss": 0.485,
      "step": 10171
    },
    {
      "epoch": 12.450428396572827,
      "grad_norm": 1.0662031085240027,
      "learning_rate": 3.4803075913010545e-07,
      "loss": 0.5913,
      "step": 10172
    },
    {
      "epoch": 12.451652386780905,
      "grad_norm": 2.3886765274370334,
      "learning_rate": 3.477046449912724e-07,
      "loss": 1.0716,
      "step": 10173
    },
    {
      "epoch": 12.452876376988984,
      "grad_norm": 2.911116283716038,
      "learning_rate": 3.4737867229333035e-07,
      "loss": 0.3506,
      "step": 10174
    },
    {
      "epoch": 12.454100367197062,
      "grad_norm": 1.9373305506523122,
      "learning_rate": 3.470528410576995e-07,
      "loss": 0.9972,
      "step": 10175
    },
    {
      "epoch": 12.455324357405141,
      "grad_norm": 2.0551508387200124,
      "learning_rate": 3.4672715130579263e-07,
      "loss": 0.5216,
      "step": 10176
    },
    {
      "epoch": 12.45654834761322,
      "grad_norm": 2.2226476257634373,
      "learning_rate": 3.464016030590131e-07,
      "loss": 0.5936,
      "step": 10177
    },
    {
      "epoch": 12.457772337821297,
      "grad_norm": 1.9941758554836873,
      "learning_rate": 3.4607619633875405e-07,
      "loss": 0.4529,
      "step": 10178
    },
    {
      "epoch": 12.458996328029375,
      "grad_norm": 1.7924555600920329,
      "learning_rate": 3.457509311664012e-07,
      "loss": 0.6511,
      "step": 10179
    },
    {
      "epoch": 12.460220318237454,
      "grad_norm": 1.1672275873971574,
      "learning_rate": 3.4542580756332903e-07,
      "loss": 0.5553,
      "step": 10180
    },
    {
      "epoch": 12.461444308445532,
      "grad_norm": 1.445043530368504,
      "learning_rate": 3.451008255509042e-07,
      "loss": 0.8088,
      "step": 10181
    },
    {
      "epoch": 12.462668298653611,
      "grad_norm": 1.4463485783763796,
      "learning_rate": 3.4477598515048143e-07,
      "loss": 0.9639,
      "step": 10182
    },
    {
      "epoch": 12.46389228886169,
      "grad_norm": 1.4869556007860945,
      "learning_rate": 3.4445128638340983e-07,
      "loss": 0.8051,
      "step": 10183
    },
    {
      "epoch": 12.465116279069768,
      "grad_norm": 1.4296856281724217,
      "learning_rate": 3.441267292710271e-07,
      "loss": 0.6909,
      "step": 10184
    },
    {
      "epoch": 12.466340269277845,
      "grad_norm": 1.583061208312277,
      "learning_rate": 3.4380231383466163e-07,
      "loss": 0.3564,
      "step": 10185
    },
    {
      "epoch": 12.467564259485924,
      "grad_norm": 1.2657906510496397,
      "learning_rate": 3.434780400956328e-07,
      "loss": 0.9815,
      "step": 10186
    },
    {
      "epoch": 12.468788249694002,
      "grad_norm": 2.8997391744810534,
      "learning_rate": 3.4315390807525076e-07,
      "loss": 0.4945,
      "step": 10187
    },
    {
      "epoch": 12.470012239902081,
      "grad_norm": 1.4446574939065222,
      "learning_rate": 3.4282991779481604e-07,
      "loss": 0.429,
      "step": 10188
    },
    {
      "epoch": 12.47123623011016,
      "grad_norm": 1.614542982146774,
      "learning_rate": 3.4250606927562076e-07,
      "loss": 0.4227,
      "step": 10189
    },
    {
      "epoch": 12.472460220318238,
      "grad_norm": 2.9180586206940493,
      "learning_rate": 3.4218236253894616e-07,
      "loss": 0.3485,
      "step": 10190
    },
    {
      "epoch": 12.473684210526315,
      "grad_norm": 2.027199126934295,
      "learning_rate": 3.4185879760606525e-07,
      "loss": 1.2858,
      "step": 10191
    },
    {
      "epoch": 12.474908200734394,
      "grad_norm": 2.573411157887824,
      "learning_rate": 3.4153537449824275e-07,
      "loss": 0.5319,
      "step": 10192
    },
    {
      "epoch": 12.476132190942472,
      "grad_norm": 2.0713752643275747,
      "learning_rate": 3.4121209323673104e-07,
      "loss": 0.3599,
      "step": 10193
    },
    {
      "epoch": 12.477356181150551,
      "grad_norm": 1.7903201737627987,
      "learning_rate": 3.4088895384277623e-07,
      "loss": 0.5248,
      "step": 10194
    },
    {
      "epoch": 12.47858017135863,
      "grad_norm": 1.469469020346149,
      "learning_rate": 3.405659563376124e-07,
      "loss": 0.512,
      "step": 10195
    },
    {
      "epoch": 12.479804161566708,
      "grad_norm": 2.7405915909712,
      "learning_rate": 3.4024310074246753e-07,
      "loss": 0.4847,
      "step": 10196
    },
    {
      "epoch": 12.481028151774785,
      "grad_norm": 1.508645562720883,
      "learning_rate": 3.399203870785575e-07,
      "loss": 0.7058,
      "step": 10197
    },
    {
      "epoch": 12.482252141982864,
      "grad_norm": 2.096223850795684,
      "learning_rate": 3.3959781536709006e-07,
      "loss": 0.5406,
      "step": 10198
    },
    {
      "epoch": 12.483476132190942,
      "grad_norm": 1.5835545792111225,
      "learning_rate": 3.392753856292635e-07,
      "loss": 2.2756,
      "step": 10199
    },
    {
      "epoch": 12.48470012239902,
      "grad_norm": 1.554835916673236,
      "learning_rate": 3.389530978862668e-07,
      "loss": 0.6153,
      "step": 10200
    },
    {
      "epoch": 12.4859241126071,
      "grad_norm": 1.630358496119116,
      "learning_rate": 3.38630952159279e-07,
      "loss": 0.931,
      "step": 10201
    },
    {
      "epoch": 12.487148102815178,
      "grad_norm": 2.454847688800301,
      "learning_rate": 3.38308948469471e-07,
      "loss": 0.3931,
      "step": 10202
    },
    {
      "epoch": 12.488372093023255,
      "grad_norm": 1.685144915042484,
      "learning_rate": 3.379870868380031e-07,
      "loss": 1.1655,
      "step": 10203
    },
    {
      "epoch": 12.489596083231334,
      "grad_norm": 1.8526855304761276,
      "learning_rate": 3.376653672860272e-07,
      "loss": 1.1668,
      "step": 10204
    },
    {
      "epoch": 12.490820073439412,
      "grad_norm": 2.656911714612,
      "learning_rate": 3.373437898346857e-07,
      "loss": 1.1732,
      "step": 10205
    },
    {
      "epoch": 12.49204406364749,
      "grad_norm": 1.7761932083247216,
      "learning_rate": 3.37022354505111e-07,
      "loss": 0.8204,
      "step": 10206
    },
    {
      "epoch": 12.49326805385557,
      "grad_norm": 3.3766375789728627,
      "learning_rate": 3.367010613184268e-07,
      "loss": 0.4315,
      "step": 10207
    },
    {
      "epoch": 12.494492044063648,
      "grad_norm": 3.05637134313613,
      "learning_rate": 3.3637991029574684e-07,
      "loss": 0.3911,
      "step": 10208
    },
    {
      "epoch": 12.495716034271727,
      "grad_norm": 1.4100756777620218,
      "learning_rate": 3.360589014581772e-07,
      "loss": 0.6421,
      "step": 10209
    },
    {
      "epoch": 12.496940024479803,
      "grad_norm": 2.46032745198828,
      "learning_rate": 3.35738034826813e-07,
      "loss": 0.4948,
      "step": 10210
    },
    {
      "epoch": 12.498164014687882,
      "grad_norm": 2.952971561838141,
      "learning_rate": 3.3541731042273985e-07,
      "loss": 0.9536,
      "step": 10211
    },
    {
      "epoch": 12.49938800489596,
      "grad_norm": 2.167634841699194,
      "learning_rate": 3.3509672826703393e-07,
      "loss": 0.8072,
      "step": 10212
    },
    {
      "epoch": 12.50061199510404,
      "grad_norm": 1.9527758735827099,
      "learning_rate": 3.3477628838076425e-07,
      "loss": 0.5782,
      "step": 10213
    },
    {
      "epoch": 12.501835985312118,
      "grad_norm": 3.1534958422880055,
      "learning_rate": 3.344559907849884e-07,
      "loss": 0.314,
      "step": 10214
    },
    {
      "epoch": 12.503059975520197,
      "grad_norm": 1.4039696292320003,
      "learning_rate": 3.341358355007551e-07,
      "loss": 0.3054,
      "step": 10215
    },
    {
      "epoch": 12.504283965728273,
      "grad_norm": 1.774537449025063,
      "learning_rate": 3.3381582254910366e-07,
      "loss": 0.4913,
      "step": 10216
    },
    {
      "epoch": 12.505507955936352,
      "grad_norm": 2.5376665415817894,
      "learning_rate": 3.3349595195106415e-07,
      "loss": 0.5528,
      "step": 10217
    },
    {
      "epoch": 12.50673194614443,
      "grad_norm": 2.2695178067527264,
      "learning_rate": 3.331762237276573e-07,
      "loss": 0.9481,
      "step": 10218
    },
    {
      "epoch": 12.50795593635251,
      "grad_norm": 1.1109394484900241,
      "learning_rate": 3.328566378998946e-07,
      "loss": 0.391,
      "step": 10219
    },
    {
      "epoch": 12.509179926560588,
      "grad_norm": 2.0542687459843103,
      "learning_rate": 3.3253719448877785e-07,
      "loss": 1.306,
      "step": 10220
    },
    {
      "epoch": 12.510403916768666,
      "grad_norm": 1.3247702472299907,
      "learning_rate": 3.3221789351529977e-07,
      "loss": 0.4366,
      "step": 10221
    },
    {
      "epoch": 12.511627906976745,
      "grad_norm": 1.853736696916223,
      "learning_rate": 3.318987350004438e-07,
      "loss": 0.2906,
      "step": 10222
    },
    {
      "epoch": 12.512851897184822,
      "grad_norm": 2.703554423358692,
      "learning_rate": 3.315797189651837e-07,
      "loss": 0.5024,
      "step": 10223
    },
    {
      "epoch": 12.5140758873929,
      "grad_norm": 1.9932568500055277,
      "learning_rate": 3.3126084543048386e-07,
      "loss": 0.5418,
      "step": 10224
    },
    {
      "epoch": 12.51529987760098,
      "grad_norm": 2.717140250026016,
      "learning_rate": 3.3094211441729916e-07,
      "loss": 0.5069,
      "step": 10225
    },
    {
      "epoch": 12.516523867809058,
      "grad_norm": 2.1998144023471937,
      "learning_rate": 3.306235259465765e-07,
      "loss": 0.5071,
      "step": 10226
    },
    {
      "epoch": 12.517747858017136,
      "grad_norm": 2.4944632499776778,
      "learning_rate": 3.303050800392518e-07,
      "loss": 0.4545,
      "step": 10227
    },
    {
      "epoch": 12.518971848225215,
      "grad_norm": 1.87358997801927,
      "learning_rate": 3.299867767162526e-07,
      "loss": 1.5569,
      "step": 10228
    },
    {
      "epoch": 12.520195838433292,
      "grad_norm": 1.6903712728942022,
      "learning_rate": 3.296686159984955e-07,
      "loss": 0.4845,
      "step": 10229
    },
    {
      "epoch": 12.52141982864137,
      "grad_norm": 3.495359444570635,
      "learning_rate": 3.2935059790688897e-07,
      "loss": 0.347,
      "step": 10230
    },
    {
      "epoch": 12.522643818849449,
      "grad_norm": 1.5485553251735609,
      "learning_rate": 3.29032722462333e-07,
      "loss": 1.21,
      "step": 10231
    },
    {
      "epoch": 12.523867809057528,
      "grad_norm": 1.3808182738087158,
      "learning_rate": 3.287149896857167e-07,
      "loss": 0.5487,
      "step": 10232
    },
    {
      "epoch": 12.525091799265606,
      "grad_norm": 3.1762759344252367,
      "learning_rate": 3.2839739959792034e-07,
      "loss": 0.3912,
      "step": 10233
    },
    {
      "epoch": 12.526315789473685,
      "grad_norm": 1.3457277575674944,
      "learning_rate": 3.280799522198144e-07,
      "loss": 0.7271,
      "step": 10234
    },
    {
      "epoch": 12.527539779681762,
      "grad_norm": 1.9042628230626772,
      "learning_rate": 3.277626475722609e-07,
      "loss": 0.7248,
      "step": 10235
    },
    {
      "epoch": 12.52876376988984,
      "grad_norm": 1.0935446813316578,
      "learning_rate": 3.274454856761114e-07,
      "loss": 0.5447,
      "step": 10236
    },
    {
      "epoch": 12.529987760097919,
      "grad_norm": 2.221961499070149,
      "learning_rate": 3.27128466552209e-07,
      "loss": 0.5293,
      "step": 10237
    },
    {
      "epoch": 12.531211750305998,
      "grad_norm": 2.0584157369543874,
      "learning_rate": 3.268115902213864e-07,
      "loss": 0.759,
      "step": 10238
    },
    {
      "epoch": 12.532435740514076,
      "grad_norm": 1.3907464042338014,
      "learning_rate": 3.26494856704469e-07,
      "loss": 0.3858,
      "step": 10239
    },
    {
      "epoch": 12.533659730722155,
      "grad_norm": 1.4771594051720225,
      "learning_rate": 3.261782660222701e-07,
      "loss": 0.7206,
      "step": 10240
    },
    {
      "epoch": 12.534883720930232,
      "grad_norm": 2.3269821202569636,
      "learning_rate": 3.258618181955947e-07,
      "loss": 0.3658,
      "step": 10241
    },
    {
      "epoch": 12.53610771113831,
      "grad_norm": 1.2537140834309628,
      "learning_rate": 3.2554551324523925e-07,
      "loss": 0.8393,
      "step": 10242
    },
    {
      "epoch": 12.537331701346389,
      "grad_norm": 2.7131664658702177,
      "learning_rate": 3.252293511919893e-07,
      "loss": 0.3863,
      "step": 10243
    },
    {
      "epoch": 12.538555691554468,
      "grad_norm": 2.860201957750692,
      "learning_rate": 3.2491333205662334e-07,
      "loss": 0.4331,
      "step": 10244
    },
    {
      "epoch": 12.539779681762546,
      "grad_norm": 2.303459103626772,
      "learning_rate": 3.245974558599077e-07,
      "loss": 0.4444,
      "step": 10245
    },
    {
      "epoch": 12.541003671970625,
      "grad_norm": 1.5604590317183562,
      "learning_rate": 3.2428172262260166e-07,
      "loss": 0.473,
      "step": 10246
    },
    {
      "epoch": 12.542227662178703,
      "grad_norm": 1.6539903288549973,
      "learning_rate": 3.239661323654525e-07,
      "loss": 0.3487,
      "step": 10247
    },
    {
      "epoch": 12.54345165238678,
      "grad_norm": 1.1903027959790389,
      "learning_rate": 3.236506851092008e-07,
      "loss": 0.6016,
      "step": 10248
    },
    {
      "epoch": 12.544675642594859,
      "grad_norm": 1.7384603750799568,
      "learning_rate": 3.233353808745765e-07,
      "loss": 0.4248,
      "step": 10249
    },
    {
      "epoch": 12.545899632802938,
      "grad_norm": 1.4260000743212158,
      "learning_rate": 3.230202196823001e-07,
      "loss": 0.3623,
      "step": 10250
    },
    {
      "epoch": 12.547123623011016,
      "grad_norm": 2.1568267080421815,
      "learning_rate": 3.227052015530827e-07,
      "loss": 1.3109,
      "step": 10251
    },
    {
      "epoch": 12.548347613219095,
      "grad_norm": 2.9775527525477288,
      "learning_rate": 3.223903265076259e-07,
      "loss": 0.3467,
      "step": 10252
    },
    {
      "epoch": 12.549571603427173,
      "grad_norm": 2.730084665725833,
      "learning_rate": 3.220755945666226e-07,
      "loss": 0.4554,
      "step": 10253
    },
    {
      "epoch": 12.55079559363525,
      "grad_norm": 1.5126924342657186,
      "learning_rate": 3.217610057507559e-07,
      "loss": 0.5558,
      "step": 10254
    },
    {
      "epoch": 12.552019583843329,
      "grad_norm": 1.5000183370612075,
      "learning_rate": 3.21446560080699e-07,
      "loss": 1.1125,
      "step": 10255
    },
    {
      "epoch": 12.553243574051407,
      "grad_norm": 1.4136235677865325,
      "learning_rate": 3.211322575771156e-07,
      "loss": 0.4071,
      "step": 10256
    },
    {
      "epoch": 12.554467564259486,
      "grad_norm": 1.3765333929599162,
      "learning_rate": 3.2081809826066234e-07,
      "loss": 0.6247,
      "step": 10257
    },
    {
      "epoch": 12.555691554467565,
      "grad_norm": 2.7481800011367317,
      "learning_rate": 3.2050408215198283e-07,
      "loss": 0.3799,
      "step": 10258
    },
    {
      "epoch": 12.556915544675643,
      "grad_norm": 1.7068743826124904,
      "learning_rate": 3.201902092717138e-07,
      "loss": 1.1698,
      "step": 10259
    },
    {
      "epoch": 12.55813953488372,
      "grad_norm": 1.7487390227351058,
      "learning_rate": 3.1987647964048075e-07,
      "loss": 0.7236,
      "step": 10260
    },
    {
      "epoch": 12.559363525091799,
      "grad_norm": 1.6290074102337266,
      "learning_rate": 3.195628932789027e-07,
      "loss": 0.5662,
      "step": 10261
    },
    {
      "epoch": 12.560587515299877,
      "grad_norm": 2.4785289456249444,
      "learning_rate": 3.192494502075863e-07,
      "loss": 0.5708,
      "step": 10262
    },
    {
      "epoch": 12.561811505507956,
      "grad_norm": 1.26988612588785,
      "learning_rate": 3.189361504471303e-07,
      "loss": 0.657,
      "step": 10263
    },
    {
      "epoch": 12.563035495716035,
      "grad_norm": 2.019636724527029,
      "learning_rate": 3.1862299401812236e-07,
      "loss": 0.4753,
      "step": 10264
    },
    {
      "epoch": 12.564259485924113,
      "grad_norm": 1.987795978557706,
      "learning_rate": 3.183099809411436e-07,
      "loss": 0.4873,
      "step": 10265
    },
    {
      "epoch": 12.56548347613219,
      "grad_norm": 1.239442087122034,
      "learning_rate": 3.179971112367633e-07,
      "loss": 0.499,
      "step": 10266
    },
    {
      "epoch": 12.566707466340269,
      "grad_norm": 1.1112099283641501,
      "learning_rate": 3.1768438492554204e-07,
      "loss": 0.523,
      "step": 10267
    },
    {
      "epoch": 12.567931456548347,
      "grad_norm": 1.1706234812292378,
      "learning_rate": 3.173718020280314e-07,
      "loss": 0.6369,
      "step": 10268
    },
    {
      "epoch": 12.569155446756426,
      "grad_norm": 2.2721909604122534,
      "learning_rate": 3.1705936256477274e-07,
      "loss": 0.9216,
      "step": 10269
    },
    {
      "epoch": 12.570379436964505,
      "grad_norm": 1.1850719579143436,
      "learning_rate": 3.167470665562988e-07,
      "loss": 0.4675,
      "step": 10270
    },
    {
      "epoch": 12.571603427172583,
      "grad_norm": 1.4810459344876397,
      "learning_rate": 3.1643491402313207e-07,
      "loss": 0.5339,
      "step": 10271
    },
    {
      "epoch": 12.572827417380662,
      "grad_norm": 1.5878182423511915,
      "learning_rate": 3.161229049857867e-07,
      "loss": 0.4666,
      "step": 10272
    },
    {
      "epoch": 12.574051407588739,
      "grad_norm": 1.6206296735125558,
      "learning_rate": 3.1581103946476563e-07,
      "loss": 0.6865,
      "step": 10273
    },
    {
      "epoch": 12.575275397796817,
      "grad_norm": 2.051090670910285,
      "learning_rate": 3.1549931748056483e-07,
      "loss": 1.3843,
      "step": 10274
    },
    {
      "epoch": 12.576499388004896,
      "grad_norm": 1.7239273993982327,
      "learning_rate": 3.1518773905366976e-07,
      "loss": 0.4634,
      "step": 10275
    },
    {
      "epoch": 12.577723378212974,
      "grad_norm": 2.3691404520186987,
      "learning_rate": 3.1487630420455466e-07,
      "loss": 0.3213,
      "step": 10276
    },
    {
      "epoch": 12.578947368421053,
      "grad_norm": 1.6941013900537065,
      "learning_rate": 3.145650129536862e-07,
      "loss": 0.4851,
      "step": 10277
    },
    {
      "epoch": 12.580171358629132,
      "grad_norm": 1.1725672069397968,
      "learning_rate": 3.1425386532152216e-07,
      "loss": 0.4142,
      "step": 10278
    },
    {
      "epoch": 12.581395348837209,
      "grad_norm": 3.0607638321829596,
      "learning_rate": 3.139428613285098e-07,
      "loss": 0.3354,
      "step": 10279
    },
    {
      "epoch": 12.582619339045287,
      "grad_norm": 1.9090094357253433,
      "learning_rate": 3.136320009950866e-07,
      "loss": 0.2886,
      "step": 10280
    },
    {
      "epoch": 12.583843329253366,
      "grad_norm": 1.7612608501105005,
      "learning_rate": 3.133212843416822e-07,
      "loss": 0.4122,
      "step": 10281
    },
    {
      "epoch": 12.585067319461444,
      "grad_norm": 2.7614053491711448,
      "learning_rate": 3.13010711388714e-07,
      "loss": 0.5819,
      "step": 10282
    },
    {
      "epoch": 12.586291309669523,
      "grad_norm": 2.935360182060802,
      "learning_rate": 3.1270028215659305e-07,
      "loss": 1.0536,
      "step": 10283
    },
    {
      "epoch": 12.587515299877602,
      "grad_norm": 2.3850349756811724,
      "learning_rate": 3.1238999666571955e-07,
      "loss": 0.4843,
      "step": 10284
    },
    {
      "epoch": 12.588739290085678,
      "grad_norm": 2.0554360829431215,
      "learning_rate": 3.1207985493648394e-07,
      "loss": 0.3867,
      "step": 10285
    },
    {
      "epoch": 12.589963280293757,
      "grad_norm": 1.4595596061463731,
      "learning_rate": 3.117698569892677e-07,
      "loss": 0.3956,
      "step": 10286
    },
    {
      "epoch": 12.591187270501836,
      "grad_norm": 2.058349301720611,
      "learning_rate": 3.1146000284444284e-07,
      "loss": 0.7263,
      "step": 10287
    },
    {
      "epoch": 12.592411260709914,
      "grad_norm": 1.6608164199295379,
      "learning_rate": 3.1115029252237165e-07,
      "loss": 0.4943,
      "step": 10288
    },
    {
      "epoch": 12.593635250917993,
      "grad_norm": 0.9805997184529526,
      "learning_rate": 3.1084072604340735e-07,
      "loss": 0.3877,
      "step": 10289
    },
    {
      "epoch": 12.594859241126072,
      "grad_norm": 1.3974373355979652,
      "learning_rate": 3.1053130342789254e-07,
      "loss": 0.5031,
      "step": 10290
    },
    {
      "epoch": 12.596083231334148,
      "grad_norm": 2.3650251892384966,
      "learning_rate": 3.102220246961632e-07,
      "loss": 1.0257,
      "step": 10291
    },
    {
      "epoch": 12.597307221542227,
      "grad_norm": 1.822561350385391,
      "learning_rate": 3.099128898685433e-07,
      "loss": 0.5191,
      "step": 10292
    },
    {
      "epoch": 12.598531211750306,
      "grad_norm": 1.5751133391103775,
      "learning_rate": 3.0960389896534685e-07,
      "loss": 1.1587,
      "step": 10293
    },
    {
      "epoch": 12.599755201958384,
      "grad_norm": 1.7474541483179569,
      "learning_rate": 3.0929505200688076e-07,
      "loss": 0.567,
      "step": 10294
    },
    {
      "epoch": 12.600979192166463,
      "grad_norm": 1.8757491981879895,
      "learning_rate": 3.0898634901344067e-07,
      "loss": 1.3018,
      "step": 10295
    },
    {
      "epoch": 12.602203182374542,
      "grad_norm": 1.6919072500880863,
      "learning_rate": 3.086777900053139e-07,
      "loss": 1.0619,
      "step": 10296
    },
    {
      "epoch": 12.60342717258262,
      "grad_norm": 1.8782058091601408,
      "learning_rate": 3.0836937500277775e-07,
      "loss": 1.4174,
      "step": 10297
    },
    {
      "epoch": 12.604651162790697,
      "grad_norm": 2.1007289178559927,
      "learning_rate": 3.080611040261e-07,
      "loss": 0.5868,
      "step": 10298
    },
    {
      "epoch": 12.605875152998776,
      "grad_norm": 1.4090732457599986,
      "learning_rate": 3.0775297709553895e-07,
      "loss": 0.8598,
      "step": 10299
    },
    {
      "epoch": 12.607099143206854,
      "grad_norm": 1.1705826578119698,
      "learning_rate": 3.07444994231344e-07,
      "loss": 0.6008,
      "step": 10300
    },
    {
      "epoch": 12.608323133414933,
      "grad_norm": 2.1145061694830796,
      "learning_rate": 3.0713715545375416e-07,
      "loss": 1.0651,
      "step": 10301
    },
    {
      "epoch": 12.609547123623011,
      "grad_norm": 3.390453368180344,
      "learning_rate": 3.0682946078299943e-07,
      "loss": 0.5854,
      "step": 10302
    },
    {
      "epoch": 12.61077111383109,
      "grad_norm": 0.9757744794203889,
      "learning_rate": 3.065219102393002e-07,
      "loss": 0.4379,
      "step": 10303
    },
    {
      "epoch": 12.611995104039167,
      "grad_norm": 2.6913040286280627,
      "learning_rate": 3.0621450384286906e-07,
      "loss": 0.3967,
      "step": 10304
    },
    {
      "epoch": 12.613219094247246,
      "grad_norm": 1.8690366710149815,
      "learning_rate": 3.0590724161390583e-07,
      "loss": 0.9066,
      "step": 10305
    },
    {
      "epoch": 12.614443084455324,
      "grad_norm": 1.8952469105191816,
      "learning_rate": 3.0560012357260333e-07,
      "loss": 0.4795,
      "step": 10306
    },
    {
      "epoch": 12.615667074663403,
      "grad_norm": 1.2430856523853449,
      "learning_rate": 3.0529314973914423e-07,
      "loss": 1.0381,
      "step": 10307
    },
    {
      "epoch": 12.616891064871481,
      "grad_norm": 1.3164544596758982,
      "learning_rate": 3.049863201337011e-07,
      "loss": 0.3993,
      "step": 10308
    },
    {
      "epoch": 12.61811505507956,
      "grad_norm": 1.9660933860552197,
      "learning_rate": 3.0467963477643857e-07,
      "loss": 0.6464,
      "step": 10309
    },
    {
      "epoch": 12.619339045287639,
      "grad_norm": 1.1690496558441839,
      "learning_rate": 3.043730936875114e-07,
      "loss": 0.5047,
      "step": 10310
    },
    {
      "epoch": 12.620563035495715,
      "grad_norm": 2.8894178890946747,
      "learning_rate": 3.040666968870626e-07,
      "loss": 0.356,
      "step": 10311
    },
    {
      "epoch": 12.621787025703794,
      "grad_norm": 1.0749831324777759,
      "learning_rate": 3.03760444395228e-07,
      "loss": 0.6262,
      "step": 10312
    },
    {
      "epoch": 12.623011015911873,
      "grad_norm": 2.2372521478348206,
      "learning_rate": 3.03454336232134e-07,
      "loss": 0.3707,
      "step": 10313
    },
    {
      "epoch": 12.624235006119951,
      "grad_norm": 1.667938036746886,
      "learning_rate": 3.0314837241789655e-07,
      "loss": 0.6204,
      "step": 10314
    },
    {
      "epoch": 12.62545899632803,
      "grad_norm": 1.6650047628456977,
      "learning_rate": 3.028425529726228e-07,
      "loss": 1.4817,
      "step": 10315
    },
    {
      "epoch": 12.626682986536107,
      "grad_norm": 2.3488497151823466,
      "learning_rate": 3.025368779164095e-07,
      "loss": 0.5227,
      "step": 10316
    },
    {
      "epoch": 12.627906976744185,
      "grad_norm": 2.8493640127184388,
      "learning_rate": 3.022313472693447e-07,
      "loss": 0.4987,
      "step": 10317
    },
    {
      "epoch": 12.629130966952264,
      "grad_norm": 1.7255063161140916,
      "learning_rate": 3.019259610515068e-07,
      "loss": 0.6136,
      "step": 10318
    },
    {
      "epoch": 12.630354957160343,
      "grad_norm": 1.6899967086768641,
      "learning_rate": 3.016207192829645e-07,
      "loss": 0.667,
      "step": 10319
    },
    {
      "epoch": 12.631578947368421,
      "grad_norm": 2.2719105152747594,
      "learning_rate": 3.0131562198377763e-07,
      "loss": 0.5716,
      "step": 10320
    },
    {
      "epoch": 12.6328029375765,
      "grad_norm": 2.0774410346714083,
      "learning_rate": 3.0101066917399556e-07,
      "loss": 1.0302,
      "step": 10321
    },
    {
      "epoch": 12.634026927784578,
      "grad_norm": 3.4019655005988634,
      "learning_rate": 3.0070586087365875e-07,
      "loss": 0.3358,
      "step": 10322
    },
    {
      "epoch": 12.635250917992655,
      "grad_norm": 1.7591567819844445,
      "learning_rate": 3.0040119710279816e-07,
      "loss": 1.169,
      "step": 10323
    },
    {
      "epoch": 12.636474908200734,
      "grad_norm": 2.537640717731943,
      "learning_rate": 3.0009667788143515e-07,
      "loss": 0.4539,
      "step": 10324
    },
    {
      "epoch": 12.637698898408813,
      "grad_norm": 1.6285760354130026,
      "learning_rate": 2.9979230322958103e-07,
      "loss": 0.4025,
      "step": 10325
    },
    {
      "epoch": 12.638922888616891,
      "grad_norm": 2.1707649803526086,
      "learning_rate": 2.9948807316723923e-07,
      "loss": 0.8413,
      "step": 10326
    },
    {
      "epoch": 12.64014687882497,
      "grad_norm": 1.4395334367055448,
      "learning_rate": 2.9918398771440206e-07,
      "loss": 1.086,
      "step": 10327
    },
    {
      "epoch": 12.641370869033048,
      "grad_norm": 2.6182792411096476,
      "learning_rate": 2.9888004689105383e-07,
      "loss": 0.4728,
      "step": 10328
    },
    {
      "epoch": 12.642594859241125,
      "grad_norm": 1.5789799488308673,
      "learning_rate": 2.985762507171666e-07,
      "loss": 0.5225,
      "step": 10329
    },
    {
      "epoch": 12.643818849449204,
      "grad_norm": 1.613376762717318,
      "learning_rate": 2.982725992127053e-07,
      "loss": 0.9173,
      "step": 10330
    },
    {
      "epoch": 12.645042839657282,
      "grad_norm": 1.9172354996041363,
      "learning_rate": 2.9796909239762586e-07,
      "loss": 0.7396,
      "step": 10331
    },
    {
      "epoch": 12.646266829865361,
      "grad_norm": 2.1003081355521243,
      "learning_rate": 2.9766573029187285e-07,
      "loss": 0.6882,
      "step": 10332
    },
    {
      "epoch": 12.64749082007344,
      "grad_norm": 1.7977184200964196,
      "learning_rate": 2.9736251291538206e-07,
      "loss": 0.4967,
      "step": 10333
    },
    {
      "epoch": 12.648714810281518,
      "grad_norm": 2.8832810339862465,
      "learning_rate": 2.9705944028808e-07,
      "loss": 0.8185,
      "step": 10334
    },
    {
      "epoch": 12.649938800489597,
      "grad_norm": 2.652429214618049,
      "learning_rate": 2.967565124298832e-07,
      "loss": 0.4542,
      "step": 10335
    },
    {
      "epoch": 12.651162790697674,
      "grad_norm": 2.3456830357359615,
      "learning_rate": 2.964537293606995e-07,
      "loss": 0.5505,
      "step": 10336
    },
    {
      "epoch": 12.652386780905752,
      "grad_norm": 1.981793544922854,
      "learning_rate": 2.961510911004262e-07,
      "loss": 0.5785,
      "step": 10337
    },
    {
      "epoch": 12.653610771113831,
      "grad_norm": 1.6035563791006946,
      "learning_rate": 2.9584859766895124e-07,
      "loss": 0.5638,
      "step": 10338
    },
    {
      "epoch": 12.65483476132191,
      "grad_norm": 1.8232961294457668,
      "learning_rate": 2.955462490861549e-07,
      "loss": 0.6255,
      "step": 10339
    },
    {
      "epoch": 12.656058751529988,
      "grad_norm": 1.4890747385523413,
      "learning_rate": 2.9524404537190464e-07,
      "loss": 0.5253,
      "step": 10340
    },
    {
      "epoch": 12.657282741738067,
      "grad_norm": 1.82644031509275,
      "learning_rate": 2.949419865460612e-07,
      "loss": 0.5413,
      "step": 10341
    },
    {
      "epoch": 12.658506731946144,
      "grad_norm": 1.5506534608840195,
      "learning_rate": 2.9464007262847425e-07,
      "loss": 0.5628,
      "step": 10342
    },
    {
      "epoch": 12.659730722154222,
      "grad_norm": 1.9879901570231193,
      "learning_rate": 2.943383036389841e-07,
      "loss": 0.4759,
      "step": 10343
    },
    {
      "epoch": 12.660954712362301,
      "grad_norm": 2.1319056599920847,
      "learning_rate": 2.940366795974231e-07,
      "loss": 0.5032,
      "step": 10344
    },
    {
      "epoch": 12.66217870257038,
      "grad_norm": 1.6845417707042072,
      "learning_rate": 2.9373520052361214e-07,
      "loss": 0.6344,
      "step": 10345
    },
    {
      "epoch": 12.663402692778458,
      "grad_norm": 2.9801956743631854,
      "learning_rate": 2.93433866437364e-07,
      "loss": 0.2038,
      "step": 10346
    },
    {
      "epoch": 12.664626682986537,
      "grad_norm": 1.447351253695189,
      "learning_rate": 2.931326773584797e-07,
      "loss": 1.126,
      "step": 10347
    },
    {
      "epoch": 12.665850673194614,
      "grad_norm": 1.6142866329755192,
      "learning_rate": 2.928316333067535e-07,
      "loss": 0.4589,
      "step": 10348
    },
    {
      "epoch": 12.667074663402692,
      "grad_norm": 2.115225898725521,
      "learning_rate": 2.9253073430196873e-07,
      "loss": 0.6104,
      "step": 10349
    },
    {
      "epoch": 12.668298653610771,
      "grad_norm": 1.5396703762631403,
      "learning_rate": 2.922299803638992e-07,
      "loss": 0.5562,
      "step": 10350
    },
    {
      "epoch": 12.66952264381885,
      "grad_norm": 2.8273308643215245,
      "learning_rate": 2.919293715123095e-07,
      "loss": 0.6498,
      "step": 10351
    },
    {
      "epoch": 12.670746634026928,
      "grad_norm": 2.963174918611286,
      "learning_rate": 2.9162890776695427e-07,
      "loss": 0.4018,
      "step": 10352
    },
    {
      "epoch": 12.671970624235007,
      "grad_norm": 1.7649007191956159,
      "learning_rate": 2.9132858914757894e-07,
      "loss": 0.5665,
      "step": 10353
    },
    {
      "epoch": 12.673194614443084,
      "grad_norm": 3.1007791193144523,
      "learning_rate": 2.910284156739196e-07,
      "loss": 0.522,
      "step": 10354
    },
    {
      "epoch": 12.674418604651162,
      "grad_norm": 1.6070858631534937,
      "learning_rate": 2.9072838736570243e-07,
      "loss": 0.7136,
      "step": 10355
    },
    {
      "epoch": 12.67564259485924,
      "grad_norm": 1.6252778542914683,
      "learning_rate": 2.904285042426433e-07,
      "loss": 0.812,
      "step": 10356
    },
    {
      "epoch": 12.67686658506732,
      "grad_norm": 2.0434886860885384,
      "learning_rate": 2.9012876632445146e-07,
      "loss": 0.6187,
      "step": 10357
    },
    {
      "epoch": 12.678090575275398,
      "grad_norm": 1.6117859414413007,
      "learning_rate": 2.898291736308229e-07,
      "loss": 0.5577,
      "step": 10358
    },
    {
      "epoch": 12.679314565483477,
      "grad_norm": 2.2833329329016574,
      "learning_rate": 2.8952972618144614e-07,
      "loss": 0.5053,
      "step": 10359
    },
    {
      "epoch": 12.680538555691555,
      "grad_norm": 2.7019622071424165,
      "learning_rate": 2.892304239959992e-07,
      "loss": 0.388,
      "step": 10360
    },
    {
      "epoch": 12.681762545899632,
      "grad_norm": 1.6522603562296996,
      "learning_rate": 2.8893126709415245e-07,
      "loss": 0.4695,
      "step": 10361
    },
    {
      "epoch": 12.68298653610771,
      "grad_norm": 2.0733933632395054,
      "learning_rate": 2.8863225549556423e-07,
      "loss": 0.7701,
      "step": 10362
    },
    {
      "epoch": 12.68421052631579,
      "grad_norm": 2.705157221368954,
      "learning_rate": 2.883333892198853e-07,
      "loss": 0.3832,
      "step": 10363
    },
    {
      "epoch": 12.685434516523868,
      "grad_norm": 2.014336957144505,
      "learning_rate": 2.8803466828675545e-07,
      "loss": 0.5304,
      "step": 10364
    },
    {
      "epoch": 12.686658506731947,
      "grad_norm": 1.860142208610143,
      "learning_rate": 2.877360927158057e-07,
      "loss": 0.4545,
      "step": 10365
    },
    {
      "epoch": 12.687882496940025,
      "grad_norm": 1.9939978699799252,
      "learning_rate": 2.8743766252665753e-07,
      "loss": 0.4547,
      "step": 10366
    },
    {
      "epoch": 12.689106487148102,
      "grad_norm": 2.9397784529717916,
      "learning_rate": 2.8713937773892225e-07,
      "loss": 0.356,
      "step": 10367
    },
    {
      "epoch": 12.69033047735618,
      "grad_norm": 2.1608082458690654,
      "learning_rate": 2.8684123837220246e-07,
      "loss": 0.4318,
      "step": 10368
    },
    {
      "epoch": 12.69155446756426,
      "grad_norm": 2.5436315513230587,
      "learning_rate": 2.865432444460906e-07,
      "loss": 0.3903,
      "step": 10369
    },
    {
      "epoch": 12.692778457772338,
      "grad_norm": 1.116764532938184,
      "learning_rate": 2.8624539598016946e-07,
      "loss": 0.7059,
      "step": 10370
    },
    {
      "epoch": 12.694002447980417,
      "grad_norm": 3.190529038977129,
      "learning_rate": 2.859476929940128e-07,
      "loss": 0.5041,
      "step": 10371
    },
    {
      "epoch": 12.695226438188495,
      "grad_norm": 1.8375030588905925,
      "learning_rate": 2.8565013550718473e-07,
      "loss": 0.981,
      "step": 10372
    },
    {
      "epoch": 12.696450428396572,
      "grad_norm": 1.9341826227571086,
      "learning_rate": 2.8535272353923894e-07,
      "loss": 1.0771,
      "step": 10373
    },
    {
      "epoch": 12.69767441860465,
      "grad_norm": 1.587307822760169,
      "learning_rate": 2.850554571097211e-07,
      "loss": 0.6008,
      "step": 10374
    },
    {
      "epoch": 12.69889840881273,
      "grad_norm": 2.518516297691733,
      "learning_rate": 2.8475833623816673e-07,
      "loss": 0.602,
      "step": 10375
    },
    {
      "epoch": 12.700122399020808,
      "grad_norm": 1.446319656093378,
      "learning_rate": 2.8446136094410023e-07,
      "loss": 0.4315,
      "step": 10376
    },
    {
      "epoch": 12.701346389228886,
      "grad_norm": 1.8697431661885309,
      "learning_rate": 2.841645312470381e-07,
      "loss": 0.4416,
      "step": 10377
    },
    {
      "epoch": 12.702570379436965,
      "grad_norm": 2.648712975274542,
      "learning_rate": 2.8386784716648743e-07,
      "loss": 0.3258,
      "step": 10378
    },
    {
      "epoch": 12.703794369645042,
      "grad_norm": 1.6294606004620265,
      "learning_rate": 2.835713087219452e-07,
      "loss": 0.5201,
      "step": 10379
    },
    {
      "epoch": 12.70501835985312,
      "grad_norm": 2.3321965611470654,
      "learning_rate": 2.8327491593289817e-07,
      "loss": 0.4384,
      "step": 10380
    },
    {
      "epoch": 12.7062423500612,
      "grad_norm": 2.0693141841551905,
      "learning_rate": 2.829786688188255e-07,
      "loss": 0.4061,
      "step": 10381
    },
    {
      "epoch": 12.707466340269278,
      "grad_norm": 1.1757987622546815,
      "learning_rate": 2.8268256739919343e-07,
      "loss": 0.6278,
      "step": 10382
    },
    {
      "epoch": 12.708690330477356,
      "grad_norm": 3.368876284419282,
      "learning_rate": 2.8238661169346203e-07,
      "loss": 0.4516,
      "step": 10383
    },
    {
      "epoch": 12.709914320685435,
      "grad_norm": 1.305527985518102,
      "learning_rate": 2.820908017210802e-07,
      "loss": 0.446,
      "step": 10384
    },
    {
      "epoch": 12.711138310893514,
      "grad_norm": 3.102182411675465,
      "learning_rate": 2.8179513750148743e-07,
      "loss": 0.3347,
      "step": 10385
    },
    {
      "epoch": 12.71236230110159,
      "grad_norm": 1.9869524088391952,
      "learning_rate": 2.814996190541136e-07,
      "loss": 0.4843,
      "step": 10386
    },
    {
      "epoch": 12.713586291309669,
      "grad_norm": 2.4040129500067655,
      "learning_rate": 2.812042463983794e-07,
      "loss": 0.9798,
      "step": 10387
    },
    {
      "epoch": 12.714810281517748,
      "grad_norm": 1.5394326517234969,
      "learning_rate": 2.809090195536951e-07,
      "loss": 0.5215,
      "step": 10388
    },
    {
      "epoch": 12.716034271725826,
      "grad_norm": 2.4524754745233497,
      "learning_rate": 2.806139385394621e-07,
      "loss": 0.4139,
      "step": 10389
    },
    {
      "epoch": 12.717258261933905,
      "grad_norm": 0.9239628177367463,
      "learning_rate": 2.803190033750719e-07,
      "loss": 0.3463,
      "step": 10390
    },
    {
      "epoch": 12.718482252141984,
      "grad_norm": 1.3243286408819352,
      "learning_rate": 2.8002421407990726e-07,
      "loss": 0.6532,
      "step": 10391
    },
    {
      "epoch": 12.71970624235006,
      "grad_norm": 2.5674546098805617,
      "learning_rate": 2.7972957067333995e-07,
      "loss": 0.372,
      "step": 10392
    },
    {
      "epoch": 12.720930232558139,
      "grad_norm": 1.8163925393630511,
      "learning_rate": 2.794350731747336e-07,
      "loss": 0.5345,
      "step": 10393
    },
    {
      "epoch": 12.722154222766218,
      "grad_norm": 1.4444024960937885,
      "learning_rate": 2.791407216034406e-07,
      "loss": 0.521,
      "step": 10394
    },
    {
      "epoch": 12.723378212974296,
      "grad_norm": 1.0963557190405375,
      "learning_rate": 2.788465159788042e-07,
      "loss": 0.5154,
      "step": 10395
    },
    {
      "epoch": 12.724602203182375,
      "grad_norm": 1.9318383241193127,
      "learning_rate": 2.7855245632015996e-07,
      "loss": 0.4506,
      "step": 10396
    },
    {
      "epoch": 12.725826193390454,
      "grad_norm": 1.766446251797245,
      "learning_rate": 2.782585426468318e-07,
      "loss": 0.3669,
      "step": 10397
    },
    {
      "epoch": 12.727050183598532,
      "grad_norm": 2.3834690939596204,
      "learning_rate": 2.7796477497813447e-07,
      "loss": 0.5734,
      "step": 10398
    },
    {
      "epoch": 12.728274173806609,
      "grad_norm": 1.4137717970030033,
      "learning_rate": 2.776711533333734e-07,
      "loss": 0.5517,
      "step": 10399
    },
    {
      "epoch": 12.729498164014688,
      "grad_norm": 3.8653878906188663,
      "learning_rate": 2.773776777318443e-07,
      "loss": 0.4129,
      "step": 10400
    },
    {
      "epoch": 12.730722154222766,
      "grad_norm": 1.6749436353024976,
      "learning_rate": 2.770843481928331e-07,
      "loss": 0.4659,
      "step": 10401
    },
    {
      "epoch": 12.731946144430845,
      "grad_norm": 1.7113645073805792,
      "learning_rate": 2.7679116473561695e-07,
      "loss": 0.4677,
      "step": 10402
    },
    {
      "epoch": 12.733170134638923,
      "grad_norm": 2.631374680023711,
      "learning_rate": 2.7649812737946145e-07,
      "loss": 0.4585,
      "step": 10403
    },
    {
      "epoch": 12.734394124847,
      "grad_norm": 2.5686270313780977,
      "learning_rate": 2.762052361436263e-07,
      "loss": 0.4134,
      "step": 10404
    },
    {
      "epoch": 12.735618115055079,
      "grad_norm": 1.5608715444836545,
      "learning_rate": 2.7591249104735705e-07,
      "loss": 0.4501,
      "step": 10405
    },
    {
      "epoch": 12.736842105263158,
      "grad_norm": 1.4271489844715066,
      "learning_rate": 2.7561989210989237e-07,
      "loss": 0.4822,
      "step": 10406
    },
    {
      "epoch": 12.738066095471236,
      "grad_norm": 1.0336533484210075,
      "learning_rate": 2.7532743935046125e-07,
      "loss": 0.3428,
      "step": 10407
    },
    {
      "epoch": 12.739290085679315,
      "grad_norm": 1.671856160252238,
      "learning_rate": 2.7503513278828166e-07,
      "loss": 0.4504,
      "step": 10408
    },
    {
      "epoch": 12.740514075887393,
      "grad_norm": 3.2003058212808315,
      "learning_rate": 2.747429724425643e-07,
      "loss": 0.535,
      "step": 10409
    },
    {
      "epoch": 12.741738066095472,
      "grad_norm": 1.5151763021778812,
      "learning_rate": 2.7445095833250835e-07,
      "loss": 1.2682,
      "step": 10410
    },
    {
      "epoch": 12.742962056303549,
      "grad_norm": 1.4626637940349263,
      "learning_rate": 2.741590904773034e-07,
      "loss": 0.8401,
      "step": 10411
    },
    {
      "epoch": 12.744186046511627,
      "grad_norm": 0.9979308653788985,
      "learning_rate": 2.7386736889612966e-07,
      "loss": 0.4912,
      "step": 10412
    },
    {
      "epoch": 12.745410036719706,
      "grad_norm": 2.0472708587385786,
      "learning_rate": 2.73575793608159e-07,
      "loss": 0.3835,
      "step": 10413
    },
    {
      "epoch": 12.746634026927785,
      "grad_norm": 1.7110184708344247,
      "learning_rate": 2.7328436463255226e-07,
      "loss": 0.5419,
      "step": 10414
    },
    {
      "epoch": 12.747858017135863,
      "grad_norm": 2.1113950473257583,
      "learning_rate": 2.7299308198846073e-07,
      "loss": 0.5241,
      "step": 10415
    },
    {
      "epoch": 12.749082007343942,
      "grad_norm": 1.7456887688679195,
      "learning_rate": 2.7270194569502723e-07,
      "loss": 1.2609,
      "step": 10416
    },
    {
      "epoch": 12.750305997552019,
      "grad_norm": 1.9006012563428076,
      "learning_rate": 2.7241095577138334e-07,
      "loss": 0.526,
      "step": 10417
    },
    {
      "epoch": 12.751529987760097,
      "grad_norm": 2.161932899861038,
      "learning_rate": 2.7212011223665207e-07,
      "loss": 0.3421,
      "step": 10418
    },
    {
      "epoch": 12.752753977968176,
      "grad_norm": 1.3170082134678263,
      "learning_rate": 2.718294151099468e-07,
      "loss": 0.5693,
      "step": 10419
    },
    {
      "epoch": 12.753977968176255,
      "grad_norm": 2.2288422958647303,
      "learning_rate": 2.715388644103711e-07,
      "loss": 0.7091,
      "step": 10420
    },
    {
      "epoch": 12.755201958384333,
      "grad_norm": 1.3377724377168099,
      "learning_rate": 2.712484601570181e-07,
      "loss": 0.7703,
      "step": 10421
    },
    {
      "epoch": 12.756425948592412,
      "grad_norm": 1.4952300112005785,
      "learning_rate": 2.7095820236897356e-07,
      "loss": 0.3294,
      "step": 10422
    },
    {
      "epoch": 12.75764993880049,
      "grad_norm": 3.175944171525824,
      "learning_rate": 2.706680910653109e-07,
      "loss": 0.4388,
      "step": 10423
    },
    {
      "epoch": 12.758873929008567,
      "grad_norm": 1.8564860082217187,
      "learning_rate": 2.7037812626509563e-07,
      "loss": 1.0974,
      "step": 10424
    },
    {
      "epoch": 12.760097919216646,
      "grad_norm": 2.433296081440086,
      "learning_rate": 2.7008830798738255e-07,
      "loss": 0.6629,
      "step": 10425
    },
    {
      "epoch": 12.761321909424725,
      "grad_norm": 1.062770904341979,
      "learning_rate": 2.6979863625121867e-07,
      "loss": 0.4396,
      "step": 10426
    },
    {
      "epoch": 12.762545899632803,
      "grad_norm": 2.091553738664652,
      "learning_rate": 2.695091110756393e-07,
      "loss": 0.5688,
      "step": 10427
    },
    {
      "epoch": 12.763769889840882,
      "grad_norm": 1.1559840765545346,
      "learning_rate": 2.6921973247967164e-07,
      "loss": 0.5804,
      "step": 10428
    },
    {
      "epoch": 12.76499388004896,
      "grad_norm": 2.181709499322586,
      "learning_rate": 2.6893050048233087e-07,
      "loss": 0.54,
      "step": 10429
    },
    {
      "epoch": 12.766217870257037,
      "grad_norm": 2.5825923346004993,
      "learning_rate": 2.686414151026262e-07,
      "loss": 0.9654,
      "step": 10430
    },
    {
      "epoch": 12.767441860465116,
      "grad_norm": 1.5826975781578805,
      "learning_rate": 2.6835247635955466e-07,
      "loss": 0.6196,
      "step": 10431
    },
    {
      "epoch": 12.768665850673194,
      "grad_norm": 2.2411067443447665,
      "learning_rate": 2.680636842721038e-07,
      "loss": 0.4837,
      "step": 10432
    },
    {
      "epoch": 12.769889840881273,
      "grad_norm": 2.39025091216356,
      "learning_rate": 2.6777503885925233e-07,
      "loss": 0.6006,
      "step": 10433
    },
    {
      "epoch": 12.771113831089352,
      "grad_norm": 2.8180070162518884,
      "learning_rate": 2.67486540139969e-07,
      "loss": 0.9477,
      "step": 10434
    },
    {
      "epoch": 12.77233782129743,
      "grad_norm": 1.8897208290234497,
      "learning_rate": 2.671981881332125e-07,
      "loss": 1.8745,
      "step": 10435
    },
    {
      "epoch": 12.773561811505507,
      "grad_norm": 1.6495641851822784,
      "learning_rate": 2.669099828579327e-07,
      "loss": 1.2406,
      "step": 10436
    },
    {
      "epoch": 12.774785801713586,
      "grad_norm": 2.069321488837679,
      "learning_rate": 2.666219243330692e-07,
      "loss": 0.7575,
      "step": 10437
    },
    {
      "epoch": 12.776009791921664,
      "grad_norm": 1.2002235986510243,
      "learning_rate": 2.6633401257755156e-07,
      "loss": 0.5328,
      "step": 10438
    },
    {
      "epoch": 12.777233782129743,
      "grad_norm": 2.326057044837044,
      "learning_rate": 2.6604624761030177e-07,
      "loss": 0.9596,
      "step": 10439
    },
    {
      "epoch": 12.778457772337822,
      "grad_norm": 2.0217914108938637,
      "learning_rate": 2.657586294502296e-07,
      "loss": 0.3878,
      "step": 10440
    },
    {
      "epoch": 12.7796817625459,
      "grad_norm": 1.7138880404660954,
      "learning_rate": 2.6547115811623613e-07,
      "loss": 1.6186,
      "step": 10441
    },
    {
      "epoch": 12.780905752753977,
      "grad_norm": 1.519204351357379,
      "learning_rate": 2.651838336272128e-07,
      "loss": 1.0206,
      "step": 10442
    },
    {
      "epoch": 12.782129742962056,
      "grad_norm": 1.267033236259421,
      "learning_rate": 2.6489665600204245e-07,
      "loss": 0.5275,
      "step": 10443
    },
    {
      "epoch": 12.783353733170134,
      "grad_norm": 2.4226299832521776,
      "learning_rate": 2.646096252595967e-07,
      "loss": 0.8934,
      "step": 10444
    },
    {
      "epoch": 12.784577723378213,
      "grad_norm": 0.7586572372138108,
      "learning_rate": 2.6432274141873827e-07,
      "loss": 0.2548,
      "step": 10445
    },
    {
      "epoch": 12.785801713586292,
      "grad_norm": 1.4739301033968428,
      "learning_rate": 2.6403600449832064e-07,
      "loss": 0.5402,
      "step": 10446
    },
    {
      "epoch": 12.78702570379437,
      "grad_norm": 1.679509520957207,
      "learning_rate": 2.637494145171854e-07,
      "loss": 0.6947,
      "step": 10447
    },
    {
      "epoch": 12.788249694002449,
      "grad_norm": 1.2781590217475853,
      "learning_rate": 2.634629714941678e-07,
      "loss": 0.467,
      "step": 10448
    },
    {
      "epoch": 12.789473684210526,
      "grad_norm": 2.6367825223352472,
      "learning_rate": 2.6317667544809135e-07,
      "loss": 0.3143,
      "step": 10449
    },
    {
      "epoch": 12.790697674418604,
      "grad_norm": 1.925263843742835,
      "learning_rate": 2.628905263977705e-07,
      "loss": 0.5397,
      "step": 10450
    },
    {
      "epoch": 12.791921664626683,
      "grad_norm": 1.6974673815466501,
      "learning_rate": 2.626045243620096e-07,
      "loss": 0.5044,
      "step": 10451
    },
    {
      "epoch": 12.793145654834762,
      "grad_norm": 1.561619449783848,
      "learning_rate": 2.623186693596039e-07,
      "loss": 0.7106,
      "step": 10452
    },
    {
      "epoch": 12.79436964504284,
      "grad_norm": 2.583441181430144,
      "learning_rate": 2.6203296140933864e-07,
      "loss": 0.8096,
      "step": 10453
    },
    {
      "epoch": 12.795593635250919,
      "grad_norm": 2.651230959182768,
      "learning_rate": 2.6174740052998943e-07,
      "loss": 0.943,
      "step": 10454
    },
    {
      "epoch": 12.796817625458996,
      "grad_norm": 1.9488626279522412,
      "learning_rate": 2.61461986740322e-07,
      "loss": 0.4185,
      "step": 10455
    },
    {
      "epoch": 12.798041615667074,
      "grad_norm": 1.792999412374256,
      "learning_rate": 2.611767200590934e-07,
      "loss": 0.427,
      "step": 10456
    },
    {
      "epoch": 12.799265605875153,
      "grad_norm": 1.7275846995514785,
      "learning_rate": 2.6089160050505047e-07,
      "loss": 1.0777,
      "step": 10457
    },
    {
      "epoch": 12.800489596083231,
      "grad_norm": 1.686285195434973,
      "learning_rate": 2.606066280969291e-07,
      "loss": 0.5552,
      "step": 10458
    },
    {
      "epoch": 12.80171358629131,
      "grad_norm": 1.1315262484207311,
      "learning_rate": 2.6032180285345707e-07,
      "loss": 0.5339,
      "step": 10459
    },
    {
      "epoch": 12.802937576499389,
      "grad_norm": 1.4205201942117065,
      "learning_rate": 2.6003712479335196e-07,
      "loss": 0.6025,
      "step": 10460
    },
    {
      "epoch": 12.804161566707466,
      "grad_norm": 2.9028786503639696,
      "learning_rate": 2.597525939353224e-07,
      "loss": 0.4629,
      "step": 10461
    },
    {
      "epoch": 12.805385556915544,
      "grad_norm": 1.8113087743211715,
      "learning_rate": 2.5946821029806613e-07,
      "loss": 1.1546,
      "step": 10462
    },
    {
      "epoch": 12.806609547123623,
      "grad_norm": 2.1456075011578446,
      "learning_rate": 2.591839739002719e-07,
      "loss": 1.0641,
      "step": 10463
    },
    {
      "epoch": 12.807833537331701,
      "grad_norm": 2.09766519248643,
      "learning_rate": 2.588998847606189e-07,
      "loss": 0.4019,
      "step": 10464
    },
    {
      "epoch": 12.80905752753978,
      "grad_norm": 1.2554562574320203,
      "learning_rate": 2.5861594289777606e-07,
      "loss": 0.5352,
      "step": 10465
    },
    {
      "epoch": 12.810281517747859,
      "grad_norm": 3.1676326060511406,
      "learning_rate": 2.58332148330403e-07,
      "loss": 0.5681,
      "step": 10466
    },
    {
      "epoch": 12.811505507955935,
      "grad_norm": 1.302667731454608,
      "learning_rate": 2.5804850107714996e-07,
      "loss": 0.7578,
      "step": 10467
    },
    {
      "epoch": 12.812729498164014,
      "grad_norm": 1.6141163526727345,
      "learning_rate": 2.577650011566571e-07,
      "loss": 0.374,
      "step": 10468
    },
    {
      "epoch": 12.813953488372093,
      "grad_norm": 1.4568784601022056,
      "learning_rate": 2.574816485875548e-07,
      "loss": 0.452,
      "step": 10469
    },
    {
      "epoch": 12.815177478580171,
      "grad_norm": 1.6486440355884764,
      "learning_rate": 2.5719844338846415e-07,
      "loss": 1.0384,
      "step": 10470
    },
    {
      "epoch": 12.81640146878825,
      "grad_norm": 2.470082621336716,
      "learning_rate": 2.5691538557799627e-07,
      "loss": 0.4141,
      "step": 10471
    },
    {
      "epoch": 12.817625458996329,
      "grad_norm": 2.9648848980558453,
      "learning_rate": 2.566324751747526e-07,
      "loss": 0.6802,
      "step": 10472
    },
    {
      "epoch": 12.818849449204407,
      "grad_norm": 1.8201693386458961,
      "learning_rate": 2.563497121973244e-07,
      "loss": 0.6256,
      "step": 10473
    },
    {
      "epoch": 12.820073439412484,
      "grad_norm": 1.9200878121963407,
      "learning_rate": 2.56067096664295e-07,
      "loss": 0.4672,
      "step": 10474
    },
    {
      "epoch": 12.821297429620563,
      "grad_norm": 1.1481844150688285,
      "learning_rate": 2.557846285942367e-07,
      "loss": 0.5915,
      "step": 10475
    },
    {
      "epoch": 12.822521419828641,
      "grad_norm": 1.361513155548424,
      "learning_rate": 2.555023080057115e-07,
      "loss": 0.7153,
      "step": 10476
    },
    {
      "epoch": 12.82374541003672,
      "grad_norm": 1.5242507983265545,
      "learning_rate": 2.5522013491727224e-07,
      "loss": 0.2677,
      "step": 10477
    },
    {
      "epoch": 12.824969400244798,
      "grad_norm": 3.092168843513804,
      "learning_rate": 2.5493810934746345e-07,
      "loss": 0.3964,
      "step": 10478
    },
    {
      "epoch": 12.826193390452877,
      "grad_norm": 3.145867977102824,
      "learning_rate": 2.546562313148179e-07,
      "loss": 0.6902,
      "step": 10479
    },
    {
      "epoch": 12.827417380660954,
      "grad_norm": 2.1752174438205385,
      "learning_rate": 2.543745008378601e-07,
      "loss": 1.1136,
      "step": 10480
    },
    {
      "epoch": 12.828641370869033,
      "grad_norm": 1.909358676055843,
      "learning_rate": 2.5409291793510454e-07,
      "loss": 0.4712,
      "step": 10481
    },
    {
      "epoch": 12.829865361077111,
      "grad_norm": 2.1052406315178263,
      "learning_rate": 2.5381148262505457e-07,
      "loss": 1.1474,
      "step": 10482
    },
    {
      "epoch": 12.83108935128519,
      "grad_norm": 0.987489595277372,
      "learning_rate": 2.5353019492620616e-07,
      "loss": 0.4111,
      "step": 10483
    },
    {
      "epoch": 12.832313341493268,
      "grad_norm": 1.1848560972743198,
      "learning_rate": 2.532490548570443e-07,
      "loss": 1.1955,
      "step": 10484
    },
    {
      "epoch": 12.833537331701347,
      "grad_norm": 2.778431065456724,
      "learning_rate": 2.529680624360445e-07,
      "loss": 0.3532,
      "step": 10485
    },
    {
      "epoch": 12.834761321909426,
      "grad_norm": 1.3415448985798206,
      "learning_rate": 2.526872176816722e-07,
      "loss": 0.7137,
      "step": 10486
    },
    {
      "epoch": 12.835985312117502,
      "grad_norm": 1.7233013075357044,
      "learning_rate": 2.524065206123838e-07,
      "loss": 0.5484,
      "step": 10487
    },
    {
      "epoch": 12.837209302325581,
      "grad_norm": 1.1541422018405514,
      "learning_rate": 2.521259712466256e-07,
      "loss": 0.5048,
      "step": 10488
    },
    {
      "epoch": 12.83843329253366,
      "grad_norm": 1.528146151946793,
      "learning_rate": 2.5184556960283434e-07,
      "loss": 0.6153,
      "step": 10489
    },
    {
      "epoch": 12.839657282741738,
      "grad_norm": 1.8428157353858758,
      "learning_rate": 2.515653156994366e-07,
      "loss": 1.4947,
      "step": 10490
    },
    {
      "epoch": 12.840881272949817,
      "grad_norm": 2.7280139646740187,
      "learning_rate": 2.5128520955485016e-07,
      "loss": 0.4201,
      "step": 10491
    },
    {
      "epoch": 12.842105263157894,
      "grad_norm": 1.3052925484652351,
      "learning_rate": 2.510052511874822e-07,
      "loss": 0.7456,
      "step": 10492
    },
    {
      "epoch": 12.843329253365972,
      "grad_norm": 1.575369343770074,
      "learning_rate": 2.507254406157314e-07,
      "loss": 0.5461,
      "step": 10493
    },
    {
      "epoch": 12.844553243574051,
      "grad_norm": 2.501162993322537,
      "learning_rate": 2.5044577785798466e-07,
      "loss": 0.3885,
      "step": 10494
    },
    {
      "epoch": 12.84577723378213,
      "grad_norm": 1.6459232482522708,
      "learning_rate": 2.5016626293262014e-07,
      "loss": 1.289,
      "step": 10495
    },
    {
      "epoch": 12.847001223990208,
      "grad_norm": 1.8156455932637243,
      "learning_rate": 2.498868958580081e-07,
      "loss": 0.925,
      "step": 10496
    },
    {
      "epoch": 12.848225214198287,
      "grad_norm": 1.888399909822423,
      "learning_rate": 2.496076766525063e-07,
      "loss": 0.5409,
      "step": 10497
    },
    {
      "epoch": 12.849449204406366,
      "grad_norm": 2.540899393307378,
      "learning_rate": 2.4932860533446456e-07,
      "loss": 0.5957,
      "step": 10498
    },
    {
      "epoch": 12.850673194614442,
      "grad_norm": 1.370788553514035,
      "learning_rate": 2.490496819222224e-07,
      "loss": 0.7627,
      "step": 10499
    },
    {
      "epoch": 12.851897184822521,
      "grad_norm": 1.6312869986835035,
      "learning_rate": 2.487709064341093e-07,
      "loss": 0.5189,
      "step": 10500
    },
    {
      "epoch": 12.8531211750306,
      "grad_norm": 1.593314971435857,
      "learning_rate": 2.484922788884453e-07,
      "loss": 0.3312,
      "step": 10501
    },
    {
      "epoch": 12.854345165238678,
      "grad_norm": 1.7276822283138542,
      "learning_rate": 2.4821379930354107e-07,
      "loss": 0.518,
      "step": 10502
    },
    {
      "epoch": 12.855569155446757,
      "grad_norm": 2.30133103489708,
      "learning_rate": 2.4793546769769697e-07,
      "loss": 0.8122,
      "step": 10503
    },
    {
      "epoch": 12.856793145654835,
      "grad_norm": 3.6032999609006637,
      "learning_rate": 2.476572840892047e-07,
      "loss": 0.3274,
      "step": 10504
    },
    {
      "epoch": 12.858017135862912,
      "grad_norm": 1.1271547434604146,
      "learning_rate": 2.473792484963447e-07,
      "loss": 0.3049,
      "step": 10505
    },
    {
      "epoch": 12.859241126070991,
      "grad_norm": 1.9956620240845078,
      "learning_rate": 2.4710136093738843e-07,
      "loss": 0.6066,
      "step": 10506
    },
    {
      "epoch": 12.86046511627907,
      "grad_norm": 1.7454426957749365,
      "learning_rate": 2.4682362143059803e-07,
      "loss": 0.4249,
      "step": 10507
    },
    {
      "epoch": 12.861689106487148,
      "grad_norm": 1.5005019264695287,
      "learning_rate": 2.465460299942246e-07,
      "loss": 0.6133,
      "step": 10508
    },
    {
      "epoch": 12.862913096695227,
      "grad_norm": 1.6897099562272597,
      "learning_rate": 2.462685866465117e-07,
      "loss": 1.1652,
      "step": 10509
    },
    {
      "epoch": 12.864137086903305,
      "grad_norm": 1.9929297545864118,
      "learning_rate": 2.4599129140569195e-07,
      "loss": 0.5178,
      "step": 10510
    },
    {
      "epoch": 12.865361077111384,
      "grad_norm": 1.9398297132974978,
      "learning_rate": 2.4571414428998686e-07,
      "loss": 0.7328,
      "step": 10511
    },
    {
      "epoch": 12.86658506731946,
      "grad_norm": 2.5727693582057447,
      "learning_rate": 2.4543714531760964e-07,
      "loss": 0.4931,
      "step": 10512
    },
    {
      "epoch": 12.86780905752754,
      "grad_norm": 1.4264422212105943,
      "learning_rate": 2.4516029450676495e-07,
      "loss": 1.5535,
      "step": 10513
    },
    {
      "epoch": 12.869033047735618,
      "grad_norm": 2.4755796854096714,
      "learning_rate": 2.4488359187564536e-07,
      "loss": 0.3587,
      "step": 10514
    },
    {
      "epoch": 12.870257037943697,
      "grad_norm": 1.427110474032553,
      "learning_rate": 2.44607037442435e-07,
      "loss": 0.5411,
      "step": 10515
    },
    {
      "epoch": 12.871481028151775,
      "grad_norm": 1.0955125011599471,
      "learning_rate": 2.4433063122530825e-07,
      "loss": 0.5097,
      "step": 10516
    },
    {
      "epoch": 12.872705018359854,
      "grad_norm": 1.8649534367460903,
      "learning_rate": 2.4405437324242914e-07,
      "loss": 1.0142,
      "step": 10517
    },
    {
      "epoch": 12.87392900856793,
      "grad_norm": 1.4508571560928227,
      "learning_rate": 2.4377826351195235e-07,
      "loss": 0.4994,
      "step": 10518
    },
    {
      "epoch": 12.87515299877601,
      "grad_norm": 1.6198864868970955,
      "learning_rate": 2.4350230205202303e-07,
      "loss": 0.4916,
      "step": 10519
    },
    {
      "epoch": 12.876376988984088,
      "grad_norm": 2.9992369588671206,
      "learning_rate": 2.4322648888077615e-07,
      "loss": 0.8179,
      "step": 10520
    },
    {
      "epoch": 12.877600979192167,
      "grad_norm": 2.4451044649152154,
      "learning_rate": 2.429508240163367e-07,
      "loss": 0.4643,
      "step": 10521
    },
    {
      "epoch": 12.878824969400245,
      "grad_norm": 2.23319044696988,
      "learning_rate": 2.4267530747682203e-07,
      "loss": 0.508,
      "step": 10522
    },
    {
      "epoch": 12.880048959608324,
      "grad_norm": 3.3921726224875686,
      "learning_rate": 2.4239993928033613e-07,
      "loss": 0.4624,
      "step": 10523
    },
    {
      "epoch": 12.8812729498164,
      "grad_norm": 0.9370913795152549,
      "learning_rate": 2.4212471944497607e-07,
      "loss": 0.5199,
      "step": 10524
    },
    {
      "epoch": 12.88249694002448,
      "grad_norm": 1.1891408697639916,
      "learning_rate": 2.418496479888277e-07,
      "loss": 0.567,
      "step": 10525
    },
    {
      "epoch": 12.883720930232558,
      "grad_norm": 1.6767074578609364,
      "learning_rate": 2.4157472492996853e-07,
      "loss": 0.5903,
      "step": 10526
    },
    {
      "epoch": 12.884944920440637,
      "grad_norm": 2.9273867952438675,
      "learning_rate": 2.412999502864649e-07,
      "loss": 0.7908,
      "step": 10527
    },
    {
      "epoch": 12.886168910648715,
      "grad_norm": 1.783706135714741,
      "learning_rate": 2.410253240763749e-07,
      "loss": 0.8795,
      "step": 10528
    },
    {
      "epoch": 12.887392900856794,
      "grad_norm": 1.5235444170238044,
      "learning_rate": 2.407508463177444e-07,
      "loss": 0.5839,
      "step": 10529
    },
    {
      "epoch": 12.88861689106487,
      "grad_norm": 1.5266542189808174,
      "learning_rate": 2.404765170286122e-07,
      "loss": 0.599,
      "step": 10530
    },
    {
      "epoch": 12.88984088127295,
      "grad_norm": 1.390438014565871,
      "learning_rate": 2.402023362270059e-07,
      "loss": 0.5056,
      "step": 10531
    },
    {
      "epoch": 12.891064871481028,
      "grad_norm": 2.5752864748044284,
      "learning_rate": 2.3992830393094383e-07,
      "loss": 0.5106,
      "step": 10532
    },
    {
      "epoch": 12.892288861689106,
      "grad_norm": 1.8705222777940498,
      "learning_rate": 2.3965442015843386e-07,
      "loss": 0.5793,
      "step": 10533
    },
    {
      "epoch": 12.893512851897185,
      "grad_norm": 1.7296513200779486,
      "learning_rate": 2.393806849274752e-07,
      "loss": 0.7091,
      "step": 10534
    },
    {
      "epoch": 12.894736842105264,
      "grad_norm": 2.3218670190041233,
      "learning_rate": 2.3910709825605645e-07,
      "loss": 0.4095,
      "step": 10535
    },
    {
      "epoch": 12.895960832313342,
      "grad_norm": 2.1287419400099408,
      "learning_rate": 2.3883366016215685e-07,
      "loss": 0.3997,
      "step": 10536
    },
    {
      "epoch": 12.89718482252142,
      "grad_norm": 3.2576876232308822,
      "learning_rate": 2.3856037066374545e-07,
      "loss": 0.304,
      "step": 10537
    },
    {
      "epoch": 12.898408812729498,
      "grad_norm": 1.6977532574192873,
      "learning_rate": 2.382872297787814e-07,
      "loss": 0.6332,
      "step": 10538
    },
    {
      "epoch": 12.899632802937576,
      "grad_norm": 2.3724659953480636,
      "learning_rate": 2.3801423752521645e-07,
      "loss": 0.5425,
      "step": 10539
    },
    {
      "epoch": 12.900856793145655,
      "grad_norm": 1.869397998199903,
      "learning_rate": 2.3774139392098854e-07,
      "loss": 0.5201,
      "step": 10540
    },
    {
      "epoch": 12.902080783353734,
      "grad_norm": 1.701696485345459,
      "learning_rate": 2.3746869898402886e-07,
      "loss": 0.8466,
      "step": 10541
    },
    {
      "epoch": 12.903304773561812,
      "grad_norm": 2.252302019197749,
      "learning_rate": 2.371961527322572e-07,
      "loss": 0.4623,
      "step": 10542
    },
    {
      "epoch": 12.904528763769889,
      "grad_norm": 1.7971740497007647,
      "learning_rate": 2.3692375518358512e-07,
      "loss": 1.3555,
      "step": 10543
    },
    {
      "epoch": 12.905752753977968,
      "grad_norm": 1.488954266700243,
      "learning_rate": 2.3665150635591356e-07,
      "loss": 0.6846,
      "step": 10544
    },
    {
      "epoch": 12.906976744186046,
      "grad_norm": 1.9327916358998265,
      "learning_rate": 2.3637940626713346e-07,
      "loss": 0.4516,
      "step": 10545
    },
    {
      "epoch": 12.908200734394125,
      "grad_norm": 1.7897888588441095,
      "learning_rate": 2.3610745493512667e-07,
      "loss": 0.3085,
      "step": 10546
    },
    {
      "epoch": 12.909424724602204,
      "grad_norm": 1.2003940891175295,
      "learning_rate": 2.358356523777633e-07,
      "loss": 0.69,
      "step": 10547
    },
    {
      "epoch": 12.910648714810282,
      "grad_norm": 1.3109445974497864,
      "learning_rate": 2.3556399861290713e-07,
      "loss": 0.5536,
      "step": 10548
    },
    {
      "epoch": 12.911872705018359,
      "grad_norm": 1.4158494756615523,
      "learning_rate": 2.352924936584089e-07,
      "loss": 1.5615,
      "step": 10549
    },
    {
      "epoch": 12.913096695226438,
      "grad_norm": 1.3829202192308803,
      "learning_rate": 2.350211375321118e-07,
      "loss": 0.7341,
      "step": 10550
    },
    {
      "epoch": 12.914320685434516,
      "grad_norm": 2.6358003248072666,
      "learning_rate": 2.3474993025184772e-07,
      "loss": 0.8453,
      "step": 10551
    },
    {
      "epoch": 12.915544675642595,
      "grad_norm": 1.2967009208693403,
      "learning_rate": 2.3447887183543987e-07,
      "loss": 0.8918,
      "step": 10552
    },
    {
      "epoch": 12.916768665850674,
      "grad_norm": 1.356560117364283,
      "learning_rate": 2.3420796230070098e-07,
      "loss": 0.4394,
      "step": 10553
    },
    {
      "epoch": 12.917992656058752,
      "grad_norm": 1.2161117846134966,
      "learning_rate": 2.3393720166543398e-07,
      "loss": 0.6216,
      "step": 10554
    },
    {
      "epoch": 12.919216646266829,
      "grad_norm": 1.257795741768906,
      "learning_rate": 2.3366658994743217e-07,
      "loss": 0.425,
      "step": 10555
    },
    {
      "epoch": 12.920440636474908,
      "grad_norm": 2.715195673651283,
      "learning_rate": 2.3339612716447995e-07,
      "loss": 0.3321,
      "step": 10556
    },
    {
      "epoch": 12.921664626682986,
      "grad_norm": 3.6614060244652435,
      "learning_rate": 2.3312581333435114e-07,
      "loss": 0.4176,
      "step": 10557
    },
    {
      "epoch": 12.922888616891065,
      "grad_norm": 1.3707810497381692,
      "learning_rate": 2.3285564847480874e-07,
      "loss": 0.5875,
      "step": 10558
    },
    {
      "epoch": 12.924112607099143,
      "grad_norm": 1.861083603885412,
      "learning_rate": 2.3258563260360773e-07,
      "loss": 1.1303,
      "step": 10559
    },
    {
      "epoch": 12.925336597307222,
      "grad_norm": 2.636362633974648,
      "learning_rate": 2.3231576573849168e-07,
      "loss": 0.8501,
      "step": 10560
    },
    {
      "epoch": 12.9265605875153,
      "grad_norm": 2.211656698145912,
      "learning_rate": 2.3204604789719643e-07,
      "loss": 0.5292,
      "step": 10561
    },
    {
      "epoch": 12.927784577723378,
      "grad_norm": 1.5762795810313264,
      "learning_rate": 2.317764790974464e-07,
      "loss": 0.3185,
      "step": 10562
    },
    {
      "epoch": 12.929008567931456,
      "grad_norm": 3.2132361481623555,
      "learning_rate": 2.3150705935695656e-07,
      "loss": 0.3677,
      "step": 10563
    },
    {
      "epoch": 12.930232558139535,
      "grad_norm": 1.87302501426283,
      "learning_rate": 2.3123778869343223e-07,
      "loss": 1.0046,
      "step": 10564
    },
    {
      "epoch": 12.931456548347613,
      "grad_norm": 2.650363237291908,
      "learning_rate": 2.3096866712456894e-07,
      "loss": 0.4231,
      "step": 10565
    },
    {
      "epoch": 12.932680538555692,
      "grad_norm": 2.0606690681732873,
      "learning_rate": 2.3069969466805202e-07,
      "loss": 0.7625,
      "step": 10566
    },
    {
      "epoch": 12.93390452876377,
      "grad_norm": 2.970515645798855,
      "learning_rate": 2.304308713415579e-07,
      "loss": 0.5249,
      "step": 10567
    },
    {
      "epoch": 12.935128518971847,
      "grad_norm": 1.6968367152014547,
      "learning_rate": 2.3016219716275184e-07,
      "loss": 0.4539,
      "step": 10568
    },
    {
      "epoch": 12.936352509179926,
      "grad_norm": 1.7723017715445057,
      "learning_rate": 2.2989367214929142e-07,
      "loss": 0.5137,
      "step": 10569
    },
    {
      "epoch": 12.937576499388005,
      "grad_norm": 2.886126429324236,
      "learning_rate": 2.2962529631882197e-07,
      "loss": 1.0152,
      "step": 10570
    },
    {
      "epoch": 12.938800489596083,
      "grad_norm": 1.501863389895933,
      "learning_rate": 2.2935706968898048e-07,
      "loss": 0.588,
      "step": 10571
    },
    {
      "epoch": 12.940024479804162,
      "grad_norm": 2.405042506990251,
      "learning_rate": 2.2908899227739367e-07,
      "loss": 0.4228,
      "step": 10572
    },
    {
      "epoch": 12.94124847001224,
      "grad_norm": 1.8304503699631078,
      "learning_rate": 2.288210641016786e-07,
      "loss": 1.3572,
      "step": 10573
    },
    {
      "epoch": 12.94247246022032,
      "grad_norm": 2.6294889979386293,
      "learning_rate": 2.2855328517944281e-07,
      "loss": 0.5428,
      "step": 10574
    },
    {
      "epoch": 12.943696450428396,
      "grad_norm": 2.1076150616584144,
      "learning_rate": 2.2828565552828447e-07,
      "loss": 1.0004,
      "step": 10575
    },
    {
      "epoch": 12.944920440636475,
      "grad_norm": 1.2794457662741603,
      "learning_rate": 2.280181751657895e-07,
      "loss": 0.5519,
      "step": 10576
    },
    {
      "epoch": 12.946144430844553,
      "grad_norm": 1.2207351712393135,
      "learning_rate": 2.2775084410953634e-07,
      "loss": 0.5071,
      "step": 10577
    },
    {
      "epoch": 12.947368421052632,
      "grad_norm": 2.746312882354513,
      "learning_rate": 2.2748366237709374e-07,
      "loss": 0.3975,
      "step": 10578
    },
    {
      "epoch": 12.94859241126071,
      "grad_norm": 2.2780037925545353,
      "learning_rate": 2.272166299860193e-07,
      "loss": 0.3922,
      "step": 10579
    },
    {
      "epoch": 12.949816401468787,
      "grad_norm": 2.3774665124851797,
      "learning_rate": 2.2694974695386147e-07,
      "loss": 0.6335,
      "step": 10580
    },
    {
      "epoch": 12.951040391676866,
      "grad_norm": 3.0664511622546406,
      "learning_rate": 2.26683013298159e-07,
      "loss": 0.4052,
      "step": 10581
    },
    {
      "epoch": 12.952264381884945,
      "grad_norm": 1.9841513899980527,
      "learning_rate": 2.264164290364404e-07,
      "loss": 0.3335,
      "step": 10582
    },
    {
      "epoch": 12.953488372093023,
      "grad_norm": 1.0471763052563037,
      "learning_rate": 2.261499941862247e-07,
      "loss": 0.5822,
      "step": 10583
    },
    {
      "epoch": 12.954712362301102,
      "grad_norm": 1.7670892769380695,
      "learning_rate": 2.2588370876502093e-07,
      "loss": 1.5136,
      "step": 10584
    },
    {
      "epoch": 12.95593635250918,
      "grad_norm": 1.5094888895905199,
      "learning_rate": 2.2561757279032842e-07,
      "loss": 1.3154,
      "step": 10585
    },
    {
      "epoch": 12.957160342717259,
      "grad_norm": 1.9178632616758706,
      "learning_rate": 2.2535158627963684e-07,
      "loss": 0.6065,
      "step": 10586
    },
    {
      "epoch": 12.958384332925336,
      "grad_norm": 1.8505792774125136,
      "learning_rate": 2.2508574925042547e-07,
      "loss": 1.4512,
      "step": 10587
    },
    {
      "epoch": 12.959608323133414,
      "grad_norm": 2.6822541849682167,
      "learning_rate": 2.248200617201643e-07,
      "loss": 0.4681,
      "step": 10588
    },
    {
      "epoch": 12.960832313341493,
      "grad_norm": 1.5100010878320198,
      "learning_rate": 2.245545237063135e-07,
      "loss": 0.4991,
      "step": 10589
    },
    {
      "epoch": 12.962056303549572,
      "grad_norm": 2.0420344308446134,
      "learning_rate": 2.2428913522632267e-07,
      "loss": 0.4883,
      "step": 10590
    },
    {
      "epoch": 12.96328029375765,
      "grad_norm": 1.7366508444890774,
      "learning_rate": 2.2402389629763294e-07,
      "loss": 0.6118,
      "step": 10591
    },
    {
      "epoch": 12.964504283965729,
      "grad_norm": 1.7134895484351578,
      "learning_rate": 2.237588069376745e-07,
      "loss": 1.2262,
      "step": 10592
    },
    {
      "epoch": 12.965728274173806,
      "grad_norm": 1.993709513663499,
      "learning_rate": 2.2349386716386868e-07,
      "loss": 1.1028,
      "step": 10593
    },
    {
      "epoch": 12.966952264381884,
      "grad_norm": 2.0945507311413802,
      "learning_rate": 2.2322907699362463e-07,
      "loss": 0.591,
      "step": 10594
    },
    {
      "epoch": 12.968176254589963,
      "grad_norm": 1.8921381859587465,
      "learning_rate": 2.2296443644434507e-07,
      "loss": 1.08,
      "step": 10595
    },
    {
      "epoch": 12.969400244798042,
      "grad_norm": 1.676071742013873,
      "learning_rate": 2.2269994553342028e-07,
      "loss": 1.6322,
      "step": 10596
    },
    {
      "epoch": 12.97062423500612,
      "grad_norm": 1.0963230615092272,
      "learning_rate": 2.2243560427823218e-07,
      "loss": 0.5443,
      "step": 10597
    },
    {
      "epoch": 12.971848225214199,
      "grad_norm": 1.9276678626298318,
      "learning_rate": 2.221714126961519e-07,
      "loss": 0.5207,
      "step": 10598
    },
    {
      "epoch": 12.973072215422278,
      "grad_norm": 2.167614156880599,
      "learning_rate": 2.2190737080454132e-07,
      "loss": 0.4765,
      "step": 10599
    },
    {
      "epoch": 12.974296205630354,
      "grad_norm": 1.4200751684629938,
      "learning_rate": 2.216434786207522e-07,
      "loss": 0.7358,
      "step": 10600
    },
    {
      "epoch": 12.975520195838433,
      "grad_norm": 1.1553015806363534,
      "learning_rate": 2.2137973616212672e-07,
      "loss": 0.5942,
      "step": 10601
    },
    {
      "epoch": 12.976744186046512,
      "grad_norm": 1.8695272632706061,
      "learning_rate": 2.2111614344599686e-07,
      "loss": 0.4947,
      "step": 10602
    },
    {
      "epoch": 12.97796817625459,
      "grad_norm": 2.1314823184365963,
      "learning_rate": 2.208527004896846e-07,
      "loss": 0.4023,
      "step": 10603
    },
    {
      "epoch": 12.979192166462669,
      "grad_norm": 2.1027981129461506,
      "learning_rate": 2.2058940731050387e-07,
      "loss": 0.6523,
      "step": 10604
    },
    {
      "epoch": 12.980416156670747,
      "grad_norm": 2.274066895915053,
      "learning_rate": 2.2032626392575557e-07,
      "loss": 0.5332,
      "step": 10605
    },
    {
      "epoch": 12.981640146878824,
      "grad_norm": 1.3017128160308493,
      "learning_rate": 2.200632703527336e-07,
      "loss": 0.3838,
      "step": 10606
    },
    {
      "epoch": 12.982864137086903,
      "grad_norm": 1.874907537541565,
      "learning_rate": 2.1980042660872002e-07,
      "loss": 0.9792,
      "step": 10607
    },
    {
      "epoch": 12.984088127294982,
      "grad_norm": 1.8485890232250277,
      "learning_rate": 2.1953773271098905e-07,
      "loss": 1.3976,
      "step": 10608
    },
    {
      "epoch": 12.98531211750306,
      "grad_norm": 2.5475064953115973,
      "learning_rate": 2.1927518867680352e-07,
      "loss": 1.2735,
      "step": 10609
    },
    {
      "epoch": 12.986536107711139,
      "grad_norm": 2.442569954295866,
      "learning_rate": 2.1901279452341662e-07,
      "loss": 0.5861,
      "step": 10610
    },
    {
      "epoch": 12.987760097919217,
      "grad_norm": 2.204485358159137,
      "learning_rate": 2.1875055026807257e-07,
      "loss": 0.4152,
      "step": 10611
    },
    {
      "epoch": 12.988984088127294,
      "grad_norm": 1.4890199926063017,
      "learning_rate": 2.184884559280037e-07,
      "loss": 0.6159,
      "step": 10612
    },
    {
      "epoch": 12.990208078335373,
      "grad_norm": 2.032254708132032,
      "learning_rate": 2.1822651152043516e-07,
      "loss": 0.9009,
      "step": 10613
    },
    {
      "epoch": 12.991432068543451,
      "grad_norm": 2.4731638053574807,
      "learning_rate": 2.1796471706258066e-07,
      "loss": 0.4411,
      "step": 10614
    },
    {
      "epoch": 12.99265605875153,
      "grad_norm": 1.5201754265090965,
      "learning_rate": 2.177030725716442e-07,
      "loss": 0.5404,
      "step": 10615
    },
    {
      "epoch": 12.993880048959609,
      "grad_norm": 1.553221801983763,
      "learning_rate": 2.1744157806482036e-07,
      "loss": 1.2546,
      "step": 10616
    },
    {
      "epoch": 12.995104039167687,
      "grad_norm": 1.9859296388373553,
      "learning_rate": 2.1718023355929346e-07,
      "loss": 0.3551,
      "step": 10617
    },
    {
      "epoch": 12.996328029375764,
      "grad_norm": 2.7241307574855544,
      "learning_rate": 2.1691903907223782e-07,
      "loss": 0.3402,
      "step": 10618
    },
    {
      "epoch": 12.997552019583843,
      "grad_norm": 1.8086843663778427,
      "learning_rate": 2.1665799462081833e-07,
      "loss": 0.4442,
      "step": 10619
    },
    {
      "epoch": 12.998776009791921,
      "grad_norm": 1.7687571371842425,
      "learning_rate": 2.1639710022218984e-07,
      "loss": 1.3947,
      "step": 10620
    },
    {
      "epoch": 13.0,
      "grad_norm": 1.471676692229264,
      "learning_rate": 2.1613635589349756e-07,
      "loss": 1.3069,
      "step": 10621
    },
    {
      "epoch": 13.001223990208079,
      "grad_norm": 1.4078831021812097,
      "learning_rate": 2.1587576165187724e-07,
      "loss": 0.3618,
      "step": 10622
    },
    {
      "epoch": 13.002447980416157,
      "grad_norm": 1.8268492227718474,
      "learning_rate": 2.1561531751445293e-07,
      "loss": 0.5731,
      "step": 10623
    },
    {
      "epoch": 13.003671970624236,
      "grad_norm": 1.6611127812544357,
      "learning_rate": 2.1535502349834065e-07,
      "loss": 1.1632,
      "step": 10624
    },
    {
      "epoch": 13.004895960832313,
      "grad_norm": 1.6471403203435377,
      "learning_rate": 2.1509487962064536e-07,
      "loss": 0.5938,
      "step": 10625
    },
    {
      "epoch": 13.006119951040391,
      "grad_norm": 1.9826636549166305,
      "learning_rate": 2.1483488589846363e-07,
      "loss": 0.5476,
      "step": 10626
    },
    {
      "epoch": 13.00734394124847,
      "grad_norm": 2.6413232711312062,
      "learning_rate": 2.1457504234888127e-07,
      "loss": 0.8485,
      "step": 10627
    },
    {
      "epoch": 13.008567931456549,
      "grad_norm": 2.9594860210660223,
      "learning_rate": 2.1431534898897406e-07,
      "loss": 0.5231,
      "step": 10628
    },
    {
      "epoch": 13.009791921664627,
      "grad_norm": 1.5818665449303817,
      "learning_rate": 2.140558058358072e-07,
      "loss": 0.4719,
      "step": 10629
    },
    {
      "epoch": 13.011015911872706,
      "grad_norm": 2.5109699715423726,
      "learning_rate": 2.137964129064382e-07,
      "loss": 0.5504,
      "step": 10630
    },
    {
      "epoch": 13.012239902080783,
      "grad_norm": 1.035109928704139,
      "learning_rate": 2.1353717021791288e-07,
      "loss": 0.4908,
      "step": 10631
    },
    {
      "epoch": 13.013463892288861,
      "grad_norm": 2.04294917593886,
      "learning_rate": 2.1327807778726755e-07,
      "loss": 0.5071,
      "step": 10632
    },
    {
      "epoch": 13.01468788249694,
      "grad_norm": 1.8183773564970662,
      "learning_rate": 2.1301913563152892e-07,
      "loss": 0.518,
      "step": 10633
    },
    {
      "epoch": 13.015911872705018,
      "grad_norm": 0.757797128181512,
      "learning_rate": 2.1276034376771393e-07,
      "loss": 0.2549,
      "step": 10634
    },
    {
      "epoch": 13.017135862913097,
      "grad_norm": 1.8983057876243326,
      "learning_rate": 2.1250170221282895e-07,
      "loss": 1.2993,
      "step": 10635
    },
    {
      "epoch": 13.018359853121176,
      "grad_norm": 1.4904906527404769,
      "learning_rate": 2.1224321098387151e-07,
      "loss": 1.314,
      "step": 10636
    },
    {
      "epoch": 13.019583843329253,
      "grad_norm": 1.6050771633207728,
      "learning_rate": 2.1198487009782832e-07,
      "loss": 0.8163,
      "step": 10637
    },
    {
      "epoch": 13.020807833537331,
      "grad_norm": 2.4613570334324684,
      "learning_rate": 2.117266795716763e-07,
      "loss": 0.4422,
      "step": 10638
    },
    {
      "epoch": 13.02203182374541,
      "grad_norm": 1.6004702677801592,
      "learning_rate": 2.1146863942238332e-07,
      "loss": 0.7055,
      "step": 10639
    },
    {
      "epoch": 13.023255813953488,
      "grad_norm": 1.980127400724189,
      "learning_rate": 2.1121074966690746e-07,
      "loss": 1.1804,
      "step": 10640
    },
    {
      "epoch": 13.024479804161567,
      "grad_norm": 2.829994268379481,
      "learning_rate": 2.109530103221949e-07,
      "loss": 0.5721,
      "step": 10641
    },
    {
      "epoch": 13.025703794369646,
      "grad_norm": 3.915845584897813,
      "learning_rate": 2.1069542140518346e-07,
      "loss": 0.3685,
      "step": 10642
    },
    {
      "epoch": 13.026927784577722,
      "grad_norm": 2.0273138738762957,
      "learning_rate": 2.1043798293280182e-07,
      "loss": 0.3579,
      "step": 10643
    },
    {
      "epoch": 13.028151774785801,
      "grad_norm": 1.431602169000849,
      "learning_rate": 2.101806949219673e-07,
      "loss": 0.7718,
      "step": 10644
    },
    {
      "epoch": 13.02937576499388,
      "grad_norm": 1.544755847289692,
      "learning_rate": 2.099235573895883e-07,
      "loss": 0.781,
      "step": 10645
    },
    {
      "epoch": 13.030599755201958,
      "grad_norm": 1.406490975701887,
      "learning_rate": 2.0966657035256298e-07,
      "loss": 0.4457,
      "step": 10646
    },
    {
      "epoch": 13.031823745410037,
      "grad_norm": 3.6200259111861626,
      "learning_rate": 2.0940973382777868e-07,
      "loss": 0.7537,
      "step": 10647
    },
    {
      "epoch": 13.033047735618116,
      "grad_norm": 1.1510963267752847,
      "learning_rate": 2.0915304783211437e-07,
      "loss": 0.5913,
      "step": 10648
    },
    {
      "epoch": 13.034271725826194,
      "grad_norm": 1.994459182504412,
      "learning_rate": 2.088965123824388e-07,
      "loss": 0.7015,
      "step": 10649
    },
    {
      "epoch": 13.035495716034271,
      "grad_norm": 1.4067703540831262,
      "learning_rate": 2.0864012749561013e-07,
      "loss": 0.6746,
      "step": 10650
    },
    {
      "epoch": 13.03671970624235,
      "grad_norm": 2.7371415347713195,
      "learning_rate": 2.0838389318847712e-07,
      "loss": 0.4642,
      "step": 10651
    },
    {
      "epoch": 13.037943696450428,
      "grad_norm": 2.439559826634428,
      "learning_rate": 2.081278094778788e-07,
      "loss": 0.5814,
      "step": 10652
    },
    {
      "epoch": 13.039167686658507,
      "grad_norm": 2.174299768494385,
      "learning_rate": 2.0787187638064337e-07,
      "loss": 0.3828,
      "step": 10653
    },
    {
      "epoch": 13.040391676866586,
      "grad_norm": 2.058014208605519,
      "learning_rate": 2.0761609391359043e-07,
      "loss": 0.7271,
      "step": 10654
    },
    {
      "epoch": 13.041615667074664,
      "grad_norm": 2.528971527663554,
      "learning_rate": 2.073604620935285e-07,
      "loss": 1.0081,
      "step": 10655
    },
    {
      "epoch": 13.042839657282741,
      "grad_norm": 1.3965508007436056,
      "learning_rate": 2.0710498093725745e-07,
      "loss": 0.5518,
      "step": 10656
    },
    {
      "epoch": 13.04406364749082,
      "grad_norm": 1.3189881224705073,
      "learning_rate": 2.0684965046156692e-07,
      "loss": 0.4454,
      "step": 10657
    },
    {
      "epoch": 13.045287637698898,
      "grad_norm": 1.969666254082318,
      "learning_rate": 2.0659447068323484e-07,
      "loss": 0.5956,
      "step": 10658
    },
    {
      "epoch": 13.046511627906977,
      "grad_norm": 1.3139505615526006,
      "learning_rate": 2.0633944161903147e-07,
      "loss": 0.5864,
      "step": 10659
    },
    {
      "epoch": 13.047735618115055,
      "grad_norm": 1.932891456951174,
      "learning_rate": 2.0608456328571585e-07,
      "loss": 1.482,
      "step": 10660
    },
    {
      "epoch": 13.048959608323134,
      "grad_norm": 1.5238700370529894,
      "learning_rate": 2.058298357000388e-07,
      "loss": 0.7894,
      "step": 10661
    },
    {
      "epoch": 13.050183598531211,
      "grad_norm": 1.906783963370472,
      "learning_rate": 2.0557525887873908e-07,
      "loss": 0.9194,
      "step": 10662
    },
    {
      "epoch": 13.05140758873929,
      "grad_norm": 1.0927219116472158,
      "learning_rate": 2.05320832838547e-07,
      "loss": 0.5337,
      "step": 10663
    },
    {
      "epoch": 13.052631578947368,
      "grad_norm": 1.6666857023512764,
      "learning_rate": 2.0506655759618245e-07,
      "loss": 0.5472,
      "step": 10664
    },
    {
      "epoch": 13.053855569155447,
      "grad_norm": 1.6583089346370568,
      "learning_rate": 2.0481243316835543e-07,
      "loss": 0.5179,
      "step": 10665
    },
    {
      "epoch": 13.055079559363525,
      "grad_norm": 2.9882163081165465,
      "learning_rate": 2.0455845957176563e-07,
      "loss": 0.3547,
      "step": 10666
    },
    {
      "epoch": 13.056303549571604,
      "grad_norm": 1.3776549262178153,
      "learning_rate": 2.0430463682310387e-07,
      "loss": 0.521,
      "step": 10667
    },
    {
      "epoch": 13.057527539779683,
      "grad_norm": 2.9589987933776456,
      "learning_rate": 2.0405096493904985e-07,
      "loss": 0.3443,
      "step": 10668
    },
    {
      "epoch": 13.05875152998776,
      "grad_norm": 2.437518545860974,
      "learning_rate": 2.03797443936275e-07,
      "loss": 0.4918,
      "step": 10669
    },
    {
      "epoch": 13.059975520195838,
      "grad_norm": 2.086566076103048,
      "learning_rate": 2.035440738314387e-07,
      "loss": 0.4939,
      "step": 10670
    },
    {
      "epoch": 13.061199510403917,
      "grad_norm": 2.678157025969947,
      "learning_rate": 2.0329085464119157e-07,
      "loss": 0.3955,
      "step": 10671
    },
    {
      "epoch": 13.062423500611995,
      "grad_norm": 1.9500250405987996,
      "learning_rate": 2.0303778638217503e-07,
      "loss": 0.678,
      "step": 10672
    },
    {
      "epoch": 13.063647490820074,
      "grad_norm": 3.8256931989743523,
      "learning_rate": 2.0278486907101852e-07,
      "loss": 0.4101,
      "step": 10673
    },
    {
      "epoch": 13.064871481028153,
      "grad_norm": 2.7202564353012986,
      "learning_rate": 2.0253210272434404e-07,
      "loss": 0.2703,
      "step": 10674
    },
    {
      "epoch": 13.06609547123623,
      "grad_norm": 3.656628430490908,
      "learning_rate": 2.0227948735876274e-07,
      "loss": 0.4169,
      "step": 10675
    },
    {
      "epoch": 13.067319461444308,
      "grad_norm": 2.327646762265486,
      "learning_rate": 2.020270229908741e-07,
      "loss": 0.5158,
      "step": 10676
    },
    {
      "epoch": 13.068543451652387,
      "grad_norm": 2.0065586994834166,
      "learning_rate": 2.0177470963726963e-07,
      "loss": 0.6194,
      "step": 10677
    },
    {
      "epoch": 13.069767441860465,
      "grad_norm": 1.6390633286319256,
      "learning_rate": 2.0152254731453102e-07,
      "loss": 0.4689,
      "step": 10678
    },
    {
      "epoch": 13.070991432068544,
      "grad_norm": 1.1674926757836828,
      "learning_rate": 2.0127053603922918e-07,
      "loss": 0.5045,
      "step": 10679
    },
    {
      "epoch": 13.072215422276622,
      "grad_norm": 2.6689240535714887,
      "learning_rate": 2.0101867582792533e-07,
      "loss": 0.3252,
      "step": 10680
    },
    {
      "epoch": 13.0734394124847,
      "grad_norm": 2.919849596127553,
      "learning_rate": 2.0076696669717066e-07,
      "loss": 0.3476,
      "step": 10681
    },
    {
      "epoch": 13.074663402692778,
      "grad_norm": 1.8761902479416768,
      "learning_rate": 2.0051540866350694e-07,
      "loss": 1.3393,
      "step": 10682
    },
    {
      "epoch": 13.075887392900857,
      "grad_norm": 2.167567530735371,
      "learning_rate": 2.0026400174346512e-07,
      "loss": 1.1134,
      "step": 10683
    },
    {
      "epoch": 13.077111383108935,
      "grad_norm": 2.0716018172033723,
      "learning_rate": 2.0001274595356696e-07,
      "loss": 0.3847,
      "step": 10684
    },
    {
      "epoch": 13.078335373317014,
      "grad_norm": 1.6661860952843675,
      "learning_rate": 1.9976164131032427e-07,
      "loss": 0.2911,
      "step": 10685
    },
    {
      "epoch": 13.079559363525092,
      "grad_norm": 1.5715609645077604,
      "learning_rate": 1.995106878302383e-07,
      "loss": 0.8599,
      "step": 10686
    },
    {
      "epoch": 13.080783353733171,
      "grad_norm": 1.1561336932536757,
      "learning_rate": 1.9925988552980114e-07,
      "loss": 0.6363,
      "step": 10687
    },
    {
      "epoch": 13.082007343941248,
      "grad_norm": 2.347464430123916,
      "learning_rate": 1.990092344254943e-07,
      "loss": 0.5403,
      "step": 10688
    },
    {
      "epoch": 13.083231334149326,
      "grad_norm": 1.194724387618907,
      "learning_rate": 1.9875873453378992e-07,
      "loss": 0.7416,
      "step": 10689
    },
    {
      "epoch": 13.084455324357405,
      "grad_norm": 2.9282037947895962,
      "learning_rate": 1.9850838587114923e-07,
      "loss": 0.3467,
      "step": 10690
    },
    {
      "epoch": 13.085679314565484,
      "grad_norm": 2.6825106352891033,
      "learning_rate": 1.982581884540255e-07,
      "loss": 0.3823,
      "step": 10691
    },
    {
      "epoch": 13.086903304773562,
      "grad_norm": 1.331154935044085,
      "learning_rate": 1.9800814229885974e-07,
      "loss": 0.5983,
      "step": 10692
    },
    {
      "epoch": 13.088127294981641,
      "grad_norm": 1.1690900339191943,
      "learning_rate": 1.9775824742208516e-07,
      "loss": 1.1941,
      "step": 10693
    },
    {
      "epoch": 13.089351285189718,
      "grad_norm": 2.2354395951051513,
      "learning_rate": 1.97508503840122e-07,
      "loss": 0.5278,
      "step": 10694
    },
    {
      "epoch": 13.090575275397796,
      "grad_norm": 1.8851408272278223,
      "learning_rate": 1.972589115693843e-07,
      "loss": 1.8735,
      "step": 10695
    },
    {
      "epoch": 13.091799265605875,
      "grad_norm": 2.016810099898951,
      "learning_rate": 1.9700947062627345e-07,
      "loss": 1.0416,
      "step": 10696
    },
    {
      "epoch": 13.093023255813954,
      "grad_norm": 2.244564978362661,
      "learning_rate": 1.9676018102718215e-07,
      "loss": 0.4781,
      "step": 10697
    },
    {
      "epoch": 13.094247246022032,
      "grad_norm": 2.4739186648045077,
      "learning_rate": 1.965110427884928e-07,
      "loss": 0.3572,
      "step": 10698
    },
    {
      "epoch": 13.095471236230111,
      "grad_norm": 3.6583261804831473,
      "learning_rate": 1.9626205592657765e-07,
      "loss": 0.3376,
      "step": 10699
    },
    {
      "epoch": 13.096695226438188,
      "grad_norm": 2.2021991068962166,
      "learning_rate": 1.9601322045779913e-07,
      "loss": 0.954,
      "step": 10700
    },
    {
      "epoch": 13.097919216646266,
      "grad_norm": 2.176663228708014,
      "learning_rate": 1.9576453639850972e-07,
      "loss": 1.1873,
      "step": 10701
    },
    {
      "epoch": 13.099143206854345,
      "grad_norm": 1.8008079671162243,
      "learning_rate": 1.9551600376505247e-07,
      "loss": 0.8196,
      "step": 10702
    },
    {
      "epoch": 13.100367197062424,
      "grad_norm": 1.8310770447131426,
      "learning_rate": 1.9526762257375932e-07,
      "loss": 0.6233,
      "step": 10703
    },
    {
      "epoch": 13.101591187270502,
      "grad_norm": 1.367096349125418,
      "learning_rate": 1.950193928409541e-07,
      "loss": 0.572,
      "step": 10704
    },
    {
      "epoch": 13.10281517747858,
      "grad_norm": 1.7092612781039433,
      "learning_rate": 1.9477131458294858e-07,
      "loss": 0.4943,
      "step": 10705
    },
    {
      "epoch": 13.104039167686658,
      "grad_norm": 1.5422920833297205,
      "learning_rate": 1.9452338781604573e-07,
      "loss": 0.6142,
      "step": 10706
    },
    {
      "epoch": 13.105263157894736,
      "grad_norm": 1.635020772035553,
      "learning_rate": 1.9427561255653816e-07,
      "loss": 0.5651,
      "step": 10707
    },
    {
      "epoch": 13.106487148102815,
      "grad_norm": 2.192782832537619,
      "learning_rate": 1.9402798882070944e-07,
      "loss": 0.304,
      "step": 10708
    },
    {
      "epoch": 13.107711138310894,
      "grad_norm": 2.952130952364076,
      "learning_rate": 1.9378051662483188e-07,
      "loss": 0.4929,
      "step": 10709
    },
    {
      "epoch": 13.108935128518972,
      "grad_norm": 3.01593307612884,
      "learning_rate": 1.9353319598516884e-07,
      "loss": 0.4511,
      "step": 10710
    },
    {
      "epoch": 13.11015911872705,
      "grad_norm": 1.416224838425651,
      "learning_rate": 1.9328602691797372e-07,
      "loss": 0.6734,
      "step": 10711
    },
    {
      "epoch": 13.11138310893513,
      "grad_norm": 1.4025118658974476,
      "learning_rate": 1.9303900943948767e-07,
      "loss": 0.5961,
      "step": 10712
    },
    {
      "epoch": 13.112607099143206,
      "grad_norm": 1.8666928361208635,
      "learning_rate": 1.9279214356594576e-07,
      "loss": 0.4612,
      "step": 10713
    },
    {
      "epoch": 13.113831089351285,
      "grad_norm": 1.5894334355936408,
      "learning_rate": 1.9254542931357062e-07,
      "loss": 0.4217,
      "step": 10714
    },
    {
      "epoch": 13.115055079559363,
      "grad_norm": 1.1630234833268647,
      "learning_rate": 1.922988666985748e-07,
      "loss": 0.3654,
      "step": 10715
    },
    {
      "epoch": 13.116279069767442,
      "grad_norm": 3.1195658146939125,
      "learning_rate": 1.9205245573716196e-07,
      "loss": 0.5194,
      "step": 10716
    },
    {
      "epoch": 13.11750305997552,
      "grad_norm": 2.098411736307625,
      "learning_rate": 1.9180619644552534e-07,
      "loss": 0.7683,
      "step": 10717
    },
    {
      "epoch": 13.1187270501836,
      "grad_norm": 1.3336682117626448,
      "learning_rate": 1.9156008883984805e-07,
      "loss": 0.3989,
      "step": 10718
    },
    {
      "epoch": 13.119951040391676,
      "grad_norm": 1.3091345357772137,
      "learning_rate": 1.913141329363033e-07,
      "loss": 0.4244,
      "step": 10719
    },
    {
      "epoch": 13.121175030599755,
      "grad_norm": 1.758357839776261,
      "learning_rate": 1.9106832875105398e-07,
      "loss": 0.5672,
      "step": 10720
    },
    {
      "epoch": 13.122399020807833,
      "grad_norm": 2.782687790887837,
      "learning_rate": 1.9082267630025463e-07,
      "loss": 0.7821,
      "step": 10721
    },
    {
      "epoch": 13.123623011015912,
      "grad_norm": 1.8906565899797316,
      "learning_rate": 1.9057717560004822e-07,
      "loss": 1.0789,
      "step": 10722
    },
    {
      "epoch": 13.12484700122399,
      "grad_norm": 1.4704545480534001,
      "learning_rate": 1.903318266665674e-07,
      "loss": 1.3067,
      "step": 10723
    },
    {
      "epoch": 13.12607099143207,
      "grad_norm": 2.152609883794192,
      "learning_rate": 1.9008662951593622e-07,
      "loss": 0.3779,
      "step": 10724
    },
    {
      "epoch": 13.127294981640146,
      "grad_norm": 1.907154629870227,
      "learning_rate": 1.8984158416426728e-07,
      "loss": 0.5248,
      "step": 10725
    },
    {
      "epoch": 13.128518971848225,
      "grad_norm": 1.7064674444064973,
      "learning_rate": 1.8959669062766556e-07,
      "loss": 0.4536,
      "step": 10726
    },
    {
      "epoch": 13.129742962056303,
      "grad_norm": 1.4989115303642548,
      "learning_rate": 1.893519489222234e-07,
      "loss": 0.6241,
      "step": 10727
    },
    {
      "epoch": 13.130966952264382,
      "grad_norm": 2.6946255187613417,
      "learning_rate": 1.8910735906402495e-07,
      "loss": 0.4662,
      "step": 10728
    },
    {
      "epoch": 13.13219094247246,
      "grad_norm": 1.7071063339189552,
      "learning_rate": 1.888629210691431e-07,
      "loss": 0.5546,
      "step": 10729
    },
    {
      "epoch": 13.13341493268054,
      "grad_norm": 1.0919970074102305,
      "learning_rate": 1.8861863495364202e-07,
      "loss": 0.5875,
      "step": 10730
    },
    {
      "epoch": 13.134638922888616,
      "grad_norm": 1.4362684303710538,
      "learning_rate": 1.8837450073357493e-07,
      "loss": 0.355,
      "step": 10731
    },
    {
      "epoch": 13.135862913096695,
      "grad_norm": 1.789025549299541,
      "learning_rate": 1.881305184249857e-07,
      "loss": 0.5243,
      "step": 10732
    },
    {
      "epoch": 13.137086903304773,
      "grad_norm": 2.1033087793088496,
      "learning_rate": 1.878866880439076e-07,
      "loss": 0.5227,
      "step": 10733
    },
    {
      "epoch": 13.138310893512852,
      "grad_norm": 1.5667056071262764,
      "learning_rate": 1.8764300960636446e-07,
      "loss": 1.1114,
      "step": 10734
    },
    {
      "epoch": 13.13953488372093,
      "grad_norm": 1.7347827005351377,
      "learning_rate": 1.8739948312837015e-07,
      "loss": 0.8454,
      "step": 10735
    },
    {
      "epoch": 13.140758873929009,
      "grad_norm": 1.7687830207123703,
      "learning_rate": 1.8715610862592797e-07,
      "loss": 1.5121,
      "step": 10736
    },
    {
      "epoch": 13.141982864137088,
      "grad_norm": 2.3982472282136484,
      "learning_rate": 1.8691288611503156e-07,
      "loss": 1.0244,
      "step": 10737
    },
    {
      "epoch": 13.143206854345165,
      "grad_norm": 1.5536916520950197,
      "learning_rate": 1.8666981561166447e-07,
      "loss": 1.0673,
      "step": 10738
    },
    {
      "epoch": 13.144430844553243,
      "grad_norm": 1.7798937542116158,
      "learning_rate": 1.864268971318012e-07,
      "loss": 1.2262,
      "step": 10739
    },
    {
      "epoch": 13.145654834761322,
      "grad_norm": 1.821191039712932,
      "learning_rate": 1.86184130691405e-07,
      "loss": 0.6496,
      "step": 10740
    },
    {
      "epoch": 13.1468788249694,
      "grad_norm": 2.5945754075618983,
      "learning_rate": 1.859415163064293e-07,
      "loss": 0.4008,
      "step": 10741
    },
    {
      "epoch": 13.148102815177479,
      "grad_norm": 2.8043650791734036,
      "learning_rate": 1.8569905399281773e-07,
      "loss": 0.5057,
      "step": 10742
    },
    {
      "epoch": 13.149326805385558,
      "grad_norm": 3.086942011029647,
      "learning_rate": 1.8545674376650474e-07,
      "loss": 0.2036,
      "step": 10743
    },
    {
      "epoch": 13.150550795593634,
      "grad_norm": 1.1782561549721169,
      "learning_rate": 1.8521458564341344e-07,
      "loss": 0.8444,
      "step": 10744
    },
    {
      "epoch": 13.151774785801713,
      "grad_norm": 1.9158324714069161,
      "learning_rate": 1.8497257963945803e-07,
      "loss": 0.4554,
      "step": 10745
    },
    {
      "epoch": 13.152998776009792,
      "grad_norm": 2.7168552012879483,
      "learning_rate": 1.847307257705419e-07,
      "loss": 0.9041,
      "step": 10746
    },
    {
      "epoch": 13.15422276621787,
      "grad_norm": 1.4901572475435245,
      "learning_rate": 1.8448902405255875e-07,
      "loss": 0.5262,
      "step": 10747
    },
    {
      "epoch": 13.155446756425949,
      "grad_norm": 1.6993474226866663,
      "learning_rate": 1.8424747450139252e-07,
      "loss": 0.4843,
      "step": 10748
    },
    {
      "epoch": 13.156670746634028,
      "grad_norm": 2.730684559942783,
      "learning_rate": 1.8400607713291691e-07,
      "loss": 0.7161,
      "step": 10749
    },
    {
      "epoch": 13.157894736842104,
      "grad_norm": 1.4922369982380612,
      "learning_rate": 1.837648319629956e-07,
      "loss": 0.6617,
      "step": 10750
    },
    {
      "epoch": 13.159118727050183,
      "grad_norm": 2.6154939610887364,
      "learning_rate": 1.835237390074826e-07,
      "loss": 0.809,
      "step": 10751
    },
    {
      "epoch": 13.160342717258262,
      "grad_norm": 2.253508069302998,
      "learning_rate": 1.8328279828222134e-07,
      "loss": 0.8147,
      "step": 10752
    },
    {
      "epoch": 13.16156670746634,
      "grad_norm": 1.5155161123409944,
      "learning_rate": 1.8304200980304553e-07,
      "loss": 1.4059,
      "step": 10753
    },
    {
      "epoch": 13.162790697674419,
      "grad_norm": 2.3261801653531498,
      "learning_rate": 1.8280137358577892e-07,
      "loss": 0.477,
      "step": 10754
    },
    {
      "epoch": 13.164014687882498,
      "grad_norm": 2.3070712285817128,
      "learning_rate": 1.8256088964623498e-07,
      "loss": 0.5965,
      "step": 10755
    },
    {
      "epoch": 13.165238678090576,
      "grad_norm": 1.4924585403227348,
      "learning_rate": 1.8232055800021826e-07,
      "loss": 0.62,
      "step": 10756
    },
    {
      "epoch": 13.166462668298653,
      "grad_norm": 1.9381792718964943,
      "learning_rate": 1.820803786635217e-07,
      "loss": 0.3836,
      "step": 10757
    },
    {
      "epoch": 13.167686658506732,
      "grad_norm": 1.9085052725699083,
      "learning_rate": 1.8184035165192986e-07,
      "loss": 0.7244,
      "step": 10758
    },
    {
      "epoch": 13.16891064871481,
      "grad_norm": 1.93959065510096,
      "learning_rate": 1.8160047698121519e-07,
      "loss": 0.4664,
      "step": 10759
    },
    {
      "epoch": 13.170134638922889,
      "grad_norm": 2.7749625302833376,
      "learning_rate": 1.8136075466714225e-07,
      "loss": 0.8874,
      "step": 10760
    },
    {
      "epoch": 13.171358629130967,
      "grad_norm": 2.83586223226373,
      "learning_rate": 1.811211847254643e-07,
      "loss": 0.3517,
      "step": 10761
    },
    {
      "epoch": 13.172582619339046,
      "grad_norm": 1.847277119047382,
      "learning_rate": 1.808817671719254e-07,
      "loss": 0.521,
      "step": 10762
    },
    {
      "epoch": 13.173806609547123,
      "grad_norm": 2.3333794732089617,
      "learning_rate": 1.8064250202225881e-07,
      "loss": 0.4082,
      "step": 10763
    },
    {
      "epoch": 13.175030599755202,
      "grad_norm": 1.8104690864242152,
      "learning_rate": 1.8040338929218832e-07,
      "loss": 0.3073,
      "step": 10764
    },
    {
      "epoch": 13.17625458996328,
      "grad_norm": 2.667455516796761,
      "learning_rate": 1.8016442899742747e-07,
      "loss": 0.6668,
      "step": 10765
    },
    {
      "epoch": 13.177478580171359,
      "grad_norm": 2.0485787941196,
      "learning_rate": 1.7992562115367983e-07,
      "loss": 1.0291,
      "step": 10766
    },
    {
      "epoch": 13.178702570379437,
      "grad_norm": 1.5937292172026674,
      "learning_rate": 1.7968696577663923e-07,
      "loss": 0.3183,
      "step": 10767
    },
    {
      "epoch": 13.179926560587516,
      "grad_norm": 2.0497641717052435,
      "learning_rate": 1.7944846288198842e-07,
      "loss": 0.5106,
      "step": 10768
    },
    {
      "epoch": 13.181150550795593,
      "grad_norm": 0.9812707088664636,
      "learning_rate": 1.7921011248540233e-07,
      "loss": 0.3874,
      "step": 10769
    },
    {
      "epoch": 13.182374541003671,
      "grad_norm": 2.1040564946447304,
      "learning_rate": 1.7897191460254316e-07,
      "loss": 0.4,
      "step": 10770
    },
    {
      "epoch": 13.18359853121175,
      "grad_norm": 1.7700125707148011,
      "learning_rate": 1.7873386924906476e-07,
      "loss": 0.743,
      "step": 10771
    },
    {
      "epoch": 13.184822521419829,
      "grad_norm": 2.654962935092691,
      "learning_rate": 1.784959764406105e-07,
      "loss": 0.4575,
      "step": 10772
    },
    {
      "epoch": 13.186046511627907,
      "grad_norm": 0.9687164084364793,
      "learning_rate": 1.7825823619281452e-07,
      "loss": 0.4373,
      "step": 10773
    },
    {
      "epoch": 13.187270501835986,
      "grad_norm": 1.1084454954272414,
      "learning_rate": 1.7802064852129957e-07,
      "loss": 0.504,
      "step": 10774
    },
    {
      "epoch": 13.188494492044065,
      "grad_norm": 2.076194384364367,
      "learning_rate": 1.777832134416796e-07,
      "loss": 0.5209,
      "step": 10775
    },
    {
      "epoch": 13.189718482252141,
      "grad_norm": 1.1795323473334225,
      "learning_rate": 1.775459309695571e-07,
      "loss": 0.5551,
      "step": 10776
    },
    {
      "epoch": 13.19094247246022,
      "grad_norm": 1.5935701728585197,
      "learning_rate": 1.7730880112052568e-07,
      "loss": 0.6065,
      "step": 10777
    },
    {
      "epoch": 13.192166462668299,
      "grad_norm": 1.6408053564338947,
      "learning_rate": 1.7707182391016902e-07,
      "loss": 1.2883,
      "step": 10778
    },
    {
      "epoch": 13.193390452876377,
      "grad_norm": 2.414702596140684,
      "learning_rate": 1.7683499935406023e-07,
      "loss": 0.8985,
      "step": 10779
    },
    {
      "epoch": 13.194614443084456,
      "grad_norm": 2.1913066763070503,
      "learning_rate": 1.765983274677624e-07,
      "loss": 0.8288,
      "step": 10780
    },
    {
      "epoch": 13.195838433292534,
      "grad_norm": 1.432272376149489,
      "learning_rate": 1.7636180826682892e-07,
      "loss": 0.5403,
      "step": 10781
    },
    {
      "epoch": 13.197062423500611,
      "grad_norm": 2.6400438930332313,
      "learning_rate": 1.761254417668029e-07,
      "loss": 0.6135,
      "step": 10782
    },
    {
      "epoch": 13.19828641370869,
      "grad_norm": 1.610207887302078,
      "learning_rate": 1.7588922798321723e-07,
      "loss": 0.5626,
      "step": 10783
    },
    {
      "epoch": 13.199510403916769,
      "grad_norm": 1.2489417890792995,
      "learning_rate": 1.7565316693159528e-07,
      "loss": 0.5371,
      "step": 10784
    },
    {
      "epoch": 13.200734394124847,
      "grad_norm": 1.5402933914798187,
      "learning_rate": 1.7541725862745024e-07,
      "loss": 0.4498,
      "step": 10785
    },
    {
      "epoch": 13.201958384332926,
      "grad_norm": 1.7506840936823174,
      "learning_rate": 1.751815030862844e-07,
      "loss": 0.6652,
      "step": 10786
    },
    {
      "epoch": 13.203182374541004,
      "grad_norm": 2.901722710543732,
      "learning_rate": 1.7494590032359204e-07,
      "loss": 0.437,
      "step": 10787
    },
    {
      "epoch": 13.204406364749081,
      "grad_norm": 2.080655535296931,
      "learning_rate": 1.7471045035485495e-07,
      "loss": 0.7562,
      "step": 10788
    },
    {
      "epoch": 13.20563035495716,
      "grad_norm": 2.3333946452347702,
      "learning_rate": 1.7447515319554632e-07,
      "loss": 0.6658,
      "step": 10789
    },
    {
      "epoch": 13.206854345165238,
      "grad_norm": 2.044234534582507,
      "learning_rate": 1.7424000886112875e-07,
      "loss": 0.6179,
      "step": 10790
    },
    {
      "epoch": 13.208078335373317,
      "grad_norm": 1.5621450202085896,
      "learning_rate": 1.74005017367056e-07,
      "loss": 0.8664,
      "step": 10791
    },
    {
      "epoch": 13.209302325581396,
      "grad_norm": 1.910650533699908,
      "learning_rate": 1.737701787287699e-07,
      "loss": 0.433,
      "step": 10792
    },
    {
      "epoch": 13.210526315789474,
      "grad_norm": 2.9039544718234946,
      "learning_rate": 1.735354929617042e-07,
      "loss": 0.4397,
      "step": 10793
    },
    {
      "epoch": 13.211750305997551,
      "grad_norm": 2.4771014994006273,
      "learning_rate": 1.733009600812799e-07,
      "loss": 0.391,
      "step": 10794
    },
    {
      "epoch": 13.21297429620563,
      "grad_norm": 2.086556318852155,
      "learning_rate": 1.7306658010291134e-07,
      "loss": 0.6273,
      "step": 10795
    },
    {
      "epoch": 13.214198286413708,
      "grad_norm": 2.1694867274762886,
      "learning_rate": 1.7283235304200008e-07,
      "loss": 1.0624,
      "step": 10796
    },
    {
      "epoch": 13.215422276621787,
      "grad_norm": 1.1495953584462653,
      "learning_rate": 1.725982789139391e-07,
      "loss": 0.575,
      "step": 10797
    },
    {
      "epoch": 13.216646266829866,
      "grad_norm": 1.5811994439121682,
      "learning_rate": 1.7236435773411082e-07,
      "loss": 1.0772,
      "step": 10798
    },
    {
      "epoch": 13.217870257037944,
      "grad_norm": 3.0894868780916296,
      "learning_rate": 1.7213058951788763e-07,
      "loss": 0.335,
      "step": 10799
    },
    {
      "epoch": 13.219094247246023,
      "grad_norm": 2.5962865360129244,
      "learning_rate": 1.71896974280632e-07,
      "loss": 0.5091,
      "step": 10800
    },
    {
      "epoch": 13.2203182374541,
      "grad_norm": 1.2079963450320874,
      "learning_rate": 1.7166351203769582e-07,
      "loss": 0.532,
      "step": 10801
    },
    {
      "epoch": 13.221542227662178,
      "grad_norm": 2.369377977850026,
      "learning_rate": 1.7143020280442207e-07,
      "loss": 0.4555,
      "step": 10802
    },
    {
      "epoch": 13.222766217870257,
      "grad_norm": 1.6343277011988764,
      "learning_rate": 1.711970465961421e-07,
      "loss": 0.5068,
      "step": 10803
    },
    {
      "epoch": 13.223990208078336,
      "grad_norm": 2.262014734534502,
      "learning_rate": 1.7096404342817925e-07,
      "loss": 0.5048,
      "step": 10804
    },
    {
      "epoch": 13.225214198286414,
      "grad_norm": 1.4766126463559532,
      "learning_rate": 1.7073119331584454e-07,
      "loss": 0.7934,
      "step": 10805
    },
    {
      "epoch": 13.226438188494493,
      "grad_norm": 1.6919889888114774,
      "learning_rate": 1.7049849627444076e-07,
      "loss": 1.0084,
      "step": 10806
    },
    {
      "epoch": 13.22766217870257,
      "grad_norm": 2.1710621221441975,
      "learning_rate": 1.7026595231925874e-07,
      "loss": 0.8056,
      "step": 10807
    },
    {
      "epoch": 13.228886168910648,
      "grad_norm": 1.7864428063764535,
      "learning_rate": 1.7003356146558208e-07,
      "loss": 0.4257,
      "step": 10808
    },
    {
      "epoch": 13.230110159118727,
      "grad_norm": 2.860779390789012,
      "learning_rate": 1.6980132372868135e-07,
      "loss": 0.8193,
      "step": 10809
    },
    {
      "epoch": 13.231334149326806,
      "grad_norm": 1.1441442959846486,
      "learning_rate": 1.6956923912381906e-07,
      "loss": 0.3838,
      "step": 10810
    },
    {
      "epoch": 13.232558139534884,
      "grad_norm": 1.8751782267202073,
      "learning_rate": 1.6933730766624716e-07,
      "loss": 0.9056,
      "step": 10811
    },
    {
      "epoch": 13.233782129742963,
      "grad_norm": 1.2521568035969637,
      "learning_rate": 1.69105529371206e-07,
      "loss": 0.8133,
      "step": 10812
    },
    {
      "epoch": 13.23500611995104,
      "grad_norm": 1.8123350563859886,
      "learning_rate": 1.6887390425392862e-07,
      "loss": 1.1545,
      "step": 10813
    },
    {
      "epoch": 13.236230110159118,
      "grad_norm": 2.2390308022553835,
      "learning_rate": 1.6864243232963596e-07,
      "loss": 1.1239,
      "step": 10814
    },
    {
      "epoch": 13.237454100367197,
      "grad_norm": 1.4107062863668618,
      "learning_rate": 1.684111136135394e-07,
      "loss": 0.5026,
      "step": 10815
    },
    {
      "epoch": 13.238678090575275,
      "grad_norm": 1.2846502534865445,
      "learning_rate": 1.6817994812084043e-07,
      "loss": 0.745,
      "step": 10816
    },
    {
      "epoch": 13.239902080783354,
      "grad_norm": 1.2438650669410458,
      "learning_rate": 1.6794893586673077e-07,
      "loss": 0.5796,
      "step": 10817
    },
    {
      "epoch": 13.241126070991433,
      "grad_norm": 2.6572833885915275,
      "learning_rate": 1.6771807686639107e-07,
      "loss": 0.5675,
      "step": 10818
    },
    {
      "epoch": 13.24235006119951,
      "grad_norm": 1.303866672751778,
      "learning_rate": 1.674873711349931e-07,
      "loss": 0.6422,
      "step": 10819
    },
    {
      "epoch": 13.243574051407588,
      "grad_norm": 3.220991829650694,
      "learning_rate": 1.6725681868769716e-07,
      "loss": 0.503,
      "step": 10820
    },
    {
      "epoch": 13.244798041615667,
      "grad_norm": 1.2361705384488078,
      "learning_rate": 1.6702641953965537e-07,
      "loss": 0.569,
      "step": 10821
    },
    {
      "epoch": 13.246022031823745,
      "grad_norm": 1.8394625493099772,
      "learning_rate": 1.6679617370600866e-07,
      "loss": 0.9238,
      "step": 10822
    },
    {
      "epoch": 13.247246022031824,
      "grad_norm": 1.9441899957108804,
      "learning_rate": 1.6656608120188711e-07,
      "loss": 1.2074,
      "step": 10823
    },
    {
      "epoch": 13.248470012239903,
      "grad_norm": 1.484827110309995,
      "learning_rate": 1.663361420424117e-07,
      "loss": 0.7205,
      "step": 10824
    },
    {
      "epoch": 13.249694002447981,
      "grad_norm": 2.6321178192356496,
      "learning_rate": 1.6610635624269316e-07,
      "loss": 0.5139,
      "step": 10825
    },
    {
      "epoch": 13.250917992656058,
      "grad_norm": 1.16619766981921,
      "learning_rate": 1.65876723817833e-07,
      "loss": 0.5041,
      "step": 10826
    },
    {
      "epoch": 13.252141982864137,
      "grad_norm": 2.453868361025007,
      "learning_rate": 1.6564724478292104e-07,
      "loss": 0.4116,
      "step": 10827
    },
    {
      "epoch": 13.253365973072215,
      "grad_norm": 2.097264568588874,
      "learning_rate": 1.654179191530378e-07,
      "loss": 0.2873,
      "step": 10828
    },
    {
      "epoch": 13.254589963280294,
      "grad_norm": 1.926811787043996,
      "learning_rate": 1.651887469432542e-07,
      "loss": 0.5198,
      "step": 10829
    },
    {
      "epoch": 13.255813953488373,
      "grad_norm": 2.1484456148198183,
      "learning_rate": 1.649597281686302e-07,
      "loss": 0.4348,
      "step": 10830
    },
    {
      "epoch": 13.257037943696451,
      "grad_norm": 2.460527724459208,
      "learning_rate": 1.6473086284421598e-07,
      "loss": 0.4371,
      "step": 10831
    },
    {
      "epoch": 13.258261933904528,
      "grad_norm": 1.435788568962698,
      "learning_rate": 1.6450215098505195e-07,
      "loss": 0.869,
      "step": 10832
    },
    {
      "epoch": 13.259485924112607,
      "grad_norm": 1.6495361639700936,
      "learning_rate": 1.6427359260616836e-07,
      "loss": 0.9789,
      "step": 10833
    },
    {
      "epoch": 13.260709914320685,
      "grad_norm": 1.9502321829070655,
      "learning_rate": 1.6404518772258488e-07,
      "loss": 0.4494,
      "step": 10834
    },
    {
      "epoch": 13.261933904528764,
      "grad_norm": 1.5242530374589205,
      "learning_rate": 1.638169363493114e-07,
      "loss": 1.3017,
      "step": 10835
    },
    {
      "epoch": 13.263157894736842,
      "grad_norm": 1.5662903106132338,
      "learning_rate": 1.6358883850134815e-07,
      "loss": 0.54,
      "step": 10836
    },
    {
      "epoch": 13.264381884944921,
      "grad_norm": 1.9278425193351822,
      "learning_rate": 1.6336089419368455e-07,
      "loss": 0.4834,
      "step": 10837
    },
    {
      "epoch": 13.265605875152998,
      "grad_norm": 2.9952831029568916,
      "learning_rate": 1.6313310344130002e-07,
      "loss": 0.9525,
      "step": 10838
    },
    {
      "epoch": 13.266829865361077,
      "grad_norm": 1.2754614481353925,
      "learning_rate": 1.6290546625916475e-07,
      "loss": 0.5682,
      "step": 10839
    },
    {
      "epoch": 13.268053855569155,
      "grad_norm": 1.8580386242631521,
      "learning_rate": 1.6267798266223821e-07,
      "loss": 1.1827,
      "step": 10840
    },
    {
      "epoch": 13.269277845777234,
      "grad_norm": 2.3688391700801446,
      "learning_rate": 1.6245065266546928e-07,
      "loss": 0.3197,
      "step": 10841
    },
    {
      "epoch": 13.270501835985312,
      "grad_norm": 1.1490147691059225,
      "learning_rate": 1.6222347628379708e-07,
      "loss": 0.5327,
      "step": 10842
    },
    {
      "epoch": 13.271725826193391,
      "grad_norm": 1.1454077696957563,
      "learning_rate": 1.6199645353215164e-07,
      "loss": 0.5599,
      "step": 10843
    },
    {
      "epoch": 13.27294981640147,
      "grad_norm": 0.9149384642967056,
      "learning_rate": 1.6176958442545133e-07,
      "loss": 0.3486,
      "step": 10844
    },
    {
      "epoch": 13.274173806609546,
      "grad_norm": 1.7583961126641905,
      "learning_rate": 1.6154286897860583e-07,
      "loss": 0.4895,
      "step": 10845
    },
    {
      "epoch": 13.275397796817625,
      "grad_norm": 2.4980617802245404,
      "learning_rate": 1.6131630720651327e-07,
      "loss": 0.5401,
      "step": 10846
    },
    {
      "epoch": 13.276621787025704,
      "grad_norm": 2.893561610405688,
      "learning_rate": 1.610898991240628e-07,
      "loss": 0.5546,
      "step": 10847
    },
    {
      "epoch": 13.277845777233782,
      "grad_norm": 2.0954861631110866,
      "learning_rate": 1.608636447461334e-07,
      "loss": 0.4827,
      "step": 10848
    },
    {
      "epoch": 13.279069767441861,
      "grad_norm": 1.2815612048103522,
      "learning_rate": 1.6063754408759318e-07,
      "loss": 0.6803,
      "step": 10849
    },
    {
      "epoch": 13.28029375764994,
      "grad_norm": 1.7073844131635023,
      "learning_rate": 1.604115971633008e-07,
      "loss": 0.4478,
      "step": 10850
    },
    {
      "epoch": 13.281517747858016,
      "grad_norm": 2.2804247157815505,
      "learning_rate": 1.601858039881049e-07,
      "loss": 0.532,
      "step": 10851
    },
    {
      "epoch": 13.282741738066095,
      "grad_norm": 2.349756458271896,
      "learning_rate": 1.5996016457684365e-07,
      "loss": 0.522,
      "step": 10852
    },
    {
      "epoch": 13.283965728274174,
      "grad_norm": 1.9986119536360727,
      "learning_rate": 1.597346789443449e-07,
      "loss": 0.3911,
      "step": 10853
    },
    {
      "epoch": 13.285189718482252,
      "grad_norm": 2.3793216247820506,
      "learning_rate": 1.5950934710542738e-07,
      "loss": 0.9767,
      "step": 10854
    },
    {
      "epoch": 13.286413708690331,
      "grad_norm": 2.0329442389016696,
      "learning_rate": 1.5928416907489808e-07,
      "loss": 0.5407,
      "step": 10855
    },
    {
      "epoch": 13.28763769889841,
      "grad_norm": 2.0249883014943983,
      "learning_rate": 1.5905914486755575e-07,
      "loss": 0.9887,
      "step": 10856
    },
    {
      "epoch": 13.288861689106486,
      "grad_norm": 1.5420696953007693,
      "learning_rate": 1.58834274498188e-07,
      "loss": 0.8052,
      "step": 10857
    },
    {
      "epoch": 13.290085679314565,
      "grad_norm": 2.0371489354556163,
      "learning_rate": 1.5860955798157275e-07,
      "loss": 0.9525,
      "step": 10858
    },
    {
      "epoch": 13.291309669522644,
      "grad_norm": 1.9251879668124443,
      "learning_rate": 1.5838499533247648e-07,
      "loss": 1.076,
      "step": 10859
    },
    {
      "epoch": 13.292533659730722,
      "grad_norm": 1.3714859166374462,
      "learning_rate": 1.581605865656574e-07,
      "loss": 0.5293,
      "step": 10860
    },
    {
      "epoch": 13.2937576499388,
      "grad_norm": 1.8820078626693557,
      "learning_rate": 1.5793633169586293e-07,
      "loss": 0.54,
      "step": 10861
    },
    {
      "epoch": 13.29498164014688,
      "grad_norm": 2.702883351291548,
      "learning_rate": 1.577122307378301e-07,
      "loss": 1.1724,
      "step": 10862
    },
    {
      "epoch": 13.296205630354958,
      "grad_norm": 0.7918596604935282,
      "learning_rate": 1.5748828370628578e-07,
      "loss": 0.2162,
      "step": 10863
    },
    {
      "epoch": 13.297429620563035,
      "grad_norm": 1.4395185554913117,
      "learning_rate": 1.5726449061594706e-07,
      "loss": 0.6883,
      "step": 10864
    },
    {
      "epoch": 13.298653610771114,
      "grad_norm": 1.7058986957170796,
      "learning_rate": 1.570408514815211e-07,
      "loss": 0.461,
      "step": 10865
    },
    {
      "epoch": 13.299877600979192,
      "grad_norm": 2.8504793596031384,
      "learning_rate": 1.5681736631770416e-07,
      "loss": 0.4976,
      "step": 10866
    },
    {
      "epoch": 13.30110159118727,
      "grad_norm": 2.5778118136458597,
      "learning_rate": 1.5659403513918313e-07,
      "loss": 0.4112,
      "step": 10867
    },
    {
      "epoch": 13.30232558139535,
      "grad_norm": 1.6672873047533252,
      "learning_rate": 1.5637085796063407e-07,
      "loss": 1.2398,
      "step": 10868
    },
    {
      "epoch": 13.303549571603428,
      "grad_norm": 1.3844709341649555,
      "learning_rate": 1.5614783479672436e-07,
      "loss": 1.0851,
      "step": 10869
    },
    {
      "epoch": 13.304773561811505,
      "grad_norm": 2.6574773128525058,
      "learning_rate": 1.5592496566210953e-07,
      "loss": 0.661,
      "step": 10870
    },
    {
      "epoch": 13.305997552019583,
      "grad_norm": 2.5427782694388137,
      "learning_rate": 1.5570225057143562e-07,
      "loss": 0.447,
      "step": 10871
    },
    {
      "epoch": 13.307221542227662,
      "grad_norm": 2.939229664591074,
      "learning_rate": 1.5547968953933846e-07,
      "loss": 0.8152,
      "step": 10872
    },
    {
      "epoch": 13.30844553243574,
      "grad_norm": 1.1612952353732153,
      "learning_rate": 1.5525728258044464e-07,
      "loss": 0.4919,
      "step": 10873
    },
    {
      "epoch": 13.30966952264382,
      "grad_norm": 1.7358483247606669,
      "learning_rate": 1.5503502970936972e-07,
      "loss": 0.5556,
      "step": 10874
    },
    {
      "epoch": 13.310893512851898,
      "grad_norm": 3.7617256753745183,
      "learning_rate": 1.548129309407198e-07,
      "loss": 0.4657,
      "step": 10875
    },
    {
      "epoch": 13.312117503059975,
      "grad_norm": 1.674560755268325,
      "learning_rate": 1.5459098628908931e-07,
      "loss": 1.1644,
      "step": 10876
    },
    {
      "epoch": 13.313341493268053,
      "grad_norm": 1.4391134139915083,
      "learning_rate": 1.543691957690635e-07,
      "loss": 1.5537,
      "step": 10877
    },
    {
      "epoch": 13.314565483476132,
      "grad_norm": 1.631606911415783,
      "learning_rate": 1.541475593952188e-07,
      "loss": 0.3738,
      "step": 10878
    },
    {
      "epoch": 13.31578947368421,
      "grad_norm": 2.158577088564206,
      "learning_rate": 1.5392607718211993e-07,
      "loss": 0.5895,
      "step": 10879
    },
    {
      "epoch": 13.31701346389229,
      "grad_norm": 1.5088371006107566,
      "learning_rate": 1.5370474914432166e-07,
      "loss": 1.1123,
      "step": 10880
    },
    {
      "epoch": 13.318237454100368,
      "grad_norm": 2.4734886079116154,
      "learning_rate": 1.534835752963687e-07,
      "loss": 0.4394,
      "step": 10881
    },
    {
      "epoch": 13.319461444308445,
      "grad_norm": 2.269118948809782,
      "learning_rate": 1.532625556527964e-07,
      "loss": 0.391,
      "step": 10882
    },
    {
      "epoch": 13.320685434516523,
      "grad_norm": 1.0006367676963155,
      "learning_rate": 1.530416902281287e-07,
      "loss": 0.3424,
      "step": 10883
    },
    {
      "epoch": 13.321909424724602,
      "grad_norm": 2.1837726295426343,
      "learning_rate": 1.5282097903688036e-07,
      "loss": 1.1129,
      "step": 10884
    },
    {
      "epoch": 13.32313341493268,
      "grad_norm": 1.1791248680404671,
      "learning_rate": 1.5260042209355534e-07,
      "loss": 0.563,
      "step": 10885
    },
    {
      "epoch": 13.32435740514076,
      "grad_norm": 1.9618569987731278,
      "learning_rate": 1.5238001941264847e-07,
      "loss": 0.9575,
      "step": 10886
    },
    {
      "epoch": 13.325581395348838,
      "grad_norm": 1.7291278723524608,
      "learning_rate": 1.5215977100864394e-07,
      "loss": 0.5302,
      "step": 10887
    },
    {
      "epoch": 13.326805385556916,
      "grad_norm": 1.8326640205649831,
      "learning_rate": 1.5193967689601462e-07,
      "loss": 1.0893,
      "step": 10888
    },
    {
      "epoch": 13.328029375764993,
      "grad_norm": 1.5086102203100962,
      "learning_rate": 1.517197370892251e-07,
      "loss": 0.5524,
      "step": 10889
    },
    {
      "epoch": 13.329253365973072,
      "grad_norm": 2.1180461916230415,
      "learning_rate": 1.5149995160272824e-07,
      "loss": 1.3102,
      "step": 10890
    },
    {
      "epoch": 13.33047735618115,
      "grad_norm": 2.7575883979639357,
      "learning_rate": 1.5128032045096858e-07,
      "loss": 0.5806,
      "step": 10891
    },
    {
      "epoch": 13.331701346389229,
      "grad_norm": 1.87411483411944,
      "learning_rate": 1.5106084364837875e-07,
      "loss": 0.4941,
      "step": 10892
    },
    {
      "epoch": 13.332925336597308,
      "grad_norm": 2.5273256166629086,
      "learning_rate": 1.508415212093828e-07,
      "loss": 0.4523,
      "step": 10893
    },
    {
      "epoch": 13.334149326805386,
      "grad_norm": 3.7759644214310994,
      "learning_rate": 1.506223531483922e-07,
      "loss": 0.2642,
      "step": 10894
    },
    {
      "epoch": 13.335373317013463,
      "grad_norm": 1.3263639223442347,
      "learning_rate": 1.5040333947981127e-07,
      "loss": 0.5531,
      "step": 10895
    },
    {
      "epoch": 13.336597307221542,
      "grad_norm": 1.8516890541158635,
      "learning_rate": 1.5018448021803217e-07,
      "loss": 0.9479,
      "step": 10896
    },
    {
      "epoch": 13.33782129742962,
      "grad_norm": 1.5211796951401029,
      "learning_rate": 1.499657753774378e-07,
      "loss": 0.3507,
      "step": 10897
    },
    {
      "epoch": 13.339045287637699,
      "grad_norm": 2.1092140960958656,
      "learning_rate": 1.497472249724005e-07,
      "loss": 0.7614,
      "step": 10898
    },
    {
      "epoch": 13.340269277845778,
      "grad_norm": 2.2737466794586583,
      "learning_rate": 1.495288290172825e-07,
      "loss": 0.9141,
      "step": 10899
    },
    {
      "epoch": 13.341493268053856,
      "grad_norm": 1.2471647266876746,
      "learning_rate": 1.4931058752643613e-07,
      "loss": 0.8388,
      "step": 10900
    },
    {
      "epoch": 13.342717258261933,
      "grad_norm": 2.21695813910818,
      "learning_rate": 1.4909250051420327e-07,
      "loss": 0.6029,
      "step": 10901
    },
    {
      "epoch": 13.343941248470012,
      "grad_norm": 3.151954403779326,
      "learning_rate": 1.4887456799491577e-07,
      "loss": 0.3893,
      "step": 10902
    },
    {
      "epoch": 13.34516523867809,
      "grad_norm": 2.063810627218216,
      "learning_rate": 1.486567899828953e-07,
      "loss": 0.7256,
      "step": 10903
    },
    {
      "epoch": 13.346389228886169,
      "grad_norm": 1.0871145753586322,
      "learning_rate": 1.484391664924542e-07,
      "loss": 0.5089,
      "step": 10904
    },
    {
      "epoch": 13.347613219094248,
      "grad_norm": 1.4548310648503975,
      "learning_rate": 1.4822169753789305e-07,
      "loss": 0.4309,
      "step": 10905
    },
    {
      "epoch": 13.348837209302326,
      "grad_norm": 1.464659419604835,
      "learning_rate": 1.4800438313350318e-07,
      "loss": 0.5335,
      "step": 10906
    },
    {
      "epoch": 13.350061199510403,
      "grad_norm": 1.6966142281407584,
      "learning_rate": 1.4778722329356538e-07,
      "loss": 0.3804,
      "step": 10907
    },
    {
      "epoch": 13.351285189718482,
      "grad_norm": 1.8406945873731702,
      "learning_rate": 1.4757021803235156e-07,
      "loss": 1.0965,
      "step": 10908
    },
    {
      "epoch": 13.35250917992656,
      "grad_norm": 1.7291889961421663,
      "learning_rate": 1.4735336736412198e-07,
      "loss": 0.5028,
      "step": 10909
    },
    {
      "epoch": 13.353733170134639,
      "grad_norm": 1.9655327365906838,
      "learning_rate": 1.471366713031272e-07,
      "loss": 1.0997,
      "step": 10910
    },
    {
      "epoch": 13.354957160342718,
      "grad_norm": 1.6880825842042297,
      "learning_rate": 1.4692012986360775e-07,
      "loss": 1.5524,
      "step": 10911
    },
    {
      "epoch": 13.356181150550796,
      "grad_norm": 1.5405336092491348,
      "learning_rate": 1.4670374305979397e-07,
      "loss": 0.7457,
      "step": 10912
    },
    {
      "epoch": 13.357405140758875,
      "grad_norm": 2.5163713221162776,
      "learning_rate": 1.464875109059058e-07,
      "loss": 0.4792,
      "step": 10913
    },
    {
      "epoch": 13.358629130966952,
      "grad_norm": 2.522892827791178,
      "learning_rate": 1.462714334161533e-07,
      "loss": 0.601,
      "step": 10914
    },
    {
      "epoch": 13.35985312117503,
      "grad_norm": 1.4193905928615667,
      "learning_rate": 1.4605551060473678e-07,
      "loss": 1.0845,
      "step": 10915
    },
    {
      "epoch": 13.361077111383109,
      "grad_norm": 3.5633840878752276,
      "learning_rate": 1.4583974248584516e-07,
      "loss": 0.3883,
      "step": 10916
    },
    {
      "epoch": 13.362301101591187,
      "grad_norm": 1.6037680563010934,
      "learning_rate": 1.456241290736582e-07,
      "loss": 0.9245,
      "step": 10917
    },
    {
      "epoch": 13.363525091799266,
      "grad_norm": 1.875026209907363,
      "learning_rate": 1.454086703823454e-07,
      "loss": 0.5392,
      "step": 10918
    },
    {
      "epoch": 13.364749082007345,
      "grad_norm": 1.8234691534869811,
      "learning_rate": 1.4519336642606546e-07,
      "loss": 0.6257,
      "step": 10919
    },
    {
      "epoch": 13.365973072215422,
      "grad_norm": 1.6184075088374692,
      "learning_rate": 1.449782172189676e-07,
      "loss": 0.9168,
      "step": 10920
    },
    {
      "epoch": 13.3671970624235,
      "grad_norm": 2.2621672547693352,
      "learning_rate": 1.4476322277519105e-07,
      "loss": 0.5649,
      "step": 10921
    },
    {
      "epoch": 13.368421052631579,
      "grad_norm": 2.5811845520218846,
      "learning_rate": 1.4454838310886427e-07,
      "loss": 0.4928,
      "step": 10922
    },
    {
      "epoch": 13.369645042839657,
      "grad_norm": 2.3669033587335067,
      "learning_rate": 1.443336982341051e-07,
      "loss": 0.6247,
      "step": 10923
    },
    {
      "epoch": 13.370869033047736,
      "grad_norm": 1.4067260982400014,
      "learning_rate": 1.4411916816502224e-07,
      "loss": 0.5106,
      "step": 10924
    },
    {
      "epoch": 13.372093023255815,
      "grad_norm": 1.7540784786546804,
      "learning_rate": 1.439047929157139e-07,
      "loss": 0.5134,
      "step": 10925
    },
    {
      "epoch": 13.373317013463891,
      "grad_norm": 3.6618990093218686,
      "learning_rate": 1.4369057250026824e-07,
      "loss": 0.3273,
      "step": 10926
    },
    {
      "epoch": 13.37454100367197,
      "grad_norm": 2.754121058678357,
      "learning_rate": 1.434765069327626e-07,
      "loss": 0.3953,
      "step": 10927
    },
    {
      "epoch": 13.375764993880049,
      "grad_norm": 2.1193150408409713,
      "learning_rate": 1.4326259622726462e-07,
      "loss": 0.8471,
      "step": 10928
    },
    {
      "epoch": 13.376988984088127,
      "grad_norm": 1.213814937239059,
      "learning_rate": 1.4304884039783224e-07,
      "loss": 0.879,
      "step": 10929
    },
    {
      "epoch": 13.378212974296206,
      "grad_norm": 1.8967576318743815,
      "learning_rate": 1.4283523945851197e-07,
      "loss": 1.5842,
      "step": 10930
    },
    {
      "epoch": 13.379436964504285,
      "grad_norm": 2.0255844152529336,
      "learning_rate": 1.426217934233412e-07,
      "loss": 1.3822,
      "step": 10931
    },
    {
      "epoch": 13.380660954712361,
      "grad_norm": 1.1853242545499996,
      "learning_rate": 1.4240850230634708e-07,
      "loss": 0.4133,
      "step": 10932
    },
    {
      "epoch": 13.38188494492044,
      "grad_norm": 1.6154801037507642,
      "learning_rate": 1.4219536612154556e-07,
      "loss": 0.8119,
      "step": 10933
    },
    {
      "epoch": 13.383108935128519,
      "grad_norm": 1.865362818518297,
      "learning_rate": 1.4198238488294436e-07,
      "loss": 1.1661,
      "step": 10934
    },
    {
      "epoch": 13.384332925336597,
      "grad_norm": 3.0120903672731205,
      "learning_rate": 1.4176955860453868e-07,
      "loss": 0.6491,
      "step": 10935
    },
    {
      "epoch": 13.385556915544676,
      "grad_norm": 1.736528876777888,
      "learning_rate": 1.4155688730031536e-07,
      "loss": 0.978,
      "step": 10936
    },
    {
      "epoch": 13.386780905752754,
      "grad_norm": 2.9131605047881637,
      "learning_rate": 1.4134437098424992e-07,
      "loss": 0.4614,
      "step": 10937
    },
    {
      "epoch": 13.388004895960833,
      "grad_norm": 2.9086776144812116,
      "learning_rate": 1.411320096703081e-07,
      "loss": 1.0509,
      "step": 10938
    },
    {
      "epoch": 13.38922888616891,
      "grad_norm": 1.6150453333181418,
      "learning_rate": 1.4091980337244626e-07,
      "loss": 0.6006,
      "step": 10939
    },
    {
      "epoch": 13.390452876376989,
      "grad_norm": 2.4380176437477896,
      "learning_rate": 1.4070775210460936e-07,
      "loss": 1.0704,
      "step": 10940
    },
    {
      "epoch": 13.391676866585067,
      "grad_norm": 2.4082467773789933,
      "learning_rate": 1.4049585588073233e-07,
      "loss": 0.633,
      "step": 10941
    },
    {
      "epoch": 13.392900856793146,
      "grad_norm": 2.908076008955064,
      "learning_rate": 1.402841147147402e-07,
      "loss": 0.5711,
      "step": 10942
    },
    {
      "epoch": 13.394124847001224,
      "grad_norm": 1.6047706103248838,
      "learning_rate": 1.4007252862054816e-07,
      "loss": 0.6773,
      "step": 10943
    },
    {
      "epoch": 13.395348837209303,
      "grad_norm": 3.369858722812164,
      "learning_rate": 1.3986109761206097e-07,
      "loss": 0.3685,
      "step": 10944
    },
    {
      "epoch": 13.39657282741738,
      "grad_norm": 1.9973847632431856,
      "learning_rate": 1.3964982170317303e-07,
      "loss": 0.6055,
      "step": 10945
    },
    {
      "epoch": 13.397796817625458,
      "grad_norm": 2.1590781389777547,
      "learning_rate": 1.3943870090776829e-07,
      "loss": 0.4271,
      "step": 10946
    },
    {
      "epoch": 13.399020807833537,
      "grad_norm": 2.7149098844063673,
      "learning_rate": 1.3922773523972115e-07,
      "loss": 0.5009,
      "step": 10947
    },
    {
      "epoch": 13.400244798041616,
      "grad_norm": 1.6521008980740193,
      "learning_rate": 1.39016924712895e-07,
      "loss": 1.061,
      "step": 10948
    },
    {
      "epoch": 13.401468788249694,
      "grad_norm": 2.03225120288025,
      "learning_rate": 1.3880626934114432e-07,
      "loss": 0.7209,
      "step": 10949
    },
    {
      "epoch": 13.402692778457773,
      "grad_norm": 1.2077853607205484,
      "learning_rate": 1.3859576913831192e-07,
      "loss": 0.5577,
      "step": 10950
    },
    {
      "epoch": 13.403916768665852,
      "grad_norm": 1.3739859444511033,
      "learning_rate": 1.383854241182314e-07,
      "loss": 0.7148,
      "step": 10951
    },
    {
      "epoch": 13.405140758873928,
      "grad_norm": 3.3987758441834894,
      "learning_rate": 1.3817523429472597e-07,
      "loss": 0.5841,
      "step": 10952
    },
    {
      "epoch": 13.406364749082007,
      "grad_norm": 1.4687778066785078,
      "learning_rate": 1.3796519968160809e-07,
      "loss": 0.3613,
      "step": 10953
    },
    {
      "epoch": 13.407588739290086,
      "grad_norm": 2.5539251738271247,
      "learning_rate": 1.3775532029268096e-07,
      "loss": 0.5947,
      "step": 10954
    },
    {
      "epoch": 13.408812729498164,
      "grad_norm": 1.2102666829624382,
      "learning_rate": 1.375455961417363e-07,
      "loss": 0.4957,
      "step": 10955
    },
    {
      "epoch": 13.410036719706243,
      "grad_norm": 1.2957113073653597,
      "learning_rate": 1.3733602724255723e-07,
      "loss": 0.9811,
      "step": 10956
    },
    {
      "epoch": 13.411260709914322,
      "grad_norm": 1.3401143634531203,
      "learning_rate": 1.3712661360891582e-07,
      "loss": 0.7132,
      "step": 10957
    },
    {
      "epoch": 13.412484700122398,
      "grad_norm": 2.4748375246819325,
      "learning_rate": 1.3691735525457384e-07,
      "loss": 1.193,
      "step": 10958
    },
    {
      "epoch": 13.413708690330477,
      "grad_norm": 1.5736608449679688,
      "learning_rate": 1.3670825219328222e-07,
      "loss": 0.5685,
      "step": 10959
    },
    {
      "epoch": 13.414932680538556,
      "grad_norm": 1.5096504213822937,
      "learning_rate": 1.364993044387833e-07,
      "loss": 0.5871,
      "step": 10960
    },
    {
      "epoch": 13.416156670746634,
      "grad_norm": 2.5570932917277602,
      "learning_rate": 1.362905120048083e-07,
      "loss": 0.4789,
      "step": 10961
    },
    {
      "epoch": 13.417380660954713,
      "grad_norm": 1.084047464143069,
      "learning_rate": 1.3608187490507795e-07,
      "loss": 0.6254,
      "step": 10962
    },
    {
      "epoch": 13.418604651162791,
      "grad_norm": 2.7767564885696947,
      "learning_rate": 1.358733931533035e-07,
      "loss": 0.3909,
      "step": 10963
    },
    {
      "epoch": 13.419828641370868,
      "grad_norm": 1.6922774145086323,
      "learning_rate": 1.356650667631851e-07,
      "loss": 1.5492,
      "step": 10964
    },
    {
      "epoch": 13.421052631578947,
      "grad_norm": 2.149805042951668,
      "learning_rate": 1.3545689574841341e-07,
      "loss": 0.6095,
      "step": 10965
    },
    {
      "epoch": 13.422276621787026,
      "grad_norm": 1.8493065806451887,
      "learning_rate": 1.3524888012266867e-07,
      "loss": 0.2891,
      "step": 10966
    },
    {
      "epoch": 13.423500611995104,
      "grad_norm": 1.9137664744956175,
      "learning_rate": 1.3504101989962132e-07,
      "loss": 0.7388,
      "step": 10967
    },
    {
      "epoch": 13.424724602203183,
      "grad_norm": 3.179601886558288,
      "learning_rate": 1.3483331509292984e-07,
      "loss": 0.6886,
      "step": 10968
    },
    {
      "epoch": 13.425948592411261,
      "grad_norm": 2.465401736900861,
      "learning_rate": 1.3462576571624587e-07,
      "loss": 0.891,
      "step": 10969
    },
    {
      "epoch": 13.427172582619338,
      "grad_norm": 1.3553563169147569,
      "learning_rate": 1.3441837178320732e-07,
      "loss": 0.5795,
      "step": 10970
    },
    {
      "epoch": 13.428396572827417,
      "grad_norm": 1.9522204248379809,
      "learning_rate": 1.3421113330744333e-07,
      "loss": 0.9957,
      "step": 10971
    },
    {
      "epoch": 13.429620563035495,
      "grad_norm": 1.6244637135657432,
      "learning_rate": 1.340040503025733e-07,
      "loss": 0.7409,
      "step": 10972
    },
    {
      "epoch": 13.430844553243574,
      "grad_norm": 1.6761914402883717,
      "learning_rate": 1.3379712278220574e-07,
      "loss": 0.853,
      "step": 10973
    },
    {
      "epoch": 13.432068543451653,
      "grad_norm": 1.9013548062027816,
      "learning_rate": 1.3359035075993982e-07,
      "loss": 0.4658,
      "step": 10974
    },
    {
      "epoch": 13.433292533659731,
      "grad_norm": 1.5524312175296473,
      "learning_rate": 1.3338373424936297e-07,
      "loss": 1.3961,
      "step": 10975
    },
    {
      "epoch": 13.43451652386781,
      "grad_norm": 2.7426981605857397,
      "learning_rate": 1.3317727326405406e-07,
      "loss": 0.3777,
      "step": 10976
    },
    {
      "epoch": 13.435740514075887,
      "grad_norm": 1.7748861384118169,
      "learning_rate": 1.3297096781757973e-07,
      "loss": 0.4951,
      "step": 10977
    },
    {
      "epoch": 13.436964504283965,
      "grad_norm": 2.008637090290666,
      "learning_rate": 1.327648179234986e-07,
      "loss": 0.7394,
      "step": 10978
    },
    {
      "epoch": 13.438188494492044,
      "grad_norm": 1.7388775520228492,
      "learning_rate": 1.3255882359535787e-07,
      "loss": 0.9068,
      "step": 10979
    },
    {
      "epoch": 13.439412484700123,
      "grad_norm": 1.8033364089340687,
      "learning_rate": 1.3235298484669478e-07,
      "loss": 0.671,
      "step": 10980
    },
    {
      "epoch": 13.440636474908201,
      "grad_norm": 2.5626874032664584,
      "learning_rate": 1.321473016910363e-07,
      "loss": 0.5301,
      "step": 10981
    },
    {
      "epoch": 13.44186046511628,
      "grad_norm": 1.7587953276228914,
      "learning_rate": 1.3194177414189907e-07,
      "loss": 1.1681,
      "step": 10982
    },
    {
      "epoch": 13.443084455324357,
      "grad_norm": 1.9339332360326953,
      "learning_rate": 1.3173640221278954e-07,
      "loss": 0.4995,
      "step": 10983
    },
    {
      "epoch": 13.444308445532435,
      "grad_norm": 1.3999172731370204,
      "learning_rate": 1.3153118591720388e-07,
      "loss": 0.3847,
      "step": 10984
    },
    {
      "epoch": 13.445532435740514,
      "grad_norm": 1.3252218332946255,
      "learning_rate": 1.313261252686282e-07,
      "loss": 0.6528,
      "step": 10985
    },
    {
      "epoch": 13.446756425948593,
      "grad_norm": 2.6754146586997156,
      "learning_rate": 1.311212202805387e-07,
      "loss": 0.4229,
      "step": 10986
    },
    {
      "epoch": 13.447980416156671,
      "grad_norm": 2.340049657431051,
      "learning_rate": 1.3091647096640127e-07,
      "loss": 0.6118,
      "step": 10987
    },
    {
      "epoch": 13.44920440636475,
      "grad_norm": 1.8830703203020929,
      "learning_rate": 1.307118773396701e-07,
      "loss": 0.5788,
      "step": 10988
    },
    {
      "epoch": 13.450428396572827,
      "grad_norm": 1.6498102203809497,
      "learning_rate": 1.3050743941379092e-07,
      "loss": 1.0375,
      "step": 10989
    },
    {
      "epoch": 13.451652386780905,
      "grad_norm": 2.1805949127854976,
      "learning_rate": 1.3030315720219871e-07,
      "loss": 0.7075,
      "step": 10990
    },
    {
      "epoch": 13.452876376988984,
      "grad_norm": 1.4148859998193026,
      "learning_rate": 1.300990307183181e-07,
      "loss": 0.3053,
      "step": 10991
    },
    {
      "epoch": 13.454100367197062,
      "grad_norm": 2.18311957944349,
      "learning_rate": 1.2989505997556357e-07,
      "loss": 0.5395,
      "step": 10992
    },
    {
      "epoch": 13.455324357405141,
      "grad_norm": 2.1356930153170657,
      "learning_rate": 1.296912449873397e-07,
      "loss": 0.5024,
      "step": 10993
    },
    {
      "epoch": 13.45654834761322,
      "grad_norm": 1.1693039927819981,
      "learning_rate": 1.294875857670394e-07,
      "loss": 0.6062,
      "step": 10994
    },
    {
      "epoch": 13.457772337821297,
      "grad_norm": 1.849100880946639,
      "learning_rate": 1.292840823280475e-07,
      "loss": 1.356,
      "step": 10995
    },
    {
      "epoch": 13.458996328029375,
      "grad_norm": 2.932136399962728,
      "learning_rate": 1.2908073468373694e-07,
      "loss": 0.3555,
      "step": 10996
    },
    {
      "epoch": 13.460220318237454,
      "grad_norm": 2.46997952055088,
      "learning_rate": 1.288775428474709e-07,
      "loss": 0.5683,
      "step": 10997
    },
    {
      "epoch": 13.461444308445532,
      "grad_norm": 1.6369068877774033,
      "learning_rate": 1.2867450683260285e-07,
      "loss": 1.5381,
      "step": 10998
    },
    {
      "epoch": 13.462668298653611,
      "grad_norm": 1.7107786616999563,
      "learning_rate": 1.284716266524749e-07,
      "loss": 0.5412,
      "step": 10999
    },
    {
      "epoch": 13.46389228886169,
      "grad_norm": 1.5109899691383375,
      "learning_rate": 1.2826890232042032e-07,
      "loss": 0.5252,
      "step": 11000
    },
    {
      "epoch": 13.465116279069768,
      "grad_norm": 2.001823404510624,
      "learning_rate": 1.2806633384976092e-07,
      "loss": 0.5168,
      "step": 11001
    },
    {
      "epoch": 13.466340269277845,
      "grad_norm": 2.5199662994255627,
      "learning_rate": 1.278639212538091e-07,
      "loss": 0.8541,
      "step": 11002
    },
    {
      "epoch": 13.467564259485924,
      "grad_norm": 2.1984264397760884,
      "learning_rate": 1.2766166454586588e-07,
      "loss": 0.9633,
      "step": 11003
    },
    {
      "epoch": 13.468788249694002,
      "grad_norm": 2.469468730934486,
      "learning_rate": 1.2745956373922369e-07,
      "loss": 0.4635,
      "step": 11004
    },
    {
      "epoch": 13.470012239902081,
      "grad_norm": 1.285610803026136,
      "learning_rate": 1.2725761884716387e-07,
      "loss": 0.4681,
      "step": 11005
    },
    {
      "epoch": 13.47123623011016,
      "grad_norm": 1.6630595811907822,
      "learning_rate": 1.2705582988295716e-07,
      "loss": 0.4937,
      "step": 11006
    },
    {
      "epoch": 13.472460220318238,
      "grad_norm": 1.2646272828577558,
      "learning_rate": 1.2685419685986383e-07,
      "loss": 0.4986,
      "step": 11007
    },
    {
      "epoch": 13.473684210526315,
      "grad_norm": 3.1921743450758253,
      "learning_rate": 1.2665271979113518e-07,
      "loss": 0.3654,
      "step": 11008
    },
    {
      "epoch": 13.474908200734394,
      "grad_norm": 1.1997157101357585,
      "learning_rate": 1.2645139869001177e-07,
      "loss": 0.5068,
      "step": 11009
    },
    {
      "epoch": 13.476132190942472,
      "grad_norm": 2.1227874106563256,
      "learning_rate": 1.26250233569723e-07,
      "loss": 0.5395,
      "step": 11010
    },
    {
      "epoch": 13.477356181150551,
      "grad_norm": 1.6879048542159112,
      "learning_rate": 1.260492244434891e-07,
      "loss": 0.4679,
      "step": 11011
    },
    {
      "epoch": 13.47858017135863,
      "grad_norm": 1.079361865176921,
      "learning_rate": 1.2584837132451955e-07,
      "loss": 0.3946,
      "step": 11012
    },
    {
      "epoch": 13.479804161566708,
      "grad_norm": 2.1668191513827812,
      "learning_rate": 1.2564767422601348e-07,
      "loss": 0.4217,
      "step": 11013
    },
    {
      "epoch": 13.481028151774785,
      "grad_norm": 1.3442914102829693,
      "learning_rate": 1.2544713316116036e-07,
      "loss": 0.4737,
      "step": 11014
    },
    {
      "epoch": 13.482252141982864,
      "grad_norm": 2.2813284660389415,
      "learning_rate": 1.2524674814313853e-07,
      "loss": 0.572,
      "step": 11015
    },
    {
      "epoch": 13.483476132190942,
      "grad_norm": 3.424324989436231,
      "learning_rate": 1.2504651918511717e-07,
      "loss": 0.3367,
      "step": 11016
    },
    {
      "epoch": 13.48470012239902,
      "grad_norm": 1.997653441280003,
      "learning_rate": 1.2484644630025384e-07,
      "loss": 0.4531,
      "step": 11017
    },
    {
      "epoch": 13.4859241126071,
      "grad_norm": 1.3849121262170272,
      "learning_rate": 1.2464652950169714e-07,
      "loss": 0.7071,
      "step": 11018
    },
    {
      "epoch": 13.487148102815178,
      "grad_norm": 2.3506183969178367,
      "learning_rate": 1.2444676880258493e-07,
      "loss": 0.5731,
      "step": 11019
    },
    {
      "epoch": 13.488372093023255,
      "grad_norm": 3.1537248051712954,
      "learning_rate": 1.2424716421604393e-07,
      "loss": 0.5648,
      "step": 11020
    },
    {
      "epoch": 13.489596083231334,
      "grad_norm": 2.665260513011284,
      "learning_rate": 1.2404771575519226e-07,
      "loss": 0.9418,
      "step": 11021
    },
    {
      "epoch": 13.490820073439412,
      "grad_norm": 2.0015986711530944,
      "learning_rate": 1.238484234331372e-07,
      "loss": 0.452,
      "step": 11022
    },
    {
      "epoch": 13.49204406364749,
      "grad_norm": 1.479618104621981,
      "learning_rate": 1.2364928726297465e-07,
      "loss": 0.4257,
      "step": 11023
    },
    {
      "epoch": 13.49326805385557,
      "grad_norm": 1.6699964202185855,
      "learning_rate": 1.2345030725779084e-07,
      "loss": 0.6192,
      "step": 11024
    },
    {
      "epoch": 13.494492044063648,
      "grad_norm": 2.0540829431355325,
      "learning_rate": 1.2325148343066308e-07,
      "loss": 0.487,
      "step": 11025
    },
    {
      "epoch": 13.495716034271727,
      "grad_norm": 1.5949102498545895,
      "learning_rate": 1.23052815794657e-07,
      "loss": 0.5189,
      "step": 11026
    },
    {
      "epoch": 13.496940024479803,
      "grad_norm": 1.3786093163391744,
      "learning_rate": 1.2285430436282803e-07,
      "loss": 0.3903,
      "step": 11027
    },
    {
      "epoch": 13.498164014687882,
      "grad_norm": 1.6371275718321252,
      "learning_rate": 1.2265594914822154e-07,
      "loss": 0.9306,
      "step": 11028
    },
    {
      "epoch": 13.49938800489596,
      "grad_norm": 1.0881442193893551,
      "learning_rate": 1.2245775016387297e-07,
      "loss": 0.5912,
      "step": 11029
    },
    {
      "epoch": 13.50061199510404,
      "grad_norm": 1.165154814360197,
      "learning_rate": 1.222597074228074e-07,
      "loss": 0.4669,
      "step": 11030
    },
    {
      "epoch": 13.501835985312118,
      "grad_norm": 1.8679464607479288,
      "learning_rate": 1.2206182093803892e-07,
      "loss": 0.5605,
      "step": 11031
    },
    {
      "epoch": 13.503059975520197,
      "grad_norm": 1.1497059790945496,
      "learning_rate": 1.218640907225721e-07,
      "loss": 0.3052,
      "step": 11032
    },
    {
      "epoch": 13.504283965728273,
      "grad_norm": 1.202216139200076,
      "learning_rate": 1.21666516789401e-07,
      "loss": 0.6899,
      "step": 11033
    },
    {
      "epoch": 13.505507955936352,
      "grad_norm": 2.137118527357587,
      "learning_rate": 1.2146909915151022e-07,
      "loss": 0.399,
      "step": 11034
    },
    {
      "epoch": 13.50673194614443,
      "grad_norm": 1.3712583578475774,
      "learning_rate": 1.2127183782187246e-07,
      "loss": 0.5476,
      "step": 11035
    },
    {
      "epoch": 13.50795593635251,
      "grad_norm": 1.8853655876978586,
      "learning_rate": 1.2107473281345095e-07,
      "loss": 0.9794,
      "step": 11036
    },
    {
      "epoch": 13.509179926560588,
      "grad_norm": 3.211805267485466,
      "learning_rate": 1.2087778413919866e-07,
      "loss": 0.2947,
      "step": 11037
    },
    {
      "epoch": 13.510403916768666,
      "grad_norm": 1.7578810763873924,
      "learning_rate": 1.206809918120591e-07,
      "loss": 0.3663,
      "step": 11038
    },
    {
      "epoch": 13.511627906976745,
      "grad_norm": 2.9776727394898277,
      "learning_rate": 1.204843558449642e-07,
      "loss": 0.3993,
      "step": 11039
    },
    {
      "epoch": 13.512851897184822,
      "grad_norm": 1.1884347813775973,
      "learning_rate": 1.2028787625083666e-07,
      "loss": 0.4433,
      "step": 11040
    },
    {
      "epoch": 13.5140758873929,
      "grad_norm": 2.5140022197185536,
      "learning_rate": 1.2009155304258753e-07,
      "loss": 0.9646,
      "step": 11041
    },
    {
      "epoch": 13.51529987760098,
      "grad_norm": 2.3445721046948944,
      "learning_rate": 1.1989538623311842e-07,
      "loss": 0.3651,
      "step": 11042
    },
    {
      "epoch": 13.516523867809058,
      "grad_norm": 2.4037120788572968,
      "learning_rate": 1.1969937583532153e-07,
      "loss": 0.4214,
      "step": 11043
    },
    {
      "epoch": 13.517747858017136,
      "grad_norm": 1.7271730059638564,
      "learning_rate": 1.195035218620777e-07,
      "loss": 1.6187,
      "step": 11044
    },
    {
      "epoch": 13.518971848225215,
      "grad_norm": 1.6932436968850926,
      "learning_rate": 1.1930782432625742e-07,
      "loss": 0.3478,
      "step": 11045
    },
    {
      "epoch": 13.520195838433292,
      "grad_norm": 2.4314593590689135,
      "learning_rate": 1.1911228324072127e-07,
      "loss": 0.6619,
      "step": 11046
    },
    {
      "epoch": 13.52141982864137,
      "grad_norm": 2.9731999809742207,
      "learning_rate": 1.1891689861831979e-07,
      "loss": 0.6782,
      "step": 11047
    },
    {
      "epoch": 13.522643818849449,
      "grad_norm": 2.5255570479128338,
      "learning_rate": 1.1872167047189242e-07,
      "loss": 0.3868,
      "step": 11048
    },
    {
      "epoch": 13.523867809057528,
      "grad_norm": 1.6624506975068105,
      "learning_rate": 1.1852659881426947e-07,
      "loss": 0.5726,
      "step": 11049
    },
    {
      "epoch": 13.525091799265606,
      "grad_norm": 2.531557536686078,
      "learning_rate": 1.183316836582693e-07,
      "loss": 0.4944,
      "step": 11050
    },
    {
      "epoch": 13.526315789473685,
      "grad_norm": 3.151626575717831,
      "learning_rate": 1.1813692501670276e-07,
      "loss": 0.4373,
      "step": 11051
    },
    {
      "epoch": 13.527539779681762,
      "grad_norm": 1.0608006595112094,
      "learning_rate": 1.1794232290236685e-07,
      "loss": 0.4395,
      "step": 11052
    },
    {
      "epoch": 13.52876376988984,
      "grad_norm": 1.8586674311708151,
      "learning_rate": 1.1774787732805104e-07,
      "loss": 0.703,
      "step": 11053
    },
    {
      "epoch": 13.529987760097919,
      "grad_norm": 1.837849111109393,
      "learning_rate": 1.1755358830653346e-07,
      "loss": 0.4022,
      "step": 11054
    },
    {
      "epoch": 13.531211750305998,
      "grad_norm": 1.4649750724375565,
      "learning_rate": 1.1735945585058167e-07,
      "loss": 1.1249,
      "step": 11055
    },
    {
      "epoch": 13.532435740514076,
      "grad_norm": 2.9287670869870612,
      "learning_rate": 1.1716547997295408e-07,
      "loss": 0.7908,
      "step": 11056
    },
    {
      "epoch": 13.533659730722155,
      "grad_norm": 2.0548497291930476,
      "learning_rate": 1.1697166068639743e-07,
      "loss": 0.92,
      "step": 11057
    },
    {
      "epoch": 13.534883720930232,
      "grad_norm": 1.3188057107502962,
      "learning_rate": 1.1677799800364958e-07,
      "loss": 0.4361,
      "step": 11058
    },
    {
      "epoch": 13.53610771113831,
      "grad_norm": 2.558051427838892,
      "learning_rate": 1.1658449193743593e-07,
      "loss": 0.9634,
      "step": 11059
    },
    {
      "epoch": 13.537331701346389,
      "grad_norm": 2.0189747947917303,
      "learning_rate": 1.1639114250047406e-07,
      "loss": 0.4092,
      "step": 11060
    },
    {
      "epoch": 13.538555691554468,
      "grad_norm": 2.071330675261987,
      "learning_rate": 1.1619794970547022e-07,
      "loss": 0.6874,
      "step": 11061
    },
    {
      "epoch": 13.539779681762546,
      "grad_norm": 2.4573087528509485,
      "learning_rate": 1.1600491356511978e-07,
      "loss": 0.5856,
      "step": 11062
    },
    {
      "epoch": 13.541003671970625,
      "grad_norm": 1.6084071741882107,
      "learning_rate": 1.1581203409210873e-07,
      "loss": 0.218,
      "step": 11063
    },
    {
      "epoch": 13.542227662178703,
      "grad_norm": 1.3979429688890508,
      "learning_rate": 1.1561931129911247e-07,
      "loss": 1.3268,
      "step": 11064
    },
    {
      "epoch": 13.54345165238678,
      "grad_norm": 1.9781212357335967,
      "learning_rate": 1.1542674519879587e-07,
      "loss": 0.4182,
      "step": 11065
    },
    {
      "epoch": 13.544675642594859,
      "grad_norm": 2.029963936984083,
      "learning_rate": 1.1523433580381327e-07,
      "loss": 0.4343,
      "step": 11066
    },
    {
      "epoch": 13.545899632802938,
      "grad_norm": 1.093803055772265,
      "learning_rate": 1.1504208312680981e-07,
      "loss": 0.5147,
      "step": 11067
    },
    {
      "epoch": 13.547123623011016,
      "grad_norm": 2.147833891602053,
      "learning_rate": 1.1484998718041902e-07,
      "loss": 0.4005,
      "step": 11068
    },
    {
      "epoch": 13.548347613219095,
      "grad_norm": 1.9165143638009656,
      "learning_rate": 1.1465804797726549e-07,
      "loss": 0.4469,
      "step": 11069
    },
    {
      "epoch": 13.549571603427173,
      "grad_norm": 1.1693761687961362,
      "learning_rate": 1.1446626552996192e-07,
      "loss": 0.627,
      "step": 11070
    },
    {
      "epoch": 13.55079559363525,
      "grad_norm": 1.7246288790326252,
      "learning_rate": 1.1427463985111187e-07,
      "loss": 1.0767,
      "step": 11071
    },
    {
      "epoch": 13.552019583843329,
      "grad_norm": 1.9761759469607916,
      "learning_rate": 1.1408317095330829e-07,
      "loss": 0.5775,
      "step": 11072
    },
    {
      "epoch": 13.553243574051407,
      "grad_norm": 2.705594610690152,
      "learning_rate": 1.1389185884913389e-07,
      "loss": 0.5052,
      "step": 11073
    },
    {
      "epoch": 13.554467564259486,
      "grad_norm": 1.607683081303067,
      "learning_rate": 1.1370070355116086e-07,
      "loss": 0.503,
      "step": 11074
    },
    {
      "epoch": 13.555691554467565,
      "grad_norm": 2.006463631450044,
      "learning_rate": 1.1350970507195136e-07,
      "loss": 0.4755,
      "step": 11075
    },
    {
      "epoch": 13.556915544675643,
      "grad_norm": 2.390755351967212,
      "learning_rate": 1.133188634240573e-07,
      "loss": 0.4425,
      "step": 11076
    },
    {
      "epoch": 13.55813953488372,
      "grad_norm": 1.7881894060918522,
      "learning_rate": 1.1312817862001945e-07,
      "loss": 0.7248,
      "step": 11077
    },
    {
      "epoch": 13.559363525091799,
      "grad_norm": 1.2815362875889305,
      "learning_rate": 1.129376506723695e-07,
      "loss": 0.466,
      "step": 11078
    },
    {
      "epoch": 13.560587515299877,
      "grad_norm": 0.9449306285328828,
      "learning_rate": 1.1274727959362769e-07,
      "loss": 0.5193,
      "step": 11079
    },
    {
      "epoch": 13.561811505507956,
      "grad_norm": 3.294419479735996,
      "learning_rate": 1.1255706539630512e-07,
      "loss": 0.3015,
      "step": 11080
    },
    {
      "epoch": 13.563035495716035,
      "grad_norm": 1.8752505239388875,
      "learning_rate": 1.1236700809290152e-07,
      "loss": 0.4418,
      "step": 11081
    },
    {
      "epoch": 13.564259485924113,
      "grad_norm": 1.6532702985254657,
      "learning_rate": 1.1217710769590717e-07,
      "loss": 0.564,
      "step": 11082
    },
    {
      "epoch": 13.56548347613219,
      "grad_norm": 1.8968282654574655,
      "learning_rate": 1.1198736421780099e-07,
      "loss": 1.3756,
      "step": 11083
    },
    {
      "epoch": 13.566707466340269,
      "grad_norm": 2.097407038122065,
      "learning_rate": 1.1179777767105299e-07,
      "loss": 0.9567,
      "step": 11084
    },
    {
      "epoch": 13.567931456548347,
      "grad_norm": 1.9037381185776505,
      "learning_rate": 1.1160834806812099e-07,
      "loss": 0.539,
      "step": 11085
    },
    {
      "epoch": 13.569155446756426,
      "grad_norm": 0.9084982542741277,
      "learning_rate": 1.1141907542145475e-07,
      "loss": 0.3457,
      "step": 11086
    },
    {
      "epoch": 13.570379436964505,
      "grad_norm": 1.1643992573023785,
      "learning_rate": 1.1122995974349265e-07,
      "loss": 0.5658,
      "step": 11087
    },
    {
      "epoch": 13.571603427172583,
      "grad_norm": 1.7943913226373358,
      "learning_rate": 1.1104100104666171e-07,
      "loss": 1.4942,
      "step": 11088
    },
    {
      "epoch": 13.572827417380662,
      "grad_norm": 2.0756608679664637,
      "learning_rate": 1.1085219934337948e-07,
      "loss": 0.5258,
      "step": 11089
    },
    {
      "epoch": 13.574051407588739,
      "grad_norm": 1.3986422984943154,
      "learning_rate": 1.1066355464605438e-07,
      "loss": 0.4586,
      "step": 11090
    },
    {
      "epoch": 13.575275397796817,
      "grad_norm": 2.1326258864961214,
      "learning_rate": 1.1047506696708288e-07,
      "loss": 0.4304,
      "step": 11091
    },
    {
      "epoch": 13.576499388004896,
      "grad_norm": 1.1006660014263576,
      "learning_rate": 1.1028673631885173e-07,
      "loss": 0.5262,
      "step": 11092
    },
    {
      "epoch": 13.577723378212974,
      "grad_norm": 1.5763700166583237,
      "learning_rate": 1.1009856271373742e-07,
      "loss": 0.646,
      "step": 11093
    },
    {
      "epoch": 13.578947368421053,
      "grad_norm": 2.790531541988042,
      "learning_rate": 1.099105461641059e-07,
      "loss": 0.4263,
      "step": 11094
    },
    {
      "epoch": 13.580171358629132,
      "grad_norm": 2.753480839464336,
      "learning_rate": 1.0972268668231284e-07,
      "loss": 0.9231,
      "step": 11095
    },
    {
      "epoch": 13.581395348837209,
      "grad_norm": 3.950430962359351,
      "learning_rate": 1.0953498428070392e-07,
      "loss": 0.3441,
      "step": 11096
    },
    {
      "epoch": 13.582619339045287,
      "grad_norm": 2.058694348622937,
      "learning_rate": 1.0934743897161426e-07,
      "loss": 1.3044,
      "step": 11097
    },
    {
      "epoch": 13.583843329253366,
      "grad_norm": 2.428594556390859,
      "learning_rate": 1.0916005076736819e-07,
      "loss": 1.0078,
      "step": 11098
    },
    {
      "epoch": 13.585067319461444,
      "grad_norm": 1.6106987869284564,
      "learning_rate": 1.0897281968028056e-07,
      "loss": 0.4659,
      "step": 11099
    },
    {
      "epoch": 13.586291309669523,
      "grad_norm": 2.6258494937218058,
      "learning_rate": 1.0878574572265571e-07,
      "loss": 0.5985,
      "step": 11100
    },
    {
      "epoch": 13.587515299877602,
      "grad_norm": 2.1557763303471518,
      "learning_rate": 1.0859882890678686e-07,
      "loss": 0.3416,
      "step": 11101
    },
    {
      "epoch": 13.588739290085678,
      "grad_norm": 2.3803155656307817,
      "learning_rate": 1.0841206924495779e-07,
      "loss": 0.3663,
      "step": 11102
    },
    {
      "epoch": 13.589963280293757,
      "grad_norm": 2.755945707753369,
      "learning_rate": 1.0822546674944145e-07,
      "loss": 0.4541,
      "step": 11103
    },
    {
      "epoch": 13.591187270501836,
      "grad_norm": 1.1476632095226418,
      "learning_rate": 1.0803902143250111e-07,
      "loss": 0.5341,
      "step": 11104
    },
    {
      "epoch": 13.592411260709914,
      "grad_norm": 2.362687548023299,
      "learning_rate": 1.0785273330638946e-07,
      "loss": 1.4002,
      "step": 11105
    },
    {
      "epoch": 13.593635250917993,
      "grad_norm": 1.5413221937978268,
      "learning_rate": 1.0766660238334781e-07,
      "loss": 0.8849,
      "step": 11106
    },
    {
      "epoch": 13.594859241126072,
      "grad_norm": 2.3500682905182484,
      "learning_rate": 1.0748062867560805e-07,
      "loss": 0.8105,
      "step": 11107
    },
    {
      "epoch": 13.596083231334148,
      "grad_norm": 2.127328205160727,
      "learning_rate": 1.0729481219539262e-07,
      "loss": 1.0643,
      "step": 11108
    },
    {
      "epoch": 13.597307221542227,
      "grad_norm": 1.6994023515085155,
      "learning_rate": 1.0710915295491176e-07,
      "loss": 0.503,
      "step": 11109
    },
    {
      "epoch": 13.598531211750306,
      "grad_norm": 2.2334491392057245,
      "learning_rate": 1.0692365096636709e-07,
      "loss": 0.3686,
      "step": 11110
    },
    {
      "epoch": 13.599755201958384,
      "grad_norm": 2.699523270135167,
      "learning_rate": 1.0673830624194832e-07,
      "loss": 0.351,
      "step": 11111
    },
    {
      "epoch": 13.600979192166463,
      "grad_norm": 2.1374666312408603,
      "learning_rate": 1.0655311879383623e-07,
      "loss": 1.1458,
      "step": 11112
    },
    {
      "epoch": 13.602203182374542,
      "grad_norm": 1.785613951831227,
      "learning_rate": 1.0636808863420056e-07,
      "loss": 1.3939,
      "step": 11113
    },
    {
      "epoch": 13.60342717258262,
      "grad_norm": 2.2909132928278826,
      "learning_rate": 1.0618321577520047e-07,
      "loss": 1.0088,
      "step": 11114
    },
    {
      "epoch": 13.604651162790697,
      "grad_norm": 2.394710877948436,
      "learning_rate": 1.0599850022898539e-07,
      "loss": 0.5987,
      "step": 11115
    },
    {
      "epoch": 13.605875152998776,
      "grad_norm": 1.5469692003859765,
      "learning_rate": 1.0581394200769396e-07,
      "loss": 1.2238,
      "step": 11116
    },
    {
      "epoch": 13.607099143206854,
      "grad_norm": 3.03760536615815,
      "learning_rate": 1.056295411234548e-07,
      "loss": 0.4128,
      "step": 11117
    },
    {
      "epoch": 13.608323133414933,
      "grad_norm": 1.5814033986960192,
      "learning_rate": 1.05445297588386e-07,
      "loss": 0.6194,
      "step": 11118
    },
    {
      "epoch": 13.609547123623011,
      "grad_norm": 2.269585357125046,
      "learning_rate": 1.0526121141459567e-07,
      "loss": 0.6671,
      "step": 11119
    },
    {
      "epoch": 13.61077111383109,
      "grad_norm": 1.430924149470905,
      "learning_rate": 1.0507728261418049e-07,
      "loss": 0.6902,
      "step": 11120
    },
    {
      "epoch": 13.611995104039167,
      "grad_norm": 1.5466580541574284,
      "learning_rate": 1.048935111992283e-07,
      "loss": 0.9781,
      "step": 11121
    },
    {
      "epoch": 13.613219094247246,
      "grad_norm": 2.656740758652461,
      "learning_rate": 1.0470989718181585e-07,
      "loss": 0.8443,
      "step": 11122
    },
    {
      "epoch": 13.614443084455324,
      "grad_norm": 1.1299804642563789,
      "learning_rate": 1.0452644057400957e-07,
      "loss": 0.6768,
      "step": 11123
    },
    {
      "epoch": 13.615667074663403,
      "grad_norm": 1.5897359392885195,
      "learning_rate": 1.0434314138786484e-07,
      "loss": 0.4868,
      "step": 11124
    },
    {
      "epoch": 13.616891064871481,
      "grad_norm": 1.506412187285538,
      "learning_rate": 1.0415999963542811e-07,
      "loss": 1.0196,
      "step": 11125
    },
    {
      "epoch": 13.61811505507956,
      "grad_norm": 1.6920213796950074,
      "learning_rate": 1.0397701532873478e-07,
      "loss": 1.6318,
      "step": 11126
    },
    {
      "epoch": 13.619339045287639,
      "grad_norm": 1.435410574452025,
      "learning_rate": 1.0379418847980965e-07,
      "loss": 0.496,
      "step": 11127
    },
    {
      "epoch": 13.620563035495715,
      "grad_norm": 2.768201150720236,
      "learning_rate": 1.0361151910066758e-07,
      "loss": 0.9433,
      "step": 11128
    },
    {
      "epoch": 13.621787025703794,
      "grad_norm": 1.4125285107617327,
      "learning_rate": 1.0342900720331283e-07,
      "loss": 0.6013,
      "step": 11129
    },
    {
      "epoch": 13.623011015911873,
      "grad_norm": 1.224265081732127,
      "learning_rate": 1.0324665279973944e-07,
      "loss": 0.6109,
      "step": 11130
    },
    {
      "epoch": 13.624235006119951,
      "grad_norm": 2.089226802465181,
      "learning_rate": 1.0306445590193087e-07,
      "loss": 1.0,
      "step": 11131
    },
    {
      "epoch": 13.62545899632803,
      "grad_norm": 1.735695421842296,
      "learning_rate": 1.0288241652186087e-07,
      "loss": 0.6129,
      "step": 11132
    },
    {
      "epoch": 13.626682986536107,
      "grad_norm": 2.2306231259373894,
      "learning_rate": 1.0270053467149154e-07,
      "loss": 1.0983,
      "step": 11133
    },
    {
      "epoch": 13.627906976744185,
      "grad_norm": 2.1118858987132976,
      "learning_rate": 1.0251881036277694e-07,
      "loss": 0.4743,
      "step": 11134
    },
    {
      "epoch": 13.629130966952264,
      "grad_norm": 2.8997207745554756,
      "learning_rate": 1.0233724360765807e-07,
      "loss": 1.0147,
      "step": 11135
    },
    {
      "epoch": 13.630354957160343,
      "grad_norm": 1.9254803527581499,
      "learning_rate": 1.0215583441806703e-07,
      "loss": 0.2872,
      "step": 11136
    },
    {
      "epoch": 13.631578947368421,
      "grad_norm": 2.475735130990915,
      "learning_rate": 1.0197458280592542e-07,
      "loss": 0.4135,
      "step": 11137
    },
    {
      "epoch": 13.6328029375765,
      "grad_norm": 1.8751890066268249,
      "learning_rate": 1.0179348878314483e-07,
      "loss": 0.4959,
      "step": 11138
    },
    {
      "epoch": 13.634026927784578,
      "grad_norm": 2.000967733894036,
      "learning_rate": 1.0161255236162571e-07,
      "loss": 0.3335,
      "step": 11139
    },
    {
      "epoch": 13.635250917992655,
      "grad_norm": 1.0293833336746747,
      "learning_rate": 1.0143177355325912e-07,
      "loss": 0.4443,
      "step": 11140
    },
    {
      "epoch": 13.636474908200734,
      "grad_norm": 3.0716746299435833,
      "learning_rate": 1.0125115236992416e-07,
      "loss": 0.3882,
      "step": 11141
    },
    {
      "epoch": 13.637698898408813,
      "grad_norm": 1.2440689960082065,
      "learning_rate": 1.0107068882349108e-07,
      "loss": 0.8415,
      "step": 11142
    },
    {
      "epoch": 13.638922888616891,
      "grad_norm": 1.692351104435305,
      "learning_rate": 1.0089038292581926e-07,
      "loss": 0.6342,
      "step": 11143
    },
    {
      "epoch": 13.64014687882497,
      "grad_norm": 2.0310225406138547,
      "learning_rate": 1.0071023468875813e-07,
      "loss": 0.4741,
      "step": 11144
    },
    {
      "epoch": 13.641370869033048,
      "grad_norm": 3.2274155885282334,
      "learning_rate": 1.0053024412414569e-07,
      "loss": 0.4427,
      "step": 11145
    },
    {
      "epoch": 13.642594859241125,
      "grad_norm": 2.4020147746963416,
      "learning_rate": 1.0035041124381084e-07,
      "loss": 0.2798,
      "step": 11146
    },
    {
      "epoch": 13.643818849449204,
      "grad_norm": 1.0774078090802617,
      "learning_rate": 1.0017073605957106e-07,
      "loss": 0.5075,
      "step": 11147
    },
    {
      "epoch": 13.645042839657282,
      "grad_norm": 1.2661809390689824,
      "learning_rate": 9.99912185832344e-08,
      "loss": 1.0385,
      "step": 11148
    },
    {
      "epoch": 13.646266829865361,
      "grad_norm": 2.755154886578883,
      "learning_rate": 9.981185882659756e-08,
      "loss": 0.4188,
      "step": 11149
    },
    {
      "epoch": 13.64749082007344,
      "grad_norm": 2.2288398355490555,
      "learning_rate": 9.963265680144774e-08,
      "loss": 0.5923,
      "step": 11150
    },
    {
      "epoch": 13.648714810281518,
      "grad_norm": 1.512706888677791,
      "learning_rate": 9.945361251956137e-08,
      "loss": 0.614,
      "step": 11151
    },
    {
      "epoch": 13.649938800489597,
      "grad_norm": 2.743536722908378,
      "learning_rate": 9.927472599270488e-08,
      "loss": 0.3391,
      "step": 11152
    },
    {
      "epoch": 13.651162790697674,
      "grad_norm": 1.8235457878013934,
      "learning_rate": 9.909599723263358e-08,
      "loss": 1.1872,
      "step": 11153
    },
    {
      "epoch": 13.652386780905752,
      "grad_norm": 2.440807889050476,
      "learning_rate": 9.89174262510928e-08,
      "loss": 0.5102,
      "step": 11154
    },
    {
      "epoch": 13.653610771113831,
      "grad_norm": 1.640130233099999,
      "learning_rate": 9.873901305981787e-08,
      "loss": 0.4593,
      "step": 11155
    },
    {
      "epoch": 13.65483476132191,
      "grad_norm": 1.4241379165698853,
      "learning_rate": 9.85607576705333e-08,
      "loss": 0.552,
      "step": 11156
    },
    {
      "epoch": 13.656058751529988,
      "grad_norm": 3.084291972110577,
      "learning_rate": 9.838266009495362e-08,
      "loss": 0.4821,
      "step": 11157
    },
    {
      "epoch": 13.657282741738067,
      "grad_norm": 1.45300594811808,
      "learning_rate": 9.820472034478251e-08,
      "loss": 0.4987,
      "step": 11158
    },
    {
      "epoch": 13.658506731946144,
      "grad_norm": 1.6877245262100105,
      "learning_rate": 9.80269384317134e-08,
      "loss": 1.224,
      "step": 11159
    },
    {
      "epoch": 13.659730722154222,
      "grad_norm": 2.045992411529776,
      "learning_rate": 9.784931436742973e-08,
      "loss": 1.2858,
      "step": 11160
    },
    {
      "epoch": 13.660954712362301,
      "grad_norm": 1.709239527867135,
      "learning_rate": 9.767184816360409e-08,
      "loss": 0.6318,
      "step": 11161
    },
    {
      "epoch": 13.66217870257038,
      "grad_norm": 2.128085614219623,
      "learning_rate": 9.74945398318991e-08,
      "loss": 0.5858,
      "step": 11162
    },
    {
      "epoch": 13.663402692778458,
      "grad_norm": 1.504783155218763,
      "learning_rate": 9.731738938396657e-08,
      "loss": 0.6155,
      "step": 11163
    },
    {
      "epoch": 13.664626682986537,
      "grad_norm": 1.7919409033866889,
      "learning_rate": 9.714039683144855e-08,
      "loss": 1.6098,
      "step": 11164
    },
    {
      "epoch": 13.665850673194614,
      "grad_norm": 1.7781825859072111,
      "learning_rate": 9.696356218597575e-08,
      "loss": 0.6541,
      "step": 11165
    },
    {
      "epoch": 13.667074663402692,
      "grad_norm": 1.6331418162693239,
      "learning_rate": 9.67868854591697e-08,
      "loss": 0.4014,
      "step": 11166
    },
    {
      "epoch": 13.668298653610771,
      "grad_norm": 1.2664878009220473,
      "learning_rate": 9.661036666264057e-08,
      "loss": 0.5341,
      "step": 11167
    },
    {
      "epoch": 13.66952264381885,
      "grad_norm": 1.262166609576095,
      "learning_rate": 9.643400580798823e-08,
      "loss": 0.575,
      "step": 11168
    },
    {
      "epoch": 13.670746634026928,
      "grad_norm": 1.5834442316659936,
      "learning_rate": 9.62578029068037e-08,
      "loss": 0.3554,
      "step": 11169
    },
    {
      "epoch": 13.671970624235007,
      "grad_norm": 1.792114240577857,
      "learning_rate": 9.608175797066494e-08,
      "loss": 0.993,
      "step": 11170
    },
    {
      "epoch": 13.673194614443084,
      "grad_norm": 1.7783690354574087,
      "learning_rate": 9.590587101114185e-08,
      "loss": 0.3001,
      "step": 11171
    },
    {
      "epoch": 13.674418604651162,
      "grad_norm": 1.5414809564837917,
      "learning_rate": 9.573014203979241e-08,
      "loss": 0.267,
      "step": 11172
    },
    {
      "epoch": 13.67564259485924,
      "grad_norm": 1.8189540231636432,
      "learning_rate": 9.555457106816546e-08,
      "loss": 0.8787,
      "step": 11173
    },
    {
      "epoch": 13.67686658506732,
      "grad_norm": 3.0240749504166944,
      "learning_rate": 9.537915810779868e-08,
      "loss": 0.7308,
      "step": 11174
    },
    {
      "epoch": 13.678090575275398,
      "grad_norm": 1.47360594273982,
      "learning_rate": 9.520390317021954e-08,
      "loss": 0.4508,
      "step": 11175
    },
    {
      "epoch": 13.679314565483477,
      "grad_norm": 1.6039112060372953,
      "learning_rate": 9.502880626694522e-08,
      "loss": 0.3986,
      "step": 11176
    },
    {
      "epoch": 13.680538555691555,
      "grad_norm": 1.7884197286317915,
      "learning_rate": 9.485386740948232e-08,
      "loss": 0.7231,
      "step": 11177
    },
    {
      "epoch": 13.681762545899632,
      "grad_norm": 3.08791482579676,
      "learning_rate": 9.467908660932724e-08,
      "loss": 0.9014,
      "step": 11178
    },
    {
      "epoch": 13.68298653610771,
      "grad_norm": 1.3893882452562059,
      "learning_rate": 9.450446387796603e-08,
      "loss": 0.7341,
      "step": 11179
    },
    {
      "epoch": 13.68421052631579,
      "grad_norm": 3.1796499171678563,
      "learning_rate": 9.432999922687397e-08,
      "loss": 0.6366,
      "step": 11180
    },
    {
      "epoch": 13.685434516523868,
      "grad_norm": 2.4557351719134695,
      "learning_rate": 9.415569266751662e-08,
      "loss": 0.4627,
      "step": 11181
    },
    {
      "epoch": 13.686658506731947,
      "grad_norm": 2.6569542664934405,
      "learning_rate": 9.398154421134842e-08,
      "loss": 0.5427,
      "step": 11182
    },
    {
      "epoch": 13.687882496940025,
      "grad_norm": 1.9690902273147446,
      "learning_rate": 9.380755386981383e-08,
      "loss": 0.6455,
      "step": 11183
    },
    {
      "epoch": 13.689106487148102,
      "grad_norm": 1.490958929363694,
      "learning_rate": 9.363372165434703e-08,
      "loss": 0.9957,
      "step": 11184
    },
    {
      "epoch": 13.69033047735618,
      "grad_norm": 2.0553051903768003,
      "learning_rate": 9.346004757637112e-08,
      "loss": 0.571,
      "step": 11185
    },
    {
      "epoch": 13.69155446756426,
      "grad_norm": 2.2717346631280195,
      "learning_rate": 9.328653164730001e-08,
      "loss": 0.9466,
      "step": 11186
    },
    {
      "epoch": 13.692778457772338,
      "grad_norm": 3.4444210647351374,
      "learning_rate": 9.31131738785368e-08,
      "loss": 0.4599,
      "step": 11187
    },
    {
      "epoch": 13.694002447980417,
      "grad_norm": 2.712392405383955,
      "learning_rate": 9.293997428147267e-08,
      "loss": 0.7778,
      "step": 11188
    },
    {
      "epoch": 13.695226438188495,
      "grad_norm": 1.2337734793324764,
      "learning_rate": 9.276693286749017e-08,
      "loss": 0.6565,
      "step": 11189
    },
    {
      "epoch": 13.696450428396572,
      "grad_norm": 2.635160157722294,
      "learning_rate": 9.259404964796131e-08,
      "loss": 0.4702,
      "step": 11190
    },
    {
      "epoch": 13.69767441860465,
      "grad_norm": 1.3109972168699626,
      "learning_rate": 9.24213246342473e-08,
      "loss": 0.3829,
      "step": 11191
    },
    {
      "epoch": 13.69889840881273,
      "grad_norm": 1.8022580737796436,
      "learning_rate": 9.224875783769876e-08,
      "loss": 0.5112,
      "step": 11192
    },
    {
      "epoch": 13.700122399020808,
      "grad_norm": 1.387148926839083,
      "learning_rate": 9.207634926965608e-08,
      "loss": 0.6241,
      "step": 11193
    },
    {
      "epoch": 13.701346389228886,
      "grad_norm": 1.3231476733769352,
      "learning_rate": 9.190409894144964e-08,
      "loss": 0.7636,
      "step": 11194
    },
    {
      "epoch": 13.702570379436965,
      "grad_norm": 1.8653468522807153,
      "learning_rate": 9.173200686439871e-08,
      "loss": 1.4503,
      "step": 11195
    },
    {
      "epoch": 13.703794369645042,
      "grad_norm": 1.8258687047261888,
      "learning_rate": 9.156007304981285e-08,
      "loss": 0.41,
      "step": 11196
    },
    {
      "epoch": 13.70501835985312,
      "grad_norm": 2.3024978378874583,
      "learning_rate": 9.138829750899081e-08,
      "loss": 0.9586,
      "step": 11197
    },
    {
      "epoch": 13.7062423500612,
      "grad_norm": 1.939164829313745,
      "learning_rate": 9.121668025322078e-08,
      "loss": 0.7334,
      "step": 11198
    },
    {
      "epoch": 13.707466340269278,
      "grad_norm": 2.219053075379834,
      "learning_rate": 9.104522129378124e-08,
      "loss": 0.5601,
      "step": 11199
    },
    {
      "epoch": 13.708690330477356,
      "grad_norm": 1.486148050302697,
      "learning_rate": 9.087392064193956e-08,
      "loss": 0.4814,
      "step": 11200
    },
    {
      "epoch": 13.709914320685435,
      "grad_norm": 0.9516978437332777,
      "learning_rate": 9.070277830895313e-08,
      "loss": 0.4612,
      "step": 11201
    },
    {
      "epoch": 13.711138310893514,
      "grad_norm": 1.750241164016142,
      "learning_rate": 9.053179430606851e-08,
      "loss": 0.9061,
      "step": 11202
    },
    {
      "epoch": 13.71236230110159,
      "grad_norm": 1.519069105048366,
      "learning_rate": 9.036096864452254e-08,
      "loss": 0.5547,
      "step": 11203
    },
    {
      "epoch": 13.713586291309669,
      "grad_norm": 1.6830575315119183,
      "learning_rate": 9.019030133554125e-08,
      "loss": 0.4649,
      "step": 11204
    },
    {
      "epoch": 13.714810281517748,
      "grad_norm": 2.20055440116889,
      "learning_rate": 9.00197923903401e-08,
      "loss": 1.1812,
      "step": 11205
    },
    {
      "epoch": 13.716034271725826,
      "grad_norm": 1.900838892132819,
      "learning_rate": 8.984944182012428e-08,
      "loss": 0.5399,
      "step": 11206
    },
    {
      "epoch": 13.717258261933905,
      "grad_norm": 1.4199244936646243,
      "learning_rate": 8.96792496360882e-08,
      "loss": 0.8594,
      "step": 11207
    },
    {
      "epoch": 13.718482252141984,
      "grad_norm": 2.5383854393917824,
      "learning_rate": 8.95092158494168e-08,
      "loss": 0.7034,
      "step": 11208
    },
    {
      "epoch": 13.71970624235006,
      "grad_norm": 1.7131423122616922,
      "learning_rate": 8.933934047128417e-08,
      "loss": 0.7083,
      "step": 11209
    },
    {
      "epoch": 13.720930232558139,
      "grad_norm": 3.4164906536505866,
      "learning_rate": 8.916962351285363e-08,
      "loss": 0.513,
      "step": 11210
    },
    {
      "epoch": 13.722154222766218,
      "grad_norm": 2.2426104845853296,
      "learning_rate": 8.900006498527818e-08,
      "loss": 0.5011,
      "step": 11211
    },
    {
      "epoch": 13.723378212974296,
      "grad_norm": 1.5378521874351199,
      "learning_rate": 8.883066489970116e-08,
      "loss": 0.6429,
      "step": 11212
    },
    {
      "epoch": 13.724602203182375,
      "grad_norm": 1.568545376445181,
      "learning_rate": 8.866142326725418e-08,
      "loss": 0.7096,
      "step": 11213
    },
    {
      "epoch": 13.725826193390454,
      "grad_norm": 1.6225240102562688,
      "learning_rate": 8.849234009905977e-08,
      "loss": 0.6852,
      "step": 11214
    },
    {
      "epoch": 13.727050183598532,
      "grad_norm": 2.730391092490124,
      "learning_rate": 8.832341540622902e-08,
      "loss": 0.3848,
      "step": 11215
    },
    {
      "epoch": 13.728274173806609,
      "grad_norm": 1.572513732017943,
      "learning_rate": 8.815464919986361e-08,
      "loss": 0.5452,
      "step": 11216
    },
    {
      "epoch": 13.729498164014688,
      "grad_norm": 0.8724213273101712,
      "learning_rate": 8.798604149105355e-08,
      "loss": 0.4308,
      "step": 11217
    },
    {
      "epoch": 13.730722154222766,
      "grad_norm": 2.6005425468932173,
      "learning_rate": 8.781759229087944e-08,
      "loss": 1.0176,
      "step": 11218
    },
    {
      "epoch": 13.731946144430845,
      "grad_norm": 1.7810398952573052,
      "learning_rate": 8.76493016104113e-08,
      "loss": 0.4112,
      "step": 11219
    },
    {
      "epoch": 13.733170134638923,
      "grad_norm": 1.9378270540014673,
      "learning_rate": 8.748116946070778e-08,
      "loss": 1.13,
      "step": 11220
    },
    {
      "epoch": 13.734394124847,
      "grad_norm": 1.4912517612074776,
      "learning_rate": 8.731319585281922e-08,
      "loss": 0.3289,
      "step": 11221
    },
    {
      "epoch": 13.735618115055079,
      "grad_norm": 2.3266439838181245,
      "learning_rate": 8.714538079778318e-08,
      "loss": 0.4438,
      "step": 11222
    },
    {
      "epoch": 13.736842105263158,
      "grad_norm": 1.4956494747935003,
      "learning_rate": 8.697772430662859e-08,
      "loss": 0.6841,
      "step": 11223
    },
    {
      "epoch": 13.738066095471236,
      "grad_norm": 1.8122725188949516,
      "learning_rate": 8.681022639037223e-08,
      "loss": 0.5337,
      "step": 11224
    },
    {
      "epoch": 13.739290085679315,
      "grad_norm": 1.2961342162176972,
      "learning_rate": 8.664288706002221e-08,
      "loss": 0.5268,
      "step": 11225
    },
    {
      "epoch": 13.740514075887393,
      "grad_norm": 2.225226826893795,
      "learning_rate": 8.647570632657532e-08,
      "loss": 0.4142,
      "step": 11226
    },
    {
      "epoch": 13.741738066095472,
      "grad_norm": 1.490725715656985,
      "learning_rate": 8.630868420101828e-08,
      "loss": 0.9434,
      "step": 11227
    },
    {
      "epoch": 13.742962056303549,
      "grad_norm": 1.8802394142161059,
      "learning_rate": 8.614182069432653e-08,
      "loss": 1.4165,
      "step": 11228
    },
    {
      "epoch": 13.744186046511627,
      "grad_norm": 2.035592533579679,
      "learning_rate": 8.597511581746626e-08,
      "loss": 0.7217,
      "step": 11229
    },
    {
      "epoch": 13.745410036719706,
      "grad_norm": 2.2619173839453026,
      "learning_rate": 8.580856958139233e-08,
      "loss": 0.4793,
      "step": 11230
    },
    {
      "epoch": 13.746634026927785,
      "grad_norm": 1.663392148463811,
      "learning_rate": 8.564218199704988e-08,
      "loss": 0.3306,
      "step": 11231
    },
    {
      "epoch": 13.747858017135863,
      "grad_norm": 3.360561051178189,
      "learning_rate": 8.54759530753732e-08,
      "loss": 0.4503,
      "step": 11232
    },
    {
      "epoch": 13.749082007343942,
      "grad_norm": 1.1509994841871396,
      "learning_rate": 8.53098828272858e-08,
      "loss": 0.6587,
      "step": 11233
    },
    {
      "epoch": 13.750305997552019,
      "grad_norm": 1.7638802781686982,
      "learning_rate": 8.5143971263702e-08,
      "loss": 0.901,
      "step": 11234
    },
    {
      "epoch": 13.751529987760097,
      "grad_norm": 1.3775878624300648,
      "learning_rate": 8.497821839552445e-08,
      "loss": 0.7619,
      "step": 11235
    },
    {
      "epoch": 13.752753977968176,
      "grad_norm": 1.3821126011356417,
      "learning_rate": 8.481262423364584e-08,
      "loss": 0.6667,
      "step": 11236
    },
    {
      "epoch": 13.753977968176255,
      "grad_norm": 1.4390776143915307,
      "learning_rate": 8.4647188788948e-08,
      "loss": 0.735,
      "step": 11237
    },
    {
      "epoch": 13.755201958384333,
      "grad_norm": 1.6701094589895662,
      "learning_rate": 8.448191207230338e-08,
      "loss": 0.6961,
      "step": 11238
    },
    {
      "epoch": 13.756425948592412,
      "grad_norm": 1.831365441121446,
      "learning_rate": 8.4316794094573e-08,
      "loss": 0.4442,
      "step": 11239
    },
    {
      "epoch": 13.75764993880049,
      "grad_norm": 1.7086748091591495,
      "learning_rate": 8.415183486660844e-08,
      "loss": 0.4838,
      "step": 11240
    },
    {
      "epoch": 13.758873929008567,
      "grad_norm": 1.1150375698558823,
      "learning_rate": 8.39870343992491e-08,
      "loss": 0.5912,
      "step": 11241
    },
    {
      "epoch": 13.760097919216646,
      "grad_norm": 1.3566758870436664,
      "learning_rate": 8.382239270332604e-08,
      "loss": 0.5048,
      "step": 11242
    },
    {
      "epoch": 13.761321909424725,
      "grad_norm": 2.508705691549121,
      "learning_rate": 8.365790978965809e-08,
      "loss": 0.4523,
      "step": 11243
    },
    {
      "epoch": 13.762545899632803,
      "grad_norm": 1.0666988475740022,
      "learning_rate": 8.349358566905524e-08,
      "loss": 0.5814,
      "step": 11244
    },
    {
      "epoch": 13.763769889840882,
      "grad_norm": 2.884520767458708,
      "learning_rate": 8.332942035231578e-08,
      "loss": 0.4319,
      "step": 11245
    },
    {
      "epoch": 13.76499388004896,
      "grad_norm": 1.6465984700217884,
      "learning_rate": 8.31654138502283e-08,
      "loss": 0.4914,
      "step": 11246
    },
    {
      "epoch": 13.766217870257037,
      "grad_norm": 2.0520875542986756,
      "learning_rate": 8.30015661735703e-08,
      "loss": 0.4067,
      "step": 11247
    },
    {
      "epoch": 13.767441860465116,
      "grad_norm": 2.332151182115059,
      "learning_rate": 8.283787733310982e-08,
      "loss": 1.3273,
      "step": 11248
    },
    {
      "epoch": 13.768665850673194,
      "grad_norm": 1.7544132327652926,
      "learning_rate": 8.267434733960356e-08,
      "loss": 0.923,
      "step": 11249
    },
    {
      "epoch": 13.769889840881273,
      "grad_norm": 2.2954964838209992,
      "learning_rate": 8.25109762037976e-08,
      "loss": 0.437,
      "step": 11250
    },
    {
      "epoch": 13.771113831089352,
      "grad_norm": 1.538516468946373,
      "learning_rate": 8.234776393642895e-08,
      "loss": 0.5833,
      "step": 11251
    },
    {
      "epoch": 13.77233782129743,
      "grad_norm": 1.344462372373999,
      "learning_rate": 8.218471054822346e-08,
      "loss": 0.7269,
      "step": 11252
    },
    {
      "epoch": 13.773561811505507,
      "grad_norm": 3.1099072082345507,
      "learning_rate": 8.202181604989535e-08,
      "loss": 0.3327,
      "step": 11253
    },
    {
      "epoch": 13.774785801713586,
      "grad_norm": 1.7717170213846847,
      "learning_rate": 8.185908045215019e-08,
      "loss": 1.2594,
      "step": 11254
    },
    {
      "epoch": 13.776009791921664,
      "grad_norm": 3.367380278525698,
      "learning_rate": 8.169650376568195e-08,
      "loss": 0.3403,
      "step": 11255
    },
    {
      "epoch": 13.777233782129743,
      "grad_norm": 1.9291897063529257,
      "learning_rate": 8.153408600117485e-08,
      "loss": 0.4113,
      "step": 11256
    },
    {
      "epoch": 13.778457772337822,
      "grad_norm": 1.0919906767746577,
      "learning_rate": 8.137182716930258e-08,
      "loss": 0.5438,
      "step": 11257
    },
    {
      "epoch": 13.7796817625459,
      "grad_norm": 2.308410322136566,
      "learning_rate": 8.120972728072801e-08,
      "loss": 0.4675,
      "step": 11258
    },
    {
      "epoch": 13.780905752753977,
      "grad_norm": 2.469636843237229,
      "learning_rate": 8.104778634610316e-08,
      "loss": 0.4072,
      "step": 11259
    },
    {
      "epoch": 13.782129742962056,
      "grad_norm": 1.2976619544979011,
      "learning_rate": 8.08860043760712e-08,
      "loss": 0.5519,
      "step": 11260
    },
    {
      "epoch": 13.783353733170134,
      "grad_norm": 1.860781253602644,
      "learning_rate": 8.072438138126304e-08,
      "loss": 1.004,
      "step": 11261
    },
    {
      "epoch": 13.784577723378213,
      "grad_norm": 1.7592327200116582,
      "learning_rate": 8.056291737230049e-08,
      "loss": 0.435,
      "step": 11262
    },
    {
      "epoch": 13.785801713586292,
      "grad_norm": 2.4100294244383256,
      "learning_rate": 8.040161235979394e-08,
      "loss": 0.4744,
      "step": 11263
    },
    {
      "epoch": 13.78702570379437,
      "grad_norm": 1.675046000284579,
      "learning_rate": 8.024046635434408e-08,
      "loss": 1.1653,
      "step": 11264
    },
    {
      "epoch": 13.788249694002449,
      "grad_norm": 1.602945930924442,
      "learning_rate": 8.007947936654075e-08,
      "loss": 1.2546,
      "step": 11265
    },
    {
      "epoch": 13.789473684210526,
      "grad_norm": 1.5035631731103103,
      "learning_rate": 7.991865140696331e-08,
      "loss": 0.9637,
      "step": 11266
    },
    {
      "epoch": 13.790697674418604,
      "grad_norm": 2.2749651163408195,
      "learning_rate": 7.975798248618077e-08,
      "loss": 0.4622,
      "step": 11267
    },
    {
      "epoch": 13.791921664626683,
      "grad_norm": 1.674879102971246,
      "learning_rate": 7.959747261475164e-08,
      "loss": 0.4497,
      "step": 11268
    },
    {
      "epoch": 13.793145654834762,
      "grad_norm": 2.4103904666035643,
      "learning_rate": 7.943712180322417e-08,
      "loss": 0.4904,
      "step": 11269
    },
    {
      "epoch": 13.79436964504284,
      "grad_norm": 1.7627491610209542,
      "learning_rate": 7.927693006213655e-08,
      "loss": 0.4247,
      "step": 11270
    },
    {
      "epoch": 13.795593635250919,
      "grad_norm": 1.1001336314844625,
      "learning_rate": 7.911689740201511e-08,
      "loss": 0.7054,
      "step": 11271
    },
    {
      "epoch": 13.796817625458996,
      "grad_norm": 3.3325472427400245,
      "learning_rate": 7.895702383337673e-08,
      "loss": 0.5334,
      "step": 11272
    },
    {
      "epoch": 13.798041615667074,
      "grad_norm": 1.907054229585781,
      "learning_rate": 7.879730936672797e-08,
      "loss": 0.5198,
      "step": 11273
    },
    {
      "epoch": 13.799265605875153,
      "grad_norm": 2.09141300634952,
      "learning_rate": 7.86377540125649e-08,
      "loss": 0.3541,
      "step": 11274
    },
    {
      "epoch": 13.800489596083231,
      "grad_norm": 1.9463603753461127,
      "learning_rate": 7.847835778137247e-08,
      "loss": 0.5771,
      "step": 11275
    },
    {
      "epoch": 13.80171358629131,
      "grad_norm": 1.5421689500669706,
      "learning_rate": 7.831912068362563e-08,
      "loss": 0.555,
      "step": 11276
    },
    {
      "epoch": 13.802937576499389,
      "grad_norm": 1.4706651677242522,
      "learning_rate": 7.816004272978933e-08,
      "loss": 0.4298,
      "step": 11277
    },
    {
      "epoch": 13.804161566707466,
      "grad_norm": 1.152633078863864,
      "learning_rate": 7.800112393031716e-08,
      "loss": 0.3584,
      "step": 11278
    },
    {
      "epoch": 13.805385556915544,
      "grad_norm": 1.3160925078997552,
      "learning_rate": 7.784236429565273e-08,
      "loss": 0.8918,
      "step": 11279
    },
    {
      "epoch": 13.806609547123623,
      "grad_norm": 2.064412176778618,
      "learning_rate": 7.768376383622905e-08,
      "loss": 0.7572,
      "step": 11280
    },
    {
      "epoch": 13.807833537331701,
      "grad_norm": 2.191335214769516,
      "learning_rate": 7.752532256246892e-08,
      "loss": 0.528,
      "step": 11281
    },
    {
      "epoch": 13.80905752753978,
      "grad_norm": 1.1235996988741317,
      "learning_rate": 7.736704048478482e-08,
      "loss": 0.3908,
      "step": 11282
    },
    {
      "epoch": 13.810281517747859,
      "grad_norm": 1.9112348438002085,
      "learning_rate": 7.720891761357763e-08,
      "loss": 0.4161,
      "step": 11283
    },
    {
      "epoch": 13.811505507955935,
      "grad_norm": 1.60703189359901,
      "learning_rate": 7.705095395923956e-08,
      "loss": 0.5867,
      "step": 11284
    },
    {
      "epoch": 13.812729498164014,
      "grad_norm": 1.8652604509031379,
      "learning_rate": 7.689314953215038e-08,
      "loss": 0.4586,
      "step": 11285
    },
    {
      "epoch": 13.813953488372093,
      "grad_norm": 1.4277351008869217,
      "learning_rate": 7.673550434268123e-08,
      "loss": 1.5613,
      "step": 11286
    },
    {
      "epoch": 13.815177478580171,
      "grad_norm": 2.3294003884701646,
      "learning_rate": 7.657801840119189e-08,
      "loss": 0.5491,
      "step": 11287
    },
    {
      "epoch": 13.81640146878825,
      "grad_norm": 2.5567025444351237,
      "learning_rate": 7.642069171803157e-08,
      "loss": 1.2722,
      "step": 11288
    },
    {
      "epoch": 13.817625458996329,
      "grad_norm": 1.7517269064418863,
      "learning_rate": 7.626352430353867e-08,
      "loss": 0.5656,
      "step": 11289
    },
    {
      "epoch": 13.818849449204407,
      "grad_norm": 2.6469350368992917,
      "learning_rate": 7.610651616804271e-08,
      "loss": 0.3126,
      "step": 11290
    },
    {
      "epoch": 13.820073439412484,
      "grad_norm": 1.4006151936728661,
      "learning_rate": 7.594966732186071e-08,
      "loss": 0.5869,
      "step": 11291
    },
    {
      "epoch": 13.821297429620563,
      "grad_norm": 1.480070344664168,
      "learning_rate": 7.57929777753011e-08,
      "loss": 0.4283,
      "step": 11292
    },
    {
      "epoch": 13.822521419828641,
      "grad_norm": 2.699380648826637,
      "learning_rate": 7.563644753866006e-08,
      "loss": 0.3721,
      "step": 11293
    },
    {
      "epoch": 13.82374541003672,
      "grad_norm": 1.1625575784773936,
      "learning_rate": 7.548007662222495e-08,
      "loss": 0.5938,
      "step": 11294
    },
    {
      "epoch": 13.824969400244798,
      "grad_norm": 2.5079313912740138,
      "learning_rate": 7.532386503627143e-08,
      "loss": 0.5353,
      "step": 11295
    },
    {
      "epoch": 13.826193390452877,
      "grad_norm": 2.1516572802943834,
      "learning_rate": 7.516781279106516e-08,
      "loss": 1.1984,
      "step": 11296
    },
    {
      "epoch": 13.827417380660954,
      "grad_norm": 1.7422491090045111,
      "learning_rate": 7.50119198968613e-08,
      "loss": 0.4623,
      "step": 11297
    },
    {
      "epoch": 13.828641370869033,
      "grad_norm": 1.946064688095308,
      "learning_rate": 7.485618636390468e-08,
      "loss": 0.4799,
      "step": 11298
    },
    {
      "epoch": 13.829865361077111,
      "grad_norm": 1.7403013159273397,
      "learning_rate": 7.470061220242964e-08,
      "loss": 0.4234,
      "step": 11299
    },
    {
      "epoch": 13.83108935128519,
      "grad_norm": 3.0485113638642,
      "learning_rate": 7.454519742265992e-08,
      "loss": 0.4042,
      "step": 11300
    },
    {
      "epoch": 13.832313341493268,
      "grad_norm": 1.417618530882634,
      "learning_rate": 7.438994203480848e-08,
      "loss": 0.6415,
      "step": 11301
    },
    {
      "epoch": 13.833537331701347,
      "grad_norm": 2.1692204852078487,
      "learning_rate": 7.423484604907827e-08,
      "loss": 0.6511,
      "step": 11302
    },
    {
      "epoch": 13.834761321909426,
      "grad_norm": 2.7369393588293454,
      "learning_rate": 7.407990947566169e-08,
      "loss": 0.4512,
      "step": 11303
    },
    {
      "epoch": 13.835985312117502,
      "grad_norm": 1.5623553734423097,
      "learning_rate": 7.392513232474086e-08,
      "loss": 1.2097,
      "step": 11304
    },
    {
      "epoch": 13.837209302325581,
      "grad_norm": 2.3045123012626463,
      "learning_rate": 7.377051460648682e-08,
      "loss": 0.2963,
      "step": 11305
    },
    {
      "epoch": 13.83843329253366,
      "grad_norm": 1.5091614863591931,
      "learning_rate": 7.361605633106061e-08,
      "loss": 0.5401,
      "step": 11306
    },
    {
      "epoch": 13.839657282741738,
      "grad_norm": 1.1682973219792911,
      "learning_rate": 7.346175750861217e-08,
      "loss": 0.6364,
      "step": 11307
    },
    {
      "epoch": 13.840881272949817,
      "grad_norm": 2.115705477665773,
      "learning_rate": 7.330761814928228e-08,
      "loss": 0.3145,
      "step": 11308
    },
    {
      "epoch": 13.842105263157894,
      "grad_norm": 2.4938549887531862,
      "learning_rate": 7.315363826320005e-08,
      "loss": 0.5307,
      "step": 11309
    },
    {
      "epoch": 13.843329253365972,
      "grad_norm": 1.8298109488374228,
      "learning_rate": 7.299981786048432e-08,
      "loss": 0.4701,
      "step": 11310
    },
    {
      "epoch": 13.844553243574051,
      "grad_norm": 2.082035596738621,
      "learning_rate": 7.28461569512437e-08,
      "loss": 0.7186,
      "step": 11311
    },
    {
      "epoch": 13.84577723378213,
      "grad_norm": 1.7645570068555108,
      "learning_rate": 7.269265554557619e-08,
      "loss": 0.5179,
      "step": 11312
    },
    {
      "epoch": 13.847001223990208,
      "grad_norm": 1.6623289775199221,
      "learning_rate": 7.253931365356931e-08,
      "loss": 0.4607,
      "step": 11313
    },
    {
      "epoch": 13.848225214198287,
      "grad_norm": 1.7917325734921157,
      "learning_rate": 7.238613128530053e-08,
      "loss": 0.927,
      "step": 11314
    },
    {
      "epoch": 13.849449204406366,
      "grad_norm": 1.6149798525629688,
      "learning_rate": 7.223310845083542e-08,
      "loss": 2.2751,
      "step": 11315
    },
    {
      "epoch": 13.850673194614442,
      "grad_norm": 2.194400597270533,
      "learning_rate": 7.208024516023149e-08,
      "loss": 0.6316,
      "step": 11316
    },
    {
      "epoch": 13.851897184822521,
      "grad_norm": 1.691738178363653,
      "learning_rate": 7.19275414235332e-08,
      "loss": 0.6943,
      "step": 11317
    },
    {
      "epoch": 13.8531211750306,
      "grad_norm": 1.4712867618376417,
      "learning_rate": 7.177499725077613e-08,
      "loss": 0.3942,
      "step": 11318
    },
    {
      "epoch": 13.854345165238678,
      "grad_norm": 1.6126733293461826,
      "learning_rate": 7.162261265198478e-08,
      "loss": 0.5573,
      "step": 11319
    },
    {
      "epoch": 13.855569155446757,
      "grad_norm": 1.522073705428832,
      "learning_rate": 7.147038763717334e-08,
      "loss": 0.5209,
      "step": 11320
    },
    {
      "epoch": 13.856793145654835,
      "grad_norm": 1.1747999485172196,
      "learning_rate": 7.13183222163455e-08,
      "loss": 0.5228,
      "step": 11321
    },
    {
      "epoch": 13.858017135862912,
      "grad_norm": 2.2216112166802855,
      "learning_rate": 7.116641639949462e-08,
      "loss": 0.5683,
      "step": 11322
    },
    {
      "epoch": 13.859241126070991,
      "grad_norm": 1.6570630187848416,
      "learning_rate": 7.101467019660357e-08,
      "loss": 0.4683,
      "step": 11323
    },
    {
      "epoch": 13.86046511627907,
      "grad_norm": 1.9926977919090887,
      "learning_rate": 7.086308361764355e-08,
      "loss": 0.4515,
      "step": 11324
    },
    {
      "epoch": 13.861689106487148,
      "grad_norm": 1.7319392475175175,
      "learning_rate": 7.071165667257712e-08,
      "loss": 0.5479,
      "step": 11325
    },
    {
      "epoch": 13.862913096695227,
      "grad_norm": 2.022781484744085,
      "learning_rate": 7.056038937135551e-08,
      "loss": 0.3526,
      "step": 11326
    },
    {
      "epoch": 13.864137086903305,
      "grad_norm": 1.2054124609764831,
      "learning_rate": 7.040928172391908e-08,
      "loss": 0.4719,
      "step": 11327
    },
    {
      "epoch": 13.865361077111384,
      "grad_norm": 2.4174537972286716,
      "learning_rate": 7.025833374019825e-08,
      "loss": 0.5414,
      "step": 11328
    },
    {
      "epoch": 13.86658506731946,
      "grad_norm": 1.4973860410822903,
      "learning_rate": 7.010754543011256e-08,
      "loss": 1.1557,
      "step": 11329
    },
    {
      "epoch": 13.86780905752754,
      "grad_norm": 1.5811063261293088,
      "learning_rate": 6.995691680357159e-08,
      "loss": 0.5625,
      "step": 11330
    },
    {
      "epoch": 13.869033047735618,
      "grad_norm": 2.247074048148165,
      "learning_rate": 6.980644787047408e-08,
      "loss": 0.5073,
      "step": 11331
    },
    {
      "epoch": 13.870257037943697,
      "grad_norm": 1.4152562984974963,
      "learning_rate": 6.965613864070797e-08,
      "loss": 0.4062,
      "step": 11332
    },
    {
      "epoch": 13.871481028151775,
      "grad_norm": 2.4895883604674087,
      "learning_rate": 6.95059891241509e-08,
      "loss": 0.4154,
      "step": 11333
    },
    {
      "epoch": 13.872705018359854,
      "grad_norm": 1.3504034538665053,
      "learning_rate": 6.935599933067111e-08,
      "loss": 0.4873,
      "step": 11334
    },
    {
      "epoch": 13.87392900856793,
      "grad_norm": 2.2170234548160552,
      "learning_rate": 6.920616927012431e-08,
      "loss": 1.169,
      "step": 11335
    },
    {
      "epoch": 13.87515299877601,
      "grad_norm": 2.069847693577311,
      "learning_rate": 6.905649895235732e-08,
      "loss": 1.133,
      "step": 11336
    },
    {
      "epoch": 13.876376988984088,
      "grad_norm": 2.0423452010544807,
      "learning_rate": 6.890698838720538e-08,
      "loss": 0.9001,
      "step": 11337
    },
    {
      "epoch": 13.877600979192167,
      "grad_norm": 1.4855737470131296,
      "learning_rate": 6.875763758449477e-08,
      "loss": 0.8089,
      "step": 11338
    },
    {
      "epoch": 13.878824969400245,
      "grad_norm": 0.5925390683598846,
      "learning_rate": 6.86084465540393e-08,
      "loss": 0.1285,
      "step": 11339
    },
    {
      "epoch": 13.880048959608324,
      "grad_norm": 1.527874505755996,
      "learning_rate": 6.845941530564393e-08,
      "loss": 0.6098,
      "step": 11340
    },
    {
      "epoch": 13.8812729498164,
      "grad_norm": 1.803660106069246,
      "learning_rate": 6.831054384910191e-08,
      "loss": 1.7954,
      "step": 11341
    },
    {
      "epoch": 13.88249694002448,
      "grad_norm": 1.4030742991866738,
      "learning_rate": 6.81618321941968e-08,
      "loss": 0.4386,
      "step": 11342
    },
    {
      "epoch": 13.883720930232558,
      "grad_norm": 1.794211667536993,
      "learning_rate": 6.801328035070138e-08,
      "loss": 1.3551,
      "step": 11343
    },
    {
      "epoch": 13.884944920440637,
      "grad_norm": 1.7935326346910867,
      "learning_rate": 6.786488832837778e-08,
      "loss": 0.6161,
      "step": 11344
    },
    {
      "epoch": 13.886168910648715,
      "grad_norm": 2.7406554899264495,
      "learning_rate": 6.771665613697798e-08,
      "loss": 0.3303,
      "step": 11345
    },
    {
      "epoch": 13.887392900856794,
      "grad_norm": 1.3169698604985118,
      "learning_rate": 6.756858378624303e-08,
      "loss": 0.892,
      "step": 11346
    },
    {
      "epoch": 13.88861689106487,
      "grad_norm": 1.5700670762773787,
      "learning_rate": 6.74206712859038e-08,
      "loss": 0.6652,
      "step": 11347
    },
    {
      "epoch": 13.88984088127295,
      "grad_norm": 2.705008982452678,
      "learning_rate": 6.72729186456808e-08,
      "loss": 0.387,
      "step": 11348
    },
    {
      "epoch": 13.891064871481028,
      "grad_norm": 1.924009363087987,
      "learning_rate": 6.712532587528326e-08,
      "loss": 0.6378,
      "step": 11349
    },
    {
      "epoch": 13.892288861689106,
      "grad_norm": 1.4697418233957091,
      "learning_rate": 6.69778929844106e-08,
      "loss": 1.1437,
      "step": 11350
    },
    {
      "epoch": 13.893512851897185,
      "grad_norm": 2.856751314721035,
      "learning_rate": 6.683061998275204e-08,
      "loss": 0.9453,
      "step": 11351
    },
    {
      "epoch": 13.894736842105264,
      "grad_norm": 1.734408007696091,
      "learning_rate": 6.668350687998565e-08,
      "loss": 0.6112,
      "step": 11352
    },
    {
      "epoch": 13.895960832313342,
      "grad_norm": 3.297411093070813,
      "learning_rate": 6.653655368577872e-08,
      "loss": 0.3991,
      "step": 11353
    },
    {
      "epoch": 13.89718482252142,
      "grad_norm": 1.8406464647651577,
      "learning_rate": 6.638976040978851e-08,
      "loss": 0.6831,
      "step": 11354
    },
    {
      "epoch": 13.898408812729498,
      "grad_norm": 1.1994948894757536,
      "learning_rate": 6.624312706166203e-08,
      "loss": 0.567,
      "step": 11355
    },
    {
      "epoch": 13.899632802937576,
      "grad_norm": 1.1913793572217108,
      "learning_rate": 6.609665365103546e-08,
      "loss": 0.3803,
      "step": 11356
    },
    {
      "epoch": 13.900856793145655,
      "grad_norm": 2.03621561868549,
      "learning_rate": 6.595034018753444e-08,
      "loss": 0.5341,
      "step": 11357
    },
    {
      "epoch": 13.902080783353734,
      "grad_norm": 2.092696066439,
      "learning_rate": 6.580418668077459e-08,
      "loss": 0.5894,
      "step": 11358
    },
    {
      "epoch": 13.903304773561812,
      "grad_norm": 2.226065846505523,
      "learning_rate": 6.565819314035937e-08,
      "loss": 0.4828,
      "step": 11359
    },
    {
      "epoch": 13.904528763769889,
      "grad_norm": 1.8621954827750984,
      "learning_rate": 6.551235957588414e-08,
      "loss": 1.3972,
      "step": 11360
    },
    {
      "epoch": 13.905752753977968,
      "grad_norm": 1.4081685003313074,
      "learning_rate": 6.53666859969318e-08,
      "loss": 0.5197,
      "step": 11361
    },
    {
      "epoch": 13.906976744186046,
      "grad_norm": 1.6376567849722556,
      "learning_rate": 6.522117241307608e-08,
      "loss": 0.6838,
      "step": 11362
    },
    {
      "epoch": 13.908200734394125,
      "grad_norm": 1.3112630290640406,
      "learning_rate": 6.507581883387904e-08,
      "loss": 0.9707,
      "step": 11363
    },
    {
      "epoch": 13.909424724602204,
      "grad_norm": 1.93070692078817,
      "learning_rate": 6.493062526889305e-08,
      "loss": 1.004,
      "step": 11364
    },
    {
      "epoch": 13.910648714810282,
      "grad_norm": 3.073420871307899,
      "learning_rate": 6.478559172765991e-08,
      "loss": 0.3943,
      "step": 11365
    },
    {
      "epoch": 13.911872705018359,
      "grad_norm": 1.4732247519403472,
      "learning_rate": 6.464071821971035e-08,
      "loss": 0.8396,
      "step": 11366
    },
    {
      "epoch": 13.913096695226438,
      "grad_norm": 1.2378981120865593,
      "learning_rate": 6.449600475456453e-08,
      "loss": 0.44,
      "step": 11367
    },
    {
      "epoch": 13.914320685434516,
      "grad_norm": 1.0389271561267996,
      "learning_rate": 6.435145134173343e-08,
      "loss": 0.4366,
      "step": 11368
    },
    {
      "epoch": 13.915544675642595,
      "grad_norm": 1.724829882381599,
      "learning_rate": 6.420705799071613e-08,
      "loss": 1.3278,
      "step": 11369
    },
    {
      "epoch": 13.916768665850674,
      "grad_norm": 2.005458444816427,
      "learning_rate": 6.406282471100173e-08,
      "loss": 0.4873,
      "step": 11370
    },
    {
      "epoch": 13.917992656058752,
      "grad_norm": 2.318805652315315,
      "learning_rate": 6.391875151206845e-08,
      "loss": 0.4598,
      "step": 11371
    },
    {
      "epoch": 13.919216646266829,
      "grad_norm": 1.6878687165168977,
      "learning_rate": 6.3774838403384e-08,
      "loss": 0.589,
      "step": 11372
    },
    {
      "epoch": 13.920440636474908,
      "grad_norm": 2.758391285305014,
      "learning_rate": 6.363108539440665e-08,
      "loss": 0.4814,
      "step": 11373
    },
    {
      "epoch": 13.921664626682986,
      "grad_norm": 1.1636246593683968,
      "learning_rate": 6.348749249458275e-08,
      "loss": 0.5798,
      "step": 11374
    },
    {
      "epoch": 13.922888616891065,
      "grad_norm": 2.328624874994347,
      "learning_rate": 6.334405971334861e-08,
      "loss": 0.344,
      "step": 11375
    },
    {
      "epoch": 13.924112607099143,
      "grad_norm": 1.1075233625282106,
      "learning_rate": 6.320078706013061e-08,
      "loss": 0.5443,
      "step": 11376
    },
    {
      "epoch": 13.925336597307222,
      "grad_norm": 2.8857976653910145,
      "learning_rate": 6.305767454434369e-08,
      "loss": 0.2219,
      "step": 11377
    },
    {
      "epoch": 13.9265605875153,
      "grad_norm": 1.6795459471032432,
      "learning_rate": 6.291472217539285e-08,
      "loss": 0.4497,
      "step": 11378
    },
    {
      "epoch": 13.927784577723378,
      "grad_norm": 1.8197205752387493,
      "learning_rate": 6.277192996267222e-08,
      "loss": 0.9799,
      "step": 11379
    },
    {
      "epoch": 13.929008567931456,
      "grad_norm": 0.9886663625272907,
      "learning_rate": 6.262929791556543e-08,
      "loss": 0.4101,
      "step": 11380
    },
    {
      "epoch": 13.930232558139535,
      "grad_norm": 2.766879551643058,
      "learning_rate": 6.248682604344636e-08,
      "loss": 0.5276,
      "step": 11381
    },
    {
      "epoch": 13.931456548347613,
      "grad_norm": 1.627029445158408,
      "learning_rate": 6.234451435567724e-08,
      "loss": 0.4581,
      "step": 11382
    },
    {
      "epoch": 13.932680538555692,
      "grad_norm": 3.163905943289417,
      "learning_rate": 6.22023628616103e-08,
      "loss": 0.313,
      "step": 11383
    },
    {
      "epoch": 13.93390452876377,
      "grad_norm": 2.79659800119924,
      "learning_rate": 6.206037157058725e-08,
      "loss": 0.4933,
      "step": 11384
    },
    {
      "epoch": 13.935128518971847,
      "grad_norm": 3.510631437126887,
      "learning_rate": 6.191854049193896e-08,
      "loss": 0.345,
      "step": 11385
    },
    {
      "epoch": 13.936352509179926,
      "grad_norm": 1.5012295276410548,
      "learning_rate": 6.177686963498686e-08,
      "loss": 1.2675,
      "step": 11386
    },
    {
      "epoch": 13.937576499388005,
      "grad_norm": 1.3326569757626876,
      "learning_rate": 6.163535900904044e-08,
      "loss": 0.5157,
      "step": 11387
    },
    {
      "epoch": 13.938800489596083,
      "grad_norm": 1.004282237153405,
      "learning_rate": 6.149400862339949e-08,
      "loss": 0.4251,
      "step": 11388
    },
    {
      "epoch": 13.940024479804162,
      "grad_norm": 1.4132434507715836,
      "learning_rate": 6.135281848735242e-08,
      "loss": 1.0403,
      "step": 11389
    },
    {
      "epoch": 13.94124847001224,
      "grad_norm": 2.0897579475531467,
      "learning_rate": 6.121178861017845e-08,
      "loss": 0.5669,
      "step": 11390
    },
    {
      "epoch": 13.94247246022032,
      "grad_norm": 1.8777700093535563,
      "learning_rate": 6.10709190011452e-08,
      "loss": 1.0137,
      "step": 11391
    },
    {
      "epoch": 13.943696450428396,
      "grad_norm": 1.2332177503547308,
      "learning_rate": 6.093020966951024e-08,
      "loss": 0.6219,
      "step": 11392
    },
    {
      "epoch": 13.944920440636475,
      "grad_norm": 1.8579618427342717,
      "learning_rate": 6.078966062452036e-08,
      "loss": 0.4536,
      "step": 11393
    },
    {
      "epoch": 13.946144430844553,
      "grad_norm": 1.231128116153342,
      "learning_rate": 6.064927187541202e-08,
      "loss": 0.4774,
      "step": 11394
    },
    {
      "epoch": 13.947368421052632,
      "grad_norm": 1.655350555373785,
      "learning_rate": 6.050904343141095e-08,
      "loss": 0.4708,
      "step": 11395
    },
    {
      "epoch": 13.94859241126071,
      "grad_norm": 2.0109748319878116,
      "learning_rate": 6.036897530173253e-08,
      "loss": 1.1023,
      "step": 11396
    },
    {
      "epoch": 13.949816401468787,
      "grad_norm": 1.6167401440960756,
      "learning_rate": 6.022906749558161e-08,
      "loss": 1.5683,
      "step": 11397
    },
    {
      "epoch": 13.951040391676866,
      "grad_norm": 2.3560357292357343,
      "learning_rate": 6.008932002215195e-08,
      "loss": 1.0277,
      "step": 11398
    },
    {
      "epoch": 13.952264381884945,
      "grad_norm": 1.181344427317334,
      "learning_rate": 5.994973289062789e-08,
      "loss": 0.6158,
      "step": 11399
    },
    {
      "epoch": 13.953488372093023,
      "grad_norm": 1.2144266317883052,
      "learning_rate": 5.981030611018235e-08,
      "loss": 0.6676,
      "step": 11400
    },
    {
      "epoch": 13.954712362301102,
      "grad_norm": 2.0225602199720027,
      "learning_rate": 5.967103968997745e-08,
      "loss": 0.3828,
      "step": 11401
    },
    {
      "epoch": 13.95593635250918,
      "grad_norm": 1.697990560287181,
      "learning_rate": 5.9531933639165876e-08,
      "loss": 0.7125,
      "step": 11402
    },
    {
      "epoch": 13.957160342717259,
      "grad_norm": 3.4934935001719754,
      "learning_rate": 5.939298796688892e-08,
      "loss": 0.4287,
      "step": 11403
    },
    {
      "epoch": 13.958384332925336,
      "grad_norm": 1.308279105581052,
      "learning_rate": 5.925420268227761e-08,
      "loss": 0.7577,
      "step": 11404
    },
    {
      "epoch": 13.959608323133414,
      "grad_norm": 2.3697787646343094,
      "learning_rate": 5.911557779445271e-08,
      "loss": 0.4839,
      "step": 11405
    },
    {
      "epoch": 13.960832313341493,
      "grad_norm": 2.4717233764738404,
      "learning_rate": 5.897711331252359e-08,
      "loss": 0.7261,
      "step": 11406
    },
    {
      "epoch": 13.962056303549572,
      "grad_norm": 2.141728707191067,
      "learning_rate": 5.8838809245589656e-08,
      "loss": 0.6448,
      "step": 11407
    },
    {
      "epoch": 13.96328029375765,
      "grad_norm": 2.195527976822376,
      "learning_rate": 5.8700665602740016e-08,
      "loss": 0.8397,
      "step": 11408
    },
    {
      "epoch": 13.964504283965729,
      "grad_norm": 3.3167494379496,
      "learning_rate": 5.8562682393053236e-08,
      "loss": 0.6749,
      "step": 11409
    },
    {
      "epoch": 13.965728274173806,
      "grad_norm": 2.712816720550659,
      "learning_rate": 5.842485962559652e-08,
      "loss": 0.4127,
      "step": 11410
    },
    {
      "epoch": 13.966952264381884,
      "grad_norm": 2.341582891983283,
      "learning_rate": 5.8287197309427056e-08,
      "loss": 0.3426,
      "step": 11411
    },
    {
      "epoch": 13.968176254589963,
      "grad_norm": 2.0038027248813832,
      "learning_rate": 5.814969545359206e-08,
      "loss": 0.4755,
      "step": 11412
    },
    {
      "epoch": 13.969400244798042,
      "grad_norm": 1.5096502532358407,
      "learning_rate": 5.8012354067127074e-08,
      "loss": 0.6124,
      "step": 11413
    },
    {
      "epoch": 13.97062423500612,
      "grad_norm": 1.5702929242035408,
      "learning_rate": 5.787517315905794e-08,
      "loss": 0.9104,
      "step": 11414
    },
    {
      "epoch": 13.971848225214199,
      "grad_norm": 3.6564626866674934,
      "learning_rate": 5.773815273839939e-08,
      "loss": 0.403,
      "step": 11415
    },
    {
      "epoch": 13.973072215422278,
      "grad_norm": 3.193233164324562,
      "learning_rate": 5.7601292814156715e-08,
      "loss": 0.4514,
      "step": 11416
    },
    {
      "epoch": 13.974296205630354,
      "grad_norm": 1.6521890156723305,
      "learning_rate": 5.7464593395323006e-08,
      "loss": 0.351,
      "step": 11417
    },
    {
      "epoch": 13.975520195838433,
      "grad_norm": 2.2000739713883326,
      "learning_rate": 5.732805449088191e-08,
      "loss": 0.5061,
      "step": 11418
    },
    {
      "epoch": 13.976744186046512,
      "grad_norm": 1.517983253206331,
      "learning_rate": 5.719167610980597e-08,
      "loss": 0.4984,
      "step": 11419
    },
    {
      "epoch": 13.97796817625459,
      "grad_norm": 1.1583438625693545,
      "learning_rate": 5.7055458261058013e-08,
      "loss": 0.5998,
      "step": 11420
    },
    {
      "epoch": 13.979192166462669,
      "grad_norm": 1.6624847300663583,
      "learning_rate": 5.69194009535895e-08,
      "loss": 1.4804,
      "step": 11421
    },
    {
      "epoch": 13.980416156670747,
      "grad_norm": 1.5986708742956313,
      "learning_rate": 5.67835041963416e-08,
      "loss": 0.5215,
      "step": 11422
    },
    {
      "epoch": 13.981640146878824,
      "grad_norm": 1.9878811813985708,
      "learning_rate": 5.664776799824523e-08,
      "loss": 0.5843,
      "step": 11423
    },
    {
      "epoch": 13.982864137086903,
      "grad_norm": 1.9072267580272728,
      "learning_rate": 5.651219236821964e-08,
      "loss": 0.6062,
      "step": 11424
    },
    {
      "epoch": 13.984088127294982,
      "grad_norm": 1.460633168781365,
      "learning_rate": 5.637677731517521e-08,
      "loss": 1.2778,
      "step": 11425
    },
    {
      "epoch": 13.98531211750306,
      "grad_norm": 1.5332182996584836,
      "learning_rate": 5.6241522848010656e-08,
      "loss": 0.5986,
      "step": 11426
    },
    {
      "epoch": 13.986536107711139,
      "grad_norm": 2.160740066542416,
      "learning_rate": 5.610642897561441e-08,
      "loss": 0.6614,
      "step": 11427
    },
    {
      "epoch": 13.987760097919217,
      "grad_norm": 1.1822734559054673,
      "learning_rate": 5.597149570686439e-08,
      "loss": 0.6012,
      "step": 11428
    },
    {
      "epoch": 13.988984088127294,
      "grad_norm": 1.9712204945594751,
      "learning_rate": 5.583672305062765e-08,
      "loss": 0.5293,
      "step": 11429
    },
    {
      "epoch": 13.990208078335373,
      "grad_norm": 2.010657863222377,
      "learning_rate": 5.570211101576128e-08,
      "loss": 0.5254,
      "step": 11430
    },
    {
      "epoch": 13.991432068543451,
      "grad_norm": 1.3384451482999957,
      "learning_rate": 5.5567659611111546e-08,
      "loss": 0.7699,
      "step": 11431
    },
    {
      "epoch": 13.99265605875153,
      "grad_norm": 1.7468952987084188,
      "learning_rate": 5.543336884551359e-08,
      "loss": 1.4465,
      "step": 11432
    },
    {
      "epoch": 13.993880048959609,
      "grad_norm": 1.7071536204035782,
      "learning_rate": 5.529923872779258e-08,
      "loss": 1.1686,
      "step": 11433
    },
    {
      "epoch": 13.995104039167687,
      "grad_norm": 1.6064996242212264,
      "learning_rate": 5.516526926676397e-08,
      "loss": 1.1584,
      "step": 11434
    },
    {
      "epoch": 13.996328029375764,
      "grad_norm": 2.0240950602542838,
      "learning_rate": 5.503146047123043e-08,
      "loss": 0.3872,
      "step": 11435
    },
    {
      "epoch": 13.997552019583843,
      "grad_norm": 1.889461247733897,
      "learning_rate": 5.489781234998631e-08,
      "loss": 1.556,
      "step": 11436
    },
    {
      "epoch": 13.998776009791921,
      "grad_norm": 1.920988715923497,
      "learning_rate": 5.476432491181377e-08,
      "loss": 0.3464,
      "step": 11437
    },
    {
      "epoch": 14.0,
      "grad_norm": 3.785202186943911,
      "learning_rate": 5.463099816548578e-08,
      "loss": 0.3989,
      "step": 11438
    },
    {
      "epoch": 14.001223990208079,
      "grad_norm": 2.039578043424495,
      "learning_rate": 5.449783211976395e-08,
      "loss": 0.5069,
      "step": 11439
    },
    {
      "epoch": 14.002447980416157,
      "grad_norm": 1.8235320904536156,
      "learning_rate": 5.436482678339905e-08,
      "loss": 1.1535,
      "step": 11440
    },
    {
      "epoch": 14.003671970624236,
      "grad_norm": 1.6192206504975701,
      "learning_rate": 5.423198216513214e-08,
      "loss": 0.8117,
      "step": 11441
    },
    {
      "epoch": 14.004895960832313,
      "grad_norm": 3.079595824774245,
      "learning_rate": 5.4099298273692925e-08,
      "loss": 0.3891,
      "step": 11442
    },
    {
      "epoch": 14.006119951040391,
      "grad_norm": 1.5938909656903524,
      "learning_rate": 5.396677511780135e-08,
      "loss": 0.6835,
      "step": 11443
    },
    {
      "epoch": 14.00734394124847,
      "grad_norm": 2.1627160714429965,
      "learning_rate": 5.383441270616574e-08,
      "loss": 0.4332,
      "step": 11444
    },
    {
      "epoch": 14.008567931456549,
      "grad_norm": 2.979533393528313,
      "learning_rate": 5.370221104748496e-08,
      "loss": 0.3441,
      "step": 11445
    },
    {
      "epoch": 14.009791921664627,
      "grad_norm": 2.1500958959057863,
      "learning_rate": 5.35701701504468e-08,
      "loss": 0.528,
      "step": 11446
    },
    {
      "epoch": 14.011015911872706,
      "grad_norm": 1.425045375955601,
      "learning_rate": 5.343829002372791e-08,
      "loss": 0.601,
      "step": 11447
    },
    {
      "epoch": 14.012239902080783,
      "grad_norm": 1.9796971356028275,
      "learning_rate": 5.330657067599582e-08,
      "loss": 0.6457,
      "step": 11448
    },
    {
      "epoch": 14.013463892288861,
      "grad_norm": 2.0627247609467623,
      "learning_rate": 5.317501211590581e-08,
      "loss": 1.3042,
      "step": 11449
    },
    {
      "epoch": 14.01468788249694,
      "grad_norm": 1.432195581201928,
      "learning_rate": 5.304361435210404e-08,
      "loss": 0.6367,
      "step": 11450
    },
    {
      "epoch": 14.015911872705018,
      "grad_norm": 2.335243922415969,
      "learning_rate": 5.291237739322525e-08,
      "loss": 0.6109,
      "step": 11451
    },
    {
      "epoch": 14.017135862913097,
      "grad_norm": 1.4125734884482832,
      "learning_rate": 5.2781301247893934e-08,
      "loss": 1.0404,
      "step": 11452
    },
    {
      "epoch": 14.018359853121176,
      "grad_norm": 1.0624051299811235,
      "learning_rate": 5.2650385924724045e-08,
      "loss": 0.4369,
      "step": 11453
    },
    {
      "epoch": 14.019583843329253,
      "grad_norm": 1.7001860454816549,
      "learning_rate": 5.251963143231814e-08,
      "loss": 0.6329,
      "step": 11454
    },
    {
      "epoch": 14.020807833537331,
      "grad_norm": 1.6852605757930093,
      "learning_rate": 5.2389037779269625e-08,
      "loss": 1.5519,
      "step": 11455
    },
    {
      "epoch": 14.02203182374541,
      "grad_norm": 2.1322827932502864,
      "learning_rate": 5.22586049741608e-08,
      "loss": 0.406,
      "step": 11456
    },
    {
      "epoch": 14.023255813953488,
      "grad_norm": 2.0905221049587204,
      "learning_rate": 5.212833302556258e-08,
      "loss": 0.5698,
      "step": 11457
    },
    {
      "epoch": 14.024479804161567,
      "grad_norm": 1.562124019001566,
      "learning_rate": 5.199822194203646e-08,
      "loss": 0.9787,
      "step": 11458
    },
    {
      "epoch": 14.025703794369646,
      "grad_norm": 2.148848377831356,
      "learning_rate": 5.1868271732132546e-08,
      "loss": 0.6446,
      "step": 11459
    },
    {
      "epoch": 14.026927784577722,
      "grad_norm": 2.8022867618528133,
      "learning_rate": 5.1738482404390666e-08,
      "loss": 0.6472,
      "step": 11460
    },
    {
      "epoch": 14.028151774785801,
      "grad_norm": 2.2162888108042402,
      "learning_rate": 5.160885396734039e-08,
      "loss": 0.5678,
      "step": 11461
    },
    {
      "epoch": 14.02937576499388,
      "grad_norm": 1.9095854328970123,
      "learning_rate": 5.1479386429500455e-08,
      "loss": 0.6377,
      "step": 11462
    },
    {
      "epoch": 14.030599755201958,
      "grad_norm": 1.2799407444053532,
      "learning_rate": 5.1350079799378505e-08,
      "loss": 0.4662,
      "step": 11463
    },
    {
      "epoch": 14.031823745410037,
      "grad_norm": 1.7113868187885042,
      "learning_rate": 5.1220934085472465e-08,
      "loss": 0.4833,
      "step": 11464
    },
    {
      "epoch": 14.033047735618116,
      "grad_norm": 1.8036133321081538,
      "learning_rate": 5.109194929626915e-08,
      "loss": 1.6092,
      "step": 11465
    },
    {
      "epoch": 14.034271725826194,
      "grad_norm": 2.6846284697772664,
      "learning_rate": 5.096312544024512e-08,
      "loss": 0.3809,
      "step": 11466
    },
    {
      "epoch": 14.035495716034271,
      "grad_norm": 2.837767339294805,
      "learning_rate": 5.0834462525866104e-08,
      "loss": 0.351,
      "step": 11467
    },
    {
      "epoch": 14.03671970624235,
      "grad_norm": 1.5084833628440928,
      "learning_rate": 5.0705960561587276e-08,
      "loss": 1.0195,
      "step": 11468
    },
    {
      "epoch": 14.037943696450428,
      "grad_norm": 1.2509477250390353,
      "learning_rate": 5.057761955585355e-08,
      "loss": 0.6215,
      "step": 11469
    },
    {
      "epoch": 14.039167686658507,
      "grad_norm": 1.6146665421754045,
      "learning_rate": 5.044943951709902e-08,
      "loss": 1.5676,
      "step": 11470
    },
    {
      "epoch": 14.040391676866586,
      "grad_norm": 1.8880110175775144,
      "learning_rate": 5.032142045374694e-08,
      "loss": 0.5792,
      "step": 11471
    },
    {
      "epoch": 14.041615667074664,
      "grad_norm": 1.7856066396330448,
      "learning_rate": 5.019356237421002e-08,
      "loss": 0.6159,
      "step": 11472
    },
    {
      "epoch": 14.042839657282741,
      "grad_norm": 2.5983947823962814,
      "learning_rate": 5.0065865286891555e-08,
      "loss": 0.808,
      "step": 11473
    },
    {
      "epoch": 14.04406364749082,
      "grad_norm": 2.1647748139221985,
      "learning_rate": 4.993832920018232e-08,
      "loss": 1.1135,
      "step": 11474
    },
    {
      "epoch": 14.045287637698898,
      "grad_norm": 1.9089547069748665,
      "learning_rate": 4.981095412246423e-08,
      "loss": 0.5245,
      "step": 11475
    },
    {
      "epoch": 14.046511627906977,
      "grad_norm": 1.517343281139389,
      "learning_rate": 4.968374006210752e-08,
      "loss": 0.5546,
      "step": 11476
    },
    {
      "epoch": 14.047735618115055,
      "grad_norm": 1.660802960716323,
      "learning_rate": 4.9556687027472185e-08,
      "loss": 0.4604,
      "step": 11477
    },
    {
      "epoch": 14.048959608323134,
      "grad_norm": 1.5113278088346342,
      "learning_rate": 4.9429795026907925e-08,
      "loss": 0.5522,
      "step": 11478
    },
    {
      "epoch": 14.050183598531211,
      "grad_norm": 1.573569846579871,
      "learning_rate": 4.9303064068753627e-08,
      "loss": 0.4861,
      "step": 11479
    },
    {
      "epoch": 14.05140758873929,
      "grad_norm": 2.9822326342336405,
      "learning_rate": 4.9176494161337074e-08,
      "loss": 0.3987,
      "step": 11480
    },
    {
      "epoch": 14.052631578947368,
      "grad_norm": 0.9446770010512163,
      "learning_rate": 4.905008531297661e-08,
      "loss": 0.5192,
      "step": 11481
    },
    {
      "epoch": 14.053855569155447,
      "grad_norm": 1.3091344584162352,
      "learning_rate": 4.8923837531979214e-08,
      "loss": 0.891,
      "step": 11482
    },
    {
      "epoch": 14.055079559363525,
      "grad_norm": 1.464953740085889,
      "learning_rate": 4.879775082664129e-08,
      "loss": 0.3619,
      "step": 11483
    },
    {
      "epoch": 14.056303549571604,
      "grad_norm": 2.1351515784926636,
      "learning_rate": 4.867182520524871e-08,
      "loss": 0.3974,
      "step": 11484
    },
    {
      "epoch": 14.057527539779683,
      "grad_norm": 1.4892104814436362,
      "learning_rate": 4.854606067607681e-08,
      "loss": 0.9952,
      "step": 11485
    },
    {
      "epoch": 14.05875152998776,
      "grad_norm": 1.1866602615022195,
      "learning_rate": 4.842045724739064e-08,
      "loss": 0.6159,
      "step": 11486
    },
    {
      "epoch": 14.059975520195838,
      "grad_norm": 1.7495963305716633,
      "learning_rate": 4.829501492744415e-08,
      "loss": 1.4457,
      "step": 11487
    },
    {
      "epoch": 14.061199510403917,
      "grad_norm": 1.6713414890648837,
      "learning_rate": 4.816973372448131e-08,
      "loss": 0.2906,
      "step": 11488
    },
    {
      "epoch": 14.062423500611995,
      "grad_norm": 1.502220429597136,
      "learning_rate": 4.8044613646734686e-08,
      "loss": 0.8086,
      "step": 11489
    },
    {
      "epoch": 14.063647490820074,
      "grad_norm": 1.2621520684575402,
      "learning_rate": 4.791965470242688e-08,
      "loss": 0.4239,
      "step": 11490
    },
    {
      "epoch": 14.064871481028153,
      "grad_norm": 2.335134681481364,
      "learning_rate": 4.779485689976965e-08,
      "loss": 0.5147,
      "step": 11491
    },
    {
      "epoch": 14.06609547123623,
      "grad_norm": 2.8666228033949386,
      "learning_rate": 4.767022024696477e-08,
      "loss": 0.4929,
      "step": 11492
    },
    {
      "epoch": 14.067319461444308,
      "grad_norm": 1.212321480545358,
      "learning_rate": 4.754574475220208e-08,
      "loss": 0.878,
      "step": 11493
    },
    {
      "epoch": 14.068543451652387,
      "grad_norm": 1.3946061585871428,
      "learning_rate": 4.7421430423662266e-08,
      "loss": 0.7068,
      "step": 11494
    },
    {
      "epoch": 14.069767441860465,
      "grad_norm": 2.0316620882976966,
      "learning_rate": 4.729727726951433e-08,
      "loss": 0.382,
      "step": 11495
    },
    {
      "epoch": 14.070991432068544,
      "grad_norm": 1.72891736112303,
      "learning_rate": 4.717328529791759e-08,
      "loss": 0.9783,
      "step": 11496
    },
    {
      "epoch": 14.072215422276622,
      "grad_norm": 1.7660666583421578,
      "learning_rate": 4.7049454517020254e-08,
      "loss": 1.2593,
      "step": 11497
    },
    {
      "epoch": 14.0734394124847,
      "grad_norm": 1.516154234676482,
      "learning_rate": 4.6925784934959695e-08,
      "loss": 0.3544,
      "step": 11498
    },
    {
      "epoch": 14.074663402692778,
      "grad_norm": 2.5972292388337834,
      "learning_rate": 4.680227655986358e-08,
      "loss": 0.4465,
      "step": 11499
    },
    {
      "epoch": 14.075887392900857,
      "grad_norm": 2.345389270560507,
      "learning_rate": 4.6678929399847916e-08,
      "loss": 1.3283,
      "step": 11500
    },
    {
      "epoch": 14.077111383108935,
      "grad_norm": 1.9117511671342071,
      "learning_rate": 4.655574346301872e-08,
      "loss": 0.7232,
      "step": 11501
    },
    {
      "epoch": 14.078335373317014,
      "grad_norm": 2.4801037887617725,
      "learning_rate": 4.643271875747118e-08,
      "loss": 0.3904,
      "step": 11502
    },
    {
      "epoch": 14.079559363525092,
      "grad_norm": 1.376050470079658,
      "learning_rate": 4.630985529129051e-08,
      "loss": 0.5481,
      "step": 11503
    },
    {
      "epoch": 14.080783353733171,
      "grad_norm": 1.3421354612085616,
      "learning_rate": 4.618715307255078e-08,
      "loss": 0.5159,
      "step": 11504
    },
    {
      "epoch": 14.082007343941248,
      "grad_norm": 1.8784380183108278,
      "learning_rate": 4.606461210931529e-08,
      "loss": 1.3395,
      "step": 11505
    },
    {
      "epoch": 14.083231334149326,
      "grad_norm": 0.9873884261568006,
      "learning_rate": 4.5942232409636743e-08,
      "loss": 0.4101,
      "step": 11506
    },
    {
      "epoch": 14.084455324357405,
      "grad_norm": 1.181490898321425,
      "learning_rate": 4.582001398155789e-08,
      "loss": 0.6105,
      "step": 11507
    },
    {
      "epoch": 14.085679314565484,
      "grad_norm": 2.3654117717561425,
      "learning_rate": 4.569795683311035e-08,
      "loss": 0.5402,
      "step": 11508
    },
    {
      "epoch": 14.086903304773562,
      "grad_norm": 1.9457352337801574,
      "learning_rate": 4.5576060972315205e-08,
      "loss": 1.2063,
      "step": 11509
    },
    {
      "epoch": 14.088127294981641,
      "grad_norm": 1.515194784322062,
      "learning_rate": 4.5454326407183e-08,
      "loss": 0.6424,
      "step": 11510
    },
    {
      "epoch": 14.089351285189718,
      "grad_norm": 3.4834668813856684,
      "learning_rate": 4.5332753145713725e-08,
      "loss": 0.5129,
      "step": 11511
    },
    {
      "epoch": 14.090575275397796,
      "grad_norm": 1.3763958824624418,
      "learning_rate": 4.521134119589682e-08,
      "loss": 0.521,
      "step": 11512
    },
    {
      "epoch": 14.091799265605875,
      "grad_norm": 2.7290058238189037,
      "learning_rate": 4.5090090565711174e-08,
      "loss": 0.5267,
      "step": 11513
    },
    {
      "epoch": 14.093023255813954,
      "grad_norm": 1.6910953284631638,
      "learning_rate": 4.4969001263124314e-08,
      "loss": 0.3476,
      "step": 11514
    },
    {
      "epoch": 14.094247246022032,
      "grad_norm": 1.2546829030093793,
      "learning_rate": 4.484807329609403e-08,
      "loss": 0.5755,
      "step": 11515
    },
    {
      "epoch": 14.095471236230111,
      "grad_norm": 2.075031722358038,
      "learning_rate": 4.4727306672567585e-08,
      "loss": 0.7177,
      "step": 11516
    },
    {
      "epoch": 14.096695226438188,
      "grad_norm": 1.1125701421456615,
      "learning_rate": 4.46067014004814e-08,
      "loss": 0.7051,
      "step": 11517
    },
    {
      "epoch": 14.097919216646266,
      "grad_norm": 2.5244349394204577,
      "learning_rate": 4.448625748776081e-08,
      "loss": 0.4934,
      "step": 11518
    },
    {
      "epoch": 14.099143206854345,
      "grad_norm": 2.469658290140717,
      "learning_rate": 4.4365974942320865e-08,
      "loss": 0.7263,
      "step": 11519
    },
    {
      "epoch": 14.100367197062424,
      "grad_norm": 2.9827934857524454,
      "learning_rate": 4.424585377206636e-08,
      "loss": 0.6774,
      "step": 11520
    },
    {
      "epoch": 14.101591187270502,
      "grad_norm": 1.782733044709766,
      "learning_rate": 4.412589398489098e-08,
      "loss": 0.4247,
      "step": 11521
    },
    {
      "epoch": 14.10281517747858,
      "grad_norm": 1.9010187821802655,
      "learning_rate": 4.400609558867841e-08,
      "loss": 1.1657,
      "step": 11522
    },
    {
      "epoch": 14.104039167686658,
      "grad_norm": 2.9195448351279243,
      "learning_rate": 4.388645859130153e-08,
      "loss": 1.0144,
      "step": 11523
    },
    {
      "epoch": 14.105263157894736,
      "grad_norm": 2.442437479576611,
      "learning_rate": 4.376698300062127e-08,
      "loss": 0.8908,
      "step": 11524
    },
    {
      "epoch": 14.106487148102815,
      "grad_norm": 1.9348220106633436,
      "learning_rate": 4.364766882449023e-08,
      "loss": 0.4831,
      "step": 11525
    },
    {
      "epoch": 14.107711138310894,
      "grad_norm": 1.6985035927867322,
      "learning_rate": 4.352851607074909e-08,
      "loss": 1.0085,
      "step": 11526
    },
    {
      "epoch": 14.108935128518972,
      "grad_norm": 1.9175424425077368,
      "learning_rate": 4.3409524747227985e-08,
      "loss": 0.4325,
      "step": 11527
    },
    {
      "epoch": 14.11015911872705,
      "grad_norm": 2.359400189948757,
      "learning_rate": 4.3290694861746474e-08,
      "loss": 0.5727,
      "step": 11528
    },
    {
      "epoch": 14.11138310893513,
      "grad_norm": 1.8321517974819657,
      "learning_rate": 4.317202642211388e-08,
      "loss": 0.4101,
      "step": 11529
    },
    {
      "epoch": 14.112607099143206,
      "grad_norm": 1.4453695770484185,
      "learning_rate": 4.3053519436128674e-08,
      "loss": 0.4984,
      "step": 11530
    },
    {
      "epoch": 14.113831089351285,
      "grad_norm": 1.692696691799215,
      "learning_rate": 4.2935173911578253e-08,
      "loss": 0.4641,
      "step": 11531
    },
    {
      "epoch": 14.115055079559363,
      "grad_norm": 2.5104911860689088,
      "learning_rate": 4.2816989856240273e-08,
      "loss": 0.4784,
      "step": 11532
    },
    {
      "epoch": 14.116279069767442,
      "grad_norm": 3.156164252006626,
      "learning_rate": 4.2698967277881033e-08,
      "loss": 0.3114,
      "step": 11533
    },
    {
      "epoch": 14.11750305997552,
      "grad_norm": 2.009846448144662,
      "learning_rate": 4.258110618425737e-08,
      "loss": 0.987,
      "step": 11534
    },
    {
      "epoch": 14.1187270501836,
      "grad_norm": 2.346603970296224,
      "learning_rate": 4.246340658311365e-08,
      "loss": 0.4367,
      "step": 11535
    },
    {
      "epoch": 14.119951040391676,
      "grad_norm": 2.693151025265512,
      "learning_rate": 4.2345868482184806e-08,
      "loss": 0.4511,
      "step": 11536
    },
    {
      "epoch": 14.121175030599755,
      "grad_norm": 1.3111137201625636,
      "learning_rate": 4.2228491889195475e-08,
      "loss": 0.7634,
      "step": 11537
    },
    {
      "epoch": 14.122399020807833,
      "grad_norm": 2.459296345421133,
      "learning_rate": 4.2111276811858946e-08,
      "loss": 0.4112,
      "step": 11538
    },
    {
      "epoch": 14.123623011015912,
      "grad_norm": 1.702879181582311,
      "learning_rate": 4.199422325787822e-08,
      "loss": 0.467,
      "step": 11539
    },
    {
      "epoch": 14.12484700122399,
      "grad_norm": 1.6802103097334484,
      "learning_rate": 4.187733123494575e-08,
      "loss": 0.4926,
      "step": 11540
    },
    {
      "epoch": 14.12607099143207,
      "grad_norm": 1.628003829424904,
      "learning_rate": 4.1760600750742894e-08,
      "loss": 0.7405,
      "step": 11541
    },
    {
      "epoch": 14.127294981640146,
      "grad_norm": 2.224913109908127,
      "learning_rate": 4.164403181294102e-08,
      "loss": 1.1694,
      "step": 11542
    },
    {
      "epoch": 14.128518971848225,
      "grad_norm": 2.7267319616298766,
      "learning_rate": 4.1527624429200666e-08,
      "loss": 0.4179,
      "step": 11543
    },
    {
      "epoch": 14.129742962056303,
      "grad_norm": 1.4274047845659668,
      "learning_rate": 4.1411378607171257e-08,
      "loss": 0.5398,
      "step": 11544
    },
    {
      "epoch": 14.130966952264382,
      "grad_norm": 1.4160259733062652,
      "learning_rate": 4.1295294354492534e-08,
      "loss": 0.8598,
      "step": 11545
    },
    {
      "epoch": 14.13219094247246,
      "grad_norm": 1.285984688967624,
      "learning_rate": 4.11793716787931e-08,
      "loss": 0.5685,
      "step": 11546
    },
    {
      "epoch": 14.13341493268054,
      "grad_norm": 1.357954799674962,
      "learning_rate": 4.106361058769076e-08,
      "loss": 0.5287,
      "step": 11547
    },
    {
      "epoch": 14.134638922888616,
      "grad_norm": 1.6086501826621533,
      "learning_rate": 4.094801108879304e-08,
      "loss": 0.8594,
      "step": 11548
    },
    {
      "epoch": 14.135862913096695,
      "grad_norm": 1.5187195778415952,
      "learning_rate": 4.0832573189696356e-08,
      "loss": 0.4986,
      "step": 11549
    },
    {
      "epoch": 14.137086903304773,
      "grad_norm": 2.027910966186808,
      "learning_rate": 4.071729689798715e-08,
      "loss": 0.7196,
      "step": 11550
    },
    {
      "epoch": 14.138310893512852,
      "grad_norm": 1.705699092980649,
      "learning_rate": 4.0602182221241016e-08,
      "loss": 1.1633,
      "step": 11551
    },
    {
      "epoch": 14.13953488372093,
      "grad_norm": 2.176808643240439,
      "learning_rate": 4.0487229167023026e-08,
      "loss": 0.9539,
      "step": 11552
    },
    {
      "epoch": 14.140758873929009,
      "grad_norm": 1.295708271090009,
      "learning_rate": 4.0372437742887126e-08,
      "loss": 0.5682,
      "step": 11553
    },
    {
      "epoch": 14.141982864137088,
      "grad_norm": 1.086415595318528,
      "learning_rate": 4.025780795637674e-08,
      "loss": 0.5913,
      "step": 11554
    },
    {
      "epoch": 14.143206854345165,
      "grad_norm": 1.782111477689603,
      "learning_rate": 4.014333981502555e-08,
      "loss": 0.9918,
      "step": 11555
    },
    {
      "epoch": 14.144430844553243,
      "grad_norm": 1.7792593054169954,
      "learning_rate": 4.00290333263556e-08,
      "loss": 0.9261,
      "step": 11556
    },
    {
      "epoch": 14.145654834761322,
      "grad_norm": 2.2731073810412106,
      "learning_rate": 3.991488849787867e-08,
      "loss": 0.9458,
      "step": 11557
    },
    {
      "epoch": 14.1468788249694,
      "grad_norm": 2.4126094281333232,
      "learning_rate": 3.980090533709624e-08,
      "loss": 0.5093,
      "step": 11558
    },
    {
      "epoch": 14.148102815177479,
      "grad_norm": 1.9044766735889704,
      "learning_rate": 3.968708385149872e-08,
      "loss": 1.299,
      "step": 11559
    },
    {
      "epoch": 14.149326805385558,
      "grad_norm": 1.5128046278689606,
      "learning_rate": 3.9573424048565676e-08,
      "loss": 0.3502,
      "step": 11560
    },
    {
      "epoch": 14.150550795593634,
      "grad_norm": 1.4215092097734507,
      "learning_rate": 3.945992593576698e-08,
      "loss": 0.6417,
      "step": 11561
    },
    {
      "epoch": 14.151774785801713,
      "grad_norm": 1.878073222297094,
      "learning_rate": 3.9346589520561084e-08,
      "loss": 0.5385,
      "step": 11562
    },
    {
      "epoch": 14.152998776009792,
      "grad_norm": 2.219573074885244,
      "learning_rate": 3.9233414810395944e-08,
      "loss": 0.4136,
      "step": 11563
    },
    {
      "epoch": 14.15422276621787,
      "grad_norm": 2.298919082921737,
      "learning_rate": 3.912040181270893e-08,
      "loss": 0.4672,
      "step": 11564
    },
    {
      "epoch": 14.155446756425949,
      "grad_norm": 1.6592771490220868,
      "learning_rate": 3.9007550534926886e-08,
      "loss": 1.4808,
      "step": 11565
    },
    {
      "epoch": 14.156670746634028,
      "grad_norm": 2.540451332907024,
      "learning_rate": 3.88948609844661e-08,
      "loss": 0.5341,
      "step": 11566
    },
    {
      "epoch": 14.157894736842104,
      "grad_norm": 1.7861241915990238,
      "learning_rate": 3.8782333168732033e-08,
      "loss": 0.4896,
      "step": 11567
    },
    {
      "epoch": 14.159118727050183,
      "grad_norm": 1.1605425198447181,
      "learning_rate": 3.866996709511961e-08,
      "loss": 0.5037,
      "step": 11568
    },
    {
      "epoch": 14.160342717258262,
      "grad_norm": 1.89199162528628,
      "learning_rate": 3.855776277101319e-08,
      "loss": 1.4151,
      "step": 11569
    },
    {
      "epoch": 14.16156670746634,
      "grad_norm": 2.7777913591483094,
      "learning_rate": 3.844572020378662e-08,
      "loss": 0.325,
      "step": 11570
    },
    {
      "epoch": 14.162790697674419,
      "grad_norm": 1.3159594571678437,
      "learning_rate": 3.833383940080232e-08,
      "loss": 0.586,
      "step": 11571
    },
    {
      "epoch": 14.164014687882498,
      "grad_norm": 1.5766500178813465,
      "learning_rate": 3.8222120369413027e-08,
      "loss": 1.1583,
      "step": 11572
    },
    {
      "epoch": 14.165238678090576,
      "grad_norm": 2.2545169239406846,
      "learning_rate": 3.811056311696065e-08,
      "loss": 0.9134,
      "step": 11573
    },
    {
      "epoch": 14.166462668298653,
      "grad_norm": 1.9921955048625328,
      "learning_rate": 3.7999167650775994e-08,
      "loss": 1.4809,
      "step": 11574
    },
    {
      "epoch": 14.167686658506732,
      "grad_norm": 1.4924615581117442,
      "learning_rate": 3.7887933978179856e-08,
      "loss": 0.5262,
      "step": 11575
    },
    {
      "epoch": 14.16891064871481,
      "grad_norm": 1.4634206618207768,
      "learning_rate": 3.777686210648196e-08,
      "loss": 0.8393,
      "step": 11576
    },
    {
      "epoch": 14.170134638922889,
      "grad_norm": 1.7380511598766002,
      "learning_rate": 3.7665952042981455e-08,
      "loss": 0.8453,
      "step": 11577
    },
    {
      "epoch": 14.171358629130967,
      "grad_norm": 3.9753989450083833,
      "learning_rate": 3.7555203794967244e-08,
      "loss": 0.3303,
      "step": 11578
    },
    {
      "epoch": 14.172582619339046,
      "grad_norm": 3.0736544747473604,
      "learning_rate": 3.744461736971683e-08,
      "loss": 0.4034,
      "step": 11579
    },
    {
      "epoch": 14.173806609547123,
      "grad_norm": 1.5679606316155348,
      "learning_rate": 3.7334192774497734e-08,
      "loss": 0.6647,
      "step": 11580
    },
    {
      "epoch": 14.175030599755202,
      "grad_norm": 2.0983529190998373,
      "learning_rate": 3.722393001656721e-08,
      "loss": 1.284,
      "step": 11581
    },
    {
      "epoch": 14.17625458996328,
      "grad_norm": 2.08734653699696,
      "learning_rate": 3.711382910317057e-08,
      "loss": 0.385,
      "step": 11582
    },
    {
      "epoch": 14.177478580171359,
      "grad_norm": 1.4435749213001259,
      "learning_rate": 3.700389004154342e-08,
      "loss": 1.0847,
      "step": 11583
    },
    {
      "epoch": 14.178702570379437,
      "grad_norm": 2.1886935730768635,
      "learning_rate": 3.689411283891053e-08,
      "loss": 0.3037,
      "step": 11584
    },
    {
      "epoch": 14.179926560587516,
      "grad_norm": 1.8790090013118523,
      "learning_rate": 3.678449750248614e-08,
      "loss": 1.8729,
      "step": 11585
    },
    {
      "epoch": 14.181150550795593,
      "grad_norm": 1.8661664510236045,
      "learning_rate": 3.6675044039473916e-08,
      "loss": 1.3964,
      "step": 11586
    },
    {
      "epoch": 14.182374541003671,
      "grad_norm": 1.9924873567101151,
      "learning_rate": 3.6565752457066726e-08,
      "loss": 0.6046,
      "step": 11587
    },
    {
      "epoch": 14.18359853121175,
      "grad_norm": 1.9356725677062565,
      "learning_rate": 3.64566227624466e-08,
      "loss": 0.4486,
      "step": 11588
    },
    {
      "epoch": 14.184822521419829,
      "grad_norm": 2.6941735227047383,
      "learning_rate": 3.6347654962785027e-08,
      "loss": 0.3506,
      "step": 11589
    },
    {
      "epoch": 14.186046511627907,
      "grad_norm": 1.178597833456061,
      "learning_rate": 3.623884906524322e-08,
      "loss": 0.5658,
      "step": 11590
    },
    {
      "epoch": 14.187270501835986,
      "grad_norm": 1.1034270059029487,
      "learning_rate": 3.613020507697157e-08,
      "loss": 0.5144,
      "step": 11591
    },
    {
      "epoch": 14.188494492044065,
      "grad_norm": 2.230667106297832,
      "learning_rate": 3.6021723005109367e-08,
      "loss": 0.5008,
      "step": 11592
    },
    {
      "epoch": 14.189718482252141,
      "grad_norm": 2.3795204005223076,
      "learning_rate": 3.591340285678618e-08,
      "loss": 0.5417,
      "step": 11593
    },
    {
      "epoch": 14.19094247246022,
      "grad_norm": 1.7173081556664576,
      "learning_rate": 3.580524463911994e-08,
      "loss": 1.3272,
      "step": 11594
    },
    {
      "epoch": 14.192166462668299,
      "grad_norm": 1.7794697219047417,
      "learning_rate": 3.569724835921856e-08,
      "loss": 1.5124,
      "step": 11595
    },
    {
      "epoch": 14.193390452876377,
      "grad_norm": 1.059273739554263,
      "learning_rate": 3.5589414024179424e-08,
      "loss": 0.5819,
      "step": 11596
    },
    {
      "epoch": 14.194614443084456,
      "grad_norm": 2.006767433860227,
      "learning_rate": 3.548174164108853e-08,
      "loss": 1.3819,
      "step": 11597
    },
    {
      "epoch": 14.195838433292534,
      "grad_norm": 1.8067232898076888,
      "learning_rate": 3.537423121702188e-08,
      "loss": 1.1676,
      "step": 11598
    },
    {
      "epoch": 14.197062423500611,
      "grad_norm": 1.844198293921561,
      "learning_rate": 3.526688275904522e-08,
      "loss": 0.6823,
      "step": 11599
    },
    {
      "epoch": 14.19828641370869,
      "grad_norm": 1.7536055468013614,
      "learning_rate": 3.515969627421234e-08,
      "loss": 0.3654,
      "step": 11600
    },
    {
      "epoch": 14.199510403916769,
      "grad_norm": 1.5160245454205818,
      "learning_rate": 3.505267176956734e-08,
      "loss": 1.4056,
      "step": 11601
    },
    {
      "epoch": 14.200734394124847,
      "grad_norm": 2.0253107207626075,
      "learning_rate": 3.494580925214347e-08,
      "loss": 0.8996,
      "step": 11602
    },
    {
      "epoch": 14.201958384332926,
      "grad_norm": 2.2306124659748567,
      "learning_rate": 3.4839108728963456e-08,
      "loss": 0.5593,
      "step": 11603
    },
    {
      "epoch": 14.203182374541004,
      "grad_norm": 1.7133365424344351,
      "learning_rate": 3.473257020703946e-08,
      "loss": 0.4838,
      "step": 11604
    },
    {
      "epoch": 14.204406364749081,
      "grad_norm": 2.674147104253107,
      "learning_rate": 3.4626193693372555e-08,
      "loss": 0.661,
      "step": 11605
    },
    {
      "epoch": 14.20563035495716,
      "grad_norm": 1.2984869896595264,
      "learning_rate": 3.451997919495298e-08,
      "loss": 0.9812,
      "step": 11606
    },
    {
      "epoch": 14.206854345165238,
      "grad_norm": 1.7839045610799191,
      "learning_rate": 3.4413926718761536e-08,
      "loss": 0.654,
      "step": 11607
    },
    {
      "epoch": 14.208078335373317,
      "grad_norm": 2.3983645678281746,
      "learning_rate": 3.43080362717671e-08,
      "loss": 0.8395,
      "step": 11608
    },
    {
      "epoch": 14.209302325581396,
      "grad_norm": 1.7899105272480045,
      "learning_rate": 3.420230786092854e-08,
      "loss": 0.6705,
      "step": 11609
    },
    {
      "epoch": 14.210526315789474,
      "grad_norm": 2.5871150656806163,
      "learning_rate": 3.4096741493194196e-08,
      "loss": 0.4109,
      "step": 11610
    },
    {
      "epoch": 14.211750305997551,
      "grad_norm": 1.1561780863248206,
      "learning_rate": 3.399133717550101e-08,
      "loss": 0.4919,
      "step": 11611
    },
    {
      "epoch": 14.21297429620563,
      "grad_norm": 1.417475593075374,
      "learning_rate": 3.3886094914776234e-08,
      "loss": 0.5101,
      "step": 11612
    },
    {
      "epoch": 14.214198286413708,
      "grad_norm": 1.1838885648081627,
      "learning_rate": 3.378101471793543e-08,
      "loss": 0.5548,
      "step": 11613
    },
    {
      "epoch": 14.215422276621787,
      "grad_norm": 1.8051408579744028,
      "learning_rate": 3.3676096591884754e-08,
      "loss": 0.5333,
      "step": 11614
    },
    {
      "epoch": 14.216646266829866,
      "grad_norm": 1.546798399095004,
      "learning_rate": 3.35713405435184e-08,
      "loss": 0.5984,
      "step": 11615
    },
    {
      "epoch": 14.217870257037944,
      "grad_norm": 2.434809116122178,
      "learning_rate": 3.346674657972116e-08,
      "loss": 0.6317,
      "step": 11616
    },
    {
      "epoch": 14.219094247246023,
      "grad_norm": 2.3299915232780894,
      "learning_rate": 3.336231470736639e-08,
      "loss": 0.4435,
      "step": 11617
    },
    {
      "epoch": 14.2203182374541,
      "grad_norm": 1.1982419714059898,
      "learning_rate": 3.3258044933316415e-08,
      "loss": 0.6674,
      "step": 11618
    },
    {
      "epoch": 14.221542227662178,
      "grad_norm": 1.6924698154810476,
      "learning_rate": 3.315393726442406e-08,
      "loss": 0.5466,
      "step": 11619
    },
    {
      "epoch": 14.222766217870257,
      "grad_norm": 2.006685676341781,
      "learning_rate": 3.304999170753054e-08,
      "loss": 0.4751,
      "step": 11620
    },
    {
      "epoch": 14.223990208078336,
      "grad_norm": 2.367508537765273,
      "learning_rate": 3.2946208269467316e-08,
      "loss": 0.4839,
      "step": 11621
    },
    {
      "epoch": 14.225214198286414,
      "grad_norm": 1.0091807194140585,
      "learning_rate": 3.284258695705394e-08,
      "loss": 0.4248,
      "step": 11622
    },
    {
      "epoch": 14.226438188494493,
      "grad_norm": 1.8851653798644707,
      "learning_rate": 3.273912777710081e-08,
      "loss": 0.4416,
      "step": 11623
    },
    {
      "epoch": 14.22766217870257,
      "grad_norm": 1.4621903372359608,
      "learning_rate": 3.2635830736406084e-08,
      "loss": 1.1437,
      "step": 11624
    },
    {
      "epoch": 14.228886168910648,
      "grad_norm": 2.246366029496762,
      "learning_rate": 3.25326958417585e-08,
      "loss": 0.5078,
      "step": 11625
    },
    {
      "epoch": 14.230110159118727,
      "grad_norm": 2.7910140967340133,
      "learning_rate": 3.2429723099935416e-08,
      "loss": 0.8857,
      "step": 11626
    },
    {
      "epoch": 14.231334149326806,
      "grad_norm": 2.7582763345124524,
      "learning_rate": 3.23269125177042e-08,
      "loss": 0.942,
      "step": 11627
    },
    {
      "epoch": 14.232558139534884,
      "grad_norm": 1.2417303717553203,
      "learning_rate": 3.222426410182111e-08,
      "loss": 0.8412,
      "step": 11628
    },
    {
      "epoch": 14.233782129742963,
      "grad_norm": 1.7119093260851967,
      "learning_rate": 3.2121777859031586e-08,
      "loss": 0.5125,
      "step": 11629
    },
    {
      "epoch": 14.23500611995104,
      "grad_norm": 1.2008851054970762,
      "learning_rate": 3.20194537960708e-08,
      "loss": 0.532,
      "step": 11630
    },
    {
      "epoch": 14.236230110159118,
      "grad_norm": 1.8617456707875482,
      "learning_rate": 3.191729191966308e-08,
      "loss": 0.2887,
      "step": 11631
    },
    {
      "epoch": 14.237454100367197,
      "grad_norm": 1.0472858500837887,
      "learning_rate": 3.1815292236521964e-08,
      "loss": 0.507,
      "step": 11632
    },
    {
      "epoch": 14.238678090575275,
      "grad_norm": 2.66187275135374,
      "learning_rate": 3.171345475335069e-08,
      "loss": 0.9408,
      "step": 11633
    },
    {
      "epoch": 14.239902080783354,
      "grad_norm": 1.9635684735716827,
      "learning_rate": 3.161177947684169e-08,
      "loss": 0.4794,
      "step": 11634
    },
    {
      "epoch": 14.241126070991433,
      "grad_norm": 2.083226513637041,
      "learning_rate": 3.1510266413676824e-08,
      "loss": 0.4935,
      "step": 11635
    },
    {
      "epoch": 14.24235006119951,
      "grad_norm": 2.8529470991982198,
      "learning_rate": 3.1408915570526886e-08,
      "loss": 0.818,
      "step": 11636
    },
    {
      "epoch": 14.243574051407588,
      "grad_norm": 2.651350017845024,
      "learning_rate": 3.1307726954052096e-08,
      "loss": 0.5419,
      "step": 11637
    },
    {
      "epoch": 14.244798041615667,
      "grad_norm": 2.2751694882032103,
      "learning_rate": 3.1206700570902416e-08,
      "loss": 1.0085,
      "step": 11638
    },
    {
      "epoch": 14.246022031823745,
      "grad_norm": 2.1477813456902917,
      "learning_rate": 3.110583642771725e-08,
      "loss": 0.5855,
      "step": 11639
    },
    {
      "epoch": 14.247246022031824,
      "grad_norm": 1.2614400898889264,
      "learning_rate": 3.100513453112436e-08,
      "loss": 0.5345,
      "step": 11640
    },
    {
      "epoch": 14.248470012239903,
      "grad_norm": 2.061670784508604,
      "learning_rate": 3.090459488774206e-08,
      "loss": 1.0281,
      "step": 11641
    },
    {
      "epoch": 14.249694002447981,
      "grad_norm": 1.1692629557981942,
      "learning_rate": 3.080421750417728e-08,
      "loss": 0.6359,
      "step": 11642
    },
    {
      "epoch": 14.250917992656058,
      "grad_norm": 1.8710895142469839,
      "learning_rate": 3.0704002387026144e-08,
      "loss": 0.4581,
      "step": 11643
    },
    {
      "epoch": 14.252141982864137,
      "grad_norm": 2.0934255053045745,
      "learning_rate": 3.0603949542875044e-08,
      "loss": 0.7678,
      "step": 11644
    },
    {
      "epoch": 14.253365973072215,
      "grad_norm": 1.1561156403568145,
      "learning_rate": 3.050405897829817e-08,
      "loss": 0.3584,
      "step": 11645
    },
    {
      "epoch": 14.254589963280294,
      "grad_norm": 1.8956602501359068,
      "learning_rate": 3.0404330699861097e-08,
      "loss": 0.7012,
      "step": 11646
    },
    {
      "epoch": 14.255813953488373,
      "grad_norm": 1.1679160510875775,
      "learning_rate": 3.030476471411664e-08,
      "loss": 0.5618,
      "step": 11647
    },
    {
      "epoch": 14.257037943696451,
      "grad_norm": 3.326333466463565,
      "learning_rate": 3.020536102760846e-08,
      "loss": 0.4488,
      "step": 11648
    },
    {
      "epoch": 14.258261933904528,
      "grad_norm": 3.1786477860261986,
      "learning_rate": 3.0106119646868825e-08,
      "loss": 0.6886,
      "step": 11649
    },
    {
      "epoch": 14.259485924112607,
      "grad_norm": 1.3481901641623955,
      "learning_rate": 3.000704057841919e-08,
      "loss": 0.5045,
      "step": 11650
    },
    {
      "epoch": 14.260709914320685,
      "grad_norm": 1.6214457424078033,
      "learning_rate": 2.990812382877128e-08,
      "loss": 0.9166,
      "step": 11651
    },
    {
      "epoch": 14.261933904528764,
      "grad_norm": 1.3075330202586593,
      "learning_rate": 2.980936940442547e-08,
      "loss": 0.9705,
      "step": 11652
    },
    {
      "epoch": 14.263157894736842,
      "grad_norm": 1.7263405859211527,
      "learning_rate": 2.9710777311871e-08,
      "loss": 1.225,
      "step": 11653
    },
    {
      "epoch": 14.264381884944921,
      "grad_norm": 2.7406729117748556,
      "learning_rate": 2.961234755758713e-08,
      "loss": 0.3837,
      "step": 11654
    },
    {
      "epoch": 14.265605875152998,
      "grad_norm": 2.0452176854971413,
      "learning_rate": 2.951408014804258e-08,
      "loss": 0.5401,
      "step": 11655
    },
    {
      "epoch": 14.266829865361077,
      "grad_norm": 1.9630203490975757,
      "learning_rate": 2.9415975089694958e-08,
      "loss": 0.6869,
      "step": 11656
    },
    {
      "epoch": 14.268053855569155,
      "grad_norm": 2.6266351401127044,
      "learning_rate": 2.9318032388991336e-08,
      "loss": 0.5979,
      "step": 11657
    },
    {
      "epoch": 14.269277845777234,
      "grad_norm": 1.0761070303267677,
      "learning_rate": 2.9220252052368226e-08,
      "loss": 0.6256,
      "step": 11658
    },
    {
      "epoch": 14.270501835985312,
      "grad_norm": 2.758011128934539,
      "learning_rate": 2.91226340862516e-08,
      "loss": 0.4539,
      "step": 11659
    },
    {
      "epoch": 14.271725826193391,
      "grad_norm": 2.0305520162915047,
      "learning_rate": 2.902517849705605e-08,
      "loss": 0.9523,
      "step": 11660
    },
    {
      "epoch": 14.27294981640147,
      "grad_norm": 1.0392629944070975,
      "learning_rate": 2.8927885291186454e-08,
      "loss": 0.3421,
      "step": 11661
    },
    {
      "epoch": 14.274173806609546,
      "grad_norm": 1.566445192548322,
      "learning_rate": 2.8830754475036306e-08,
      "loss": 0.5453,
      "step": 11662
    },
    {
      "epoch": 14.275397796817625,
      "grad_norm": 1.8481900540335332,
      "learning_rate": 2.8733786054988554e-08,
      "loss": 0.4956,
      "step": 11663
    },
    {
      "epoch": 14.276621787025704,
      "grad_norm": 2.023655132184568,
      "learning_rate": 2.8636980037416162e-08,
      "loss": 0.409,
      "step": 11664
    },
    {
      "epoch": 14.277845777233782,
      "grad_norm": 1.8633761956167059,
      "learning_rate": 2.8540336428680427e-08,
      "loss": 0.4528,
      "step": 11665
    },
    {
      "epoch": 14.279069767441861,
      "grad_norm": 2.155644878627489,
      "learning_rate": 2.844385523513238e-08,
      "loss": 0.4007,
      "step": 11666
    },
    {
      "epoch": 14.28029375764994,
      "grad_norm": 3.0185114540935554,
      "learning_rate": 2.8347536463112502e-08,
      "loss": 0.4122,
      "step": 11667
    },
    {
      "epoch": 14.281517747858016,
      "grad_norm": 1.923739816677876,
      "learning_rate": 2.8251380118950456e-08,
      "loss": 0.5188,
      "step": 11668
    },
    {
      "epoch": 14.282741738066095,
      "grad_norm": 2.190569808509688,
      "learning_rate": 2.815538620896535e-08,
      "loss": 0.5384,
      "step": 11669
    },
    {
      "epoch": 14.283965728274174,
      "grad_norm": 1.8312989130086386,
      "learning_rate": 2.8059554739465756e-08,
      "loss": 0.9797,
      "step": 11670
    },
    {
      "epoch": 14.285189718482252,
      "grad_norm": 1.6146703087694012,
      "learning_rate": 2.7963885716748852e-08,
      "loss": 0.5304,
      "step": 11671
    },
    {
      "epoch": 14.286413708690331,
      "grad_norm": 1.4179713287610263,
      "learning_rate": 2.7868379147102108e-08,
      "loss": 0.6746,
      "step": 11672
    },
    {
      "epoch": 14.28763769889841,
      "grad_norm": 1.3252836946576443,
      "learning_rate": 2.77730350368019e-08,
      "loss": 0.3982,
      "step": 11673
    },
    {
      "epoch": 14.288861689106486,
      "grad_norm": 1.3988438935532834,
      "learning_rate": 2.7677853392113486e-08,
      "loss": 0.384,
      "step": 11674
    },
    {
      "epoch": 14.290085679314565,
      "grad_norm": 1.5162230799891747,
      "learning_rate": 2.758283421929214e-08,
      "loss": 0.789,
      "step": 11675
    },
    {
      "epoch": 14.291309669522644,
      "grad_norm": 2.5323268335299303,
      "learning_rate": 2.7487977524582043e-08,
      "loss": 0.3862,
      "step": 11676
    },
    {
      "epoch": 14.292533659730722,
      "grad_norm": 2.6920520886903447,
      "learning_rate": 2.739328331421681e-08,
      "loss": 0.4998,
      "step": 11677
    },
    {
      "epoch": 14.2937576499388,
      "grad_norm": 1.705254791363702,
      "learning_rate": 2.7298751594419527e-08,
      "loss": 0.3803,
      "step": 11678
    },
    {
      "epoch": 14.29498164014688,
      "grad_norm": 1.1733221769327755,
      "learning_rate": 2.7204382371402437e-08,
      "loss": 0.3651,
      "step": 11679
    },
    {
      "epoch": 14.296205630354958,
      "grad_norm": 2.015471664894745,
      "learning_rate": 2.71101756513667e-08,
      "loss": 0.4745,
      "step": 11680
    },
    {
      "epoch": 14.297429620563035,
      "grad_norm": 2.390686038277476,
      "learning_rate": 2.7016131440504302e-08,
      "loss": 0.296,
      "step": 11681
    },
    {
      "epoch": 14.298653610771114,
      "grad_norm": 2.031379861002272,
      "learning_rate": 2.6922249744994466e-08,
      "loss": 0.3875,
      "step": 11682
    },
    {
      "epoch": 14.299877600979192,
      "grad_norm": 2.165849881958494,
      "learning_rate": 2.6828530571006983e-08,
      "loss": 0.6617,
      "step": 11683
    },
    {
      "epoch": 14.30110159118727,
      "grad_norm": 1.5262334771236559,
      "learning_rate": 2.6734973924700815e-08,
      "loss": 0.6095,
      "step": 11684
    },
    {
      "epoch": 14.30232558139535,
      "grad_norm": 1.2255133684029256,
      "learning_rate": 2.6641579812224373e-08,
      "loss": 0.4723,
      "step": 11685
    },
    {
      "epoch": 14.303549571603428,
      "grad_norm": 1.434240934499408,
      "learning_rate": 2.6548348239714972e-08,
      "loss": 1.5531,
      "step": 11686
    },
    {
      "epoch": 14.304773561811505,
      "grad_norm": 1.0822822409555777,
      "learning_rate": 2.6455279213299655e-08,
      "loss": 0.3941,
      "step": 11687
    },
    {
      "epoch": 14.305997552019583,
      "grad_norm": 3.3835073281953574,
      "learning_rate": 2.6362372739094365e-08,
      "loss": 0.3683,
      "step": 11688
    },
    {
      "epoch": 14.307221542227662,
      "grad_norm": 1.3153805424307832,
      "learning_rate": 2.626962882320422e-08,
      "loss": 0.4445,
      "step": 11689
    },
    {
      "epoch": 14.30844553243574,
      "grad_norm": 1.6171247350862432,
      "learning_rate": 2.6177047471724337e-08,
      "loss": 0.5649,
      "step": 11690
    },
    {
      "epoch": 14.30966952264382,
      "grad_norm": 1.676473528603012,
      "learning_rate": 2.60846286907393e-08,
      "loss": 0.6957,
      "step": 11691
    },
    {
      "epoch": 14.310893512851898,
      "grad_norm": 1.203470415982053,
      "learning_rate": 2.5992372486321748e-08,
      "loss": 0.6893,
      "step": 11692
    },
    {
      "epoch": 14.312117503059975,
      "grad_norm": 2.6076008675748636,
      "learning_rate": 2.5900278864534888e-08,
      "loss": 0.9651,
      "step": 11693
    },
    {
      "epoch": 14.313341493268053,
      "grad_norm": 2.2134449266285725,
      "learning_rate": 2.5808347831430547e-08,
      "loss": 0.9627,
      "step": 11694
    },
    {
      "epoch": 14.314565483476132,
      "grad_norm": 1.601688109821816,
      "learning_rate": 2.571657939305028e-08,
      "loss": 0.5187,
      "step": 11695
    },
    {
      "epoch": 14.31578947368421,
      "grad_norm": 2.070009315730931,
      "learning_rate": 2.5624973555424816e-08,
      "loss": 0.4734,
      "step": 11696
    },
    {
      "epoch": 14.31701346389229,
      "grad_norm": 2.9623710786535082,
      "learning_rate": 2.5533530324573507e-08,
      "loss": 0.4923,
      "step": 11697
    },
    {
      "epoch": 14.318237454100368,
      "grad_norm": 2.661553502018858,
      "learning_rate": 2.544224970650655e-08,
      "loss": 0.5678,
      "step": 11698
    },
    {
      "epoch": 14.319461444308445,
      "grad_norm": 1.428219079259293,
      "learning_rate": 2.5351131707222197e-08,
      "loss": 0.4956,
      "step": 11699
    },
    {
      "epoch": 14.320685434516523,
      "grad_norm": 2.6141243373854324,
      "learning_rate": 2.5260176332708443e-08,
      "loss": 0.2217,
      "step": 11700
    },
    {
      "epoch": 14.321909424724602,
      "grad_norm": 1.878326016154184,
      "learning_rate": 2.5169383588942443e-08,
      "loss": 1.4938,
      "step": 11701
    },
    {
      "epoch": 14.32313341493268,
      "grad_norm": 1.8312593042792251,
      "learning_rate": 2.507875348189054e-08,
      "loss": 0.5722,
      "step": 11702
    },
    {
      "epoch": 14.32435740514076,
      "grad_norm": 1.536563667054103,
      "learning_rate": 2.49882860175088e-08,
      "loss": 1.0675,
      "step": 11703
    },
    {
      "epoch": 14.325581395348838,
      "grad_norm": 1.425857044544653,
      "learning_rate": 2.489798120174275e-08,
      "loss": 0.3611,
      "step": 11704
    },
    {
      "epoch": 14.326805385556916,
      "grad_norm": 1.9717003876714219,
      "learning_rate": 2.4807839040526527e-08,
      "loss": 0.9571,
      "step": 11705
    },
    {
      "epoch": 14.328029375764993,
      "grad_norm": 1.9549049337500595,
      "learning_rate": 2.4717859539784006e-08,
      "loss": 0.4509,
      "step": 11706
    },
    {
      "epoch": 14.329253365973072,
      "grad_norm": 3.000171286213234,
      "learning_rate": 2.462804270542851e-08,
      "loss": 0.951,
      "step": 11707
    },
    {
      "epoch": 14.33047735618115,
      "grad_norm": 2.096143845073606,
      "learning_rate": 2.453838854336199e-08,
      "loss": 0.7605,
      "step": 11708
    },
    {
      "epoch": 14.331701346389229,
      "grad_norm": 2.183883198097502,
      "learning_rate": 2.444889705947667e-08,
      "loss": 0.4002,
      "step": 11709
    },
    {
      "epoch": 14.332925336597308,
      "grad_norm": 1.5109545455747402,
      "learning_rate": 2.435956825965341e-08,
      "loss": 0.6144,
      "step": 11710
    },
    {
      "epoch": 14.334149326805386,
      "grad_norm": 1.7705403756815545,
      "learning_rate": 2.4270402149762506e-08,
      "loss": 1.3939,
      "step": 11711
    },
    {
      "epoch": 14.335373317013463,
      "grad_norm": 1.255307885829857,
      "learning_rate": 2.4181398735663442e-08,
      "loss": 0.5066,
      "step": 11712
    },
    {
      "epoch": 14.336597307221542,
      "grad_norm": 2.1887185896535266,
      "learning_rate": 2.4092558023205703e-08,
      "loss": 0.8053,
      "step": 11713
    },
    {
      "epoch": 14.33782129742962,
      "grad_norm": 2.947993249459628,
      "learning_rate": 2.40038800182274e-08,
      "loss": 0.347,
      "step": 11714
    },
    {
      "epoch": 14.339045287637699,
      "grad_norm": 1.7214416567497126,
      "learning_rate": 2.391536472655581e-08,
      "loss": 0.5546,
      "step": 11715
    },
    {
      "epoch": 14.340269277845778,
      "grad_norm": 1.094444799582268,
      "learning_rate": 2.3827012154007957e-08,
      "loss": 0.587,
      "step": 11716
    },
    {
      "epoch": 14.341493268053856,
      "grad_norm": 1.9003232496011773,
      "learning_rate": 2.3738822306390575e-08,
      "loss": 0.6058,
      "step": 11717
    },
    {
      "epoch": 14.342717258261933,
      "grad_norm": 2.0278507527153016,
      "learning_rate": 2.3650795189498478e-08,
      "loss": 1.1802,
      "step": 11718
    },
    {
      "epoch": 14.343941248470012,
      "grad_norm": 2.2058271188254324,
      "learning_rate": 2.356293080911648e-08,
      "loss": 1.0622,
      "step": 11719
    },
    {
      "epoch": 14.34516523867809,
      "grad_norm": 1.3933374008279338,
      "learning_rate": 2.3475229171019132e-08,
      "loss": 0.4582,
      "step": 11720
    },
    {
      "epoch": 14.346389228886169,
      "grad_norm": 1.6051290188013752,
      "learning_rate": 2.3387690280969876e-08,
      "loss": 0.9236,
      "step": 11721
    },
    {
      "epoch": 14.347613219094248,
      "grad_norm": 2.463631406373448,
      "learning_rate": 2.330031414472106e-08,
      "loss": 0.581,
      "step": 11722
    },
    {
      "epoch": 14.348837209302326,
      "grad_norm": 2.5316313477642587,
      "learning_rate": 2.3213100768015308e-08,
      "loss": 0.9192,
      "step": 11723
    },
    {
      "epoch": 14.350061199510403,
      "grad_norm": 1.652838283588307,
      "learning_rate": 2.3126050156583034e-08,
      "loss": 0.4709,
      "step": 11724
    },
    {
      "epoch": 14.351285189718482,
      "grad_norm": 2.0461511259403755,
      "learning_rate": 2.303916231614578e-08,
      "loss": 0.5108,
      "step": 11725
    },
    {
      "epoch": 14.35250917992656,
      "grad_norm": 2.6728897673054326,
      "learning_rate": 2.2952437252412863e-08,
      "loss": 0.6141,
      "step": 11726
    },
    {
      "epoch": 14.353733170134639,
      "grad_norm": 1.9605931347826409,
      "learning_rate": 2.286587497108389e-08,
      "loss": 0.6778,
      "step": 11727
    },
    {
      "epoch": 14.354957160342718,
      "grad_norm": 1.411126660408548,
      "learning_rate": 2.2779475477847368e-08,
      "loss": 0.4454,
      "step": 11728
    },
    {
      "epoch": 14.356181150550796,
      "grad_norm": 2.3295372341044795,
      "learning_rate": 2.2693238778381254e-08,
      "loss": 0.8104,
      "step": 11729
    },
    {
      "epoch": 14.357405140758875,
      "grad_norm": 1.5240227712258276,
      "learning_rate": 2.2607164878352405e-08,
      "loss": 1.3019,
      "step": 11730
    },
    {
      "epoch": 14.358629130966952,
      "grad_norm": 1.2428271538947602,
      "learning_rate": 2.2521253783417406e-08,
      "loss": 0.5789,
      "step": 11731
    },
    {
      "epoch": 14.35985312117503,
      "grad_norm": 2.326192198932301,
      "learning_rate": 2.2435505499222023e-08,
      "loss": 0.476,
      "step": 11732
    },
    {
      "epoch": 14.361077111383109,
      "grad_norm": 3.8985033997031717,
      "learning_rate": 2.2349920031401463e-08,
      "loss": 0.4136,
      "step": 11733
    },
    {
      "epoch": 14.362301101591187,
      "grad_norm": 2.005983974230653,
      "learning_rate": 2.226449738558012e-08,
      "loss": 0.3536,
      "step": 11734
    },
    {
      "epoch": 14.363525091799266,
      "grad_norm": 1.3399811877289287,
      "learning_rate": 2.217923756737128e-08,
      "loss": 0.4874,
      "step": 11735
    },
    {
      "epoch": 14.364749082007345,
      "grad_norm": 1.3984865811348874,
      "learning_rate": 2.209414058237824e-08,
      "loss": 0.5025,
      "step": 11736
    },
    {
      "epoch": 14.365973072215422,
      "grad_norm": 2.2487304932718466,
      "learning_rate": 2.200920643619292e-08,
      "loss": 0.8143,
      "step": 11737
    },
    {
      "epoch": 14.3671970624235,
      "grad_norm": 2.4424634801401615,
      "learning_rate": 2.1924435134397237e-08,
      "loss": 0.4624,
      "step": 11738
    },
    {
      "epoch": 14.368421052631579,
      "grad_norm": 1.8771345416346272,
      "learning_rate": 2.1839826682562015e-08,
      "loss": 0.4934,
      "step": 11739
    },
    {
      "epoch": 14.369645042839657,
      "grad_norm": 1.242761941608854,
      "learning_rate": 2.175538108624725e-08,
      "loss": 0.8384,
      "step": 11740
    },
    {
      "epoch": 14.370869033047736,
      "grad_norm": 1.0988167645982703,
      "learning_rate": 2.1671098351002394e-08,
      "loss": 0.5038,
      "step": 11741
    },
    {
      "epoch": 14.372093023255815,
      "grad_norm": 1.7175959759758357,
      "learning_rate": 2.1586978482366072e-08,
      "loss": 0.5412,
      "step": 11742
    },
    {
      "epoch": 14.373317013463891,
      "grad_norm": 1.8701375613710491,
      "learning_rate": 2.1503021485866914e-08,
      "loss": 1.003,
      "step": 11743
    },
    {
      "epoch": 14.37454100367197,
      "grad_norm": 1.2633508770820023,
      "learning_rate": 2.1419227367021624e-08,
      "loss": 0.4675,
      "step": 11744
    },
    {
      "epoch": 14.375764993880049,
      "grad_norm": 3.6372614061318895,
      "learning_rate": 2.13355961313369e-08,
      "loss": 0.4024,
      "step": 11745
    },
    {
      "epoch": 14.376988984088127,
      "grad_norm": 2.522834002658433,
      "learning_rate": 2.1252127784309183e-08,
      "loss": 0.6006,
      "step": 11746
    },
    {
      "epoch": 14.378212974296206,
      "grad_norm": 1.1406569759017657,
      "learning_rate": 2.1168822331423255e-08,
      "loss": 0.5339,
      "step": 11747
    },
    {
      "epoch": 14.379436964504285,
      "grad_norm": 1.9533712713502616,
      "learning_rate": 2.1085679778153622e-08,
      "loss": 0.9957,
      "step": 11748
    },
    {
      "epoch": 14.380660954712361,
      "grad_norm": 2.5569573918808204,
      "learning_rate": 2.1002700129964527e-08,
      "loss": 0.5301,
      "step": 11749
    },
    {
      "epoch": 14.38188494492044,
      "grad_norm": 2.598140699947736,
      "learning_rate": 2.0919883392308548e-08,
      "loss": 0.5079,
      "step": 11750
    },
    {
      "epoch": 14.383108935128519,
      "grad_norm": 1.5535158980739432,
      "learning_rate": 2.0837229570628283e-08,
      "loss": 0.8665,
      "step": 11751
    },
    {
      "epoch": 14.384332925336597,
      "grad_norm": 3.010698931818374,
      "learning_rate": 2.0754738670355767e-08,
      "loss": 0.8149,
      "step": 11752
    },
    {
      "epoch": 14.385556915544676,
      "grad_norm": 1.6567035073889302,
      "learning_rate": 2.067241069691167e-08,
      "loss": 0.3504,
      "step": 11753
    },
    {
      "epoch": 14.386780905752754,
      "grad_norm": 1.337612912356196,
      "learning_rate": 2.0590245655706386e-08,
      "loss": 0.598,
      "step": 11754
    },
    {
      "epoch": 14.388004895960833,
      "grad_norm": 2.458712013143854,
      "learning_rate": 2.0508243552139206e-08,
      "loss": 0.491,
      "step": 11755
    },
    {
      "epoch": 14.38922888616891,
      "grad_norm": 1.5971211357046111,
      "learning_rate": 2.0426404391599432e-08,
      "loss": 1.1109,
      "step": 11756
    },
    {
      "epoch": 14.390452876376989,
      "grad_norm": 1.4175683273732314,
      "learning_rate": 2.0344728179464977e-08,
      "loss": 0.3542,
      "step": 11757
    },
    {
      "epoch": 14.391676866585067,
      "grad_norm": 2.202986575845754,
      "learning_rate": 2.0263214921103503e-08,
      "loss": 0.5061,
      "step": 11758
    },
    {
      "epoch": 14.392900856793146,
      "grad_norm": 1.5589195402199008,
      "learning_rate": 2.018186462187155e-08,
      "loss": 0.5397,
      "step": 11759
    },
    {
      "epoch": 14.394124847001224,
      "grad_norm": 2.210875263394031,
      "learning_rate": 2.01006772871154e-08,
      "loss": 0.7073,
      "step": 11760
    },
    {
      "epoch": 14.395348837209303,
      "grad_norm": 1.4550626386564818,
      "learning_rate": 2.0019652922169953e-08,
      "loss": 1.1252,
      "step": 11761
    },
    {
      "epoch": 14.39657282741738,
      "grad_norm": 1.60376420276922,
      "learning_rate": 1.993879153236039e-08,
      "loss": 0.4212,
      "step": 11762
    },
    {
      "epoch": 14.397796817625458,
      "grad_norm": 2.267791986931107,
      "learning_rate": 1.9858093122999965e-08,
      "loss": 0.4612,
      "step": 11763
    },
    {
      "epoch": 14.399020807833537,
      "grad_norm": 1.1373166688957719,
      "learning_rate": 1.977755769939277e-08,
      "loss": 0.3834,
      "step": 11764
    },
    {
      "epoch": 14.400244798041616,
      "grad_norm": 2.7349403066868194,
      "learning_rate": 1.9697185266830397e-08,
      "loss": 0.7153,
      "step": 11765
    },
    {
      "epoch": 14.401468788249694,
      "grad_norm": 1.996663612541804,
      "learning_rate": 1.961697583059502e-08,
      "loss": 1.0988,
      "step": 11766
    },
    {
      "epoch": 14.402692778457773,
      "grad_norm": 2.707934784550659,
      "learning_rate": 1.9536929395957693e-08,
      "loss": 0.3385,
      "step": 11767
    },
    {
      "epoch": 14.403916768665852,
      "grad_norm": 3.08854396899621,
      "learning_rate": 1.9457045968178655e-08,
      "loss": 0.3931,
      "step": 11768
    },
    {
      "epoch": 14.405140758873928,
      "grad_norm": 1.7061962053028807,
      "learning_rate": 1.9377325552507875e-08,
      "loss": 0.5041,
      "step": 11769
    },
    {
      "epoch": 14.406364749082007,
      "grad_norm": 1.6234895462022758,
      "learning_rate": 1.9297768154184217e-08,
      "loss": 1.5378,
      "step": 11770
    },
    {
      "epoch": 14.407588739290086,
      "grad_norm": 0.8689854084722297,
      "learning_rate": 1.9218373778435438e-08,
      "loss": 0.3484,
      "step": 11771
    },
    {
      "epoch": 14.408812729498164,
      "grad_norm": 2.0296730210407166,
      "learning_rate": 1.9139142430479596e-08,
      "loss": 1.0414,
      "step": 11772
    },
    {
      "epoch": 14.410036719706243,
      "grad_norm": 2.4730986206468595,
      "learning_rate": 1.9060074115523075e-08,
      "loss": 0.4559,
      "step": 11773
    },
    {
      "epoch": 14.411260709914322,
      "grad_norm": 1.368271376042541,
      "learning_rate": 1.8981168838762276e-08,
      "loss": 0.39,
      "step": 11774
    },
    {
      "epoch": 14.412484700122398,
      "grad_norm": 1.764557077058254,
      "learning_rate": 1.890242660538222e-08,
      "loss": 0.6112,
      "step": 11775
    },
    {
      "epoch": 14.413708690330477,
      "grad_norm": 2.8675963537706797,
      "learning_rate": 1.8823847420557928e-08,
      "loss": 0.4314,
      "step": 11776
    },
    {
      "epoch": 14.414932680538556,
      "grad_norm": 1.8984365108407846,
      "learning_rate": 1.8745431289453054e-08,
      "loss": 0.5396,
      "step": 11777
    },
    {
      "epoch": 14.416156670746634,
      "grad_norm": 1.1476624182569166,
      "learning_rate": 1.8667178217220972e-08,
      "loss": 0.5751,
      "step": 11778
    },
    {
      "epoch": 14.417380660954713,
      "grad_norm": 2.005957403813264,
      "learning_rate": 1.8589088209003957e-08,
      "loss": 0.5164,
      "step": 11779
    },
    {
      "epoch": 14.418604651162791,
      "grad_norm": 1.2404171614270372,
      "learning_rate": 1.851116126993402e-08,
      "loss": 0.4981,
      "step": 11780
    },
    {
      "epoch": 14.419828641370868,
      "grad_norm": 1.1480485945897576,
      "learning_rate": 1.8433397405132336e-08,
      "loss": 0.6582,
      "step": 11781
    },
    {
      "epoch": 14.421052631578947,
      "grad_norm": 1.8835107570680054,
      "learning_rate": 1.8355796619708988e-08,
      "loss": 0.978,
      "step": 11782
    },
    {
      "epoch": 14.422276621787026,
      "grad_norm": 1.4946091528276997,
      "learning_rate": 1.8278358918763507e-08,
      "loss": 0.7933,
      "step": 11783
    },
    {
      "epoch": 14.423500611995104,
      "grad_norm": 2.0859782857346985,
      "learning_rate": 1.8201084307385152e-08,
      "loss": 0.5897,
      "step": 11784
    },
    {
      "epoch": 14.424724602203183,
      "grad_norm": 1.7032582920274157,
      "learning_rate": 1.812397279065209e-08,
      "loss": 1.2233,
      "step": 11785
    },
    {
      "epoch": 14.425948592411261,
      "grad_norm": 3.686123481091353,
      "learning_rate": 1.804702437363165e-08,
      "loss": 0.3361,
      "step": 11786
    },
    {
      "epoch": 14.427172582619338,
      "grad_norm": 2.2311141265503185,
      "learning_rate": 1.79702390613809e-08,
      "loss": 1.0975,
      "step": 11787
    },
    {
      "epoch": 14.428396572827417,
      "grad_norm": 1.3050973454066233,
      "learning_rate": 1.7893616858945527e-08,
      "loss": 0.7575,
      "step": 11788
    },
    {
      "epoch": 14.429620563035495,
      "grad_norm": 1.613549286430271,
      "learning_rate": 1.7817157771360672e-08,
      "loss": 0.5569,
      "step": 11789
    },
    {
      "epoch": 14.430844553243574,
      "grad_norm": 1.726274143875277,
      "learning_rate": 1.774086180365148e-08,
      "loss": 0.9227,
      "step": 11790
    },
    {
      "epoch": 14.432068543451653,
      "grad_norm": 3.199461748609514,
      "learning_rate": 1.766472896083171e-08,
      "loss": 0.3655,
      "step": 11791
    },
    {
      "epoch": 14.433292533659731,
      "grad_norm": 2.38988802438142,
      "learning_rate": 1.7588759247904318e-08,
      "loss": 0.366,
      "step": 11792
    },
    {
      "epoch": 14.43451652386781,
      "grad_norm": 1.8449225782756353,
      "learning_rate": 1.751295266986197e-08,
      "loss": 0.4016,
      "step": 11793
    },
    {
      "epoch": 14.435740514075887,
      "grad_norm": 1.249866727807637,
      "learning_rate": 1.743730923168624e-08,
      "loss": 0.5958,
      "step": 11794
    },
    {
      "epoch": 14.436964504283965,
      "grad_norm": 1.7649587090711278,
      "learning_rate": 1.7361828938348146e-08,
      "loss": 0.424,
      "step": 11795
    },
    {
      "epoch": 14.438188494492044,
      "grad_norm": 1.901882166204179,
      "learning_rate": 1.728651179480789e-08,
      "loss": 1.584,
      "step": 11796
    },
    {
      "epoch": 14.439412484700123,
      "grad_norm": 1.136078275604049,
      "learning_rate": 1.7211357806015128e-08,
      "loss": 0.6762,
      "step": 11797
    },
    {
      "epoch": 14.440636474908201,
      "grad_norm": 1.9187996126188502,
      "learning_rate": 1.713636697690896e-08,
      "loss": 1.0033,
      "step": 11798
    },
    {
      "epoch": 14.44186046511628,
      "grad_norm": 2.4646574683137126,
      "learning_rate": 1.7061539312417107e-08,
      "loss": 0.4362,
      "step": 11799
    },
    {
      "epoch": 14.443084455324357,
      "grad_norm": 2.512580202228061,
      "learning_rate": 1.698687481745731e-08,
      "loss": 0.4129,
      "step": 11800
    },
    {
      "epoch": 14.444308445532435,
      "grad_norm": 2.1536094623987103,
      "learning_rate": 1.691237349693592e-08,
      "loss": 0.5839,
      "step": 11801
    },
    {
      "epoch": 14.445532435740514,
      "grad_norm": 1.3789985928296056,
      "learning_rate": 1.6838035355749017e-08,
      "loss": 0.5712,
      "step": 11802
    },
    {
      "epoch": 14.446756425948593,
      "grad_norm": 3.4174334077344595,
      "learning_rate": 1.6763860398781863e-08,
      "loss": 0.334,
      "step": 11803
    },
    {
      "epoch": 14.447980416156671,
      "grad_norm": 1.7336034070909205,
      "learning_rate": 1.6689848630909166e-08,
      "loss": 0.5473,
      "step": 11804
    },
    {
      "epoch": 14.44920440636475,
      "grad_norm": 2.375920381884651,
      "learning_rate": 1.6616000056994264e-08,
      "loss": 0.4302,
      "step": 11805
    },
    {
      "epoch": 14.450428396572827,
      "grad_norm": 2.4810399807205457,
      "learning_rate": 1.654231468189049e-08,
      "loss": 0.5688,
      "step": 11806
    },
    {
      "epoch": 14.451652386780905,
      "grad_norm": 1.4394338433680323,
      "learning_rate": 1.6468792510440368e-08,
      "loss": 0.7348,
      "step": 11807
    },
    {
      "epoch": 14.452876376988984,
      "grad_norm": 1.8965097914352775,
      "learning_rate": 1.6395433547475025e-08,
      "loss": 0.5376,
      "step": 11808
    },
    {
      "epoch": 14.454100367197062,
      "grad_norm": 2.1341089191605405,
      "learning_rate": 1.6322237797815332e-08,
      "loss": 1.0636,
      "step": 11809
    },
    {
      "epoch": 14.455324357405141,
      "grad_norm": 2.0684190606030652,
      "learning_rate": 1.624920526627216e-08,
      "loss": 0.526,
      "step": 11810
    },
    {
      "epoch": 14.45654834761322,
      "grad_norm": 1.7125353162899737,
      "learning_rate": 1.6176335957644174e-08,
      "loss": 0.503,
      "step": 11811
    },
    {
      "epoch": 14.457772337821297,
      "grad_norm": 1.4007636509966792,
      "learning_rate": 1.6103629876720318e-08,
      "loss": 0.8919,
      "step": 11812
    },
    {
      "epoch": 14.458996328029375,
      "grad_norm": 3.234711150044288,
      "learning_rate": 1.6031087028278712e-08,
      "loss": 0.3323,
      "step": 11813
    },
    {
      "epoch": 14.460220318237454,
      "grad_norm": 1.4460708387005283,
      "learning_rate": 1.595870741708666e-08,
      "loss": 0.6733,
      "step": 11814
    },
    {
      "epoch": 14.461444308445532,
      "grad_norm": 1.4269761311710913,
      "learning_rate": 1.5886491047900076e-08,
      "loss": 0.6901,
      "step": 11815
    },
    {
      "epoch": 14.462668298653611,
      "grad_norm": 3.9653498934754503,
      "learning_rate": 1.581443792546544e-08,
      "loss": 0.3425,
      "step": 11816
    },
    {
      "epoch": 14.46389228886169,
      "grad_norm": 1.308485454500981,
      "learning_rate": 1.574254805451786e-08,
      "loss": 0.5265,
      "step": 11817
    },
    {
      "epoch": 14.465116279069768,
      "grad_norm": 1.266865276911998,
      "learning_rate": 1.567082143978105e-08,
      "loss": 0.474,
      "step": 11818
    },
    {
      "epoch": 14.466340269277845,
      "grad_norm": 3.1993644843224747,
      "learning_rate": 1.5599258085969016e-08,
      "loss": 0.4501,
      "step": 11819
    },
    {
      "epoch": 14.467564259485924,
      "grad_norm": 2.712878966537097,
      "learning_rate": 1.5527857997784668e-08,
      "loss": 0.3871,
      "step": 11820
    },
    {
      "epoch": 14.468788249694002,
      "grad_norm": 1.6318394739688387,
      "learning_rate": 1.5456621179920084e-08,
      "loss": 0.506,
      "step": 11821
    },
    {
      "epoch": 14.470012239902081,
      "grad_norm": 1.7845639001870235,
      "learning_rate": 1.5385547637056518e-08,
      "loss": 0.2996,
      "step": 11822
    },
    {
      "epoch": 14.47123623011016,
      "grad_norm": 0.9936944391624243,
      "learning_rate": 1.531463737386496e-08,
      "loss": 0.4905,
      "step": 11823
    },
    {
      "epoch": 14.472460220318238,
      "grad_norm": 1.1731937438076323,
      "learning_rate": 1.5243890395005012e-08,
      "loss": 0.5994,
      "step": 11824
    },
    {
      "epoch": 14.473684210526315,
      "grad_norm": 1.5332970122576097,
      "learning_rate": 1.517330670512629e-08,
      "loss": 0.7045,
      "step": 11825
    },
    {
      "epoch": 14.474908200734394,
      "grad_norm": 2.467300168526655,
      "learning_rate": 1.5102886308867027e-08,
      "loss": 0.5305,
      "step": 11826
    },
    {
      "epoch": 14.476132190942472,
      "grad_norm": 2.6390742502230475,
      "learning_rate": 1.5032629210855188e-08,
      "loss": 0.3123,
      "step": 11827
    },
    {
      "epoch": 14.477356181150551,
      "grad_norm": 1.7540093059336526,
      "learning_rate": 1.4962535415707634e-08,
      "loss": 0.5173,
      "step": 11828
    },
    {
      "epoch": 14.47858017135863,
      "grad_norm": 0.972641891192008,
      "learning_rate": 1.4892604928030685e-08,
      "loss": 0.4371,
      "step": 11829
    },
    {
      "epoch": 14.479804161566708,
      "grad_norm": 1.5555448608481006,
      "learning_rate": 1.4822837752420106e-08,
      "loss": 1.396,
      "step": 11830
    },
    {
      "epoch": 14.481028151774785,
      "grad_norm": 2.1405433933521834,
      "learning_rate": 1.4753233893460561e-08,
      "loss": 0.4216,
      "step": 11831
    },
    {
      "epoch": 14.482252141982864,
      "grad_norm": 2.9165378635399026,
      "learning_rate": 1.4683793355725895e-08,
      "loss": 0.2022,
      "step": 11832
    },
    {
      "epoch": 14.483476132190942,
      "grad_norm": 2.901719301365012,
      "learning_rate": 1.4614516143780233e-08,
      "loss": 1.051,
      "step": 11833
    },
    {
      "epoch": 14.48470012239902,
      "grad_norm": 3.0236898440651006,
      "learning_rate": 1.454540226217549e-08,
      "loss": 0.7308,
      "step": 11834
    },
    {
      "epoch": 14.4859241126071,
      "grad_norm": 2.3528956915174,
      "learning_rate": 1.4476451715453865e-08,
      "loss": 0.5211,
      "step": 11835
    },
    {
      "epoch": 14.487148102815178,
      "grad_norm": 1.7657759389375858,
      "learning_rate": 1.4407664508146456e-08,
      "loss": 0.9009,
      "step": 11836
    },
    {
      "epoch": 14.488372093023255,
      "grad_norm": 1.4764207347625293,
      "learning_rate": 1.4339040644774094e-08,
      "loss": 0.4252,
      "step": 11837
    },
    {
      "epoch": 14.489596083231334,
      "grad_norm": 1.156354300069392,
      "learning_rate": 1.4270580129845946e-08,
      "loss": 0.6266,
      "step": 11838
    },
    {
      "epoch": 14.490820073439412,
      "grad_norm": 1.5976375105950849,
      "learning_rate": 1.4202282967861192e-08,
      "loss": 0.5217,
      "step": 11839
    },
    {
      "epoch": 14.49204406364749,
      "grad_norm": 1.8725890922041513,
      "learning_rate": 1.4134149163307908e-08,
      "loss": 0.9049,
      "step": 11840
    },
    {
      "epoch": 14.49326805385557,
      "grad_norm": 2.325899575857879,
      "learning_rate": 1.4066178720663904e-08,
      "loss": 0.5483,
      "step": 11841
    },
    {
      "epoch": 14.494492044063648,
      "grad_norm": 1.7233622950815752,
      "learning_rate": 1.399837164439588e-08,
      "loss": 0.7079,
      "step": 11842
    },
    {
      "epoch": 14.495716034271727,
      "grad_norm": 1.6373487622783311,
      "learning_rate": 1.393072793895972e-08,
      "loss": 0.3732,
      "step": 11843
    },
    {
      "epoch": 14.496940024479803,
      "grad_norm": 1.5393461812813325,
      "learning_rate": 1.3863247608800756e-08,
      "loss": 0.7452,
      "step": 11844
    },
    {
      "epoch": 14.498164014687882,
      "grad_norm": 1.5735384743488203,
      "learning_rate": 1.3795930658353495e-08,
      "loss": 0.5683,
      "step": 11845
    },
    {
      "epoch": 14.49938800489596,
      "grad_norm": 1.4071899566763624,
      "learning_rate": 1.3728777092042179e-08,
      "loss": 0.7336,
      "step": 11846
    },
    {
      "epoch": 14.50061199510404,
      "grad_norm": 1.1644197447096472,
      "learning_rate": 1.3661786914279384e-08,
      "loss": 0.4126,
      "step": 11847
    },
    {
      "epoch": 14.501835985312118,
      "grad_norm": 1.9215345178696646,
      "learning_rate": 1.3594960129467427e-08,
      "loss": 0.4465,
      "step": 11848
    },
    {
      "epoch": 14.503059975520197,
      "grad_norm": 1.4384645458134662,
      "learning_rate": 1.352829674199807e-08,
      "loss": 1.5609,
      "step": 11849
    },
    {
      "epoch": 14.504283965728273,
      "grad_norm": 1.4054785686746254,
      "learning_rate": 1.3461796756252531e-08,
      "loss": 0.4362,
      "step": 11850
    },
    {
      "epoch": 14.505507955936352,
      "grad_norm": 2.728758969139984,
      "learning_rate": 1.339546017660065e-08,
      "loss": 0.5048,
      "step": 11851
    },
    {
      "epoch": 14.50673194614443,
      "grad_norm": 1.477308892430505,
      "learning_rate": 1.3329287007401715e-08,
      "loss": 1.3063,
      "step": 11852
    },
    {
      "epoch": 14.50795593635251,
      "grad_norm": 1.6010467500319885,
      "learning_rate": 1.326327725300447e-08,
      "loss": 0.3175,
      "step": 11853
    },
    {
      "epoch": 14.509179926560588,
      "grad_norm": 1.9201227859345615,
      "learning_rate": 1.3197430917746557e-08,
      "loss": 0.7383,
      "step": 11854
    },
    {
      "epoch": 14.510403916768666,
      "grad_norm": 1.4805472450593995,
      "learning_rate": 1.3131748005955903e-08,
      "loss": 0.5399,
      "step": 11855
    },
    {
      "epoch": 14.511627906976745,
      "grad_norm": 1.38502471310643,
      "learning_rate": 1.3066228521948221e-08,
      "loss": 0.6665,
      "step": 11856
    },
    {
      "epoch": 14.512851897184822,
      "grad_norm": 2.5982048097192707,
      "learning_rate": 1.3000872470029513e-08,
      "loss": 0.4914,
      "step": 11857
    },
    {
      "epoch": 14.5140758873929,
      "grad_norm": 1.5884836363063317,
      "learning_rate": 1.2935679854494677e-08,
      "loss": 0.6063,
      "step": 11858
    },
    {
      "epoch": 14.51529987760098,
      "grad_norm": 2.111370198701462,
      "learning_rate": 1.2870650679628061e-08,
      "loss": 0.3139,
      "step": 11859
    },
    {
      "epoch": 14.516523867809058,
      "grad_norm": 3.0783092234918326,
      "learning_rate": 1.2805784949702916e-08,
      "loss": 0.3329,
      "step": 11860
    },
    {
      "epoch": 14.517747858017136,
      "grad_norm": 1.8290989754725602,
      "learning_rate": 1.2741082668982219e-08,
      "loss": 1.0947,
      "step": 11861
    },
    {
      "epoch": 14.518971848225215,
      "grad_norm": 1.6199585479820948,
      "learning_rate": 1.267654384171757e-08,
      "loss": 2.2747,
      "step": 11862
    },
    {
      "epoch": 14.520195838433292,
      "grad_norm": 2.420293015904689,
      "learning_rate": 1.2612168472150577e-08,
      "loss": 0.4903,
      "step": 11863
    },
    {
      "epoch": 14.52141982864137,
      "grad_norm": 1.3422236150452398,
      "learning_rate": 1.254795656451202e-08,
      "loss": 0.77,
      "step": 11864
    },
    {
      "epoch": 14.522643818849449,
      "grad_norm": 1.5390193212206982,
      "learning_rate": 1.2483908123021027e-08,
      "loss": 0.9435,
      "step": 11865
    },
    {
      "epoch": 14.523867809057528,
      "grad_norm": 1.8621453150722713,
      "learning_rate": 1.2420023151886729e-08,
      "loss": 0.5602,
      "step": 11866
    },
    {
      "epoch": 14.525091799265606,
      "grad_norm": 2.8605052832350903,
      "learning_rate": 1.2356301655307711e-08,
      "loss": 0.4965,
      "step": 11867
    },
    {
      "epoch": 14.526315789473685,
      "grad_norm": 1.54081937197799,
      "learning_rate": 1.2292743637471461e-08,
      "loss": 0.8848,
      "step": 11868
    },
    {
      "epoch": 14.527539779681762,
      "grad_norm": 1.5147516651083055,
      "learning_rate": 1.2229349102554633e-08,
      "loss": 0.524,
      "step": 11869
    },
    {
      "epoch": 14.52876376988984,
      "grad_norm": 1.4956979398730028,
      "learning_rate": 1.216611805472362e-08,
      "loss": 1.1114,
      "step": 11870
    },
    {
      "epoch": 14.529987760097919,
      "grad_norm": 1.1498299068457392,
      "learning_rate": 1.2103050498133151e-08,
      "loss": 0.5324,
      "step": 11871
    },
    {
      "epoch": 14.531211750305998,
      "grad_norm": 3.6704461662364696,
      "learning_rate": 1.2040146436928247e-08,
      "loss": 0.4145,
      "step": 11872
    },
    {
      "epoch": 14.532435740514076,
      "grad_norm": 1.690375561608401,
      "learning_rate": 1.1977405875242543e-08,
      "loss": 0.4685,
      "step": 11873
    },
    {
      "epoch": 14.533659730722155,
      "grad_norm": 1.625648613037077,
      "learning_rate": 1.1914828817199131e-08,
      "loss": 0.4713,
      "step": 11874
    },
    {
      "epoch": 14.534883720930232,
      "grad_norm": 1.87470599450634,
      "learning_rate": 1.1852415266910555e-08,
      "loss": 1.0122,
      "step": 11875
    },
    {
      "epoch": 14.53610771113831,
      "grad_norm": 1.1687518117412161,
      "learning_rate": 1.1790165228477979e-08,
      "loss": 0.8442,
      "step": 11876
    },
    {
      "epoch": 14.537331701346389,
      "grad_norm": 2.268592143382809,
      "learning_rate": 1.1728078705992574e-08,
      "loss": 0.4772,
      "step": 11877
    },
    {
      "epoch": 14.538555691554468,
      "grad_norm": 2.388133926293711,
      "learning_rate": 1.166615570353441e-08,
      "loss": 0.3201,
      "step": 11878
    },
    {
      "epoch": 14.539779681762546,
      "grad_norm": 2.7157227569830136,
      "learning_rate": 1.1604396225173008e-08,
      "loss": 0.3893,
      "step": 11879
    },
    {
      "epoch": 14.541003671970625,
      "grad_norm": 1.7440937413547302,
      "learning_rate": 1.1542800274966238e-08,
      "loss": 0.4353,
      "step": 11880
    },
    {
      "epoch": 14.542227662178703,
      "grad_norm": 3.377514025499077,
      "learning_rate": 1.1481367856962799e-08,
      "loss": 0.5826,
      "step": 11881
    },
    {
      "epoch": 14.54345165238678,
      "grad_norm": 3.2201798006659974,
      "learning_rate": 1.1420098975199467e-08,
      "loss": 0.2938,
      "step": 11882
    },
    {
      "epoch": 14.544675642594859,
      "grad_norm": 2.1574853631045228,
      "learning_rate": 1.1358993633702742e-08,
      "loss": 0.3777,
      "step": 11883
    },
    {
      "epoch": 14.545899632802938,
      "grad_norm": 1.5698418038342512,
      "learning_rate": 1.1298051836487744e-08,
      "loss": 1.209,
      "step": 11884
    },
    {
      "epoch": 14.547123623011016,
      "grad_norm": 2.0541512849196963,
      "learning_rate": 1.1237273587559883e-08,
      "loss": 0.6165,
      "step": 11885
    },
    {
      "epoch": 14.548347613219095,
      "grad_norm": 2.025850129381251,
      "learning_rate": 1.1176658890913183e-08,
      "loss": 0.7398,
      "step": 11886
    },
    {
      "epoch": 14.549571603427173,
      "grad_norm": 2.5164853719986344,
      "learning_rate": 1.1116207750530849e-08,
      "loss": 0.4526,
      "step": 11887
    },
    {
      "epoch": 14.55079559363525,
      "grad_norm": 2.5613033165525025,
      "learning_rate": 1.1055920170385815e-08,
      "loss": 0.5504,
      "step": 11888
    },
    {
      "epoch": 14.552019583843329,
      "grad_norm": 2.1078180559131594,
      "learning_rate": 1.0995796154439353e-08,
      "loss": 0.5385,
      "step": 11889
    },
    {
      "epoch": 14.553243574051407,
      "grad_norm": 1.7492266003685322,
      "learning_rate": 1.0935835706643028e-08,
      "loss": 1.0756,
      "step": 11890
    },
    {
      "epoch": 14.554467564259486,
      "grad_norm": 2.5540251074783464,
      "learning_rate": 1.0876038830937296e-08,
      "loss": 1.2731,
      "step": 11891
    },
    {
      "epoch": 14.555691554467565,
      "grad_norm": 1.8211669417351077,
      "learning_rate": 1.0816405531251794e-08,
      "loss": 0.6246,
      "step": 11892
    },
    {
      "epoch": 14.556915544675643,
      "grad_norm": 2.0677350341790794,
      "learning_rate": 1.0756935811505054e-08,
      "loss": 0.6275,
      "step": 11893
    },
    {
      "epoch": 14.55813953488372,
      "grad_norm": 1.7977723299605575,
      "learning_rate": 1.0697629675605614e-08,
      "loss": 0.8195,
      "step": 11894
    },
    {
      "epoch": 14.559363525091799,
      "grad_norm": 2.756020941164699,
      "learning_rate": 1.0638487127450636e-08,
      "loss": 0.3941,
      "step": 11895
    },
    {
      "epoch": 14.560587515299877,
      "grad_norm": 2.7268879397644956,
      "learning_rate": 1.0579508170926734e-08,
      "loss": 0.3765,
      "step": 11896
    },
    {
      "epoch": 14.561811505507956,
      "grad_norm": 2.5307430494653174,
      "learning_rate": 1.0520692809909693e-08,
      "loss": 0.5388,
      "step": 11897
    },
    {
      "epoch": 14.563035495716035,
      "grad_norm": 3.660668411193295,
      "learning_rate": 1.0462041048264759e-08,
      "loss": 0.7531,
      "step": 11898
    },
    {
      "epoch": 14.564259485924113,
      "grad_norm": 2.650056337336155,
      "learning_rate": 1.0403552889846624e-08,
      "loss": 0.8471,
      "step": 11899
    },
    {
      "epoch": 14.56548347613219,
      "grad_norm": 1.787920617403508,
      "learning_rate": 1.0345228338498325e-08,
      "loss": 0.4947,
      "step": 11900
    },
    {
      "epoch": 14.566707466340269,
      "grad_norm": 1.4050997129385918,
      "learning_rate": 1.0287067398053185e-08,
      "loss": 1.3264,
      "step": 11901
    },
    {
      "epoch": 14.567931456548347,
      "grad_norm": 1.556164719124341,
      "learning_rate": 1.0229070072332869e-08,
      "loss": 0.7102,
      "step": 11902
    },
    {
      "epoch": 14.569155446756426,
      "grad_norm": 1.7056940569171484,
      "learning_rate": 1.0171236365149329e-08,
      "loss": 0.4475,
      "step": 11903
    },
    {
      "epoch": 14.570379436964505,
      "grad_norm": 3.093813583841516,
      "learning_rate": 1.0113566280302855e-08,
      "loss": 0.8997,
      "step": 11904
    },
    {
      "epoch": 14.571603427172583,
      "grad_norm": 2.3997118108888413,
      "learning_rate": 1.0056059821583475e-08,
      "loss": 0.5994,
      "step": 11905
    },
    {
      "epoch": 14.572827417380662,
      "grad_norm": 2.1259527008949064,
      "learning_rate": 9.998716992770107e-09,
      "loss": 0.6087,
      "step": 11906
    },
    {
      "epoch": 14.574051407588739,
      "grad_norm": 1.6915588935346049,
      "learning_rate": 9.941537797631129e-09,
      "loss": 1.6306,
      "step": 11907
    },
    {
      "epoch": 14.575275397796817,
      "grad_norm": 2.5364224234102584,
      "learning_rate": 9.884522239924088e-09,
      "loss": 0.7041,
      "step": 11908
    },
    {
      "epoch": 14.576499388004896,
      "grad_norm": 1.616157646868152,
      "learning_rate": 9.827670323395988e-09,
      "loss": 0.7118,
      "step": 11909
    },
    {
      "epoch": 14.577723378212974,
      "grad_norm": 2.0087992094288336,
      "learning_rate": 9.770982051783006e-09,
      "loss": 1.1018,
      "step": 11910
    },
    {
      "epoch": 14.578947368421053,
      "grad_norm": 2.273333147234298,
      "learning_rate": 9.71445742881022e-09,
      "loss": 0.5036,
      "step": 11911
    },
    {
      "epoch": 14.580171358629132,
      "grad_norm": 2.106530816330172,
      "learning_rate": 9.658096458192435e-09,
      "loss": 0.8463,
      "step": 11912
    },
    {
      "epoch": 14.581395348837209,
      "grad_norm": 2.1911559966597234,
      "learning_rate": 9.60189914363363e-09,
      "loss": 0.5279,
      "step": 11913
    },
    {
      "epoch": 14.582619339045287,
      "grad_norm": 1.6590976725928632,
      "learning_rate": 9.545865488826411e-09,
      "loss": 0.5939,
      "step": 11914
    },
    {
      "epoch": 14.583843329253366,
      "grad_norm": 1.5869074758027664,
      "learning_rate": 9.489995497453387e-09,
      "loss": 1.2542,
      "step": 11915
    },
    {
      "epoch": 14.585067319461444,
      "grad_norm": 0.8688571252077006,
      "learning_rate": 9.434289173186062e-09,
      "loss": 0.4307,
      "step": 11916
    },
    {
      "epoch": 14.586291309669523,
      "grad_norm": 1.3842279746139179,
      "learning_rate": 9.378746519685399e-09,
      "loss": 1.0849,
      "step": 11917
    },
    {
      "epoch": 14.587515299877602,
      "grad_norm": 2.421870413006182,
      "learning_rate": 9.323367540601259e-09,
      "loss": 0.2802,
      "step": 11918
    },
    {
      "epoch": 14.588739290085678,
      "grad_norm": 1.8400634518648125,
      "learning_rate": 9.268152239572947e-09,
      "loss": 0.5393,
      "step": 11919
    },
    {
      "epoch": 14.589963280293757,
      "grad_norm": 1.79413541723308,
      "learning_rate": 9.213100620229231e-09,
      "loss": 1.3544,
      "step": 11920
    },
    {
      "epoch": 14.591187270501836,
      "grad_norm": 2.710439080452024,
      "learning_rate": 9.158212686187773e-09,
      "loss": 0.9037,
      "step": 11921
    },
    {
      "epoch": 14.592411260709914,
      "grad_norm": 1.8426638880890254,
      "learning_rate": 9.103488441055408e-09,
      "loss": 0.9238,
      "step": 11922
    },
    {
      "epoch": 14.593635250917993,
      "grad_norm": 2.1596562994816866,
      "learning_rate": 9.048927888428981e-09,
      "loss": 1.1979,
      "step": 11923
    },
    {
      "epoch": 14.594859241126072,
      "grad_norm": 3.234753870796488,
      "learning_rate": 8.994531031893406e-09,
      "loss": 0.5026,
      "step": 11924
    },
    {
      "epoch": 14.596083231334148,
      "grad_norm": 2.232730173339315,
      "learning_rate": 8.940297875023596e-09,
      "loss": 0.3686,
      "step": 11925
    },
    {
      "epoch": 14.597307221542227,
      "grad_norm": 1.7933304550002303,
      "learning_rate": 8.886228421383647e-09,
      "loss": 0.523,
      "step": 11926
    },
    {
      "epoch": 14.598531211750306,
      "grad_norm": 2.070882561295051,
      "learning_rate": 8.832322674526827e-09,
      "loss": 0.7251,
      "step": 11927
    },
    {
      "epoch": 14.599755201958384,
      "grad_norm": 2.234513769421013,
      "learning_rate": 8.778580637995305e-09,
      "loss": 1.1796,
      "step": 11928
    },
    {
      "epoch": 14.600979192166463,
      "grad_norm": 2.5890045777317274,
      "learning_rate": 8.725002315321528e-09,
      "loss": 0.3987,
      "step": 11929
    },
    {
      "epoch": 14.602203182374542,
      "grad_norm": 2.177484019942961,
      "learning_rate": 8.671587710025742e-09,
      "loss": 0.382,
      "step": 11930
    },
    {
      "epoch": 14.60342717258262,
      "grad_norm": 2.0575025051918394,
      "learning_rate": 8.618336825618468e-09,
      "loss": 0.4862,
      "step": 11931
    },
    {
      "epoch": 14.604651162790697,
      "grad_norm": 2.5488984398304217,
      "learning_rate": 8.565249665598851e-09,
      "loss": 0.452,
      "step": 11932
    },
    {
      "epoch": 14.605875152998776,
      "grad_norm": 2.3688001762611353,
      "learning_rate": 8.512326233456048e-09,
      "loss": 1.0275,
      "step": 11933
    },
    {
      "epoch": 14.607099143206854,
      "grad_norm": 1.5351374298177984,
      "learning_rate": 8.459566532668107e-09,
      "loss": 0.2666,
      "step": 11934
    },
    {
      "epoch": 14.608323133414933,
      "grad_norm": 1.292612311880505,
      "learning_rate": 8.4069705667017e-09,
      "loss": 0.5516,
      "step": 11935
    },
    {
      "epoch": 14.609547123623011,
      "grad_norm": 1.4053273472957217,
      "learning_rate": 8.35453833901323e-09,
      "loss": 0.6197,
      "step": 11936
    },
    {
      "epoch": 14.61077111383109,
      "grad_norm": 1.1754848663813604,
      "learning_rate": 8.30226985304855e-09,
      "loss": 1.1947,
      "step": 11937
    },
    {
      "epoch": 14.611995104039167,
      "grad_norm": 1.4964674015147679,
      "learning_rate": 8.250165112242692e-09,
      "loss": 0.3289,
      "step": 11938
    },
    {
      "epoch": 14.613219094247246,
      "grad_norm": 1.1005721013641638,
      "learning_rate": 8.198224120019583e-09,
      "loss": 0.533,
      "step": 11939
    },
    {
      "epoch": 14.614443084455324,
      "grad_norm": 1.0656451590078104,
      "learning_rate": 8.146446879792602e-09,
      "loss": 0.439,
      "step": 11940
    },
    {
      "epoch": 14.615667074663403,
      "grad_norm": 1.5423375834387363,
      "learning_rate": 8.094833394964585e-09,
      "loss": 0.5554,
      "step": 11941
    },
    {
      "epoch": 14.616891064871481,
      "grad_norm": 1.323191775369342,
      "learning_rate": 8.043383668926985e-09,
      "loss": 0.6524,
      "step": 11942
    },
    {
      "epoch": 14.61811505507956,
      "grad_norm": 1.444818577042891,
      "learning_rate": 7.992097705061263e-09,
      "loss": 0.4302,
      "step": 11943
    },
    {
      "epoch": 14.619339045287639,
      "grad_norm": 1.2418132988363064,
      "learning_rate": 7.940975506737502e-09,
      "loss": 0.5371,
      "step": 11944
    },
    {
      "epoch": 14.620563035495715,
      "grad_norm": 2.1517068161596042,
      "learning_rate": 7.890017077315237e-09,
      "loss": 1.3102,
      "step": 11945
    },
    {
      "epoch": 14.621787025703794,
      "grad_norm": 1.6180776595586872,
      "learning_rate": 7.839222420143456e-09,
      "loss": 0.5866,
      "step": 11946
    },
    {
      "epoch": 14.623011015911873,
      "grad_norm": 1.6568905927271154,
      "learning_rate": 7.78859153856032e-09,
      "loss": 0.4684,
      "step": 11947
    },
    {
      "epoch": 14.624235006119951,
      "grad_norm": 1.847092959139734,
      "learning_rate": 7.738124435892614e-09,
      "loss": 0.5175,
      "step": 11948
    },
    {
      "epoch": 14.62545899632803,
      "grad_norm": 1.6705355619277302,
      "learning_rate": 7.687821115457129e-09,
      "loss": 0.4495,
      "step": 11949
    },
    {
      "epoch": 14.626682986536107,
      "grad_norm": 3.367147213086265,
      "learning_rate": 7.637681580559831e-09,
      "loss": 0.3399,
      "step": 11950
    },
    {
      "epoch": 14.627906976744185,
      "grad_norm": 1.5482650158698146,
      "learning_rate": 7.587705834495306e-09,
      "loss": 0.5029,
      "step": 11951
    },
    {
      "epoch": 14.629130966952264,
      "grad_norm": 2.1559075335298328,
      "learning_rate": 7.537893880548153e-09,
      "loss": 0.427,
      "step": 11952
    },
    {
      "epoch": 14.630354957160343,
      "grad_norm": 1.2637290871920543,
      "learning_rate": 7.488245721991582e-09,
      "loss": 0.3047,
      "step": 11953
    },
    {
      "epoch": 14.631578947368421,
      "grad_norm": 2.0732105228290028,
      "learning_rate": 7.438761362087987e-09,
      "loss": 0.7268,
      "step": 11954
    },
    {
      "epoch": 14.6328029375765,
      "grad_norm": 1.346906025070435,
      "learning_rate": 7.3894408040900424e-09,
      "loss": 0.5786,
      "step": 11955
    },
    {
      "epoch": 14.634026927784578,
      "grad_norm": 1.7215283619514976,
      "learning_rate": 7.340284051238489e-09,
      "loss": 0.4531,
      "step": 11956
    },
    {
      "epoch": 14.635250917992655,
      "grad_norm": 1.7928761879910258,
      "learning_rate": 7.291291106763521e-09,
      "loss": 0.5101,
      "step": 11957
    },
    {
      "epoch": 14.636474908200734,
      "grad_norm": 2.074789328630792,
      "learning_rate": 7.2424619738853394e-09,
      "loss": 1.133,
      "step": 11958
    },
    {
      "epoch": 14.637698898408813,
      "grad_norm": 1.6102854322773874,
      "learning_rate": 7.193796655812213e-09,
      "loss": 0.8154,
      "step": 11959
    },
    {
      "epoch": 14.638922888616891,
      "grad_norm": 1.8628115225171258,
      "learning_rate": 7.145295155742693e-09,
      "loss": 1.356,
      "step": 11960
    },
    {
      "epoch": 14.64014687882497,
      "grad_norm": 2.4821720912681857,
      "learning_rate": 7.096957476863953e-09,
      "loss": 1.192,
      "step": 11961
    },
    {
      "epoch": 14.641370869033048,
      "grad_norm": 1.7010688543696035,
      "learning_rate": 7.04878362235234e-09,
      "loss": 1.0613,
      "step": 11962
    },
    {
      "epoch": 14.642594859241125,
      "grad_norm": 1.6979937111115437,
      "learning_rate": 7.000773595373933e-09,
      "loss": 1.1685,
      "step": 11963
    },
    {
      "epoch": 14.643818849449204,
      "grad_norm": 1.2074746733778579,
      "learning_rate": 6.952927399083986e-09,
      "loss": 0.5668,
      "step": 11964
    },
    {
      "epoch": 14.645042839657282,
      "grad_norm": 3.266282170561454,
      "learning_rate": 6.905245036626374e-09,
      "loss": 0.4428,
      "step": 11965
    },
    {
      "epoch": 14.646266829865361,
      "grad_norm": 1.717595010715417,
      "learning_rate": 6.8577265111344216e-09,
      "loss": 0.6313,
      "step": 11966
    },
    {
      "epoch": 14.64749082007344,
      "grad_norm": 1.8427544513034466,
      "learning_rate": 6.810371825731465e-09,
      "loss": 0.3074,
      "step": 11967
    },
    {
      "epoch": 14.648714810281518,
      "grad_norm": 1.9572611390514756,
      "learning_rate": 6.763180983529183e-09,
      "loss": 1.0754,
      "step": 11968
    },
    {
      "epoch": 14.649938800489597,
      "grad_norm": 1.4871658253088362,
      "learning_rate": 6.716153987628982e-09,
      "loss": 1.2778,
      "step": 11969
    },
    {
      "epoch": 14.651162790697674,
      "grad_norm": 1.2142159911734456,
      "learning_rate": 6.66929084112089e-09,
      "loss": 0.5572,
      "step": 11970
    },
    {
      "epoch": 14.652386780905752,
      "grad_norm": 3.190358159888911,
      "learning_rate": 6.622591547084667e-09,
      "loss": 0.533,
      "step": 11971
    },
    {
      "epoch": 14.653610771113831,
      "grad_norm": 2.5459206174972855,
      "learning_rate": 6.576056108589801e-09,
      "loss": 0.3707,
      "step": 11972
    },
    {
      "epoch": 14.65483476132191,
      "grad_norm": 2.3497110115447692,
      "learning_rate": 6.529684528693569e-09,
      "loss": 1.3994,
      "step": 11973
    },
    {
      "epoch": 14.656058751529988,
      "grad_norm": 2.2195152468493013,
      "learning_rate": 6.483476810444089e-09,
      "loss": 0.5922,
      "step": 11974
    },
    {
      "epoch": 14.657282741738067,
      "grad_norm": 2.8520197263322316,
      "learning_rate": 6.437432956877543e-09,
      "loss": 0.5711,
      "step": 11975
    },
    {
      "epoch": 14.658506731946144,
      "grad_norm": 1.4445986387827394,
      "learning_rate": 6.391552971019843e-09,
      "loss": 0.4815,
      "step": 11976
    },
    {
      "epoch": 14.659730722154222,
      "grad_norm": 1.2059594119922876,
      "learning_rate": 6.345836855885801e-09,
      "loss": 0.4399,
      "step": 11977
    },
    {
      "epoch": 14.660954712362301,
      "grad_norm": 2.082890381616836,
      "learning_rate": 6.300284614480234e-09,
      "loss": 0.9559,
      "step": 11978
    },
    {
      "epoch": 14.66217870257038,
      "grad_norm": 2.3515845652890675,
      "learning_rate": 6.254896249796305e-09,
      "loss": 0.6654,
      "step": 11979
    },
    {
      "epoch": 14.663402692778458,
      "grad_norm": 1.9182379912512766,
      "learning_rate": 6.2096717648169044e-09,
      "loss": 0.4548,
      "step": 11980
    },
    {
      "epoch": 14.664626682986537,
      "grad_norm": 1.3247897748085322,
      "learning_rate": 6.1646111625141006e-09,
      "loss": 0.5527,
      "step": 11981
    },
    {
      "epoch": 14.665850673194614,
      "grad_norm": 1.6460759888894458,
      "learning_rate": 6.1197144458488565e-09,
      "loss": 1.288,
      "step": 11982
    },
    {
      "epoch": 14.667074663402692,
      "grad_norm": 1.683770761948297,
      "learning_rate": 6.074981617771869e-09,
      "loss": 0.6941,
      "step": 11983
    },
    {
      "epoch": 14.668298653610771,
      "grad_norm": 1.916964652302,
      "learning_rate": 6.030412681222453e-09,
      "loss": 0.4109,
      "step": 11984
    },
    {
      "epoch": 14.66952264381885,
      "grad_norm": 1.8092482515901291,
      "learning_rate": 5.986007639130209e-09,
      "loss": 0.8527,
      "step": 11985
    },
    {
      "epoch": 14.670746634026928,
      "grad_norm": 1.498960594800365,
      "learning_rate": 5.941766494412526e-09,
      "loss": 1.2674,
      "step": 11986
    },
    {
      "epoch": 14.671970624235007,
      "grad_norm": 2.484691166491592,
      "learning_rate": 5.897689249977079e-09,
      "loss": 0.4151,
      "step": 11987
    },
    {
      "epoch": 14.673194614443084,
      "grad_norm": 1.110358911457998,
      "learning_rate": 5.853775908720438e-09,
      "loss": 0.3908,
      "step": 11988
    },
    {
      "epoch": 14.674418604651162,
      "grad_norm": 1.5586375318363053,
      "learning_rate": 5.8100264735283516e-09,
      "loss": 1.0768,
      "step": 11989
    },
    {
      "epoch": 14.67564259485924,
      "grad_norm": 1.7083323992340866,
      "learning_rate": 5.76644094727602e-09,
      "loss": 0.493,
      "step": 11990
    },
    {
      "epoch": 14.67686658506732,
      "grad_norm": 1.8417214245081257,
      "learning_rate": 5.723019332827817e-09,
      "loss": 1.4501,
      "step": 11991
    },
    {
      "epoch": 14.678090575275398,
      "grad_norm": 1.6729142902705374,
      "learning_rate": 5.6797616330370175e-09,
      "loss": 0.4605,
      "step": 11992
    },
    {
      "epoch": 14.679314565483477,
      "grad_norm": 1.6610347680549726,
      "learning_rate": 5.636667850746347e-09,
      "loss": 1.0375,
      "step": 11993
    },
    {
      "epoch": 14.680538555691555,
      "grad_norm": 1.3865362252421265,
      "learning_rate": 5.593737988787984e-09,
      "loss": 0.6234,
      "step": 11994
    },
    {
      "epoch": 14.681762545899632,
      "grad_norm": 2.5947741402192563,
      "learning_rate": 5.550972049982728e-09,
      "loss": 1.0169,
      "step": 11995
    },
    {
      "epoch": 14.68298653610771,
      "grad_norm": 2.623945129456273,
      "learning_rate": 5.5083700371413865e-09,
      "loss": 0.4698,
      "step": 11996
    },
    {
      "epoch": 14.68421052631579,
      "grad_norm": 1.5753835654399828,
      "learning_rate": 5.465931953063663e-09,
      "loss": 0.6456,
      "step": 11997
    },
    {
      "epoch": 14.685434516523868,
      "grad_norm": 2.2905436304190707,
      "learning_rate": 5.4236578005378845e-09,
      "loss": 0.571,
      "step": 11998
    },
    {
      "epoch": 14.686658506731947,
      "grad_norm": 2.384636399693884,
      "learning_rate": 5.3815475823429385e-09,
      "loss": 0.4426,
      "step": 11999
    },
    {
      "epoch": 14.687882496940025,
      "grad_norm": 3.0899334764713244,
      "learning_rate": 5.3396013012455e-09,
      "loss": 0.4811,
      "step": 12000
    },
    {
      "epoch": 14.689106487148102,
      "grad_norm": 3.4129137468504642,
      "learning_rate": 5.297818960002255e-09,
      "loss": 0.4286,
      "step": 12001
    },
    {
      "epoch": 14.69033047735618,
      "grad_norm": 1.1749579390202942,
      "learning_rate": 5.256200561359337e-09,
      "loss": 0.4431,
      "step": 12002
    },
    {
      "epoch": 14.69155446756426,
      "grad_norm": 1.6701097562477463,
      "learning_rate": 5.214746108051505e-09,
      "loss": 0.4494,
      "step": 12003
    },
    {
      "epoch": 14.692778457772338,
      "grad_norm": 1.7745840209709787,
      "learning_rate": 5.1734556028026905e-09,
      "loss": 0.4106,
      "step": 12004
    },
    {
      "epoch": 14.694002447980417,
      "grad_norm": 3.940483915286823,
      "learning_rate": 5.132329048326834e-09,
      "loss": 0.388,
      "step": 12005
    },
    {
      "epoch": 14.695226438188495,
      "grad_norm": 2.0111992093033293,
      "learning_rate": 5.091366447326495e-09,
      "loss": 0.5253,
      "step": 12006
    },
    {
      "epoch": 14.696450428396572,
      "grad_norm": 2.1689262865218373,
      "learning_rate": 5.0505678024934115e-09,
      "loss": 1.1861,
      "step": 12007
    },
    {
      "epoch": 14.69767441860465,
      "grad_norm": 1.6165103064162882,
      "learning_rate": 5.00993311650877e-09,
      "loss": 0.5625,
      "step": 12008
    },
    {
      "epoch": 14.69889840881273,
      "grad_norm": 1.4941545553563675,
      "learning_rate": 4.969462392043211e-09,
      "loss": 1.155,
      "step": 12009
    },
    {
      "epoch": 14.700122399020808,
      "grad_norm": 2.045676560298923,
      "learning_rate": 4.9291556317559995e-09,
      "loss": 0.3533,
      "step": 12010
    },
    {
      "epoch": 14.701346389228886,
      "grad_norm": 1.796473087557276,
      "learning_rate": 4.8890128382961255e-09,
      "loss": 1.089,
      "step": 12011
    },
    {
      "epoch": 14.702570379436965,
      "grad_norm": 3.1126240470788797,
      "learning_rate": 4.849034014301479e-09,
      "loss": 0.5186,
      "step": 12012
    },
    {
      "epoch": 14.703794369645042,
      "grad_norm": 2.071772835442174,
      "learning_rate": 4.809219162399403e-09,
      "loss": 0.4337,
      "step": 12013
    },
    {
      "epoch": 14.70501835985312,
      "grad_norm": 1.974740397526902,
      "learning_rate": 4.769568285206139e-09,
      "loss": 0.4173,
      "step": 12014
    },
    {
      "epoch": 14.7062423500612,
      "grad_norm": 2.155642759295847,
      "learning_rate": 4.730081385327934e-09,
      "loss": 0.8275,
      "step": 12015
    },
    {
      "epoch": 14.707466340269278,
      "grad_norm": 2.651671253206143,
      "learning_rate": 4.6907584653593795e-09,
      "loss": 0.8438,
      "step": 12016
    },
    {
      "epoch": 14.708690330477356,
      "grad_norm": 2.456385601623163,
      "learning_rate": 4.6515995278845204e-09,
      "loss": 0.5846,
      "step": 12017
    },
    {
      "epoch": 14.709914320685435,
      "grad_norm": 2.0122952947416355,
      "learning_rate": 4.612604575477131e-09,
      "loss": 0.4755,
      "step": 12018
    },
    {
      "epoch": 14.711138310893514,
      "grad_norm": 1.4237014690514598,
      "learning_rate": 4.573773610699328e-09,
      "loss": 0.3048,
      "step": 12019
    },
    {
      "epoch": 14.71236230110159,
      "grad_norm": 1.1272883569587864,
      "learning_rate": 4.535106636103237e-09,
      "loss": 0.5222,
      "step": 12020
    },
    {
      "epoch": 14.713586291309669,
      "grad_norm": 1.3285399796581532,
      "learning_rate": 4.496603654229881e-09,
      "loss": 0.5866,
      "step": 12021
    },
    {
      "epoch": 14.714810281517748,
      "grad_norm": 0.9573492710695023,
      "learning_rate": 4.458264667609458e-09,
      "loss": 0.4613,
      "step": 12022
    },
    {
      "epoch": 14.716034271725826,
      "grad_norm": 2.4529661752430694,
      "learning_rate": 4.420089678761619e-09,
      "loss": 1.0072,
      "step": 12023
    },
    {
      "epoch": 14.717258261933905,
      "grad_norm": 1.7090606276554066,
      "learning_rate": 4.382078690194913e-09,
      "loss": 0.589,
      "step": 12024
    },
    {
      "epoch": 14.718482252141984,
      "grad_norm": 1.4383684640349346,
      "learning_rate": 4.3442317044073425e-09,
      "loss": 0.4389,
      "step": 12025
    },
    {
      "epoch": 14.71970624235006,
      "grad_norm": 1.1818214551203328,
      "learning_rate": 4.306548723886084e-09,
      "loss": 0.6066,
      "step": 12026
    },
    {
      "epoch": 14.720930232558139,
      "grad_norm": 2.227678613992446,
      "learning_rate": 4.269029751107489e-09,
      "loss": 0.6025,
      "step": 12027
    },
    {
      "epoch": 14.722154222766218,
      "grad_norm": 1.6451130107259915,
      "learning_rate": 4.231674788537088e-09,
      "loss": 0.5173,
      "step": 12028
    },
    {
      "epoch": 14.723378212974296,
      "grad_norm": 1.841379568337515,
      "learning_rate": 4.194483838630137e-09,
      "loss": 0.6235,
      "step": 12029
    },
    {
      "epoch": 14.724602203182375,
      "grad_norm": 1.7273954034842762,
      "learning_rate": 4.157456903829959e-09,
      "loss": 0.6123,
      "step": 12030
    },
    {
      "epoch": 14.725826193390454,
      "grad_norm": 2.181940262109109,
      "learning_rate": 4.120593986570443e-09,
      "loss": 0.6303,
      "step": 12031
    },
    {
      "epoch": 14.727050183598532,
      "grad_norm": 1.8206841947665204,
      "learning_rate": 4.083895089273815e-09,
      "loss": 0.47,
      "step": 12032
    },
    {
      "epoch": 14.728274173806609,
      "grad_norm": 1.4337594383819938,
      "learning_rate": 4.0473602143520365e-09,
      "loss": 0.7708,
      "step": 12033
    },
    {
      "epoch": 14.729498164014688,
      "grad_norm": 1.5297240679015143,
      "learning_rate": 4.0109893642056866e-09,
      "loss": 0.6231,
      "step": 12034
    },
    {
      "epoch": 14.730722154222766,
      "grad_norm": 2.0518213106638745,
      "learning_rate": 3.9747825412250755e-09,
      "loss": 0.358,
      "step": 12035
    },
    {
      "epoch": 14.731946144430845,
      "grad_norm": 1.3506714254113898,
      "learning_rate": 3.9387397477896885e-09,
      "loss": 0.7262,
      "step": 12036
    },
    {
      "epoch": 14.733170134638923,
      "grad_norm": 0.7910128903737157,
      "learning_rate": 3.902860986267909e-09,
      "loss": 0.216,
      "step": 12037
    },
    {
      "epoch": 14.734394124847,
      "grad_norm": 1.872401694921444,
      "learning_rate": 3.867146259017851e-09,
      "loss": 1.1821,
      "step": 12038
    },
    {
      "epoch": 14.735618115055079,
      "grad_norm": 1.9329840944187948,
      "learning_rate": 3.831595568386248e-09,
      "loss": 0.4656,
      "step": 12039
    },
    {
      "epoch": 14.736842105263158,
      "grad_norm": 1.175646503572795,
      "learning_rate": 3.796208916709565e-09,
      "loss": 0.4665,
      "step": 12040
    },
    {
      "epoch": 14.738066095471236,
      "grad_norm": 1.7022426068931982,
      "learning_rate": 3.7609863063131635e-09,
      "loss": 0.5556,
      "step": 12041
    },
    {
      "epoch": 14.739290085679315,
      "grad_norm": 2.048106299018896,
      "learning_rate": 3.725927739511581e-09,
      "loss": 0.4828,
      "step": 12042
    },
    {
      "epoch": 14.740514075887393,
      "grad_norm": 1.637698322371853,
      "learning_rate": 3.6910332186093635e-09,
      "loss": 0.9305,
      "step": 12043
    },
    {
      "epoch": 14.741738066095472,
      "grad_norm": 1.5876905463233346,
      "learning_rate": 3.6563027458991205e-09,
      "loss": 0.4657,
      "step": 12044
    },
    {
      "epoch": 14.742962056303549,
      "grad_norm": 1.7489582337449632,
      "learning_rate": 3.6217363236631943e-09,
      "loss": 0.7226,
      "step": 12045
    },
    {
      "epoch": 14.744186046511627,
      "grad_norm": 1.7586886267498139,
      "learning_rate": 3.5873339541733775e-09,
      "loss": 0.5657,
      "step": 12046
    },
    {
      "epoch": 14.745410036719706,
      "grad_norm": 2.4085866413787382,
      "learning_rate": 3.553095639690363e-09,
      "loss": 0.8971,
      "step": 12047
    },
    {
      "epoch": 14.746634026927785,
      "grad_norm": 2.7538762763649465,
      "learning_rate": 3.5190213824640164e-09,
      "loss": 0.2691,
      "step": 12048
    },
    {
      "epoch": 14.747858017135863,
      "grad_norm": 1.1374786839385027,
      "learning_rate": 3.4851111847339357e-09,
      "loss": 0.5039,
      "step": 12049
    },
    {
      "epoch": 14.749082007343942,
      "grad_norm": 1.5575306930777009,
      "learning_rate": 3.451365048728339e-09,
      "loss": 0.6768,
      "step": 12050
    },
    {
      "epoch": 14.750305997552019,
      "grad_norm": 1.838356351293293,
      "learning_rate": 3.417782976664896e-09,
      "loss": 0.9063,
      "step": 12051
    },
    {
      "epoch": 14.751529987760097,
      "grad_norm": 2.922553590812241,
      "learning_rate": 3.384364970750731e-09,
      "loss": 0.4372,
      "step": 12052
    },
    {
      "epoch": 14.752753977968176,
      "grad_norm": 1.4723203170658377,
      "learning_rate": 3.3511110331815867e-09,
      "loss": 0.3946,
      "step": 12053
    },
    {
      "epoch": 14.753977968176255,
      "grad_norm": 1.8983366420931678,
      "learning_rate": 3.318021166142937e-09,
      "loss": 0.4157,
      "step": 12054
    },
    {
      "epoch": 14.755201958384333,
      "grad_norm": 1.6420109865714743,
      "learning_rate": 3.2850953718094324e-09,
      "loss": 0.3307,
      "step": 12055
    },
    {
      "epoch": 14.756425948592412,
      "grad_norm": 2.123250641195415,
      "learning_rate": 3.252333652344619e-09,
      "loss": 1.1456,
      "step": 12056
    },
    {
      "epoch": 14.75764993880049,
      "grad_norm": 2.028019907887872,
      "learning_rate": 3.219736009901775e-09,
      "loss": 0.7208,
      "step": 12057
    },
    {
      "epoch": 14.758873929008567,
      "grad_norm": 1.9170852764164041,
      "learning_rate": 3.1873024466227976e-09,
      "loss": 0.9184,
      "step": 12058
    },
    {
      "epoch": 14.760097919216646,
      "grad_norm": 1.4843253161436196,
      "learning_rate": 3.1550329646393152e-09,
      "loss": 0.7201,
      "step": 12059
    },
    {
      "epoch": 14.761321909424725,
      "grad_norm": 2.9491646637516444,
      "learning_rate": 3.122927566071854e-09,
      "loss": 0.3541,
      "step": 12060
    },
    {
      "epoch": 14.762545899632803,
      "grad_norm": 1.2256035735028528,
      "learning_rate": 3.0909862530301147e-09,
      "loss": 0.4769,
      "step": 12061
    },
    {
      "epoch": 14.763769889840882,
      "grad_norm": 2.936739804706395,
      "learning_rate": 3.05920902761353e-09,
      "loss": 0.3458,
      "step": 12062
    },
    {
      "epoch": 14.76499388004896,
      "grad_norm": 2.420783951128229,
      "learning_rate": 3.027595891910151e-09,
      "loss": 0.4738,
      "step": 12063
    },
    {
      "epoch": 14.766217870257037,
      "grad_norm": 1.6331076879900313,
      "learning_rate": 2.996146847997483e-09,
      "loss": 0.4587,
      "step": 12064
    },
    {
      "epoch": 14.767441860465116,
      "grad_norm": 1.92520803216434,
      "learning_rate": 2.9648618979424836e-09,
      "loss": 0.4989,
      "step": 12065
    },
    {
      "epoch": 14.768665850673194,
      "grad_norm": 1.6014861423912945,
      "learning_rate": 2.9337410438007306e-09,
      "loss": 0.563,
      "step": 12066
    },
    {
      "epoch": 14.769889840881273,
      "grad_norm": 1.3779677310953304,
      "learning_rate": 2.902784287617255e-09,
      "loss": 0.7145,
      "step": 12067
    },
    {
      "epoch": 14.771113831089352,
      "grad_norm": 1.30988462885473,
      "learning_rate": 2.871991631426818e-09,
      "loss": 0.3832,
      "step": 12068
    },
    {
      "epoch": 14.77233782129743,
      "grad_norm": 1.4130832172974648,
      "learning_rate": 2.8413630772530786e-09,
      "loss": 0.5191,
      "step": 12069
    },
    {
      "epoch": 14.773561811505507,
      "grad_norm": 1.6468348058531797,
      "learning_rate": 2.810898627108316e-09,
      "loss": 0.9782,
      "step": 12070
    },
    {
      "epoch": 14.774785801713586,
      "grad_norm": 2.3334597617261887,
      "learning_rate": 2.780598282995095e-09,
      "loss": 0.9581,
      "step": 12071
    },
    {
      "epoch": 14.776009791921664,
      "grad_norm": 1.8136525688234348,
      "learning_rate": 2.7504620469040456e-09,
      "loss": 0.8794,
      "step": 12072
    },
    {
      "epoch": 14.777233782129743,
      "grad_norm": 1.4085511647361788,
      "learning_rate": 2.7204899208160828e-09,
      "loss": 0.551,
      "step": 12073
    },
    {
      "epoch": 14.778457772337822,
      "grad_norm": 1.589010917623209,
      "learning_rate": 2.69068190670102e-09,
      "loss": 0.6186,
      "step": 12074
    },
    {
      "epoch": 14.7796817625459,
      "grad_norm": 2.492124256032937,
      "learning_rate": 2.661038006517014e-09,
      "loss": 0.4632,
      "step": 12075
    },
    {
      "epoch": 14.780905752753977,
      "grad_norm": 2.303652306306227,
      "learning_rate": 2.631558222212782e-09,
      "loss": 0.5955,
      "step": 12076
    },
    {
      "epoch": 14.782129742962056,
      "grad_norm": 1.8006199562568233,
      "learning_rate": 2.602242555725387e-09,
      "loss": 1.7956,
      "step": 12077
    },
    {
      "epoch": 14.783353733170134,
      "grad_norm": 2.097502637413556,
      "learning_rate": 2.5730910089813432e-09,
      "loss": 0.4858,
      "step": 12078
    },
    {
      "epoch": 14.784577723378213,
      "grad_norm": 3.9322304699397472,
      "learning_rate": 2.5441035838966177e-09,
      "loss": 0.3676,
      "step": 12079
    },
    {
      "epoch": 14.785801713586292,
      "grad_norm": 1.7135196509592694,
      "learning_rate": 2.515280282375521e-09,
      "loss": 1.5488,
      "step": 12080
    },
    {
      "epoch": 14.78702570379437,
      "grad_norm": 1.2745926845952955,
      "learning_rate": 2.486621106312925e-09,
      "loss": 0.6801,
      "step": 12081
    },
    {
      "epoch": 14.788249694002449,
      "grad_norm": 1.4761354796202726,
      "learning_rate": 2.4581260575920473e-09,
      "loss": 0.4279,
      "step": 12082
    },
    {
      "epoch": 14.789473684210526,
      "grad_norm": 2.4985856694385586,
      "learning_rate": 2.429795138085278e-09,
      "loss": 0.3572,
      "step": 12083
    },
    {
      "epoch": 14.790697674418604,
      "grad_norm": 1.0926703644246472,
      "learning_rate": 2.4016283496544614e-09,
      "loss": 0.5258,
      "step": 12084
    },
    {
      "epoch": 14.791921664626683,
      "grad_norm": 3.29754122662569,
      "learning_rate": 2.3736256941506164e-09,
      "loss": 0.3985,
      "step": 12085
    },
    {
      "epoch": 14.793145654834762,
      "grad_norm": 1.9205216032894543,
      "learning_rate": 2.345787173413938e-09,
      "loss": 0.2866,
      "step": 12086
    },
    {
      "epoch": 14.79436964504284,
      "grad_norm": 1.5141406215281468,
      "learning_rate": 2.318112789274074e-09,
      "loss": 0.6123,
      "step": 12087
    },
    {
      "epoch": 14.795593635250919,
      "grad_norm": 1.6336229696842457,
      "learning_rate": 2.2906025435495694e-09,
      "loss": 0.4015,
      "step": 12088
    },
    {
      "epoch": 14.796817625458996,
      "grad_norm": 1.470674357324275,
      "learning_rate": 2.2632564380481447e-09,
      "loss": 0.4291,
      "step": 12089
    },
    {
      "epoch": 14.798041615667074,
      "grad_norm": 2.60864405655822,
      "learning_rate": 2.2360744745669737e-09,
      "loss": 0.5127,
      "step": 12090
    },
    {
      "epoch": 14.799265605875153,
      "grad_norm": 2.707511124003092,
      "learning_rate": 2.2090566548926827e-09,
      "loss": 0.4216,
      "step": 12091
    },
    {
      "epoch": 14.800489596083231,
      "grad_norm": 2.37247533698625,
      "learning_rate": 2.182202980800241e-09,
      "loss": 0.976,
      "step": 12092
    },
    {
      "epoch": 14.80171358629131,
      "grad_norm": 1.7697433928768147,
      "learning_rate": 2.155513454054625e-09,
      "loss": 0.4794,
      "step": 12093
    },
    {
      "epoch": 14.802937576499389,
      "grad_norm": 2.9133471710146006,
      "learning_rate": 2.128988076409988e-09,
      "loss": 0.5694,
      "step": 12094
    },
    {
      "epoch": 14.804161566707466,
      "grad_norm": 2.544908183901255,
      "learning_rate": 2.1026268496091018e-09,
      "loss": 0.8543,
      "step": 12095
    },
    {
      "epoch": 14.805385556915544,
      "grad_norm": 1.4402036684224415,
      "learning_rate": 2.0764297753844698e-09,
      "loss": 0.8688,
      "step": 12096
    },
    {
      "epoch": 14.806609547123623,
      "grad_norm": 1.6484440044821693,
      "learning_rate": 2.05039685545777e-09,
      "loss": 0.5635,
      "step": 12097
    },
    {
      "epoch": 14.807833537331701,
      "grad_norm": 1.1169493879033598,
      "learning_rate": 2.024528091539857e-09,
      "loss": 0.5908,
      "step": 12098
    },
    {
      "epoch": 14.80905752753978,
      "grad_norm": 0.5911293090855352,
      "learning_rate": 1.998823485330481e-09,
      "loss": 0.1283,
      "step": 12099
    },
    {
      "epoch": 14.810281517747859,
      "grad_norm": 1.504717898552085,
      "learning_rate": 1.9732830385188474e-09,
      "loss": 1.3141,
      "step": 12100
    },
    {
      "epoch": 14.811505507955935,
      "grad_norm": 1.1131744411137092,
      "learning_rate": 1.9479067527836124e-09,
      "loss": 0.5437,
      "step": 12101
    },
    {
      "epoch": 14.812729498164014,
      "grad_norm": 1.1572848853061175,
      "learning_rate": 1.9226946297920545e-09,
      "loss": 0.5797,
      "step": 12102
    },
    {
      "epoch": 14.813953488372093,
      "grad_norm": 2.5622719318110545,
      "learning_rate": 1.8976466712014586e-09,
      "loss": 0.9624,
      "step": 12103
    },
    {
      "epoch": 14.815177478580171,
      "grad_norm": 1.7476371751757434,
      "learning_rate": 1.87276287865773e-09,
      "loss": 0.4235,
      "step": 12104
    },
    {
      "epoch": 14.81640146878825,
      "grad_norm": 2.4872121374879037,
      "learning_rate": 1.848043253796228e-09,
      "loss": 0.4401,
      "step": 12105
    },
    {
      "epoch": 14.817625458996329,
      "grad_norm": 1.6213591940632264,
      "learning_rate": 1.8234877982409305e-09,
      "loss": 0.4908,
      "step": 12106
    },
    {
      "epoch": 14.818849449204407,
      "grad_norm": 2.2973941492091443,
      "learning_rate": 1.7990965136061022e-09,
      "loss": 0.3903,
      "step": 12107
    },
    {
      "epoch": 14.820073439412484,
      "grad_norm": 1.5306668125827514,
      "learning_rate": 1.77486940149435e-09,
      "loss": 0.5833,
      "step": 12108
    },
    {
      "epoch": 14.821297429620563,
      "grad_norm": 1.9527980338400983,
      "learning_rate": 1.7508064634980116e-09,
      "loss": 0.3458,
      "step": 12109
    },
    {
      "epoch": 14.822521419828641,
      "grad_norm": 1.9038430495331289,
      "learning_rate": 1.7269077011983215e-09,
      "loss": 0.4654,
      "step": 12110
    },
    {
      "epoch": 14.82374541003672,
      "grad_norm": 1.2611722333494386,
      "learning_rate": 1.7031731161656906e-09,
      "loss": 0.6562,
      "step": 12111
    },
    {
      "epoch": 14.824969400244798,
      "grad_norm": 1.4983964768160178,
      "learning_rate": 1.6796027099599822e-09,
      "loss": 0.9635,
      "step": 12112
    },
    {
      "epoch": 14.826193390452877,
      "grad_norm": 1.419291008784469,
      "learning_rate": 1.6561964841299573e-09,
      "loss": 0.4061,
      "step": 12113
    },
    {
      "epoch": 14.827417380660954,
      "grad_norm": 1.1844347429774829,
      "learning_rate": 1.6329544402141074e-09,
      "loss": 0.7414,
      "step": 12114
    },
    {
      "epoch": 14.828641370869033,
      "grad_norm": 1.5006779837142645,
      "learning_rate": 1.609876579739822e-09,
      "loss": 0.6152,
      "step": 12115
    },
    {
      "epoch": 14.829865361077111,
      "grad_norm": 1.51535588543413,
      "learning_rate": 1.5869629042233881e-09,
      "loss": 0.6847,
      "step": 12116
    },
    {
      "epoch": 14.83108935128519,
      "grad_norm": 1.9269026655554389,
      "learning_rate": 1.5642134151708233e-09,
      "loss": 0.7323,
      "step": 12117
    },
    {
      "epoch": 14.832313341493268,
      "grad_norm": 1.6874168731787287,
      "learning_rate": 1.5416281140770428e-09,
      "loss": 1.165,
      "step": 12118
    },
    {
      "epoch": 14.833537331701347,
      "grad_norm": 2.44621958737006,
      "learning_rate": 1.5192070024261373e-09,
      "loss": 0.4414,
      "step": 12119
    },
    {
      "epoch": 14.834761321909426,
      "grad_norm": 2.544850546660493,
      "learning_rate": 1.496950081691928e-09,
      "loss": 1.0075,
      "step": 12120
    },
    {
      "epoch": 14.835985312117502,
      "grad_norm": 2.08706063020094,
      "learning_rate": 1.4748573533368559e-09,
      "loss": 0.7565,
      "step": 12121
    },
    {
      "epoch": 14.837209302325581,
      "grad_norm": 2.76413655421455,
      "learning_rate": 1.452928818812538e-09,
      "loss": 0.4627,
      "step": 12122
    },
    {
      "epoch": 14.83843329253366,
      "grad_norm": 0.9176992532751239,
      "learning_rate": 1.4311644795605983e-09,
      "loss": 0.3457,
      "step": 12123
    },
    {
      "epoch": 14.839657282741738,
      "grad_norm": 2.093914345734106,
      "learning_rate": 1.409564337010727e-09,
      "loss": 0.4833,
      "step": 12124
    },
    {
      "epoch": 14.840881272949817,
      "grad_norm": 1.71893280075994,
      "learning_rate": 1.3881283925826218e-09,
      "loss": 1.6187,
      "step": 12125
    },
    {
      "epoch": 14.842105263157894,
      "grad_norm": 1.425465714589987,
      "learning_rate": 1.3668566476848777e-09,
      "loss": 0.5523,
      "step": 12126
    },
    {
      "epoch": 14.843329253365972,
      "grad_norm": 3.581411937168354,
      "learning_rate": 1.345749103715821e-09,
      "loss": 0.326,
      "step": 12127
    },
    {
      "epoch": 14.844553243574051,
      "grad_norm": 2.700769460940968,
      "learning_rate": 1.3248057620621201e-09,
      "loss": 0.6665,
      "step": 12128
    },
    {
      "epoch": 14.84577723378213,
      "grad_norm": 1.1615428445747722,
      "learning_rate": 1.3040266241001742e-09,
      "loss": 0.6358,
      "step": 12129
    },
    {
      "epoch": 14.847001223990208,
      "grad_norm": 2.715689852981607,
      "learning_rate": 1.2834116911955573e-09,
      "loss": 0.4253,
      "step": 12130
    },
    {
      "epoch": 14.848225214198287,
      "grad_norm": 1.660760693237752,
      "learning_rate": 1.2629609647030194e-09,
      "loss": 1.1632,
      "step": 12131
    },
    {
      "epoch": 14.849449204406366,
      "grad_norm": 3.4769599616557496,
      "learning_rate": 1.2426744459664853e-09,
      "loss": 0.398,
      "step": 12132
    },
    {
      "epoch": 14.850673194614442,
      "grad_norm": 1.8096389108700857,
      "learning_rate": 1.2225521363193326e-09,
      "loss": 0.7256,
      "step": 12133
    },
    {
      "epoch": 14.851897184822521,
      "grad_norm": 2.811508015445776,
      "learning_rate": 1.2025940370832823e-09,
      "loss": 0.5045,
      "step": 12134
    },
    {
      "epoch": 14.8531211750306,
      "grad_norm": 1.1644296693343723,
      "learning_rate": 1.1828001495708952e-09,
      "loss": 0.5939,
      "step": 12135
    },
    {
      "epoch": 14.854345165238678,
      "grad_norm": 2.0175096479105363,
      "learning_rate": 1.163170475081965e-09,
      "loss": 0.4513,
      "step": 12136
    },
    {
      "epoch": 14.855569155446757,
      "grad_norm": 2.6949738456902383,
      "learning_rate": 1.143705014907126e-09,
      "loss": 0.3948,
      "step": 12137
    },
    {
      "epoch": 14.856793145654835,
      "grad_norm": 2.1770994380904005,
      "learning_rate": 1.1244037703250776e-09,
      "loss": 0.3411,
      "step": 12138
    },
    {
      "epoch": 14.858017135862912,
      "grad_norm": 3.1748044322195215,
      "learning_rate": 1.1052667426048048e-09,
      "loss": 0.4374,
      "step": 12139
    },
    {
      "epoch": 14.859241126070991,
      "grad_norm": 1.694281725978661,
      "learning_rate": 1.0862939330036348e-09,
      "loss": 0.665,
      "step": 12140
    },
    {
      "epoch": 14.86046511627907,
      "grad_norm": 2.1338522287398587,
      "learning_rate": 1.0674853427683484e-09,
      "loss": 0.5023,
      "step": 12141
    },
    {
      "epoch": 14.861689106487148,
      "grad_norm": 1.106307045578746,
      "learning_rate": 1.0488409731349015e-09,
      "loss": 0.5437,
      "step": 12142
    },
    {
      "epoch": 14.862913096695227,
      "grad_norm": 2.801346877059512,
      "learning_rate": 1.0303608253287023e-09,
      "loss": 0.7777,
      "step": 12143
    },
    {
      "epoch": 14.864137086903305,
      "grad_norm": 2.7479443552578964,
      "learning_rate": 1.0120449005640576e-09,
      "loss": 0.4815,
      "step": 12144
    },
    {
      "epoch": 14.865361077111384,
      "grad_norm": 2.5565814458690688,
      "learning_rate": 9.938932000450042e-10,
      "loss": 0.5948,
      "step": 12145
    },
    {
      "epoch": 14.86658506731946,
      "grad_norm": 1.99935208349718,
      "learning_rate": 9.759057249639214e-10,
      "loss": 0.6177,
      "step": 12146
    },
    {
      "epoch": 14.86780905752754,
      "grad_norm": 1.1990070637721293,
      "learning_rate": 9.580824765029196e-10,
      "loss": 0.4955,
      "step": 12147
    },
    {
      "epoch": 14.869033047735618,
      "grad_norm": 2.0996915065452786,
      "learning_rate": 9.404234558335612e-10,
      "loss": 0.9994,
      "step": 12148
    },
    {
      "epoch": 14.870257037943697,
      "grad_norm": 0.9781065508599823,
      "learning_rate": 9.229286641160296e-10,
      "loss": 0.387,
      "step": 12149
    },
    {
      "epoch": 14.871481028151775,
      "grad_norm": 0.7527029392547504,
      "learning_rate": 9.055981025005156e-10,
      "loss": 0.2544,
      "step": 12150
    },
    {
      "epoch": 14.872705018359854,
      "grad_norm": 2.0286136317249053,
      "learning_rate": 8.884317721252755e-10,
      "loss": 0.5338,
      "step": 12151
    },
    {
      "epoch": 14.87392900856793,
      "grad_norm": 1.8928283831463266,
      "learning_rate": 8.714296741188511e-10,
      "loss": 1.5565,
      "step": 12152
    },
    {
      "epoch": 14.87515299877601,
      "grad_norm": 1.6086632483617815,
      "learning_rate": 8.545918095984041e-10,
      "loss": 0.3983,
      "step": 12153
    },
    {
      "epoch": 14.876376988984088,
      "grad_norm": 1.9313992628794288,
      "learning_rate": 8.379181796705494e-10,
      "loss": 0.3838,
      "step": 12154
    },
    {
      "epoch": 14.877600979192167,
      "grad_norm": 1.6495793177562212,
      "learning_rate": 8.214087854307995e-10,
      "loss": 0.4579,
      "step": 12155
    },
    {
      "epoch": 14.878824969400245,
      "grad_norm": 1.6670821775479912,
      "learning_rate": 8.050636279643975e-10,
      "loss": 0.6191,
      "step": 12156
    },
    {
      "epoch": 14.880048959608324,
      "grad_norm": 1.3146688933626793,
      "learning_rate": 7.888827083452066e-10,
      "loss": 0.642,
      "step": 12157
    },
    {
      "epoch": 14.8812729498164,
      "grad_norm": 1.3693291798353748,
      "learning_rate": 7.728660276368205e-10,
      "loss": 0.7618,
      "step": 12158
    },
    {
      "epoch": 14.88249694002448,
      "grad_norm": 2.9188593606646474,
      "learning_rate": 7.570135868917305e-10,
      "loss": 0.3537,
      "step": 12159
    },
    {
      "epoch": 14.883720930232558,
      "grad_norm": 1.5304508702577193,
      "learning_rate": 7.413253871516036e-10,
      "loss": 0.5208,
      "step": 12160
    },
    {
      "epoch": 14.884944920440637,
      "grad_norm": 2.1933457741337814,
      "learning_rate": 7.258014294472815e-10,
      "loss": 0.5896,
      "step": 12161
    },
    {
      "epoch": 14.886168910648715,
      "grad_norm": 2.2341117489606988,
      "learning_rate": 7.10441714799337e-10,
      "loss": 1.123,
      "step": 12162
    },
    {
      "epoch": 14.887392900856794,
      "grad_norm": 3.0100885533785338,
      "learning_rate": 6.952462442166852e-10,
      "loss": 0.5218,
      "step": 12163
    },
    {
      "epoch": 14.88861689106487,
      "grad_norm": 2.357731993440913,
      "learning_rate": 6.802150186985268e-10,
      "loss": 0.3423,
      "step": 12164
    },
    {
      "epoch": 14.88984088127295,
      "grad_norm": 2.3382865116353932,
      "learning_rate": 6.653480392318501e-10,
      "loss": 0.3433,
      "step": 12165
    },
    {
      "epoch": 14.891064871481028,
      "grad_norm": 3.18343263901569,
      "learning_rate": 6.50645306794484e-10,
      "loss": 0.3886,
      "step": 12166
    },
    {
      "epoch": 14.892288861689106,
      "grad_norm": 1.8372837847887986,
      "learning_rate": 6.36106822352045e-10,
      "loss": 1.187,
      "step": 12167
    },
    {
      "epoch": 14.893512851897185,
      "grad_norm": 1.570964071747566,
      "learning_rate": 6.217325868601576e-10,
      "loss": 0.9105,
      "step": 12168
    },
    {
      "epoch": 14.894736842105264,
      "grad_norm": 2.176347643471491,
      "learning_rate": 6.075226012636215e-10,
      "loss": 1.1125,
      "step": 12169
    },
    {
      "epoch": 14.895960832313342,
      "grad_norm": 2.3496200916455345,
      "learning_rate": 5.934768664958568e-10,
      "loss": 0.3643,
      "step": 12170
    },
    {
      "epoch": 14.89718482252142,
      "grad_norm": 2.7002069853474455,
      "learning_rate": 5.795953834802914e-10,
      "loss": 1.1717,
      "step": 12171
    },
    {
      "epoch": 14.898408812729498,
      "grad_norm": 1.8913277098298553,
      "learning_rate": 5.658781531289737e-10,
      "loss": 0.7032,
      "step": 12172
    },
    {
      "epoch": 14.899632802937576,
      "grad_norm": 1.7650691811391428,
      "learning_rate": 5.52325176343127e-10,
      "loss": 0.5663,
      "step": 12173
    },
    {
      "epoch": 14.900856793145655,
      "grad_norm": 2.0122160071963524,
      "learning_rate": 5.38936454013983e-10,
      "loss": 0.4432,
      "step": 12174
    },
    {
      "epoch": 14.902080783353734,
      "grad_norm": 1.7710164578573233,
      "learning_rate": 5.257119870208382e-10,
      "loss": 0.7418,
      "step": 12175
    },
    {
      "epoch": 14.903304773561812,
      "grad_norm": 2.3170921207536046,
      "learning_rate": 5.12651776233275e-10,
      "loss": 0.4596,
      "step": 12176
    },
    {
      "epoch": 14.904528763769889,
      "grad_norm": 1.500459527370147,
      "learning_rate": 4.997558225089405e-10,
      "loss": 0.4506,
      "step": 12177
    },
    {
      "epoch": 14.905752753977968,
      "grad_norm": 1.156096024014779,
      "learning_rate": 4.870241266957676e-10,
      "loss": 0.5907,
      "step": 12178
    },
    {
      "epoch": 14.906976744186046,
      "grad_norm": 1.8320985481279137,
      "learning_rate": 4.744566896305869e-10,
      "loss": 0.6494,
      "step": 12179
    },
    {
      "epoch": 14.908200734394125,
      "grad_norm": 2.7613927846731827,
      "learning_rate": 4.6205351213857166e-10,
      "loss": 0.7804,
      "step": 12180
    },
    {
      "epoch": 14.909424724602204,
      "grad_norm": 2.416877500235346,
      "learning_rate": 4.498145950354582e-10,
      "loss": 0.4209,
      "step": 12181
    },
    {
      "epoch": 14.910648714810282,
      "grad_norm": 2.757327476424989,
      "learning_rate": 4.3773993912532565e-10,
      "loss": 0.5797,
      "step": 12182
    },
    {
      "epoch": 14.911872705018359,
      "grad_norm": 2.083272355374593,
      "learning_rate": 4.2582954520170583e-10,
      "loss": 0.758,
      "step": 12183
    },
    {
      "epoch": 14.913096695226438,
      "grad_norm": 2.2805726620547135,
      "learning_rate": 4.14083414047306e-10,
      "loss": 0.6667,
      "step": 12184
    },
    {
      "epoch": 14.914320685434516,
      "grad_norm": 1.0471311613156458,
      "learning_rate": 4.0250154643400875e-10,
      "loss": 0.4442,
      "step": 12185
    },
    {
      "epoch": 14.915544675642595,
      "grad_norm": 2.0996702165943355,
      "learning_rate": 3.910839431228719e-10,
      "loss": 0.6508,
      "step": 12186
    },
    {
      "epoch": 14.916768665850674,
      "grad_norm": 2.323210710765075,
      "learning_rate": 3.798306048641287e-10,
      "loss": 0.4069,
      "step": 12187
    },
    {
      "epoch": 14.917992656058752,
      "grad_norm": 2.34444065538037,
      "learning_rate": 3.687415323977428e-10,
      "loss": 0.6244,
      "step": 12188
    },
    {
      "epoch": 14.919216646266829,
      "grad_norm": 1.9993310813210843,
      "learning_rate": 3.578167264520205e-10,
      "loss": 0.5283,
      "step": 12189
    },
    {
      "epoch": 14.920440636474908,
      "grad_norm": 1.2944814995705447,
      "learning_rate": 3.470561877449985e-10,
      "loss": 0.7448,
      "step": 12190
    },
    {
      "epoch": 14.921664626682986,
      "grad_norm": 2.92691211330855,
      "learning_rate": 3.364599169841665e-10,
      "loss": 0.7883,
      "step": 12191
    },
    {
      "epoch": 14.922888616891065,
      "grad_norm": 1.102546285043171,
      "learning_rate": 3.260279148653567e-10,
      "loss": 0.5082,
      "step": 12192
    },
    {
      "epoch": 14.924112607099143,
      "grad_norm": 1.53369750188442,
      "learning_rate": 3.157601820744094e-10,
      "loss": 0.4495,
      "step": 12193
    },
    {
      "epoch": 14.925336597307222,
      "grad_norm": 2.1260879794021674,
      "learning_rate": 3.056567192857851e-10,
      "loss": 0.5674,
      "step": 12194
    },
    {
      "epoch": 14.9265605875153,
      "grad_norm": 1.8500510113772122,
      "learning_rate": 2.957175271639523e-10,
      "loss": 0.5194,
      "step": 12195
    },
    {
      "epoch": 14.927784577723378,
      "grad_norm": 1.663544358762414,
      "learning_rate": 2.859426063617221e-10,
      "loss": 1.2389,
      "step": 12196
    },
    {
      "epoch": 14.929008567931456,
      "grad_norm": 2.737260708682882,
      "learning_rate": 2.7633195752135853e-10,
      "loss": 0.9231,
      "step": 12197
    },
    {
      "epoch": 14.930232558139535,
      "grad_norm": 2.000415977064803,
      "learning_rate": 2.668855812748561e-10,
      "loss": 0.4528,
      "step": 12198
    },
    {
      "epoch": 14.931456548347613,
      "grad_norm": 3.7732259497610237,
      "learning_rate": 2.576034782425518e-10,
      "loss": 0.464,
      "step": 12199
    },
    {
      "epoch": 14.932680538555692,
      "grad_norm": 1.2566507119353914,
      "learning_rate": 2.484856490347909e-10,
      "loss": 0.8127,
      "step": 12200
    },
    {
      "epoch": 14.93390452876377,
      "grad_norm": 1.3422610623642606,
      "learning_rate": 2.3953209425053856e-10,
      "loss": 0.7128,
      "step": 12201
    },
    {
      "epoch": 14.935128518971847,
      "grad_norm": 3.2902868750312675,
      "learning_rate": 2.3074281447821312e-10,
      "loss": 0.3016,
      "step": 12202
    },
    {
      "epoch": 14.936352509179926,
      "grad_norm": 1.9797471702062786,
      "learning_rate": 2.2211781029568558e-10,
      "loss": 0.5766,
      "step": 12203
    },
    {
      "epoch": 14.937576499388005,
      "grad_norm": 1.9549526648305577,
      "learning_rate": 2.1365708226944726e-10,
      "loss": 0.5766,
      "step": 12204
    },
    {
      "epoch": 14.938800489596083,
      "grad_norm": 2.888419011438165,
      "learning_rate": 2.0536063095544235e-10,
      "loss": 0.4356,
      "step": 12205
    },
    {
      "epoch": 14.940024479804162,
      "grad_norm": 2.518199044381515,
      "learning_rate": 1.972284568990679e-10,
      "loss": 0.4783,
      "step": 12206
    },
    {
      "epoch": 14.94124847001224,
      "grad_norm": 1.5488572801144849,
      "learning_rate": 1.892605606348963e-10,
      "loss": 1.2234,
      "step": 12207
    },
    {
      "epoch": 14.94247246022032,
      "grad_norm": 1.85600130528948,
      "learning_rate": 1.8145694268612012e-10,
      "loss": 0.4611,
      "step": 12208
    },
    {
      "epoch": 14.943696450428396,
      "grad_norm": 2.2837818478089473,
      "learning_rate": 1.7381760356594e-10,
      "loss": 0.5313,
      "step": 12209
    },
    {
      "epoch": 14.944920440636475,
      "grad_norm": 1.8332031190468716,
      "learning_rate": 1.6634254377645432e-10,
      "loss": 0.9476,
      "step": 12210
    },
    {
      "epoch": 14.946144430844553,
      "grad_norm": 3.075960561264935,
      "learning_rate": 1.590317638083816e-10,
      "loss": 0.4564,
      "step": 12211
    },
    {
      "epoch": 14.947368421052632,
      "grad_norm": 2.2696290002918187,
      "learning_rate": 1.5188526414244842e-10,
      "loss": 0.5648,
      "step": 12212
    },
    {
      "epoch": 14.94859241126071,
      "grad_norm": 3.4844274701905937,
      "learning_rate": 1.449030452485567e-10,
      "loss": 0.3437,
      "step": 12213
    },
    {
      "epoch": 14.949816401468787,
      "grad_norm": 1.6480128099897529,
      "learning_rate": 1.3808510758522854e-10,
      "loss": 0.5715,
      "step": 12214
    },
    {
      "epoch": 14.951040391676866,
      "grad_norm": 2.1077799716103023,
      "learning_rate": 1.314314516004389e-10,
      "loss": 0.5223,
      "step": 12215
    },
    {
      "epoch": 14.952264381884945,
      "grad_norm": 2.109482248832846,
      "learning_rate": 1.2494207773189327e-10,
      "loss": 0.2867,
      "step": 12216
    },
    {
      "epoch": 14.953488372093023,
      "grad_norm": 1.2497103164198702,
      "learning_rate": 1.1861698640563968e-10,
      "loss": 1.0372,
      "step": 12217
    },
    {
      "epoch": 14.954712362301102,
      "grad_norm": 1.1928900426167912,
      "learning_rate": 1.1245617803717911e-10,
      "loss": 0.3802,
      "step": 12218
    },
    {
      "epoch": 14.95593635250918,
      "grad_norm": 3.1742222852726467,
      "learning_rate": 1.0645965303202055e-10,
      "loss": 0.5662,
      "step": 12219
    },
    {
      "epoch": 14.957160342717259,
      "grad_norm": 1.9881201370155401,
      "learning_rate": 1.0062741178373802e-10,
      "loss": 0.3329,
      "step": 12220
    },
    {
      "epoch": 14.958384332925336,
      "grad_norm": 1.6440781160495652,
      "learning_rate": 9.495945467563605e-11,
      "loss": 0.6851,
      "step": 12221
    },
    {
      "epoch": 14.959608323133414,
      "grad_norm": 1.5144669466286118,
      "learning_rate": 8.945578208047201e-11,
      "loss": 0.5867,
      "step": 12222
    },
    {
      "epoch": 14.960832313341493,
      "grad_norm": 3.0332587876248644,
      "learning_rate": 8.41163943596235e-11,
      "loss": 0.4517,
      "step": 12223
    },
    {
      "epoch": 14.962056303549572,
      "grad_norm": 1.5883003699647809,
      "learning_rate": 7.894129186419852e-11,
      "loss": 0.2182,
      "step": 12224
    },
    {
      "epoch": 14.96328029375765,
      "grad_norm": 1.9958316965022371,
      "learning_rate": 7.393047493392536e-11,
      "loss": 0.3904,
      "step": 12225
    },
    {
      "epoch": 14.964504283965729,
      "grad_norm": 2.979382427301688,
      "learning_rate": 6.908394389854023e-11,
      "loss": 0.9454,
      "step": 12226
    },
    {
      "epoch": 14.965728274173806,
      "grad_norm": 3.440231529978535,
      "learning_rate": 6.440169907639959e-11,
      "loss": 0.4595,
      "step": 12227
    },
    {
      "epoch": 14.966952264381884,
      "grad_norm": 1.965075211933853,
      "learning_rate": 5.988374077503523e-11,
      "loss": 0.5942,
      "step": 12228
    },
    {
      "epoch": 14.968176254589963,
      "grad_norm": 3.8639173518875762,
      "learning_rate": 5.553006929170934e-11,
      "loss": 0.409,
      "step": 12229
    },
    {
      "epoch": 14.969400244798042,
      "grad_norm": 3.3657868069095946,
      "learning_rate": 5.134068491202682e-11,
      "loss": 0.6739,
      "step": 12230
    },
    {
      "epoch": 14.97062423500612,
      "grad_norm": 1.7191415493922453,
      "learning_rate": 4.731558791160051e-11,
      "loss": 0.4619,
      "step": 12231
    },
    {
      "epoch": 14.971848225214199,
      "grad_norm": 1.9010766832709547,
      "learning_rate": 4.345477855494107e-11,
      "loss": 1.3759,
      "step": 12232
    },
    {
      "epoch": 14.973072215422278,
      "grad_norm": 2.0757603875307797,
      "learning_rate": 3.9758257096012e-11,
      "loss": 0.5201,
      "step": 12233
    },
    {
      "epoch": 14.974296205630354,
      "grad_norm": 2.4683511732100945,
      "learning_rate": 3.622602377711948e-11,
      "loss": 0.4067,
      "step": 12234
    },
    {
      "epoch": 14.975520195838433,
      "grad_norm": 1.889939944461724,
      "learning_rate": 3.285807883085523e-11,
      "loss": 1.0785,
      "step": 12235
    },
    {
      "epoch": 14.976744186046512,
      "grad_norm": 1.6052645569748314,
      "learning_rate": 2.96544224784312e-11,
      "loss": 0.6005,
      "step": 12236
    },
    {
      "epoch": 14.97796817625459,
      "grad_norm": 2.4251814553148594,
      "learning_rate": 2.661505493023464e-11,
      "loss": 0.6619,
      "step": 12237
    },
    {
      "epoch": 14.979192166462669,
      "grad_norm": 1.1853608137145886,
      "learning_rate": 2.37399763861057e-11,
      "loss": 0.6008,
      "step": 12238
    },
    {
      "epoch": 14.980416156670747,
      "grad_norm": 1.7482523770849725,
      "learning_rate": 2.1029187035337408e-11,
      "loss": 0.9052,
      "step": 12239
    },
    {
      "epoch": 14.981640146878824,
      "grad_norm": 1.910113257656022,
      "learning_rate": 1.848268705556544e-11,
      "loss": 0.5193,
      "step": 12240
    },
    {
      "epoch": 14.982864137086903,
      "grad_norm": 1.5420679824612047,
      "learning_rate": 1.6100476614155925e-11,
      "loss": 0.804,
      "step": 12241
    },
    {
      "epoch": 14.984088127294982,
      "grad_norm": 1.472540643143315,
      "learning_rate": 1.388255586792786e-11,
      "loss": 0.5325,
      "step": 12242
    },
    {
      "epoch": 14.98531211750306,
      "grad_norm": 2.4354546752340425,
      "learning_rate": 1.1828924962598021e-11,
      "loss": 1.0698,
      "step": 12243
    },
    {
      "epoch": 14.986536107711139,
      "grad_norm": 2.707119701108476,
      "learning_rate": 9.939584033058503e-12,
      "loss": 0.4656,
      "step": 12244
    },
    {
      "epoch": 14.987760097919217,
      "grad_norm": 1.4428071224820438,
      "learning_rate": 8.214533203376729e-12,
      "loss": 0.6878,
      "step": 12245
    },
    {
      "epoch": 14.988984088127294,
      "grad_norm": 1.8374203895556762,
      "learning_rate": 6.653772587073004e-12,
      "loss": 1.1293,
      "step": 12246
    },
    {
      "epoch": 14.990208078335373,
      "grad_norm": 1.5528808681272455,
      "learning_rate": 5.2573022865654026e-12,
      "loss": 0.6141,
      "step": 12247
    },
    {
      "epoch": 14.991432068543451,
      "grad_norm": 1.4932512185123261,
      "learning_rate": 4.025122394002434e-12,
      "loss": 0.6609,
      "step": 12248
    },
    {
      "epoch": 14.99265605875153,
      "grad_norm": 2.9071477513073467,
      "learning_rate": 2.9572329895977135e-12,
      "loss": 0.5553,
      "step": 12249
    },
    {
      "epoch": 14.993880048959609,
      "grad_norm": 3.7650760860741346,
      "learning_rate": 2.053634144405514e-12,
      "loss": 0.2617,
      "step": 12250
    },
    {
      "epoch": 14.995104039167687,
      "grad_norm": 2.391589787169674,
      "learning_rate": 1.314325917267656e-12,
      "loss": 1.0234,
      "step": 12251
    },
    {
      "epoch": 14.996328029375764,
      "grad_norm": 2.916659861815552,
      "learning_rate": 7.393083567563964e-13,
      "loss": 0.4603,
      "step": 12252
    },
    {
      "epoch": 14.997552019583843,
      "grad_norm": 1.5471975959201474,
      "learning_rate": 3.28581500896874e-13,
      "loss": 0.7804,
      "step": 12253
    },
    {
      "epoch": 14.998776009791921,
      "grad_norm": 1.1298569357087687,
      "learning_rate": 8.214537661199729e-14,
      "loss": 0.5595,
      "step": 12254
    },
    {
      "epoch": 15.0,
      "grad_norm": 1.9960724675909332,
      "learning_rate": 0.0,
      "loss": 0.5467,
      "step": 12255
    },
    {
      "epoch": 15.0,
      "step": 12255,
      "total_flos": 4797777151918080.0,
      "train_loss": 0.6943354440338512,
      "train_runtime": 124613.968,
      "train_samples_per_second": 0.098,
      "train_steps_per_second": 0.098
    }
  ],
  "logging_steps": 1,
  "max_steps": 12255,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4797777151918080.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}