{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.7328691828508611,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00036643459142543056,
      "grad_norm": 2.4786333502410307,
      "learning_rate": 0.0,
      "loss": 0.664,
      "step": 1
    },
    {
      "epoch": 0.0007328691828508611,
      "grad_norm": 2.5045049576127707,
      "learning_rate": 1.2210012210012211e-08,
      "loss": 0.665,
      "step": 2
    },
    {
      "epoch": 0.0010993037742762918,
      "grad_norm": 2.396926038999881,
      "learning_rate": 2.4420024420024422e-08,
      "loss": 0.6477,
      "step": 3
    },
    {
      "epoch": 0.0014657383657017222,
      "grad_norm": 2.432708753798092,
      "learning_rate": 3.6630036630036635e-08,
      "loss": 0.6576,
      "step": 4
    },
    {
      "epoch": 0.001832172957127153,
      "grad_norm": 2.3960444299704426,
      "learning_rate": 4.8840048840048844e-08,
      "loss": 0.6556,
      "step": 5
    },
    {
      "epoch": 0.0021986075485525836,
      "grad_norm": 2.3021456880402376,
      "learning_rate": 6.105006105006105e-08,
      "loss": 0.6151,
      "step": 6
    },
    {
      "epoch": 0.002565042139978014,
      "grad_norm": 2.642068906567389,
      "learning_rate": 7.326007326007327e-08,
      "loss": 0.6746,
      "step": 7
    },
    {
      "epoch": 0.0029314767314034445,
      "grad_norm": 2.458527561204725,
      "learning_rate": 8.547008547008549e-08,
      "loss": 0.6755,
      "step": 8
    },
    {
      "epoch": 0.003297911322828875,
      "grad_norm": 2.3134377871036,
      "learning_rate": 9.768009768009769e-08,
      "loss": 0.6307,
      "step": 9
    },
    {
      "epoch": 0.003664345914254306,
      "grad_norm": 2.4097188915894847,
      "learning_rate": 1.098901098901099e-07,
      "loss": 0.6413,
      "step": 10
    },
    {
      "epoch": 0.004030780505679736,
      "grad_norm": 2.3931598955078877,
      "learning_rate": 1.221001221001221e-07,
      "loss": 0.6479,
      "step": 11
    },
    {
      "epoch": 0.004397215097105167,
      "grad_norm": 2.3041421106190056,
      "learning_rate": 1.3431013431013434e-07,
      "loss": 0.6285,
      "step": 12
    },
    {
      "epoch": 0.004763649688530597,
      "grad_norm": 2.452260472801403,
      "learning_rate": 1.4652014652014654e-07,
      "loss": 0.6485,
      "step": 13
    },
    {
      "epoch": 0.005130084279956028,
      "grad_norm": 2.400794258430855,
      "learning_rate": 1.5873015873015874e-07,
      "loss": 0.6257,
      "step": 14
    },
    {
      "epoch": 0.005496518871381458,
      "grad_norm": 2.4503807075700355,
      "learning_rate": 1.7094017094017097e-07,
      "loss": 0.66,
      "step": 15
    },
    {
      "epoch": 0.005862953462806889,
      "grad_norm": 2.316244903544412,
      "learning_rate": 1.8315018315018317e-07,
      "loss": 0.644,
      "step": 16
    },
    {
      "epoch": 0.00622938805423232,
      "grad_norm": 2.316125783374869,
      "learning_rate": 1.9536019536019538e-07,
      "loss": 0.6251,
      "step": 17
    },
    {
      "epoch": 0.00659582264565775,
      "grad_norm": 2.5206811114498104,
      "learning_rate": 2.0757020757020758e-07,
      "loss": 0.6607,
      "step": 18
    },
    {
      "epoch": 0.006962257237083181,
      "grad_norm": 2.307088703359359,
      "learning_rate": 2.197802197802198e-07,
      "loss": 0.6325,
      "step": 19
    },
    {
      "epoch": 0.007328691828508612,
      "grad_norm": 2.2848184908437044,
      "learning_rate": 2.31990231990232e-07,
      "loss": 0.6207,
      "step": 20
    },
    {
      "epoch": 0.007695126419934042,
      "grad_norm": 2.234890031415268,
      "learning_rate": 2.442002442002442e-07,
      "loss": 0.6101,
      "step": 21
    },
    {
      "epoch": 0.008061561011359472,
      "grad_norm": 2.3419067398274827,
      "learning_rate": 2.564102564102564e-07,
      "loss": 0.6449,
      "step": 22
    },
    {
      "epoch": 0.008427995602784902,
      "grad_norm": 2.2230361574254442,
      "learning_rate": 2.6862026862026867e-07,
      "loss": 0.6273,
      "step": 23
    },
    {
      "epoch": 0.008794430194210334,
      "grad_norm": 2.380204208317071,
      "learning_rate": 2.808302808302809e-07,
      "loss": 0.6588,
      "step": 24
    },
    {
      "epoch": 0.009160864785635764,
      "grad_norm": 2.3633456634285084,
      "learning_rate": 2.930402930402931e-07,
      "loss": 0.6495,
      "step": 25
    },
    {
      "epoch": 0.009527299377061194,
      "grad_norm": 2.305866310045725,
      "learning_rate": 3.052503052503053e-07,
      "loss": 0.6309,
      "step": 26
    },
    {
      "epoch": 0.009893733968486624,
      "grad_norm": 2.0316992612895484,
      "learning_rate": 3.174603174603175e-07,
      "loss": 0.6009,
      "step": 27
    },
    {
      "epoch": 0.010260168559912056,
      "grad_norm": 2.2131155106965883,
      "learning_rate": 3.296703296703297e-07,
      "loss": 0.6342,
      "step": 28
    },
    {
      "epoch": 0.010626603151337486,
      "grad_norm": 2.0127203312394157,
      "learning_rate": 3.4188034188034194e-07,
      "loss": 0.603,
      "step": 29
    },
    {
      "epoch": 0.010993037742762916,
      "grad_norm": 2.0620522943752917,
      "learning_rate": 3.540903540903541e-07,
      "loss": 0.6303,
      "step": 30
    },
    {
      "epoch": 0.011359472334188348,
      "grad_norm": 2.237929679316926,
      "learning_rate": 3.6630036630036635e-07,
      "loss": 0.66,
      "step": 31
    },
    {
      "epoch": 0.011725906925613778,
      "grad_norm": 2.0561168917799044,
      "learning_rate": 3.785103785103785e-07,
      "loss": 0.6204,
      "step": 32
    },
    {
      "epoch": 0.012092341517039208,
      "grad_norm": 2.0914008481446418,
      "learning_rate": 3.9072039072039075e-07,
      "loss": 0.6222,
      "step": 33
    },
    {
      "epoch": 0.01245877610846464,
      "grad_norm": 2.0492425666570955,
      "learning_rate": 4.02930402930403e-07,
      "loss": 0.6289,
      "step": 34
    },
    {
      "epoch": 0.01282521069989007,
      "grad_norm": 2.1055019197206004,
      "learning_rate": 4.1514041514041516e-07,
      "loss": 0.6272,
      "step": 35
    },
    {
      "epoch": 0.0131916452913155,
      "grad_norm": 2.2104657945348736,
      "learning_rate": 4.273504273504274e-07,
      "loss": 0.6588,
      "step": 36
    },
    {
      "epoch": 0.013558079882740931,
      "grad_norm": 2.066459552375068,
      "learning_rate": 4.395604395604396e-07,
      "loss": 0.6593,
      "step": 37
    },
    {
      "epoch": 0.013924514474166361,
      "grad_norm": 1.8641664031413725,
      "learning_rate": 4.517704517704518e-07,
      "loss": 0.6271,
      "step": 38
    },
    {
      "epoch": 0.014290949065591791,
      "grad_norm": 1.66226583147379,
      "learning_rate": 4.63980463980464e-07,
      "loss": 0.5951,
      "step": 39
    },
    {
      "epoch": 0.014657383657017223,
      "grad_norm": 1.697647767276202,
      "learning_rate": 4.7619047619047623e-07,
      "loss": 0.6102,
      "step": 40
    },
    {
      "epoch": 0.015023818248442653,
      "grad_norm": 1.6246998268622177,
      "learning_rate": 4.884004884004884e-07,
      "loss": 0.5835,
      "step": 41
    },
    {
      "epoch": 0.015390252839868083,
      "grad_norm": 1.6183092586382308,
      "learning_rate": 5.006105006105007e-07,
      "loss": 0.6248,
      "step": 42
    },
    {
      "epoch": 0.015756687431293513,
      "grad_norm": 1.6492790019480517,
      "learning_rate": 5.128205128205128e-07,
      "loss": 0.6318,
      "step": 43
    },
    {
      "epoch": 0.016123122022718945,
      "grad_norm": 1.5865449399026847,
      "learning_rate": 5.250305250305251e-07,
      "loss": 0.6122,
      "step": 44
    },
    {
      "epoch": 0.016489556614144377,
      "grad_norm": 1.5505379042365153,
      "learning_rate": 5.372405372405373e-07,
      "loss": 0.6032,
      "step": 45
    },
    {
      "epoch": 0.016855991205569805,
      "grad_norm": 1.724415484273136,
      "learning_rate": 5.494505494505495e-07,
      "loss": 0.6317,
      "step": 46
    },
    {
      "epoch": 0.017222425796995237,
      "grad_norm": 1.6026315688729296,
      "learning_rate": 5.616605616605618e-07,
      "loss": 0.623,
      "step": 47
    },
    {
      "epoch": 0.01758886038842067,
      "grad_norm": 1.5118701166307444,
      "learning_rate": 5.738705738705739e-07,
      "loss": 0.6169,
      "step": 48
    },
    {
      "epoch": 0.017955294979846097,
      "grad_norm": 1.5896054297443498,
      "learning_rate": 5.860805860805862e-07,
      "loss": 0.6217,
      "step": 49
    },
    {
      "epoch": 0.01832172957127153,
      "grad_norm": 1.4559587046479432,
      "learning_rate": 5.982905982905984e-07,
      "loss": 0.6162,
      "step": 50
    },
    {
      "epoch": 0.01868816416269696,
      "grad_norm": 1.3877383097055633,
      "learning_rate": 6.105006105006106e-07,
      "loss": 0.5877,
      "step": 51
    },
    {
      "epoch": 0.01905459875412239,
      "grad_norm": 1.2583675550502782,
      "learning_rate": 6.227106227106228e-07,
      "loss": 0.6262,
      "step": 52
    },
    {
      "epoch": 0.01942103334554782,
      "grad_norm": 1.0871983743940217,
      "learning_rate": 6.34920634920635e-07,
      "loss": 0.6026,
      "step": 53
    },
    {
      "epoch": 0.01978746793697325,
      "grad_norm": 1.010615864195493,
      "learning_rate": 6.471306471306472e-07,
      "loss": 0.5941,
      "step": 54
    },
    {
      "epoch": 0.02015390252839868,
      "grad_norm": 0.9955871080532015,
      "learning_rate": 6.593406593406594e-07,
      "loss": 0.5743,
      "step": 55
    },
    {
      "epoch": 0.020520337119824112,
      "grad_norm": 1.04923811040217,
      "learning_rate": 6.715506715506716e-07,
      "loss": 0.5936,
      "step": 56
    },
    {
      "epoch": 0.02088677171124954,
      "grad_norm": 0.982727228662074,
      "learning_rate": 6.837606837606839e-07,
      "loss": 0.5729,
      "step": 57
    },
    {
      "epoch": 0.021253206302674972,
      "grad_norm": 1.0569864989616775,
      "learning_rate": 6.95970695970696e-07,
      "loss": 0.5974,
      "step": 58
    },
    {
      "epoch": 0.021619640894100404,
      "grad_norm": 0.969177647224115,
      "learning_rate": 7.081807081807082e-07,
      "loss": 0.6111,
      "step": 59
    },
    {
      "epoch": 0.021986075485525832,
      "grad_norm": 1.05559778279447,
      "learning_rate": 7.203907203907203e-07,
      "loss": 0.5925,
      "step": 60
    },
    {
      "epoch": 0.022352510076951264,
      "grad_norm": 0.9078116928995486,
      "learning_rate": 7.326007326007327e-07,
      "loss": 0.6077,
      "step": 61
    },
    {
      "epoch": 0.022718944668376696,
      "grad_norm": 0.8901523347499087,
      "learning_rate": 7.448107448107448e-07,
      "loss": 0.6117,
      "step": 62
    },
    {
      "epoch": 0.023085379259802124,
      "grad_norm": 0.7769537384222807,
      "learning_rate": 7.57020757020757e-07,
      "loss": 0.5756,
      "step": 63
    },
    {
      "epoch": 0.023451813851227556,
      "grad_norm": 0.7805708330236202,
      "learning_rate": 7.692307692307694e-07,
      "loss": 0.5817,
      "step": 64
    },
    {
      "epoch": 0.023818248442652987,
      "grad_norm": 0.7216681094804467,
      "learning_rate": 7.814407814407815e-07,
      "loss": 0.5776,
      "step": 65
    },
    {
      "epoch": 0.024184683034078416,
      "grad_norm": 0.8203118176874593,
      "learning_rate": 7.936507936507937e-07,
      "loss": 0.5764,
      "step": 66
    },
    {
      "epoch": 0.024551117625503847,
      "grad_norm": 0.7037004306258906,
      "learning_rate": 8.05860805860806e-07,
      "loss": 0.6013,
      "step": 67
    },
    {
      "epoch": 0.02491755221692928,
      "grad_norm": 0.6332121106111739,
      "learning_rate": 8.180708180708182e-07,
      "loss": 0.5705,
      "step": 68
    },
    {
      "epoch": 0.025283986808354707,
      "grad_norm": 0.5891103871056083,
      "learning_rate": 8.302808302808303e-07,
      "loss": 0.5498,
      "step": 69
    },
    {
      "epoch": 0.02565042139978014,
      "grad_norm": 0.5849241404047495,
      "learning_rate": 8.424908424908426e-07,
      "loss": 0.577,
      "step": 70
    },
    {
      "epoch": 0.02601685599120557,
      "grad_norm": 0.5770560963040162,
      "learning_rate": 8.547008547008548e-07,
      "loss": 0.5803,
      "step": 71
    },
    {
      "epoch": 0.026383290582631,
      "grad_norm": 0.5067765783293705,
      "learning_rate": 8.66910866910867e-07,
      "loss": 0.5557,
      "step": 72
    },
    {
      "epoch": 0.02674972517405643,
      "grad_norm": 0.8279533074729605,
      "learning_rate": 8.791208791208792e-07,
      "loss": 0.5849,
      "step": 73
    },
    {
      "epoch": 0.027116159765481863,
      "grad_norm": 0.5775197162451533,
      "learning_rate": 8.913308913308914e-07,
      "loss": 0.5492,
      "step": 74
    },
    {
      "epoch": 0.02748259435690729,
      "grad_norm": 0.5657010760470752,
      "learning_rate": 9.035409035409036e-07,
      "loss": 0.555,
      "step": 75
    },
    {
      "epoch": 0.027849028948332723,
      "grad_norm": 0.5629166901912753,
      "learning_rate": 9.157509157509158e-07,
      "loss": 0.5975,
      "step": 76
    },
    {
      "epoch": 0.028215463539758154,
      "grad_norm": 0.6153490523003287,
      "learning_rate": 9.27960927960928e-07,
      "loss": 0.5641,
      "step": 77
    },
    {
      "epoch": 0.028581898131183583,
      "grad_norm": 0.7190636576137129,
      "learning_rate": 9.401709401709402e-07,
      "loss": 0.5643,
      "step": 78
    },
    {
      "epoch": 0.028948332722609015,
      "grad_norm": 0.5282710312078797,
      "learning_rate": 9.523809523809525e-07,
      "loss": 0.5736,
      "step": 79
    },
    {
      "epoch": 0.029314767314034446,
      "grad_norm": 0.5345231694084734,
      "learning_rate": 9.645909645909648e-07,
      "loss": 0.5433,
      "step": 80
    },
    {
      "epoch": 0.029681201905459875,
      "grad_norm": 0.6308939869525466,
      "learning_rate": 9.768009768009769e-07,
      "loss": 0.5643,
      "step": 81
    },
    {
      "epoch": 0.030047636496885306,
      "grad_norm": 0.497240077418448,
      "learning_rate": 9.890109890109891e-07,
      "loss": 0.5432,
      "step": 82
    },
    {
      "epoch": 0.030414071088310738,
      "grad_norm": 0.5038012254690507,
      "learning_rate": 1.0012210012210014e-06,
      "loss": 0.5371,
      "step": 83
    },
    {
      "epoch": 0.030780505679736166,
      "grad_norm": 0.5363381284786858,
      "learning_rate": 1.0134310134310136e-06,
      "loss": 0.561,
      "step": 84
    },
    {
      "epoch": 0.031146940271161598,
      "grad_norm": 0.5460153047624501,
      "learning_rate": 1.0256410256410257e-06,
      "loss": 0.5599,
      "step": 85
    },
    {
      "epoch": 0.031513374862587026,
      "grad_norm": 0.4488203580157306,
      "learning_rate": 1.037851037851038e-06,
      "loss": 0.5237,
      "step": 86
    },
    {
      "epoch": 0.03187980945401246,
      "grad_norm": 0.6376882547940598,
      "learning_rate": 1.0500610500610502e-06,
      "loss": 0.5573,
      "step": 87
    },
    {
      "epoch": 0.03224624404543789,
      "grad_norm": 0.5026275404694113,
      "learning_rate": 1.0622710622710622e-06,
      "loss": 0.55,
      "step": 88
    },
    {
      "epoch": 0.03261267863686332,
      "grad_norm": 0.43514217296393504,
      "learning_rate": 1.0744810744810747e-06,
      "loss": 0.5556,
      "step": 89
    },
    {
      "epoch": 0.03297911322828875,
      "grad_norm": 0.44590678131034034,
      "learning_rate": 1.0866910866910867e-06,
      "loss": 0.5427,
      "step": 90
    },
    {
      "epoch": 0.03334554781971418,
      "grad_norm": 0.3866994332341522,
      "learning_rate": 1.098901098901099e-06,
      "loss": 0.5535,
      "step": 91
    },
    {
      "epoch": 0.03371198241113961,
      "grad_norm": 0.38950413811209916,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.5526,
      "step": 92
    },
    {
      "epoch": 0.034078417002565045,
      "grad_norm": 0.4199606620668575,
      "learning_rate": 1.1233211233211235e-06,
      "loss": 0.5602,
      "step": 93
    },
    {
      "epoch": 0.03444485159399047,
      "grad_norm": 0.87568496228372,
      "learning_rate": 1.1355311355311355e-06,
      "loss": 0.5503,
      "step": 94
    },
    {
      "epoch": 0.0348112861854159,
      "grad_norm": 0.39203993419417044,
      "learning_rate": 1.1477411477411478e-06,
      "loss": 0.5475,
      "step": 95
    },
    {
      "epoch": 0.03517772077684134,
      "grad_norm": 0.3960404919233486,
      "learning_rate": 1.15995115995116e-06,
      "loss": 0.5282,
      "step": 96
    },
    {
      "epoch": 0.035544155368266765,
      "grad_norm": 0.4647754503826838,
      "learning_rate": 1.1721611721611723e-06,
      "loss": 0.5601,
      "step": 97
    },
    {
      "epoch": 0.035910589959692193,
      "grad_norm": 0.45261232183759037,
      "learning_rate": 1.1843711843711844e-06,
      "loss": 0.5512,
      "step": 98
    },
    {
      "epoch": 0.03627702455111763,
      "grad_norm": 0.3991733138335409,
      "learning_rate": 1.1965811965811968e-06,
      "loss": 0.5496,
      "step": 99
    },
    {
      "epoch": 0.03664345914254306,
      "grad_norm": 0.38542355224088154,
      "learning_rate": 1.2087912087912089e-06,
      "loss": 0.5614,
      "step": 100
    },
    {
      "epoch": 0.037009893733968485,
      "grad_norm": 0.38937387466732476,
      "learning_rate": 1.2210012210012211e-06,
      "loss": 0.5597,
      "step": 101
    },
    {
      "epoch": 0.03737632832539392,
      "grad_norm": 0.39081748257318233,
      "learning_rate": 1.2332112332112334e-06,
      "loss": 0.5656,
      "step": 102
    },
    {
      "epoch": 0.03774276291681935,
      "grad_norm": 0.3933234113196945,
      "learning_rate": 1.2454212454212456e-06,
      "loss": 0.569,
      "step": 103
    },
    {
      "epoch": 0.03810919750824478,
      "grad_norm": 0.36745002382210973,
      "learning_rate": 1.2576312576312577e-06,
      "loss": 0.5539,
      "step": 104
    },
    {
      "epoch": 0.03847563209967021,
      "grad_norm": 0.3593943103765893,
      "learning_rate": 1.26984126984127e-06,
      "loss": 0.5168,
      "step": 105
    },
    {
      "epoch": 0.03884206669109564,
      "grad_norm": 0.37473241323036766,
      "learning_rate": 1.282051282051282e-06,
      "loss": 0.5383,
      "step": 106
    },
    {
      "epoch": 0.03920850128252107,
      "grad_norm": 0.4301736128359902,
      "learning_rate": 1.2942612942612944e-06,
      "loss": 0.566,
      "step": 107
    },
    {
      "epoch": 0.0395749358739465,
      "grad_norm": 0.3699657199357681,
      "learning_rate": 1.3064713064713067e-06,
      "loss": 0.5664,
      "step": 108
    },
    {
      "epoch": 0.03994137046537193,
      "grad_norm": 0.3385938355189879,
      "learning_rate": 1.3186813186813187e-06,
      "loss": 0.5403,
      "step": 109
    },
    {
      "epoch": 0.04030780505679736,
      "grad_norm": 0.3600066069433996,
      "learning_rate": 1.330891330891331e-06,
      "loss": 0.5374,
      "step": 110
    },
    {
      "epoch": 0.04067423964822279,
      "grad_norm": 0.3555863682283024,
      "learning_rate": 1.3431013431013433e-06,
      "loss": 0.523,
      "step": 111
    },
    {
      "epoch": 0.041040674239648224,
      "grad_norm": 0.43266763991725404,
      "learning_rate": 1.3553113553113553e-06,
      "loss": 0.5535,
      "step": 112
    },
    {
      "epoch": 0.04140710883107365,
      "grad_norm": 0.36335239503287947,
      "learning_rate": 1.3675213675213678e-06,
      "loss": 0.551,
      "step": 113
    },
    {
      "epoch": 0.04177354342249908,
      "grad_norm": 0.4153413011178574,
      "learning_rate": 1.37973137973138e-06,
      "loss": 0.5573,
      "step": 114
    },
    {
      "epoch": 0.042139978013924516,
      "grad_norm": 0.37097305870920927,
      "learning_rate": 1.391941391941392e-06,
      "loss": 0.5348,
      "step": 115
    },
    {
      "epoch": 0.042506412605349944,
      "grad_norm": 0.36043828541231127,
      "learning_rate": 1.4041514041514043e-06,
      "loss": 0.5308,
      "step": 116
    },
    {
      "epoch": 0.04287284719677537,
      "grad_norm": 0.3828118608340629,
      "learning_rate": 1.4163614163614164e-06,
      "loss": 0.5421,
      "step": 117
    },
    {
      "epoch": 0.04323928178820081,
      "grad_norm": 0.4195026538308934,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.5151,
      "step": 118
    },
    {
      "epoch": 0.043605716379626236,
      "grad_norm": 0.36815761922773166,
      "learning_rate": 1.4407814407814407e-06,
      "loss": 0.5439,
      "step": 119
    },
    {
      "epoch": 0.043972150971051664,
      "grad_norm": 0.34679404339894404,
      "learning_rate": 1.4529914529914531e-06,
      "loss": 0.5557,
      "step": 120
    },
    {
      "epoch": 0.0443385855624771,
      "grad_norm": 0.3680217285092968,
      "learning_rate": 1.4652014652014654e-06,
      "loss": 0.5471,
      "step": 121
    },
    {
      "epoch": 0.04470502015390253,
      "grad_norm": 0.34997693157481913,
      "learning_rate": 1.4774114774114777e-06,
      "loss": 0.5398,
      "step": 122
    },
    {
      "epoch": 0.045071454745327956,
      "grad_norm": 0.3901630835625426,
      "learning_rate": 1.4896214896214897e-06,
      "loss": 0.5573,
      "step": 123
    },
    {
      "epoch": 0.04543788933675339,
      "grad_norm": 0.3304358496064509,
      "learning_rate": 1.501831501831502e-06,
      "loss": 0.5206,
      "step": 124
    },
    {
      "epoch": 0.04580432392817882,
      "grad_norm": 0.39773122077087064,
      "learning_rate": 1.514041514041514e-06,
      "loss": 0.5276,
      "step": 125
    },
    {
      "epoch": 0.04617075851960425,
      "grad_norm": 0.32109427289877673,
      "learning_rate": 1.5262515262515265e-06,
      "loss": 0.5363,
      "step": 126
    },
    {
      "epoch": 0.04653719311102968,
      "grad_norm": 0.6934167705522686,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 0.5466,
      "step": 127
    },
    {
      "epoch": 0.04690362770245511,
      "grad_norm": 0.3779084453393618,
      "learning_rate": 1.5506715506715508e-06,
      "loss": 0.569,
      "step": 128
    },
    {
      "epoch": 0.04727006229388054,
      "grad_norm": 0.33994513955915,
      "learning_rate": 1.562881562881563e-06,
      "loss": 0.5494,
      "step": 129
    },
    {
      "epoch": 0.047636496885305975,
      "grad_norm": 0.33530594349843995,
      "learning_rate": 1.5750915750915753e-06,
      "loss": 0.5326,
      "step": 130
    },
    {
      "epoch": 0.0480029314767314,
      "grad_norm": 0.35764295984125394,
      "learning_rate": 1.5873015873015873e-06,
      "loss": 0.508,
      "step": 131
    },
    {
      "epoch": 0.04836936606815683,
      "grad_norm": 0.3593148642247137,
      "learning_rate": 1.5995115995115996e-06,
      "loss": 0.5277,
      "step": 132
    },
    {
      "epoch": 0.04873580065958227,
      "grad_norm": 0.347344780043152,
      "learning_rate": 1.611721611721612e-06,
      "loss": 0.5325,
      "step": 133
    },
    {
      "epoch": 0.049102235251007695,
      "grad_norm": 1.4242015168137205,
      "learning_rate": 1.623931623931624e-06,
      "loss": 0.5353,
      "step": 134
    },
    {
      "epoch": 0.04946866984243312,
      "grad_norm": 0.4006425078712935,
      "learning_rate": 1.6361416361416363e-06,
      "loss": 0.5328,
      "step": 135
    },
    {
      "epoch": 0.04983510443385856,
      "grad_norm": 0.35198715548867404,
      "learning_rate": 1.6483516483516484e-06,
      "loss": 0.5543,
      "step": 136
    },
    {
      "epoch": 0.05020153902528399,
      "grad_norm": 0.33179423453246554,
      "learning_rate": 1.6605616605616606e-06,
      "loss": 0.5458,
      "step": 137
    },
    {
      "epoch": 0.050567973616709415,
      "grad_norm": 0.3577753115186785,
      "learning_rate": 1.672771672771673e-06,
      "loss": 0.5269,
      "step": 138
    },
    {
      "epoch": 0.05093440820813485,
      "grad_norm": 0.38278109855797043,
      "learning_rate": 1.6849816849816852e-06,
      "loss": 0.5608,
      "step": 139
    },
    {
      "epoch": 0.05130084279956028,
      "grad_norm": 0.39080181765686434,
      "learning_rate": 1.6971916971916974e-06,
      "loss": 0.5415,
      "step": 140
    },
    {
      "epoch": 0.05166727739098571,
      "grad_norm": 0.37290047518792696,
      "learning_rate": 1.7094017094017097e-06,
      "loss": 0.5589,
      "step": 141
    },
    {
      "epoch": 0.05203371198241114,
      "grad_norm": 0.32996121172562964,
      "learning_rate": 1.7216117216117217e-06,
      "loss": 0.5201,
      "step": 142
    },
    {
      "epoch": 0.05240014657383657,
      "grad_norm": 0.35161425954756237,
      "learning_rate": 1.733821733821734e-06,
      "loss": 0.566,
      "step": 143
    },
    {
      "epoch": 0.052766581165262,
      "grad_norm": 0.36397891094261997,
      "learning_rate": 1.746031746031746e-06,
      "loss": 0.5427,
      "step": 144
    },
    {
      "epoch": 0.053133015756687434,
      "grad_norm": 0.343308015539286,
      "learning_rate": 1.7582417582417585e-06,
      "loss": 0.5388,
      "step": 145
    },
    {
      "epoch": 0.05349945034811286,
      "grad_norm": 0.381334397994181,
      "learning_rate": 1.7704517704517707e-06,
      "loss": 0.5545,
      "step": 146
    },
    {
      "epoch": 0.05386588493953829,
      "grad_norm": 0.3608883259039496,
      "learning_rate": 1.7826617826617828e-06,
      "loss": 0.5403,
      "step": 147
    },
    {
      "epoch": 0.054232319530963725,
      "grad_norm": 0.35063011769040275,
      "learning_rate": 1.794871794871795e-06,
      "loss": 0.5468,
      "step": 148
    },
    {
      "epoch": 0.054598754122389154,
      "grad_norm": 0.338415995491298,
      "learning_rate": 1.8070818070818073e-06,
      "loss": 0.5507,
      "step": 149
    },
    {
      "epoch": 0.05496518871381458,
      "grad_norm": 0.3441392124305741,
      "learning_rate": 1.8192918192918193e-06,
      "loss": 0.5505,
      "step": 150
    },
    {
      "epoch": 0.05533162330524002,
      "grad_norm": 0.33324357923846754,
      "learning_rate": 1.8315018315018316e-06,
      "loss": 0.5363,
      "step": 151
    },
    {
      "epoch": 0.055698057896665445,
      "grad_norm": 0.4215796294421377,
      "learning_rate": 1.843711843711844e-06,
      "loss": 0.5378,
      "step": 152
    },
    {
      "epoch": 0.056064492488090874,
      "grad_norm": 0.4091386063631755,
      "learning_rate": 1.855921855921856e-06,
      "loss": 0.5664,
      "step": 153
    },
    {
      "epoch": 0.05643092707951631,
      "grad_norm": 0.3542462977884497,
      "learning_rate": 1.8681318681318684e-06,
      "loss": 0.5096,
      "step": 154
    },
    {
      "epoch": 0.05679736167094174,
      "grad_norm": 0.3528607459761826,
      "learning_rate": 1.8803418803418804e-06,
      "loss": 0.5632,
      "step": 155
    },
    {
      "epoch": 0.057163796262367166,
      "grad_norm": 0.33329033358528964,
      "learning_rate": 1.8925518925518927e-06,
      "loss": 0.5114,
      "step": 156
    },
    {
      "epoch": 0.0575302308537926,
      "grad_norm": 0.39902834210753574,
      "learning_rate": 1.904761904761905e-06,
      "loss": 0.5249,
      "step": 157
    },
    {
      "epoch": 0.05789666544521803,
      "grad_norm": 0.3890087866452553,
      "learning_rate": 1.916971916971917e-06,
      "loss": 0.5574,
      "step": 158
    },
    {
      "epoch": 0.05826310003664346,
      "grad_norm": 0.3514321869072177,
      "learning_rate": 1.9291819291819296e-06,
      "loss": 0.5313,
      "step": 159
    },
    {
      "epoch": 0.05862953462806889,
      "grad_norm": 0.3606835778795525,
      "learning_rate": 1.9413919413919417e-06,
      "loss": 0.5512,
      "step": 160
    },
    {
      "epoch": 0.05899596921949432,
      "grad_norm": 0.3560548307265957,
      "learning_rate": 1.9536019536019537e-06,
      "loss": 0.5261,
      "step": 161
    },
    {
      "epoch": 0.05936240381091975,
      "grad_norm": 0.36540490598922515,
      "learning_rate": 1.9658119658119658e-06,
      "loss": 0.5291,
      "step": 162
    },
    {
      "epoch": 0.059728838402345184,
      "grad_norm": 0.47161323204830946,
      "learning_rate": 1.9780219780219782e-06,
      "loss": 0.5242,
      "step": 163
    },
    {
      "epoch": 0.06009527299377061,
      "grad_norm": 0.3935968764089347,
      "learning_rate": 1.9902319902319903e-06,
      "loss": 0.5331,
      "step": 164
    },
    {
      "epoch": 0.06046170758519604,
      "grad_norm": 0.33786419087241454,
      "learning_rate": 2.0024420024420027e-06,
      "loss": 0.5208,
      "step": 165
    },
    {
      "epoch": 0.060828142176621476,
      "grad_norm": 0.3793668251062535,
      "learning_rate": 2.0146520146520148e-06,
      "loss": 0.555,
      "step": 166
    },
    {
      "epoch": 0.061194576768046904,
      "grad_norm": 0.3482932987976221,
      "learning_rate": 2.0268620268620273e-06,
      "loss": 0.5035,
      "step": 167
    },
    {
      "epoch": 0.06156101135947233,
      "grad_norm": 0.37515141892762793,
      "learning_rate": 2.0390720390720393e-06,
      "loss": 0.5238,
      "step": 168
    },
    {
      "epoch": 0.06192744595089777,
      "grad_norm": 0.3633414629642113,
      "learning_rate": 2.0512820512820513e-06,
      "loss": 0.5223,
      "step": 169
    },
    {
      "epoch": 0.062293880542323196,
      "grad_norm": 0.37425945254848214,
      "learning_rate": 2.0634920634920634e-06,
      "loss": 0.5243,
      "step": 170
    },
    {
      "epoch": 0.06266031513374863,
      "grad_norm": 0.33502608630070185,
      "learning_rate": 2.075702075702076e-06,
      "loss": 0.5152,
      "step": 171
    },
    {
      "epoch": 0.06302674972517405,
      "grad_norm": 0.3593712088161955,
      "learning_rate": 2.0879120879120883e-06,
      "loss": 0.562,
      "step": 172
    },
    {
      "epoch": 0.06339318431659949,
      "grad_norm": 0.35404752950984764,
      "learning_rate": 2.1001221001221004e-06,
      "loss": 0.5324,
      "step": 173
    },
    {
      "epoch": 0.06375961890802492,
      "grad_norm": 0.34895974600550317,
      "learning_rate": 2.1123321123321124e-06,
      "loss": 0.5339,
      "step": 174
    },
    {
      "epoch": 0.06412605349945034,
      "grad_norm": 0.40031194710138374,
      "learning_rate": 2.1245421245421245e-06,
      "loss": 0.528,
      "step": 175
    },
    {
      "epoch": 0.06449248809087578,
      "grad_norm": 0.3456289105929941,
      "learning_rate": 2.136752136752137e-06,
      "loss": 0.5297,
      "step": 176
    },
    {
      "epoch": 0.06485892268230121,
      "grad_norm": 0.3369134713820974,
      "learning_rate": 2.1489621489621494e-06,
      "loss": 0.5155,
      "step": 177
    },
    {
      "epoch": 0.06522535727372664,
      "grad_norm": 0.3704969402379825,
      "learning_rate": 2.1611721611721614e-06,
      "loss": 0.5031,
      "step": 178
    },
    {
      "epoch": 0.06559179186515207,
      "grad_norm": 0.40485291339411966,
      "learning_rate": 2.1733821733821735e-06,
      "loss": 0.5266,
      "step": 179
    },
    {
      "epoch": 0.0659582264565775,
      "grad_norm": 0.3792908643986255,
      "learning_rate": 2.185592185592186e-06,
      "loss": 0.519,
      "step": 180
    },
    {
      "epoch": 0.06632466104800293,
      "grad_norm": 0.36762577485052184,
      "learning_rate": 2.197802197802198e-06,
      "loss": 0.529,
      "step": 181
    },
    {
      "epoch": 0.06669109563942836,
      "grad_norm": 0.39823377978900054,
      "learning_rate": 2.21001221001221e-06,
      "loss": 0.5358,
      "step": 182
    },
    {
      "epoch": 0.0670575302308538,
      "grad_norm": 0.35712015130246566,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.5451,
      "step": 183
    },
    {
      "epoch": 0.06742396482227922,
      "grad_norm": 0.36156899659798497,
      "learning_rate": 2.2344322344322345e-06,
      "loss": 0.5395,
      "step": 184
    },
    {
      "epoch": 0.06779039941370466,
      "grad_norm": 0.5255442510793505,
      "learning_rate": 2.246642246642247e-06,
      "loss": 0.54,
      "step": 185
    },
    {
      "epoch": 0.06815683400513009,
      "grad_norm": 0.3879141917396287,
      "learning_rate": 2.258852258852259e-06,
      "loss": 0.5329,
      "step": 186
    },
    {
      "epoch": 0.06852326859655551,
      "grad_norm": 0.3966066021362682,
      "learning_rate": 2.271062271062271e-06,
      "loss": 0.5575,
      "step": 187
    },
    {
      "epoch": 0.06888970318798095,
      "grad_norm": 0.347663879787735,
      "learning_rate": 2.2832722832722836e-06,
      "loss": 0.5195,
      "step": 188
    },
    {
      "epoch": 0.06925613777940638,
      "grad_norm": 0.3573642214365536,
      "learning_rate": 2.2954822954822956e-06,
      "loss": 0.5383,
      "step": 189
    },
    {
      "epoch": 0.0696225723708318,
      "grad_norm": 0.3581984456157912,
      "learning_rate": 2.307692307692308e-06,
      "loss": 0.5275,
      "step": 190
    },
    {
      "epoch": 0.06998900696225724,
      "grad_norm": 0.38750036381408,
      "learning_rate": 2.31990231990232e-06,
      "loss": 0.524,
      "step": 191
    },
    {
      "epoch": 0.07035544155368267,
      "grad_norm": 0.40461624534271157,
      "learning_rate": 2.332112332112332e-06,
      "loss": 0.5217,
      "step": 192
    },
    {
      "epoch": 0.0707218761451081,
      "grad_norm": 0.3399037913434264,
      "learning_rate": 2.3443223443223446e-06,
      "loss": 0.5134,
      "step": 193
    },
    {
      "epoch": 0.07108831073653353,
      "grad_norm": 0.3737874764005105,
      "learning_rate": 2.3565323565323567e-06,
      "loss": 0.5478,
      "step": 194
    },
    {
      "epoch": 0.07145474532795897,
      "grad_norm": 0.35882004982582666,
      "learning_rate": 2.3687423687423687e-06,
      "loss": 0.5186,
      "step": 195
    },
    {
      "epoch": 0.07182117991938439,
      "grad_norm": 0.3503172035013749,
      "learning_rate": 2.380952380952381e-06,
      "loss": 0.5278,
      "step": 196
    },
    {
      "epoch": 0.07218761451080982,
      "grad_norm": 0.3612356698887949,
      "learning_rate": 2.3931623931623937e-06,
      "loss": 0.5273,
      "step": 197
    },
    {
      "epoch": 0.07255404910223526,
      "grad_norm": 0.347130403325483,
      "learning_rate": 2.4053724053724057e-06,
      "loss": 0.536,
      "step": 198
    },
    {
      "epoch": 0.07292048369366068,
      "grad_norm": 0.3265129867337722,
      "learning_rate": 2.4175824175824177e-06,
      "loss": 0.4871,
      "step": 199
    },
    {
      "epoch": 0.07328691828508611,
      "grad_norm": 0.5076145326861746,
      "learning_rate": 2.4297924297924298e-06,
      "loss": 0.5347,
      "step": 200
    },
    {
      "epoch": 0.07365335287651155,
      "grad_norm": 0.38938325821327724,
      "learning_rate": 2.4420024420024423e-06,
      "loss": 0.5302,
      "step": 201
    },
    {
      "epoch": 0.07401978746793697,
      "grad_norm": 0.3856232406276879,
      "learning_rate": 2.4542124542124543e-06,
      "loss": 0.5373,
      "step": 202
    },
    {
      "epoch": 0.0743862220593624,
      "grad_norm": 0.3519413080503197,
      "learning_rate": 2.4664224664224668e-06,
      "loss": 0.5273,
      "step": 203
    },
    {
      "epoch": 0.07475265665078784,
      "grad_norm": 2.2664478645913575,
      "learning_rate": 2.478632478632479e-06,
      "loss": 0.5256,
      "step": 204
    },
    {
      "epoch": 0.07511909124221326,
      "grad_norm": 0.33607652058708043,
      "learning_rate": 2.4908424908424913e-06,
      "loss": 0.5225,
      "step": 205
    },
    {
      "epoch": 0.0754855258336387,
      "grad_norm": 0.3946515046825135,
      "learning_rate": 2.5030525030525033e-06,
      "loss": 0.522,
      "step": 206
    },
    {
      "epoch": 0.07585196042506413,
      "grad_norm": 0.3644003026667947,
      "learning_rate": 2.5152625152625154e-06,
      "loss": 0.5337,
      "step": 207
    },
    {
      "epoch": 0.07621839501648955,
      "grad_norm": 0.35192580002047213,
      "learning_rate": 2.5274725274725274e-06,
      "loss": 0.5197,
      "step": 208
    },
    {
      "epoch": 0.07658482960791499,
      "grad_norm": 0.7252255544226895,
      "learning_rate": 2.53968253968254e-06,
      "loss": 0.5327,
      "step": 209
    },
    {
      "epoch": 0.07695126419934042,
      "grad_norm": 0.3663672962701081,
      "learning_rate": 2.551892551892552e-06,
      "loss": 0.5293,
      "step": 210
    },
    {
      "epoch": 0.07731769879076585,
      "grad_norm": 0.3607175185553371,
      "learning_rate": 2.564102564102564e-06,
      "loss": 0.5053,
      "step": 211
    },
    {
      "epoch": 0.07768413338219128,
      "grad_norm": 0.34868336180824805,
      "learning_rate": 2.5763125763125764e-06,
      "loss": 0.537,
      "step": 212
    },
    {
      "epoch": 0.07805056797361672,
      "grad_norm": 0.669177604599102,
      "learning_rate": 2.588522588522589e-06,
      "loss": 0.5288,
      "step": 213
    },
    {
      "epoch": 0.07841700256504214,
      "grad_norm": 0.396280830969059,
      "learning_rate": 2.600732600732601e-06,
      "loss": 0.5117,
      "step": 214
    },
    {
      "epoch": 0.07878343715646757,
      "grad_norm": 0.3844393794383178,
      "learning_rate": 2.6129426129426134e-06,
      "loss": 0.5568,
      "step": 215
    },
    {
      "epoch": 0.079149871747893,
      "grad_norm": 0.3823359148727674,
      "learning_rate": 2.6251526251526255e-06,
      "loss": 0.5415,
      "step": 216
    },
    {
      "epoch": 0.07951630633931843,
      "grad_norm": 0.5635584140626799,
      "learning_rate": 2.6373626373626375e-06,
      "loss": 0.5089,
      "step": 217
    },
    {
      "epoch": 0.07988274093074386,
      "grad_norm": 0.312564654850148,
      "learning_rate": 2.64957264957265e-06,
      "loss": 0.4747,
      "step": 218
    },
    {
      "epoch": 0.08024917552216929,
      "grad_norm": 0.396728673108171,
      "learning_rate": 2.661782661782662e-06,
      "loss": 0.494,
      "step": 219
    },
    {
      "epoch": 0.08061561011359472,
      "grad_norm": 0.3643975443922188,
      "learning_rate": 2.673992673992674e-06,
      "loss": 0.5191,
      "step": 220
    },
    {
      "epoch": 0.08098204470502016,
      "grad_norm": 0.33410327624504016,
      "learning_rate": 2.6862026862026865e-06,
      "loss": 0.5187,
      "step": 221
    },
    {
      "epoch": 0.08134847929644558,
      "grad_norm": 0.32976712237723677,
      "learning_rate": 2.6984126984126986e-06,
      "loss": 0.5342,
      "step": 222
    },
    {
      "epoch": 0.08171491388787101,
      "grad_norm": 0.35964860807817395,
      "learning_rate": 2.7106227106227106e-06,
      "loss": 0.5041,
      "step": 223
    },
    {
      "epoch": 0.08208134847929645,
      "grad_norm": 0.35815861306085817,
      "learning_rate": 2.7228327228327227e-06,
      "loss": 0.5269,
      "step": 224
    },
    {
      "epoch": 0.08244778307072187,
      "grad_norm": 0.3861349608787283,
      "learning_rate": 2.7350427350427355e-06,
      "loss": 0.5223,
      "step": 225
    },
    {
      "epoch": 0.0828142176621473,
      "grad_norm": 0.3856921998398821,
      "learning_rate": 2.7472527472527476e-06,
      "loss": 0.5181,
      "step": 226
    },
    {
      "epoch": 0.08318065225357274,
      "grad_norm": 0.35483077259618995,
      "learning_rate": 2.75946275946276e-06,
      "loss": 0.5483,
      "step": 227
    },
    {
      "epoch": 0.08354708684499816,
      "grad_norm": 0.37534012429539515,
      "learning_rate": 2.771672771672772e-06,
      "loss": 0.5262,
      "step": 228
    },
    {
      "epoch": 0.0839135214364236,
      "grad_norm": 0.34303143169997097,
      "learning_rate": 2.783882783882784e-06,
      "loss": 0.5156,
      "step": 229
    },
    {
      "epoch": 0.08427995602784903,
      "grad_norm": 0.42547057457349224,
      "learning_rate": 2.796092796092796e-06,
      "loss": 0.5458,
      "step": 230
    },
    {
      "epoch": 0.08464639061927445,
      "grad_norm": 0.3650836319312793,
      "learning_rate": 2.8083028083028087e-06,
      "loss": 0.5148,
      "step": 231
    },
    {
      "epoch": 0.08501282521069989,
      "grad_norm": 0.361806036144833,
      "learning_rate": 2.8205128205128207e-06,
      "loss": 0.5361,
      "step": 232
    },
    {
      "epoch": 0.08537925980212532,
      "grad_norm": 0.38653297634085637,
      "learning_rate": 2.8327228327228327e-06,
      "loss": 0.5283,
      "step": 233
    },
    {
      "epoch": 0.08574569439355074,
      "grad_norm": 0.3326064609674727,
      "learning_rate": 2.844932844932845e-06,
      "loss": 0.5031,
      "step": 234
    },
    {
      "epoch": 0.08611212898497618,
      "grad_norm": 0.3561815656395291,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.505,
      "step": 235
    },
    {
      "epoch": 0.08647856357640162,
      "grad_norm": 0.6148608294384966,
      "learning_rate": 2.8693528693528693e-06,
      "loss": 0.5056,
      "step": 236
    },
    {
      "epoch": 0.08684499816782704,
      "grad_norm": 0.3300250761174936,
      "learning_rate": 2.8815628815628813e-06,
      "loss": 0.5139,
      "step": 237
    },
    {
      "epoch": 0.08721143275925247,
      "grad_norm": 0.3772690796714867,
      "learning_rate": 2.8937728937728942e-06,
      "loss": 0.5189,
      "step": 238
    },
    {
      "epoch": 0.08757786735067791,
      "grad_norm": 0.32047677562245575,
      "learning_rate": 2.9059829059829063e-06,
      "loss": 0.538,
      "step": 239
    },
    {
      "epoch": 0.08794430194210333,
      "grad_norm": 0.4227586130294547,
      "learning_rate": 2.9181929181929187e-06,
      "loss": 0.5319,
      "step": 240
    },
    {
      "epoch": 0.08831073653352876,
      "grad_norm": 0.4398930428566166,
      "learning_rate": 2.930402930402931e-06,
      "loss": 0.4944,
      "step": 241
    },
    {
      "epoch": 0.0886771711249542,
      "grad_norm": 0.3261763137279799,
      "learning_rate": 2.942612942612943e-06,
      "loss": 0.5173,
      "step": 242
    },
    {
      "epoch": 0.08904360571637962,
      "grad_norm": 0.35816855541535253,
      "learning_rate": 2.9548229548229553e-06,
      "loss": 0.5243,
      "step": 243
    },
    {
      "epoch": 0.08941004030780506,
      "grad_norm": 0.3545429445657612,
      "learning_rate": 2.9670329670329673e-06,
      "loss": 0.4954,
      "step": 244
    },
    {
      "epoch": 0.08977647489923049,
      "grad_norm": 0.3317297088582488,
      "learning_rate": 2.9792429792429794e-06,
      "loss": 0.5196,
      "step": 245
    },
    {
      "epoch": 0.09014290949065591,
      "grad_norm": 0.3785026663722004,
      "learning_rate": 2.9914529914529914e-06,
      "loss": 0.5143,
      "step": 246
    },
    {
      "epoch": 0.09050934408208135,
      "grad_norm": 0.3777975939161701,
      "learning_rate": 3.003663003663004e-06,
      "loss": 0.5078,
      "step": 247
    },
    {
      "epoch": 0.09087577867350678,
      "grad_norm": 0.3454676039547994,
      "learning_rate": 3.015873015873016e-06,
      "loss": 0.5423,
      "step": 248
    },
    {
      "epoch": 0.0912422132649322,
      "grad_norm": 0.33454970200915446,
      "learning_rate": 3.028083028083028e-06,
      "loss": 0.4844,
      "step": 249
    },
    {
      "epoch": 0.09160864785635764,
      "grad_norm": 0.36725021673557684,
      "learning_rate": 3.0402930402930405e-06,
      "loss": 0.5184,
      "step": 250
    },
    {
      "epoch": 0.09197508244778307,
      "grad_norm": 0.3999063613637315,
      "learning_rate": 3.052503052503053e-06,
      "loss": 0.5178,
      "step": 251
    },
    {
      "epoch": 0.0923415170392085,
      "grad_norm": 0.360816172734895,
      "learning_rate": 3.064713064713065e-06,
      "loss": 0.5376,
      "step": 252
    },
    {
      "epoch": 0.09270795163063393,
      "grad_norm": 1.0073822568191362,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 0.5296,
      "step": 253
    },
    {
      "epoch": 0.09307438622205937,
      "grad_norm": 0.37114240846755364,
      "learning_rate": 3.0891330891330895e-06,
      "loss": 0.4947,
      "step": 254
    },
    {
      "epoch": 0.09344082081348479,
      "grad_norm": 0.4291769023817346,
      "learning_rate": 3.1013431013431015e-06,
      "loss": 0.5362,
      "step": 255
    },
    {
      "epoch": 0.09380725540491022,
      "grad_norm": 0.3283569477188691,
      "learning_rate": 3.113553113553114e-06,
      "loss": 0.5221,
      "step": 256
    },
    {
      "epoch": 0.09417368999633566,
      "grad_norm": 0.37115962316193013,
      "learning_rate": 3.125763125763126e-06,
      "loss": 0.5264,
      "step": 257
    },
    {
      "epoch": 0.09454012458776108,
      "grad_norm": 0.37176748263556864,
      "learning_rate": 3.137973137973138e-06,
      "loss": 0.5143,
      "step": 258
    },
    {
      "epoch": 0.09490655917918651,
      "grad_norm": 0.37906391783752247,
      "learning_rate": 3.1501831501831505e-06,
      "loss": 0.5289,
      "step": 259
    },
    {
      "epoch": 0.09527299377061195,
      "grad_norm": 0.3791036966506715,
      "learning_rate": 3.1623931623931626e-06,
      "loss": 0.5086,
      "step": 260
    },
    {
      "epoch": 0.09563942836203737,
      "grad_norm": 0.31177447874992353,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 0.5114,
      "step": 261
    },
    {
      "epoch": 0.0960058629534628,
      "grad_norm": 0.4043457355335449,
      "learning_rate": 3.1868131868131867e-06,
      "loss": 0.5353,
      "step": 262
    },
    {
      "epoch": 0.09637229754488824,
      "grad_norm": 0.35385450626262005,
      "learning_rate": 3.199023199023199e-06,
      "loss": 0.5188,
      "step": 263
    },
    {
      "epoch": 0.09673873213631366,
      "grad_norm": 0.41364658207516064,
      "learning_rate": 3.2112332112332116e-06,
      "loss": 0.5166,
      "step": 264
    },
    {
      "epoch": 0.0971051667277391,
      "grad_norm": 0.3426491945207554,
      "learning_rate": 3.223443223443224e-06,
      "loss": 0.5304,
      "step": 265
    },
    {
      "epoch": 0.09747160131916453,
      "grad_norm": 0.4318806862490417,
      "learning_rate": 3.235653235653236e-06,
      "loss": 0.5425,
      "step": 266
    },
    {
      "epoch": 0.09783803591058995,
      "grad_norm": 0.3543647055833978,
      "learning_rate": 3.247863247863248e-06,
      "loss": 0.5246,
      "step": 267
    },
    {
      "epoch": 0.09820447050201539,
      "grad_norm": 0.35084439456832456,
      "learning_rate": 3.2600732600732602e-06,
      "loss": 0.5074,
      "step": 268
    },
    {
      "epoch": 0.09857090509344082,
      "grad_norm": 0.36396215183869723,
      "learning_rate": 3.2722832722832727e-06,
      "loss": 0.5125,
      "step": 269
    },
    {
      "epoch": 0.09893733968486625,
      "grad_norm": 0.43814199825744815,
      "learning_rate": 3.2844932844932847e-06,
      "loss": 0.4971,
      "step": 270
    },
    {
      "epoch": 0.09930377427629168,
      "grad_norm": 0.3759597593689997,
      "learning_rate": 3.2967032967032968e-06,
      "loss": 0.5101,
      "step": 271
    },
    {
      "epoch": 0.09967020886771712,
      "grad_norm": 0.3591472818993172,
      "learning_rate": 3.3089133089133092e-06,
      "loss": 0.5166,
      "step": 272
    },
    {
      "epoch": 0.10003664345914254,
      "grad_norm": 0.3276307967273574,
      "learning_rate": 3.3211233211233213e-06,
      "loss": 0.5216,
      "step": 273
    },
    {
      "epoch": 0.10040307805056797,
      "grad_norm": 0.3446519769743233,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.5212,
      "step": 274
    },
    {
      "epoch": 0.10076951264199341,
      "grad_norm": 0.36453509255880256,
      "learning_rate": 3.345543345543346e-06,
      "loss": 0.5073,
      "step": 275
    },
    {
      "epoch": 0.10113594723341883,
      "grad_norm": 0.37334309656390274,
      "learning_rate": 3.357753357753358e-06,
      "loss": 0.4882,
      "step": 276
    },
    {
      "epoch": 0.10150238182484426,
      "grad_norm": 0.3922726958808869,
      "learning_rate": 3.3699633699633703e-06,
      "loss": 0.4901,
      "step": 277
    },
    {
      "epoch": 0.1018688164162697,
      "grad_norm": 0.3545354295533298,
      "learning_rate": 3.3821733821733828e-06,
      "loss": 0.517,
      "step": 278
    },
    {
      "epoch": 0.10223525100769512,
      "grad_norm": 0.3485623014464545,
      "learning_rate": 3.394383394383395e-06,
      "loss": 0.5125,
      "step": 279
    },
    {
      "epoch": 0.10260168559912056,
      "grad_norm": 0.3610600278115916,
      "learning_rate": 3.406593406593407e-06,
      "loss": 0.5181,
      "step": 280
    },
    {
      "epoch": 0.10296812019054599,
      "grad_norm": 0.35748159321110057,
      "learning_rate": 3.4188034188034193e-06,
      "loss": 0.5166,
      "step": 281
    },
    {
      "epoch": 0.10333455478197141,
      "grad_norm": 0.341969694182971,
      "learning_rate": 3.4310134310134314e-06,
      "loss": 0.5167,
      "step": 282
    },
    {
      "epoch": 0.10370098937339685,
      "grad_norm": 0.3634362782633189,
      "learning_rate": 3.4432234432234434e-06,
      "loss": 0.5071,
      "step": 283
    },
    {
      "epoch": 0.10406742396482228,
      "grad_norm": 0.3917607055780976,
      "learning_rate": 3.4554334554334555e-06,
      "loss": 0.5341,
      "step": 284
    },
    {
      "epoch": 0.1044338585562477,
      "grad_norm": 0.36861896550886264,
      "learning_rate": 3.467643467643468e-06,
      "loss": 0.495,
      "step": 285
    },
    {
      "epoch": 0.10480029314767314,
      "grad_norm": 0.34715617467326676,
      "learning_rate": 3.47985347985348e-06,
      "loss": 0.5067,
      "step": 286
    },
    {
      "epoch": 0.10516672773909858,
      "grad_norm": 0.4036602754309099,
      "learning_rate": 3.492063492063492e-06,
      "loss": 0.5048,
      "step": 287
    },
    {
      "epoch": 0.105533162330524,
      "grad_norm": 0.31404581692939654,
      "learning_rate": 3.5042735042735045e-06,
      "loss": 0.5026,
      "step": 288
    },
    {
      "epoch": 0.10589959692194943,
      "grad_norm": 0.42594564260540174,
      "learning_rate": 3.516483516483517e-06,
      "loss": 0.5251,
      "step": 289
    },
    {
      "epoch": 0.10626603151337487,
      "grad_norm": 0.3514547314265413,
      "learning_rate": 3.5286935286935294e-06,
      "loss": 0.504,
      "step": 290
    },
    {
      "epoch": 0.10663246610480029,
      "grad_norm": 0.3607754884717036,
      "learning_rate": 3.5409035409035415e-06,
      "loss": 0.5249,
      "step": 291
    },
    {
      "epoch": 0.10699890069622572,
      "grad_norm": 0.33139133012558614,
      "learning_rate": 3.5531135531135535e-06,
      "loss": 0.5207,
      "step": 292
    },
    {
      "epoch": 0.10736533528765116,
      "grad_norm": 0.3483037843848268,
      "learning_rate": 3.5653235653235655e-06,
      "loss": 0.507,
      "step": 293
    },
    {
      "epoch": 0.10773176987907658,
      "grad_norm": 0.3928683574793009,
      "learning_rate": 3.577533577533578e-06,
      "loss": 0.5296,
      "step": 294
    },
    {
      "epoch": 0.10809820447050202,
      "grad_norm": 0.6399127362812426,
      "learning_rate": 3.58974358974359e-06,
      "loss": 0.5052,
      "step": 295
    },
    {
      "epoch": 0.10846463906192745,
      "grad_norm": 0.3066146787672902,
      "learning_rate": 3.601953601953602e-06,
      "loss": 0.4936,
      "step": 296
    },
    {
      "epoch": 0.10883107365335287,
      "grad_norm": 0.33355708715604954,
      "learning_rate": 3.6141636141636146e-06,
      "loss": 0.5275,
      "step": 297
    },
    {
      "epoch": 0.10919750824477831,
      "grad_norm": 0.36770500447174703,
      "learning_rate": 3.6263736263736266e-06,
      "loss": 0.5053,
      "step": 298
    },
    {
      "epoch": 0.10956394283620374,
      "grad_norm": 0.37117382361133794,
      "learning_rate": 3.6385836385836387e-06,
      "loss": 0.5223,
      "step": 299
    },
    {
      "epoch": 0.10993037742762916,
      "grad_norm": 0.3047434570891219,
      "learning_rate": 3.6507936507936507e-06,
      "loss": 0.5263,
      "step": 300
    },
    {
      "epoch": 0.1102968120190546,
      "grad_norm": 0.357442418827966,
      "learning_rate": 3.663003663003663e-06,
      "loss": 0.4981,
      "step": 301
    },
    {
      "epoch": 0.11066324661048003,
      "grad_norm": 0.3417564368518373,
      "learning_rate": 3.6752136752136756e-06,
      "loss": 0.513,
      "step": 302
    },
    {
      "epoch": 0.11102968120190546,
      "grad_norm": 0.35848204083252466,
      "learning_rate": 3.687423687423688e-06,
      "loss": 0.4933,
      "step": 303
    },
    {
      "epoch": 0.11139611579333089,
      "grad_norm": 0.38837611993673493,
      "learning_rate": 3.6996336996337e-06,
      "loss": 0.4977,
      "step": 304
    },
    {
      "epoch": 0.11176255038475633,
      "grad_norm": 0.34814577723147017,
      "learning_rate": 3.711843711843712e-06,
      "loss": 0.5119,
      "step": 305
    },
    {
      "epoch": 0.11212898497618175,
      "grad_norm": 0.3347699819901902,
      "learning_rate": 3.7240537240537247e-06,
      "loss": 0.5099,
      "step": 306
    },
    {
      "epoch": 0.11249541956760718,
      "grad_norm": 0.3402634522326159,
      "learning_rate": 3.7362637362637367e-06,
      "loss": 0.5052,
      "step": 307
    },
    {
      "epoch": 0.11286185415903262,
      "grad_norm": 0.4050140632289179,
      "learning_rate": 3.7484737484737487e-06,
      "loss": 0.4921,
      "step": 308
    },
    {
      "epoch": 0.11322828875045804,
      "grad_norm": 0.3683757778678962,
      "learning_rate": 3.760683760683761e-06,
      "loss": 0.5265,
      "step": 309
    },
    {
      "epoch": 0.11359472334188347,
      "grad_norm": 0.33066821544949443,
      "learning_rate": 3.7728937728937733e-06,
      "loss": 0.5105,
      "step": 310
    },
    {
      "epoch": 0.11396115793330891,
      "grad_norm": 0.3554102507894874,
      "learning_rate": 3.7851037851037853e-06,
      "loss": 0.5322,
      "step": 311
    },
    {
      "epoch": 0.11432759252473433,
      "grad_norm": 0.36247663480765974,
      "learning_rate": 3.7973137973137973e-06,
      "loss": 0.523,
      "step": 312
    },
    {
      "epoch": 0.11469402711615977,
      "grad_norm": 0.36090951998889104,
      "learning_rate": 3.80952380952381e-06,
      "loss": 0.51,
      "step": 313
    },
    {
      "epoch": 0.1150604617075852,
      "grad_norm": 0.3963295692929509,
      "learning_rate": 3.821733821733822e-06,
      "loss": 0.5181,
      "step": 314
    },
    {
      "epoch": 0.11542689629901062,
      "grad_norm": 0.3675461339687006,
      "learning_rate": 3.833943833943834e-06,
      "loss": 0.5041,
      "step": 315
    },
    {
      "epoch": 0.11579333089043606,
      "grad_norm": 0.32562275507741045,
      "learning_rate": 3.846153846153847e-06,
      "loss": 0.5052,
      "step": 316
    },
    {
      "epoch": 0.1161597654818615,
      "grad_norm": 0.3744020711911561,
      "learning_rate": 3.858363858363859e-06,
      "loss": 0.493,
      "step": 317
    },
    {
      "epoch": 0.11652620007328691,
      "grad_norm": 0.3451440711657347,
      "learning_rate": 3.870573870573871e-06,
      "loss": 0.5036,
      "step": 318
    },
    {
      "epoch": 0.11689263466471235,
      "grad_norm": 0.37707128466999035,
      "learning_rate": 3.882783882783883e-06,
      "loss": 0.5074,
      "step": 319
    },
    {
      "epoch": 0.11725906925613779,
      "grad_norm": 0.33611219313946,
      "learning_rate": 3.894993894993895e-06,
      "loss": 0.4864,
      "step": 320
    },
    {
      "epoch": 0.1176255038475632,
      "grad_norm": 0.4197396773361997,
      "learning_rate": 3.9072039072039074e-06,
      "loss": 0.5122,
      "step": 321
    },
    {
      "epoch": 0.11799193843898864,
      "grad_norm": 0.6338102509714552,
      "learning_rate": 3.91941391941392e-06,
      "loss": 0.5118,
      "step": 322
    },
    {
      "epoch": 0.11835837303041408,
      "grad_norm": 0.3734002510438614,
      "learning_rate": 3.9316239316239315e-06,
      "loss": 0.5186,
      "step": 323
    },
    {
      "epoch": 0.1187248076218395,
      "grad_norm": 0.4303830166253478,
      "learning_rate": 3.943833943833944e-06,
      "loss": 0.5263,
      "step": 324
    },
    {
      "epoch": 0.11909124221326493,
      "grad_norm": 0.36220257082984264,
      "learning_rate": 3.9560439560439565e-06,
      "loss": 0.5161,
      "step": 325
    },
    {
      "epoch": 0.11945767680469037,
      "grad_norm": 0.41630911327298314,
      "learning_rate": 3.968253968253968e-06,
      "loss": 0.5159,
      "step": 326
    },
    {
      "epoch": 0.11982411139611579,
      "grad_norm": 0.3577722169689831,
      "learning_rate": 3.9804639804639805e-06,
      "loss": 0.5048,
      "step": 327
    },
    {
      "epoch": 0.12019054598754123,
      "grad_norm": 0.34550099625898467,
      "learning_rate": 3.992673992673993e-06,
      "loss": 0.4866,
      "step": 328
    },
    {
      "epoch": 0.12055698057896666,
      "grad_norm": 0.45949892953915134,
      "learning_rate": 4.0048840048840055e-06,
      "loss": 0.5222,
      "step": 329
    },
    {
      "epoch": 0.12092341517039208,
      "grad_norm": 0.3673202762807856,
      "learning_rate": 4.017094017094018e-06,
      "loss": 0.5141,
      "step": 330
    },
    {
      "epoch": 0.12128984976181752,
      "grad_norm": 0.3669114511660381,
      "learning_rate": 4.0293040293040296e-06,
      "loss": 0.5034,
      "step": 331
    },
    {
      "epoch": 0.12165628435324295,
      "grad_norm": 0.920703751354067,
      "learning_rate": 4.041514041514042e-06,
      "loss": 0.5181,
      "step": 332
    },
    {
      "epoch": 0.12202271894466837,
      "grad_norm": 0.39348823041750614,
      "learning_rate": 4.0537240537240545e-06,
      "loss": 0.5342,
      "step": 333
    },
    {
      "epoch": 0.12238915353609381,
      "grad_norm": 0.3748253284323868,
      "learning_rate": 4.065934065934066e-06,
      "loss": 0.504,
      "step": 334
    },
    {
      "epoch": 0.12275558812751924,
      "grad_norm": 0.3986247116515705,
      "learning_rate": 4.078144078144079e-06,
      "loss": 0.5082,
      "step": 335
    },
    {
      "epoch": 0.12312202271894467,
      "grad_norm": 0.3362694451042654,
      "learning_rate": 4.09035409035409e-06,
      "loss": 0.5096,
      "step": 336
    },
    {
      "epoch": 0.1234884573103701,
      "grad_norm": 0.4242490150102909,
      "learning_rate": 4.102564102564103e-06,
      "loss": 0.4913,
      "step": 337
    },
    {
      "epoch": 0.12385489190179554,
      "grad_norm": 0.36134267627586586,
      "learning_rate": 4.114774114774115e-06,
      "loss": 0.5175,
      "step": 338
    },
    {
      "epoch": 0.12422132649322096,
      "grad_norm": 0.3853835714947941,
      "learning_rate": 4.126984126984127e-06,
      "loss": 0.5513,
      "step": 339
    },
    {
      "epoch": 0.12458776108464639,
      "grad_norm": 0.3381897838666653,
      "learning_rate": 4.139194139194139e-06,
      "loss": 0.5158,
      "step": 340
    },
    {
      "epoch": 0.12495419567607183,
      "grad_norm": 0.3600615010663189,
      "learning_rate": 4.151404151404152e-06,
      "loss": 0.5046,
      "step": 341
    },
    {
      "epoch": 0.12532063026749726,
      "grad_norm": 0.38943558103164927,
      "learning_rate": 4.163614163614164e-06,
      "loss": 0.5248,
      "step": 342
    },
    {
      "epoch": 0.12568706485892267,
      "grad_norm": 0.3749401777689524,
      "learning_rate": 4.175824175824177e-06,
      "loss": 0.5089,
      "step": 343
    },
    {
      "epoch": 0.1260534994503481,
      "grad_norm": 0.34933169698314287,
      "learning_rate": 4.188034188034188e-06,
      "loss": 0.5233,
      "step": 344
    },
    {
      "epoch": 0.12641993404177354,
      "grad_norm": 0.3664125897104732,
      "learning_rate": 4.200244200244201e-06,
      "loss": 0.5125,
      "step": 345
    },
    {
      "epoch": 0.12678636863319898,
      "grad_norm": 0.3620158815805454,
      "learning_rate": 4.212454212454213e-06,
      "loss": 0.5138,
      "step": 346
    },
    {
      "epoch": 0.1271528032246244,
      "grad_norm": 0.4335140084252374,
      "learning_rate": 4.224664224664225e-06,
      "loss": 0.4946,
      "step": 347
    },
    {
      "epoch": 0.12751923781604985,
      "grad_norm": 0.35928323450493643,
      "learning_rate": 4.236874236874237e-06,
      "loss": 0.5053,
      "step": 348
    },
    {
      "epoch": 0.12788567240747525,
      "grad_norm": 0.42753763151753793,
      "learning_rate": 4.249084249084249e-06,
      "loss": 0.5121,
      "step": 349
    },
    {
      "epoch": 0.1282521069989007,
      "grad_norm": 0.3690549577982422,
      "learning_rate": 4.261294261294261e-06,
      "loss": 0.5216,
      "step": 350
    },
    {
      "epoch": 0.12861854159032612,
      "grad_norm": 0.35841781261032973,
      "learning_rate": 4.273504273504274e-06,
      "loss": 0.5144,
      "step": 351
    },
    {
      "epoch": 0.12898497618175156,
      "grad_norm": 0.3731386943283347,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.5066,
      "step": 352
    },
    {
      "epoch": 0.129351410773177,
      "grad_norm": 0.374644056770011,
      "learning_rate": 4.297924297924299e-06,
      "loss": 0.5075,
      "step": 353
    },
    {
      "epoch": 0.12971784536460243,
      "grad_norm": 0.8081994765546854,
      "learning_rate": 4.31013431013431e-06,
      "loss": 0.5239,
      "step": 354
    },
    {
      "epoch": 0.13008427995602784,
      "grad_norm": 0.35655054436816064,
      "learning_rate": 4.322344322344323e-06,
      "loss": 0.512,
      "step": 355
    },
    {
      "epoch": 0.13045071454745327,
      "grad_norm": 0.35843535226148043,
      "learning_rate": 4.334554334554335e-06,
      "loss": 0.5085,
      "step": 356
    },
    {
      "epoch": 0.1308171491388787,
      "grad_norm": 0.4111053608584404,
      "learning_rate": 4.346764346764347e-06,
      "loss": 0.5135,
      "step": 357
    },
    {
      "epoch": 0.13118358373030414,
      "grad_norm": 0.3717148922945872,
      "learning_rate": 4.358974358974359e-06,
      "loss": 0.5048,
      "step": 358
    },
    {
      "epoch": 0.13155001832172958,
      "grad_norm": 0.5271903806216564,
      "learning_rate": 4.371184371184372e-06,
      "loss": 0.494,
      "step": 359
    },
    {
      "epoch": 0.131916452913155,
      "grad_norm": 0.37408130623251673,
      "learning_rate": 4.3833943833943835e-06,
      "loss": 0.4959,
      "step": 360
    },
    {
      "epoch": 0.13228288750458042,
      "grad_norm": 0.3821517573508415,
      "learning_rate": 4.395604395604396e-06,
      "loss": 0.5157,
      "step": 361
    },
    {
      "epoch": 0.13264932209600586,
      "grad_norm": 0.3541956162052841,
      "learning_rate": 4.4078144078144084e-06,
      "loss": 0.5028,
      "step": 362
    },
    {
      "epoch": 0.1330157566874313,
      "grad_norm": 0.38156356976025185,
      "learning_rate": 4.42002442002442e-06,
      "loss": 0.5171,
      "step": 363
    },
    {
      "epoch": 0.13338219127885673,
      "grad_norm": 0.3647859382122239,
      "learning_rate": 4.4322344322344325e-06,
      "loss": 0.5115,
      "step": 364
    },
    {
      "epoch": 0.13374862587028216,
      "grad_norm": 0.3761611084440497,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.513,
      "step": 365
    },
    {
      "epoch": 0.1341150604617076,
      "grad_norm": 0.34832165815398175,
      "learning_rate": 4.4566544566544575e-06,
      "loss": 0.5141,
      "step": 366
    },
    {
      "epoch": 0.134481495053133,
      "grad_norm": 0.3645162828090783,
      "learning_rate": 4.468864468864469e-06,
      "loss": 0.5038,
      "step": 367
    },
    {
      "epoch": 0.13484792964455844,
      "grad_norm": 0.33472603403678,
      "learning_rate": 4.4810744810744816e-06,
      "loss": 0.495,
      "step": 368
    },
    {
      "epoch": 0.13521436423598387,
      "grad_norm": 0.3560069878873602,
      "learning_rate": 4.493284493284494e-06,
      "loss": 0.5317,
      "step": 369
    },
    {
      "epoch": 0.1355807988274093,
      "grad_norm": 0.8013974002422465,
      "learning_rate": 4.505494505494506e-06,
      "loss": 0.527,
      "step": 370
    },
    {
      "epoch": 0.13594723341883475,
      "grad_norm": 0.3386015635492836,
      "learning_rate": 4.517704517704518e-06,
      "loss": 0.5328,
      "step": 371
    },
    {
      "epoch": 0.13631366801026018,
      "grad_norm": 0.3482307238638227,
      "learning_rate": 4.5299145299145306e-06,
      "loss": 0.5008,
      "step": 372
    },
    {
      "epoch": 0.1366801026016856,
      "grad_norm": 0.3409722827362367,
      "learning_rate": 4.542124542124542e-06,
      "loss": 0.4856,
      "step": 373
    },
    {
      "epoch": 0.13704653719311102,
      "grad_norm": 0.37131321606812806,
      "learning_rate": 4.554334554334555e-06,
      "loss": 0.5161,
      "step": 374
    },
    {
      "epoch": 0.13741297178453646,
      "grad_norm": 0.4033543552314321,
      "learning_rate": 4.566544566544567e-06,
      "loss": 0.5151,
      "step": 375
    },
    {
      "epoch": 0.1377794063759619,
      "grad_norm": 0.3448107935305655,
      "learning_rate": 4.578754578754579e-06,
      "loss": 0.5163,
      "step": 376
    },
    {
      "epoch": 0.13814584096738733,
      "grad_norm": 0.3630124618037167,
      "learning_rate": 4.590964590964591e-06,
      "loss": 0.4784,
      "step": 377
    },
    {
      "epoch": 0.13851227555881276,
      "grad_norm": 0.40899532275105444,
      "learning_rate": 4.603174603174604e-06,
      "loss": 0.5234,
      "step": 378
    },
    {
      "epoch": 0.13887871015023817,
      "grad_norm": 0.4182995907488671,
      "learning_rate": 4.615384615384616e-06,
      "loss": 0.4903,
      "step": 379
    },
    {
      "epoch": 0.1392451447416636,
      "grad_norm": 0.35304162203135614,
      "learning_rate": 4.627594627594628e-06,
      "loss": 0.4915,
      "step": 380
    },
    {
      "epoch": 0.13961157933308904,
      "grad_norm": 0.3635749953601408,
      "learning_rate": 4.63980463980464e-06,
      "loss": 0.5182,
      "step": 381
    },
    {
      "epoch": 0.13997801392451448,
      "grad_norm": 0.370438249652298,
      "learning_rate": 4.652014652014653e-06,
      "loss": 0.5038,
      "step": 382
    },
    {
      "epoch": 0.1403444485159399,
      "grad_norm": 0.33840645979235445,
      "learning_rate": 4.664224664224664e-06,
      "loss": 0.5004,
      "step": 383
    },
    {
      "epoch": 0.14071088310736535,
      "grad_norm": 0.39296022448800116,
      "learning_rate": 4.676434676434677e-06,
      "loss": 0.5058,
      "step": 384
    },
    {
      "epoch": 0.14107731769879076,
      "grad_norm": 0.4802704160463043,
      "learning_rate": 4.688644688644689e-06,
      "loss": 0.4785,
      "step": 385
    },
    {
      "epoch": 0.1414437522902162,
      "grad_norm": 1.0553779339027554,
      "learning_rate": 4.700854700854701e-06,
      "loss": 0.5065,
      "step": 386
    },
    {
      "epoch": 0.14181018688164163,
      "grad_norm": 0.39647431429479657,
      "learning_rate": 4.713064713064713e-06,
      "loss": 0.4691,
      "step": 387
    },
    {
      "epoch": 0.14217662147306706,
      "grad_norm": 0.3795964376806846,
      "learning_rate": 4.725274725274726e-06,
      "loss": 0.5201,
      "step": 388
    },
    {
      "epoch": 0.1425430560644925,
      "grad_norm": 0.4493096550107767,
      "learning_rate": 4.7374847374847374e-06,
      "loss": 0.4987,
      "step": 389
    },
    {
      "epoch": 0.14290949065591793,
      "grad_norm": 0.3861440382942048,
      "learning_rate": 4.74969474969475e-06,
      "loss": 0.5211,
      "step": 390
    },
    {
      "epoch": 0.14327592524734334,
      "grad_norm": 0.36589977278316654,
      "learning_rate": 4.761904761904762e-06,
      "loss": 0.4939,
      "step": 391
    },
    {
      "epoch": 0.14364235983876877,
      "grad_norm": 0.3741600729721092,
      "learning_rate": 4.774114774114775e-06,
      "loss": 0.4906,
      "step": 392
    },
    {
      "epoch": 0.1440087944301942,
      "grad_norm": 0.36717954074667314,
      "learning_rate": 4.786324786324787e-06,
      "loss": 0.4776,
      "step": 393
    },
    {
      "epoch": 0.14437522902161964,
      "grad_norm": 0.3738851341069374,
      "learning_rate": 4.798534798534799e-06,
      "loss": 0.4963,
      "step": 394
    },
    {
      "epoch": 0.14474166361304508,
      "grad_norm": 0.3688104076618644,
      "learning_rate": 4.810744810744811e-06,
      "loss": 0.5216,
      "step": 395
    },
    {
      "epoch": 0.14510809820447051,
      "grad_norm": 0.3504505149868311,
      "learning_rate": 4.822954822954823e-06,
      "loss": 0.4836,
      "step": 396
    },
    {
      "epoch": 0.14547453279589592,
      "grad_norm": 0.34241650824809794,
      "learning_rate": 4.8351648351648355e-06,
      "loss": 0.5132,
      "step": 397
    },
    {
      "epoch": 0.14584096738732136,
      "grad_norm": 0.626217970796119,
      "learning_rate": 4.847374847374848e-06,
      "loss": 0.5026,
      "step": 398
    },
    {
      "epoch": 0.1462074019787468,
      "grad_norm": 0.44005910565783024,
      "learning_rate": 4.8595848595848596e-06,
      "loss": 0.5083,
      "step": 399
    },
    {
      "epoch": 0.14657383657017223,
      "grad_norm": 0.36224509231684804,
      "learning_rate": 4.871794871794872e-06,
      "loss": 0.5137,
      "step": 400
    },
    {
      "epoch": 0.14694027116159766,
      "grad_norm": 0.3850724910187562,
      "learning_rate": 4.8840048840048845e-06,
      "loss": 0.4894,
      "step": 401
    },
    {
      "epoch": 0.1473067057530231,
      "grad_norm": 0.3679033270369306,
      "learning_rate": 4.896214896214896e-06,
      "loss": 0.5055,
      "step": 402
    },
    {
      "epoch": 0.1476731403444485,
      "grad_norm": 0.35998123667186305,
      "learning_rate": 4.908424908424909e-06,
      "loss": 0.5063,
      "step": 403
    },
    {
      "epoch": 0.14803957493587394,
      "grad_norm": 0.340524932667634,
      "learning_rate": 4.920634920634921e-06,
      "loss": 0.5055,
      "step": 404
    },
    {
      "epoch": 0.14840600952729938,
      "grad_norm": 0.38418139578909544,
      "learning_rate": 4.9328449328449335e-06,
      "loss": 0.4985,
      "step": 405
    },
    {
      "epoch": 0.1487724441187248,
      "grad_norm": 0.6753684231306423,
      "learning_rate": 4.945054945054946e-06,
      "loss": 0.4945,
      "step": 406
    },
    {
      "epoch": 0.14913887871015025,
      "grad_norm": 0.39429778810877447,
      "learning_rate": 4.957264957264958e-06,
      "loss": 0.5156,
      "step": 407
    },
    {
      "epoch": 0.14950531330157568,
      "grad_norm": 0.3559593692132899,
      "learning_rate": 4.96947496947497e-06,
      "loss": 0.5014,
      "step": 408
    },
    {
      "epoch": 0.1498717478930011,
      "grad_norm": 0.353788147629717,
      "learning_rate": 4.9816849816849826e-06,
      "loss": 0.4972,
      "step": 409
    },
    {
      "epoch": 0.15023818248442652,
      "grad_norm": 0.49350236488602817,
      "learning_rate": 4.993894993894994e-06,
      "loss": 0.5141,
      "step": 410
    },
    {
      "epoch": 0.15060461707585196,
      "grad_norm": 0.48112209634692765,
      "learning_rate": 5.006105006105007e-06,
      "loss": 0.5024,
      "step": 411
    },
    {
      "epoch": 0.1509710516672774,
      "grad_norm": 0.367437791035824,
      "learning_rate": 5.018315018315018e-06,
      "loss": 0.4874,
      "step": 412
    },
    {
      "epoch": 0.15133748625870283,
      "grad_norm": 0.37787411803191756,
      "learning_rate": 5.030525030525031e-06,
      "loss": 0.5112,
      "step": 413
    },
    {
      "epoch": 0.15170392085012827,
      "grad_norm": 0.351722670309027,
      "learning_rate": 5.042735042735043e-06,
      "loss": 0.5011,
      "step": 414
    },
    {
      "epoch": 0.15207035544155367,
      "grad_norm": 0.3900028880032195,
      "learning_rate": 5.054945054945055e-06,
      "loss": 0.4944,
      "step": 415
    },
    {
      "epoch": 0.1524367900329791,
      "grad_norm": 0.39804944382400614,
      "learning_rate": 5.067155067155067e-06,
      "loss": 0.4967,
      "step": 416
    },
    {
      "epoch": 0.15280322462440454,
      "grad_norm": 0.3630351321415531,
      "learning_rate": 5.07936507936508e-06,
      "loss": 0.5149,
      "step": 417
    },
    {
      "epoch": 0.15316965921582998,
      "grad_norm": 0.41324060376982646,
      "learning_rate": 5.091575091575091e-06,
      "loss": 0.4953,
      "step": 418
    },
    {
      "epoch": 0.1535360938072554,
      "grad_norm": 0.3399876426318429,
      "learning_rate": 5.103785103785104e-06,
      "loss": 0.4933,
      "step": 419
    },
    {
      "epoch": 0.15390252839868085,
      "grad_norm": 0.4587729191309135,
      "learning_rate": 5.115995115995116e-06,
      "loss": 0.4708,
      "step": 420
    },
    {
      "epoch": 0.15426896299010626,
      "grad_norm": 0.40847905386562855,
      "learning_rate": 5.128205128205128e-06,
      "loss": 0.5256,
      "step": 421
    },
    {
      "epoch": 0.1546353975815317,
      "grad_norm": 0.37427956714115346,
      "learning_rate": 5.14041514041514e-06,
      "loss": 0.5001,
      "step": 422
    },
    {
      "epoch": 0.15500183217295713,
      "grad_norm": 0.42965933613161716,
      "learning_rate": 5.152625152625153e-06,
      "loss": 0.5077,
      "step": 423
    },
    {
      "epoch": 0.15536826676438256,
      "grad_norm": 0.3726305021461639,
      "learning_rate": 5.164835164835166e-06,
      "loss": 0.5082,
      "step": 424
    },
    {
      "epoch": 0.155734701355808,
      "grad_norm": 0.5733510859468256,
      "learning_rate": 5.177045177045178e-06,
      "loss": 0.5088,
      "step": 425
    },
    {
      "epoch": 0.15610113594723343,
      "grad_norm": 0.45304026926922275,
      "learning_rate": 5.18925518925519e-06,
      "loss": 0.4812,
      "step": 426
    },
    {
      "epoch": 0.15646757053865884,
      "grad_norm": 0.38442825200224695,
      "learning_rate": 5.201465201465202e-06,
      "loss": 0.4989,
      "step": 427
    },
    {
      "epoch": 0.15683400513008428,
      "grad_norm": 0.3541069206138204,
      "learning_rate": 5.213675213675214e-06,
      "loss": 0.4948,
      "step": 428
    },
    {
      "epoch": 0.1572004397215097,
      "grad_norm": 0.3730160297694107,
      "learning_rate": 5.225885225885227e-06,
      "loss": 0.5181,
      "step": 429
    },
    {
      "epoch": 0.15756687431293515,
      "grad_norm": 0.35715040000574105,
      "learning_rate": 5.2380952380952384e-06,
      "loss": 0.5039,
      "step": 430
    },
    {
      "epoch": 0.15793330890436058,
      "grad_norm": 0.3588778974395094,
      "learning_rate": 5.250305250305251e-06,
      "loss": 0.4844,
      "step": 431
    },
    {
      "epoch": 0.158299743495786,
      "grad_norm": 0.3620695516850173,
      "learning_rate": 5.262515262515263e-06,
      "loss": 0.4846,
      "step": 432
    },
    {
      "epoch": 0.15866617808721142,
      "grad_norm": 0.34081827333746656,
      "learning_rate": 5.274725274725275e-06,
      "loss": 0.5104,
      "step": 433
    },
    {
      "epoch": 0.15903261267863686,
      "grad_norm": 0.3722723414147867,
      "learning_rate": 5.2869352869352875e-06,
      "loss": 0.5176,
      "step": 434
    },
    {
      "epoch": 0.1593990472700623,
      "grad_norm": 0.6790202031898657,
      "learning_rate": 5.2991452991453e-06,
      "loss": 0.509,
      "step": 435
    },
    {
      "epoch": 0.15976548186148773,
      "grad_norm": 0.35785188019239766,
      "learning_rate": 5.3113553113553116e-06,
      "loss": 0.516,
      "step": 436
    },
    {
      "epoch": 0.16013191645291316,
      "grad_norm": 0.39035635654411727,
      "learning_rate": 5.323565323565324e-06,
      "loss": 0.5073,
      "step": 437
    },
    {
      "epoch": 0.16049835104433857,
      "grad_norm": 0.40384635847731226,
      "learning_rate": 5.3357753357753365e-06,
      "loss": 0.5204,
      "step": 438
    },
    {
      "epoch": 0.160864785635764,
      "grad_norm": 0.4199210862910261,
      "learning_rate": 5.347985347985348e-06,
      "loss": 0.5263,
      "step": 439
    },
    {
      "epoch": 0.16123122022718944,
      "grad_norm": 0.3529089231373357,
      "learning_rate": 5.360195360195361e-06,
      "loss": 0.5043,
      "step": 440
    },
    {
      "epoch": 0.16159765481861488,
      "grad_norm": 0.48900629774961196,
      "learning_rate": 5.372405372405373e-06,
      "loss": 0.504,
      "step": 441
    },
    {
      "epoch": 0.1619640894100403,
      "grad_norm": 0.40154475487288827,
      "learning_rate": 5.384615384615385e-06,
      "loss": 0.5123,
      "step": 442
    },
    {
      "epoch": 0.16233052400146575,
      "grad_norm": 0.36939445087051903,
      "learning_rate": 5.396825396825397e-06,
      "loss": 0.5142,
      "step": 443
    },
    {
      "epoch": 0.16269695859289116,
      "grad_norm": 0.35784422539254596,
      "learning_rate": 5.409035409035409e-06,
      "loss": 0.5173,
      "step": 444
    },
    {
      "epoch": 0.1630633931843166,
      "grad_norm": 0.38706118493149166,
      "learning_rate": 5.421245421245421e-06,
      "loss": 0.5184,
      "step": 445
    },
    {
      "epoch": 0.16342982777574203,
      "grad_norm": 0.367012960736475,
      "learning_rate": 5.433455433455434e-06,
      "loss": 0.4808,
      "step": 446
    },
    {
      "epoch": 0.16379626236716746,
      "grad_norm": 0.3725706197967635,
      "learning_rate": 5.445665445665445e-06,
      "loss": 0.5057,
      "step": 447
    },
    {
      "epoch": 0.1641626969585929,
      "grad_norm": 0.38063120841936027,
      "learning_rate": 5.457875457875458e-06,
      "loss": 0.5167,
      "step": 448
    },
    {
      "epoch": 0.16452913155001833,
      "grad_norm": 0.39766491439520885,
      "learning_rate": 5.470085470085471e-06,
      "loss": 0.5292,
      "step": 449
    },
    {
      "epoch": 0.16489556614144374,
      "grad_norm": 0.3982353108055876,
      "learning_rate": 5.4822954822954836e-06,
      "loss": 0.5095,
      "step": 450
    },
    {
      "epoch": 0.16526200073286917,
      "grad_norm": 0.38525265620704824,
      "learning_rate": 5.494505494505495e-06,
      "loss": 0.518,
      "step": 451
    },
    {
      "epoch": 0.1656284353242946,
      "grad_norm": 0.3748515678345235,
      "learning_rate": 5.506715506715508e-06,
      "loss": 0.4868,
      "step": 452
    },
    {
      "epoch": 0.16599486991572004,
      "grad_norm": 0.35033757356380574,
      "learning_rate": 5.51892551892552e-06,
      "loss": 0.4988,
      "step": 453
    },
    {
      "epoch": 0.16636130450714548,
      "grad_norm": 0.39136977202147827,
      "learning_rate": 5.531135531135532e-06,
      "loss": 0.4991,
      "step": 454
    },
    {
      "epoch": 0.16672773909857092,
      "grad_norm": 0.3292684818264875,
      "learning_rate": 5.543345543345544e-06,
      "loss": 0.4621,
      "step": 455
    },
    {
      "epoch": 0.16709417368999632,
      "grad_norm": 0.39088934828586164,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.514,
      "step": 456
    },
    {
      "epoch": 0.16746060828142176,
      "grad_norm": 0.4186428969700618,
      "learning_rate": 5.567765567765568e-06,
      "loss": 0.5123,
      "step": 457
    },
    {
      "epoch": 0.1678270428728472,
      "grad_norm": 0.41330129509028296,
      "learning_rate": 5.579975579975581e-06,
      "loss": 0.4843,
      "step": 458
    },
    {
      "epoch": 0.16819347746427263,
      "grad_norm": 0.36912602701457303,
      "learning_rate": 5.592185592185592e-06,
      "loss": 0.488,
      "step": 459
    },
    {
      "epoch": 0.16855991205569806,
      "grad_norm": 0.3595816945142543,
      "learning_rate": 5.604395604395605e-06,
      "loss": 0.5046,
      "step": 460
    },
    {
      "epoch": 0.1689263466471235,
      "grad_norm": 0.36576791419324994,
      "learning_rate": 5.616605616605617e-06,
      "loss": 0.4807,
      "step": 461
    },
    {
      "epoch": 0.1692927812385489,
      "grad_norm": 0.38360266794891196,
      "learning_rate": 5.628815628815629e-06,
      "loss": 0.5104,
      "step": 462
    },
    {
      "epoch": 0.16965921582997434,
      "grad_norm": 0.3679995233931312,
      "learning_rate": 5.641025641025641e-06,
      "loss": 0.5204,
      "step": 463
    },
    {
      "epoch": 0.17002565042139978,
      "grad_norm": 0.5549971701585231,
      "learning_rate": 5.653235653235654e-06,
      "loss": 0.4915,
      "step": 464
    },
    {
      "epoch": 0.1703920850128252,
      "grad_norm": 0.3634379641937027,
      "learning_rate": 5.6654456654456655e-06,
      "loss": 0.4921,
      "step": 465
    },
    {
      "epoch": 0.17075851960425065,
      "grad_norm": 1.6318911346219098,
      "learning_rate": 5.677655677655678e-06,
      "loss": 0.5046,
      "step": 466
    },
    {
      "epoch": 0.17112495419567608,
      "grad_norm": 0.3659132553552777,
      "learning_rate": 5.68986568986569e-06,
      "loss": 0.4936,
      "step": 467
    },
    {
      "epoch": 0.1714913887871015,
      "grad_norm": 0.36962283098946597,
      "learning_rate": 5.702075702075702e-06,
      "loss": 0.5068,
      "step": 468
    },
    {
      "epoch": 0.17185782337852692,
      "grad_norm": 0.3463053159476888,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.5078,
      "step": 469
    },
    {
      "epoch": 0.17222425796995236,
      "grad_norm": 0.33658632088239876,
      "learning_rate": 5.726495726495727e-06,
      "loss": 0.5143,
      "step": 470
    },
    {
      "epoch": 0.1725906925613778,
      "grad_norm": 0.544328907106882,
      "learning_rate": 5.738705738705739e-06,
      "loss": 0.5197,
      "step": 471
    },
    {
      "epoch": 0.17295712715280323,
      "grad_norm": 0.40526068947304766,
      "learning_rate": 5.750915750915751e-06,
      "loss": 0.4773,
      "step": 472
    },
    {
      "epoch": 0.17332356174422867,
      "grad_norm": 0.374044122144093,
      "learning_rate": 5.763125763125763e-06,
      "loss": 0.4979,
      "step": 473
    },
    {
      "epoch": 0.17368999633565407,
      "grad_norm": 0.3329273901497176,
      "learning_rate": 5.775335775335775e-06,
      "loss": 0.4945,
      "step": 474
    },
    {
      "epoch": 0.1740564309270795,
      "grad_norm": 0.5225284959803687,
      "learning_rate": 5.7875457875457885e-06,
      "loss": 0.4666,
      "step": 475
    },
    {
      "epoch": 0.17442286551850494,
      "grad_norm": 0.3789524714044287,
      "learning_rate": 5.799755799755801e-06,
      "loss": 0.4757,
      "step": 476
    },
    {
      "epoch": 0.17478930010993038,
      "grad_norm": 0.36086037797714726,
      "learning_rate": 5.8119658119658126e-06,
      "loss": 0.5298,
      "step": 477
    },
    {
      "epoch": 0.17515573470135581,
      "grad_norm": 0.33228011798019974,
      "learning_rate": 5.824175824175825e-06,
      "loss": 0.5015,
      "step": 478
    },
    {
      "epoch": 0.17552216929278125,
      "grad_norm": 0.3721441664271045,
      "learning_rate": 5.8363858363858375e-06,
      "loss": 0.5196,
      "step": 479
    },
    {
      "epoch": 0.17588860388420666,
      "grad_norm": 0.34432323303147777,
      "learning_rate": 5.848595848595849e-06,
      "loss": 0.497,
      "step": 480
    },
    {
      "epoch": 0.1762550384756321,
      "grad_norm": 0.3674415983917119,
      "learning_rate": 5.860805860805862e-06,
      "loss": 0.4774,
      "step": 481
    },
    {
      "epoch": 0.17662147306705753,
      "grad_norm": 0.3312766279925682,
      "learning_rate": 5.873015873015874e-06,
      "loss": 0.5096,
      "step": 482
    },
    {
      "epoch": 0.17698790765848296,
      "grad_norm": 0.40107807539720175,
      "learning_rate": 5.885225885225886e-06,
      "loss": 0.5217,
      "step": 483
    },
    {
      "epoch": 0.1773543422499084,
      "grad_norm": 0.36542707165272453,
      "learning_rate": 5.897435897435898e-06,
      "loss": 0.5032,
      "step": 484
    },
    {
      "epoch": 0.17772077684133383,
      "grad_norm": 0.45141428479022666,
      "learning_rate": 5.909645909645911e-06,
      "loss": 0.5149,
      "step": 485
    },
    {
      "epoch": 0.17808721143275924,
      "grad_norm": 0.608491950169155,
      "learning_rate": 5.921855921855922e-06,
      "loss": 0.4997,
      "step": 486
    },
    {
      "epoch": 0.17845364602418468,
      "grad_norm": 0.37540978861550406,
      "learning_rate": 5.934065934065935e-06,
      "loss": 0.4856,
      "step": 487
    },
    {
      "epoch": 0.1788200806156101,
      "grad_norm": 0.3409517509038324,
      "learning_rate": 5.946275946275946e-06,
      "loss": 0.4958,
      "step": 488
    },
    {
      "epoch": 0.17918651520703555,
      "grad_norm": 0.35470659340570865,
      "learning_rate": 5.958485958485959e-06,
      "loss": 0.4925,
      "step": 489
    },
    {
      "epoch": 0.17955294979846098,
      "grad_norm": 0.36928195076749437,
      "learning_rate": 5.970695970695971e-06,
      "loss": 0.5074,
      "step": 490
    },
    {
      "epoch": 0.17991938438988642,
      "grad_norm": 0.3401495953863161,
      "learning_rate": 5.982905982905983e-06,
      "loss": 0.4862,
      "step": 491
    },
    {
      "epoch": 0.18028581898131182,
      "grad_norm": 0.3364009447037792,
      "learning_rate": 5.995115995115995e-06,
      "loss": 0.4919,
      "step": 492
    },
    {
      "epoch": 0.18065225357273726,
      "grad_norm": 0.3879011329341925,
      "learning_rate": 6.007326007326008e-06,
      "loss": 0.4879,
      "step": 493
    },
    {
      "epoch": 0.1810186881641627,
      "grad_norm": 0.35355504130008225,
      "learning_rate": 6.019536019536019e-06,
      "loss": 0.5278,
      "step": 494
    },
    {
      "epoch": 0.18138512275558813,
      "grad_norm": 0.3627782197354617,
      "learning_rate": 6.031746031746032e-06,
      "loss": 0.5106,
      "step": 495
    },
    {
      "epoch": 0.18175155734701356,
      "grad_norm": 0.32942660241784094,
      "learning_rate": 6.043956043956044e-06,
      "loss": 0.5187,
      "step": 496
    },
    {
      "epoch": 0.182117991938439,
      "grad_norm": 0.42357619414018527,
      "learning_rate": 6.056166056166056e-06,
      "loss": 0.5065,
      "step": 497
    },
    {
      "epoch": 0.1824844265298644,
      "grad_norm": 0.35517753360491433,
      "learning_rate": 6.0683760683760684e-06,
      "loss": 0.5265,
      "step": 498
    },
    {
      "epoch": 0.18285086112128984,
      "grad_norm": 0.36310772982468736,
      "learning_rate": 6.080586080586081e-06,
      "loss": 0.4852,
      "step": 499
    },
    {
      "epoch": 0.18321729571271528,
      "grad_norm": 0.370082502489204,
      "learning_rate": 6.0927960927960925e-06,
      "loss": 0.4793,
      "step": 500
    },
    {
      "epoch": 0.1835837303041407,
      "grad_norm": 0.5252793697248789,
      "learning_rate": 6.105006105006106e-06,
      "loss": 0.4948,
      "step": 501
    },
    {
      "epoch": 0.18395016489556615,
      "grad_norm": 0.339926315822795,
      "learning_rate": 6.117216117216118e-06,
      "loss": 0.4826,
      "step": 502
    },
    {
      "epoch": 0.18431659948699158,
      "grad_norm": 0.37603796365671915,
      "learning_rate": 6.12942612942613e-06,
      "loss": 0.5029,
      "step": 503
    },
    {
      "epoch": 0.184683034078417,
      "grad_norm": 0.3883474144870869,
      "learning_rate": 6.141636141636142e-06,
      "loss": 0.4958,
      "step": 504
    },
    {
      "epoch": 0.18504946866984243,
      "grad_norm": 0.34371220708656497,
      "learning_rate": 6.153846153846155e-06,
      "loss": 0.4874,
      "step": 505
    },
    {
      "epoch": 0.18541590326126786,
      "grad_norm": 0.36953786904310587,
      "learning_rate": 6.1660561660561665e-06,
      "loss": 0.499,
      "step": 506
    },
    {
      "epoch": 0.1857823378526933,
      "grad_norm": 0.37285761397701933,
      "learning_rate": 6.178266178266179e-06,
      "loss": 0.5044,
      "step": 507
    },
    {
      "epoch": 0.18614877244411873,
      "grad_norm": 0.42183698797240804,
      "learning_rate": 6.1904761904761914e-06,
      "loss": 0.5068,
      "step": 508
    },
    {
      "epoch": 0.18651520703554417,
      "grad_norm": 0.3835772113267854,
      "learning_rate": 6.202686202686203e-06,
      "loss": 0.4963,
      "step": 509
    },
    {
      "epoch": 0.18688164162696957,
      "grad_norm": 0.37551398499094807,
      "learning_rate": 6.2148962148962155e-06,
      "loss": 0.5008,
      "step": 510
    },
    {
      "epoch": 0.187248076218395,
      "grad_norm": 0.39787535574549365,
      "learning_rate": 6.227106227106228e-06,
      "loss": 0.4869,
      "step": 511
    },
    {
      "epoch": 0.18761451080982045,
      "grad_norm": 0.3846264944781745,
      "learning_rate": 6.23931623931624e-06,
      "loss": 0.4851,
      "step": 512
    },
    {
      "epoch": 0.18798094540124588,
      "grad_norm": 2.3892762329071484,
      "learning_rate": 6.251526251526252e-06,
      "loss": 0.4538,
      "step": 513
    },
    {
      "epoch": 0.18834737999267132,
      "grad_norm": 0.3921427535779395,
      "learning_rate": 6.2637362637362645e-06,
      "loss": 0.4967,
      "step": 514
    },
    {
      "epoch": 0.18871381458409675,
      "grad_norm": 0.36088760670404235,
      "learning_rate": 6.275946275946276e-06,
      "loss": 0.4854,
      "step": 515
    },
    {
      "epoch": 0.18908024917552216,
      "grad_norm": 0.4061471472218871,
      "learning_rate": 6.288156288156289e-06,
      "loss": 0.5153,
      "step": 516
    },
    {
      "epoch": 0.1894466837669476,
      "grad_norm": 0.3372885272503332,
      "learning_rate": 6.300366300366301e-06,
      "loss": 0.4975,
      "step": 517
    },
    {
      "epoch": 0.18981311835837303,
      "grad_norm": 0.3658387785733554,
      "learning_rate": 6.312576312576313e-06,
      "loss": 0.5235,
      "step": 518
    },
    {
      "epoch": 0.19017955294979846,
      "grad_norm": 0.41077377127339915,
      "learning_rate": 6.324786324786325e-06,
      "loss": 0.4913,
      "step": 519
    },
    {
      "epoch": 0.1905459875412239,
      "grad_norm": 0.35929587574947375,
      "learning_rate": 6.336996336996337e-06,
      "loss": 0.5169,
      "step": 520
    },
    {
      "epoch": 0.19091242213264933,
      "grad_norm": 0.3585072505058271,
      "learning_rate": 6.349206349206349e-06,
      "loss": 0.5001,
      "step": 521
    },
    {
      "epoch": 0.19127885672407474,
      "grad_norm": 0.3760872052845346,
      "learning_rate": 6.361416361416362e-06,
      "loss": 0.4934,
      "step": 522
    },
    {
      "epoch": 0.19164529131550018,
      "grad_norm": 0.38439477687858997,
      "learning_rate": 6.373626373626373e-06,
      "loss": 0.5146,
      "step": 523
    },
    {
      "epoch": 0.1920117259069256,
      "grad_norm": 0.4191419998461837,
      "learning_rate": 6.385836385836386e-06,
      "loss": 0.5145,
      "step": 524
    },
    {
      "epoch": 0.19237816049835105,
      "grad_norm": 0.4318899296165643,
      "learning_rate": 6.398046398046398e-06,
      "loss": 0.4873,
      "step": 525
    },
    {
      "epoch": 0.19274459508977648,
      "grad_norm": 0.42974692828872385,
      "learning_rate": 6.410256410256412e-06,
      "loss": 0.4861,
      "step": 526
    },
    {
      "epoch": 0.19311102968120192,
      "grad_norm": 0.3725776765544941,
      "learning_rate": 6.422466422466423e-06,
      "loss": 0.5103,
      "step": 527
    },
    {
      "epoch": 0.19347746427262733,
      "grad_norm": 0.3773862134968054,
      "learning_rate": 6.434676434676436e-06,
      "loss": 0.4657,
      "step": 528
    },
    {
      "epoch": 0.19384389886405276,
      "grad_norm": 0.3691920757580188,
      "learning_rate": 6.446886446886448e-06,
      "loss": 0.4746,
      "step": 529
    },
    {
      "epoch": 0.1942103334554782,
      "grad_norm": 0.4421405930835643,
      "learning_rate": 6.45909645909646e-06,
      "loss": 0.5139,
      "step": 530
    },
    {
      "epoch": 0.19457676804690363,
      "grad_norm": 0.3354007850335288,
      "learning_rate": 6.471306471306472e-06,
      "loss": 0.4925,
      "step": 531
    },
    {
      "epoch": 0.19494320263832907,
      "grad_norm": 0.3594967614063805,
      "learning_rate": 6.483516483516485e-06,
      "loss": 0.4892,
      "step": 532
    },
    {
      "epoch": 0.1953096372297545,
      "grad_norm": 0.3524723776455673,
      "learning_rate": 6.495726495726496e-06,
      "loss": 0.4887,
      "step": 533
    },
    {
      "epoch": 0.1956760718211799,
      "grad_norm": 0.3887146953073181,
      "learning_rate": 6.507936507936509e-06,
      "loss": 0.4988,
      "step": 534
    },
    {
      "epoch": 0.19604250641260534,
      "grad_norm": 0.3676112676247605,
      "learning_rate": 6.5201465201465204e-06,
      "loss": 0.4974,
      "step": 535
    },
    {
      "epoch": 0.19640894100403078,
      "grad_norm": 0.8705156749752884,
      "learning_rate": 6.532356532356533e-06,
      "loss": 0.5031,
      "step": 536
    },
    {
      "epoch": 0.19677537559545621,
      "grad_norm": 0.3933943864897206,
      "learning_rate": 6.544566544566545e-06,
      "loss": 0.484,
      "step": 537
    },
    {
      "epoch": 0.19714181018688165,
      "grad_norm": 0.4706115448113118,
      "learning_rate": 6.556776556776557e-06,
      "loss": 0.5048,
      "step": 538
    },
    {
      "epoch": 0.19750824477830708,
      "grad_norm": 0.3537691699789792,
      "learning_rate": 6.5689865689865694e-06,
      "loss": 0.4899,
      "step": 539
    },
    {
      "epoch": 0.1978746793697325,
      "grad_norm": 0.38106346551166537,
      "learning_rate": 6.581196581196582e-06,
      "loss": 0.5019,
      "step": 540
    },
    {
      "epoch": 0.19824111396115793,
      "grad_norm": 0.36635186078111576,
      "learning_rate": 6.5934065934065935e-06,
      "loss": 0.5094,
      "step": 541
    },
    {
      "epoch": 0.19860754855258336,
      "grad_norm": 0.368215225301161,
      "learning_rate": 6.605616605616606e-06,
      "loss": 0.4802,
      "step": 542
    },
    {
      "epoch": 0.1989739831440088,
      "grad_norm": 0.43471160846690793,
      "learning_rate": 6.6178266178266185e-06,
      "loss": 0.4866,
      "step": 543
    },
    {
      "epoch": 0.19934041773543423,
      "grad_norm": 0.3508505253951781,
      "learning_rate": 6.63003663003663e-06,
      "loss": 0.513,
      "step": 544
    },
    {
      "epoch": 0.19970685232685967,
      "grad_norm": 0.3791709177460747,
      "learning_rate": 6.6422466422466426e-06,
      "loss": 0.5142,
      "step": 545
    },
    {
      "epoch": 0.20007328691828508,
      "grad_norm": 0.3734103250358158,
      "learning_rate": 6.654456654456655e-06,
      "loss": 0.4979,
      "step": 546
    },
    {
      "epoch": 0.2004397215097105,
      "grad_norm": 0.4827141680761808,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.5055,
      "step": 547
    },
    {
      "epoch": 0.20080615610113595,
      "grad_norm": 0.4280135410292611,
      "learning_rate": 6.678876678876679e-06,
      "loss": 0.5148,
      "step": 548
    },
    {
      "epoch": 0.20117259069256138,
      "grad_norm": 0.34877798083998707,
      "learning_rate": 6.691086691086692e-06,
      "loss": 0.4644,
      "step": 549
    },
    {
      "epoch": 0.20153902528398682,
      "grad_norm": 0.36801207822271265,
      "learning_rate": 6.703296703296703e-06,
      "loss": 0.4849,
      "step": 550
    },
    {
      "epoch": 0.20190545987541225,
      "grad_norm": 0.3549083835487994,
      "learning_rate": 6.715506715506716e-06,
      "loss": 0.4834,
      "step": 551
    },
    {
      "epoch": 0.20227189446683766,
      "grad_norm": 0.3424310053073312,
      "learning_rate": 6.727716727716729e-06,
      "loss": 0.4786,
      "step": 552
    },
    {
      "epoch": 0.2026383290582631,
      "grad_norm": 0.5327667646778993,
      "learning_rate": 6.739926739926741e-06,
      "loss": 0.5102,
      "step": 553
    },
    {
      "epoch": 0.20300476364968853,
      "grad_norm": 0.32661658368508917,
      "learning_rate": 6.752136752136753e-06,
      "loss": 0.4875,
      "step": 554
    },
    {
      "epoch": 0.20337119824111397,
      "grad_norm": 0.32905060004892916,
      "learning_rate": 6.7643467643467655e-06,
      "loss": 0.4862,
      "step": 555
    },
    {
      "epoch": 0.2037376328325394,
      "grad_norm": 0.36405258269602014,
      "learning_rate": 6.776556776556777e-06,
      "loss": 0.4664,
      "step": 556
    },
    {
      "epoch": 0.20410406742396484,
      "grad_norm": 0.3328421933967697,
      "learning_rate": 6.78876678876679e-06,
      "loss": 0.4675,
      "step": 557
    },
    {
      "epoch": 0.20447050201539024,
      "grad_norm": 0.43354092709616465,
      "learning_rate": 6.800976800976802e-06,
      "loss": 0.5005,
      "step": 558
    },
    {
      "epoch": 0.20483693660681568,
      "grad_norm": 0.3549726356797663,
      "learning_rate": 6.813186813186814e-06,
      "loss": 0.4664,
      "step": 559
    },
    {
      "epoch": 0.2052033711982411,
      "grad_norm": 0.3766160624833037,
      "learning_rate": 6.825396825396826e-06,
      "loss": 0.5107,
      "step": 560
    },
    {
      "epoch": 0.20556980578966655,
      "grad_norm": 0.4102199066129289,
      "learning_rate": 6.837606837606839e-06,
      "loss": 0.494,
      "step": 561
    },
    {
      "epoch": 0.20593624038109198,
      "grad_norm": 0.4069405695449329,
      "learning_rate": 6.84981684981685e-06,
      "loss": 0.4804,
      "step": 562
    },
    {
      "epoch": 0.20630267497251742,
      "grad_norm": 0.35161842508282487,
      "learning_rate": 6.862026862026863e-06,
      "loss": 0.4878,
      "step": 563
    },
    {
      "epoch": 0.20666910956394283,
      "grad_norm": 0.3742930755224748,
      "learning_rate": 6.874236874236875e-06,
      "loss": 0.4837,
      "step": 564
    },
    {
      "epoch": 0.20703554415536826,
      "grad_norm": 0.3553557625843291,
      "learning_rate": 6.886446886446887e-06,
      "loss": 0.4882,
      "step": 565
    },
    {
      "epoch": 0.2074019787467937,
      "grad_norm": 0.3449311559672515,
      "learning_rate": 6.898656898656899e-06,
      "loss": 0.4797,
      "step": 566
    },
    {
      "epoch": 0.20776841333821913,
      "grad_norm": 0.35797824632458486,
      "learning_rate": 6.910866910866911e-06,
      "loss": 0.4967,
      "step": 567
    },
    {
      "epoch": 0.20813484792964457,
      "grad_norm": 0.3706472352368549,
      "learning_rate": 6.923076923076923e-06,
      "loss": 0.5004,
      "step": 568
    },
    {
      "epoch": 0.20850128252107,
      "grad_norm": 0.3801805438028834,
      "learning_rate": 6.935286935286936e-06,
      "loss": 0.5116,
      "step": 569
    },
    {
      "epoch": 0.2088677171124954,
      "grad_norm": 0.3503130350835744,
      "learning_rate": 6.9474969474969475e-06,
      "loss": 0.5072,
      "step": 570
    },
    {
      "epoch": 0.20923415170392085,
      "grad_norm": 0.37213335531258235,
      "learning_rate": 6.95970695970696e-06,
      "loss": 0.4833,
      "step": 571
    },
    {
      "epoch": 0.20960058629534628,
      "grad_norm": 0.36167310773225614,
      "learning_rate": 6.971916971916972e-06,
      "loss": 0.4726,
      "step": 572
    },
    {
      "epoch": 0.20996702088677172,
      "grad_norm": 0.3741357749097205,
      "learning_rate": 6.984126984126984e-06,
      "loss": 0.4994,
      "step": 573
    },
    {
      "epoch": 0.21033345547819715,
      "grad_norm": 0.38234290490518297,
      "learning_rate": 6.9963369963369965e-06,
      "loss": 0.5126,
      "step": 574
    },
    {
      "epoch": 0.21069989006962256,
      "grad_norm": 0.6153210930368345,
      "learning_rate": 7.008547008547009e-06,
      "loss": 0.5182,
      "step": 575
    },
    {
      "epoch": 0.211066324661048,
      "grad_norm": 0.4079691817401987,
      "learning_rate": 7.020757020757021e-06,
      "loss": 0.5077,
      "step": 576
    },
    {
      "epoch": 0.21143275925247343,
      "grad_norm": 0.4016012941978349,
      "learning_rate": 7.032967032967034e-06,
      "loss": 0.5033,
      "step": 577
    },
    {
      "epoch": 0.21179919384389886,
      "grad_norm": 0.4205812901376417,
      "learning_rate": 7.045177045177046e-06,
      "loss": 0.4933,
      "step": 578
    },
    {
      "epoch": 0.2121656284353243,
      "grad_norm": 0.36965044496731886,
      "learning_rate": 7.057387057387059e-06,
      "loss": 0.4955,
      "step": 579
    },
    {
      "epoch": 0.21253206302674973,
      "grad_norm": 0.3548360690424224,
      "learning_rate": 7.0695970695970705e-06,
      "loss": 0.4761,
      "step": 580
    },
    {
      "epoch": 0.21289849761817514,
      "grad_norm": 0.3634505234853986,
      "learning_rate": 7.081807081807083e-06,
      "loss": 0.5151,
      "step": 581
    },
    {
      "epoch": 0.21326493220960058,
      "grad_norm": 0.3851526174727714,
      "learning_rate": 7.0940170940170945e-06,
      "loss": 0.5115,
      "step": 582
    },
    {
      "epoch": 0.213631366801026,
      "grad_norm": 0.40746590112795966,
      "learning_rate": 7.106227106227107e-06,
      "loss": 0.4794,
      "step": 583
    },
    {
      "epoch": 0.21399780139245145,
      "grad_norm": 0.38094018113987016,
      "learning_rate": 7.1184371184371195e-06,
      "loss": 0.486,
      "step": 584
    },
    {
      "epoch": 0.21436423598387688,
      "grad_norm": 0.3516532586015792,
      "learning_rate": 7.130647130647131e-06,
      "loss": 0.4802,
      "step": 585
    },
    {
      "epoch": 0.21473067057530232,
      "grad_norm": 0.36413167020940984,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.5033,
      "step": 586
    },
    {
      "epoch": 0.21509710516672773,
      "grad_norm": 0.40068673292932233,
      "learning_rate": 7.155067155067156e-06,
      "loss": 0.4805,
      "step": 587
    },
    {
      "epoch": 0.21546353975815316,
      "grad_norm": 0.3610591671212307,
      "learning_rate": 7.167277167277168e-06,
      "loss": 0.4698,
      "step": 588
    },
    {
      "epoch": 0.2158299743495786,
      "grad_norm": 0.37199923557239994,
      "learning_rate": 7.17948717948718e-06,
      "loss": 0.4637,
      "step": 589
    },
    {
      "epoch": 0.21619640894100403,
      "grad_norm": 0.37391663485492255,
      "learning_rate": 7.191697191697193e-06,
      "loss": 0.4983,
      "step": 590
    },
    {
      "epoch": 0.21656284353242947,
      "grad_norm": 0.3837290356929628,
      "learning_rate": 7.203907203907204e-06,
      "loss": 0.4766,
      "step": 591
    },
    {
      "epoch": 0.2169292781238549,
      "grad_norm": 0.38380596181428506,
      "learning_rate": 7.216117216117217e-06,
      "loss": 0.5083,
      "step": 592
    },
    {
      "epoch": 0.2172957127152803,
      "grad_norm": 0.36353059625789313,
      "learning_rate": 7.228327228327229e-06,
      "loss": 0.5156,
      "step": 593
    },
    {
      "epoch": 0.21766214730670574,
      "grad_norm": 0.37248214066947044,
      "learning_rate": 7.240537240537241e-06,
      "loss": 0.4957,
      "step": 594
    },
    {
      "epoch": 0.21802858189813118,
      "grad_norm": 0.37369306746011877,
      "learning_rate": 7.252747252747253e-06,
      "loss": 0.4957,
      "step": 595
    },
    {
      "epoch": 0.21839501648955661,
      "grad_norm": 0.4110551314252987,
      "learning_rate": 7.264957264957266e-06,
      "loss": 0.475,
      "step": 596
    },
    {
      "epoch": 0.21876145108098205,
      "grad_norm": 0.41973888175842744,
      "learning_rate": 7.277167277167277e-06,
      "loss": 0.5058,
      "step": 597
    },
    {
      "epoch": 0.21912788567240749,
      "grad_norm": 0.4100487029475357,
      "learning_rate": 7.28937728937729e-06,
      "loss": 0.5184,
      "step": 598
    },
    {
      "epoch": 0.2194943202638329,
      "grad_norm": 0.39092698118540176,
      "learning_rate": 7.301587301587301e-06,
      "loss": 0.4862,
      "step": 599
    },
    {
      "epoch": 0.21986075485525833,
      "grad_norm": 0.38169596077287504,
      "learning_rate": 7.313797313797314e-06,
      "loss": 0.4941,
      "step": 600
    },
    {
      "epoch": 0.22022718944668376,
      "grad_norm": 0.3279561644298945,
      "learning_rate": 7.326007326007326e-06,
      "loss": 0.4573,
      "step": 601
    },
    {
      "epoch": 0.2205936240381092,
      "grad_norm": 0.3650972181891811,
      "learning_rate": 7.338217338217338e-06,
      "loss": 0.4999,
      "step": 602
    },
    {
      "epoch": 0.22096005862953463,
      "grad_norm": 0.37333102269409246,
      "learning_rate": 7.350427350427351e-06,
      "loss": 0.5162,
      "step": 603
    },
    {
      "epoch": 0.22132649322096007,
      "grad_norm": 0.3822762388055442,
      "learning_rate": 7.362637362637364e-06,
      "loss": 0.5025,
      "step": 604
    },
    {
      "epoch": 0.22169292781238548,
      "grad_norm": 0.38107201837798604,
      "learning_rate": 7.374847374847376e-06,
      "loss": 0.4983,
      "step": 605
    },
    {
      "epoch": 0.2220593624038109,
      "grad_norm": 0.3518628473244443,
      "learning_rate": 7.387057387057388e-06,
      "loss": 0.5094,
      "step": 606
    },
    {
      "epoch": 0.22242579699523635,
      "grad_norm": 0.4065729659192712,
      "learning_rate": 7.3992673992674e-06,
      "loss": 0.533,
      "step": 607
    },
    {
      "epoch": 0.22279223158666178,
      "grad_norm": 0.3803879135584217,
      "learning_rate": 7.411477411477413e-06,
      "loss": 0.4872,
      "step": 608
    },
    {
      "epoch": 0.22315866617808722,
      "grad_norm": 0.5291897751112761,
      "learning_rate": 7.423687423687424e-06,
      "loss": 0.5102,
      "step": 609
    },
    {
      "epoch": 0.22352510076951265,
      "grad_norm": 0.34258697600863924,
      "learning_rate": 7.435897435897437e-06,
      "loss": 0.4867,
      "step": 610
    },
    {
      "epoch": 0.22389153536093806,
      "grad_norm": 0.4024262601359685,
      "learning_rate": 7.448107448107449e-06,
      "loss": 0.4948,
      "step": 611
    },
    {
      "epoch": 0.2242579699523635,
      "grad_norm": 0.3715769143892993,
      "learning_rate": 7.460317460317461e-06,
      "loss": 0.4757,
      "step": 612
    },
    {
      "epoch": 0.22462440454378893,
      "grad_norm": 0.3913303916996863,
      "learning_rate": 7.472527472527473e-06,
      "loss": 0.4564,
      "step": 613
    },
    {
      "epoch": 0.22499083913521437,
      "grad_norm": 0.36195617233907296,
      "learning_rate": 7.484737484737485e-06,
      "loss": 0.4868,
      "step": 614
    },
    {
      "epoch": 0.2253572737266398,
      "grad_norm": 0.33364959707073427,
      "learning_rate": 7.4969474969474975e-06,
      "loss": 0.4901,
      "step": 615
    },
    {
      "epoch": 0.22572370831806524,
      "grad_norm": 0.3807973159504809,
      "learning_rate": 7.50915750915751e-06,
      "loss": 0.4945,
      "step": 616
    },
    {
      "epoch": 0.22609014290949064,
      "grad_norm": 0.3445841045997452,
      "learning_rate": 7.521367521367522e-06,
      "loss": 0.4808,
      "step": 617
    },
    {
      "epoch": 0.22645657750091608,
      "grad_norm": 0.3945810623015093,
      "learning_rate": 7.533577533577534e-06,
      "loss": 0.4999,
      "step": 618
    },
    {
      "epoch": 0.2268230120923415,
      "grad_norm": 0.3757695302042004,
      "learning_rate": 7.5457875457875465e-06,
      "loss": 0.51,
      "step": 619
    },
    {
      "epoch": 0.22718944668376695,
      "grad_norm": 0.3838943728783627,
      "learning_rate": 7.557997557997558e-06,
      "loss": 0.5027,
      "step": 620
    },
    {
      "epoch": 0.22755588127519238,
      "grad_norm": 0.38772206861454556,
      "learning_rate": 7.570207570207571e-06,
      "loss": 0.4995,
      "step": 621
    },
    {
      "epoch": 0.22792231586661782,
      "grad_norm": 0.40769957642758475,
      "learning_rate": 7.582417582417583e-06,
      "loss": 0.5034,
      "step": 622
    },
    {
      "epoch": 0.22828875045804323,
      "grad_norm": 0.3605965316984489,
      "learning_rate": 7.594627594627595e-06,
      "loss": 0.4801,
      "step": 623
    },
    {
      "epoch": 0.22865518504946866,
      "grad_norm": 0.377583656805914,
      "learning_rate": 7.606837606837607e-06,
      "loss": 0.5108,
      "step": 624
    },
    {
      "epoch": 0.2290216196408941,
      "grad_norm": 0.4067795890033965,
      "learning_rate": 7.61904761904762e-06,
      "loss": 0.4979,
      "step": 625
    },
    {
      "epoch": 0.22938805423231953,
      "grad_norm": 0.5740805567517889,
      "learning_rate": 7.631257631257631e-06,
      "loss": 0.4985,
      "step": 626
    },
    {
      "epoch": 0.22975448882374497,
      "grad_norm": 0.3733024669730781,
      "learning_rate": 7.643467643467644e-06,
      "loss": 0.4852,
      "step": 627
    },
    {
      "epoch": 0.2301209234151704,
      "grad_norm": 0.36645603133527893,
      "learning_rate": 7.655677655677656e-06,
      "loss": 0.4876,
      "step": 628
    },
    {
      "epoch": 0.2304873580065958,
      "grad_norm": 1.0680149267970525,
      "learning_rate": 7.667887667887669e-06,
      "loss": 0.5226,
      "step": 629
    },
    {
      "epoch": 0.23085379259802125,
      "grad_norm": 0.3914890212008394,
      "learning_rate": 7.680097680097681e-06,
      "loss": 0.4899,
      "step": 630
    },
    {
      "epoch": 0.23122022718944668,
      "grad_norm": 0.3594692621965526,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.4959,
      "step": 631
    },
    {
      "epoch": 0.23158666178087212,
      "grad_norm": 0.4461062476507814,
      "learning_rate": 7.704517704517706e-06,
      "loss": 0.4958,
      "step": 632
    },
    {
      "epoch": 0.23195309637229755,
      "grad_norm": 0.3891149211084012,
      "learning_rate": 7.716727716727719e-06,
      "loss": 0.5092,
      "step": 633
    },
    {
      "epoch": 0.232319530963723,
      "grad_norm": 0.3412106659061824,
      "learning_rate": 7.72893772893773e-06,
      "loss": 0.4779,
      "step": 634
    },
    {
      "epoch": 0.2326859655551484,
      "grad_norm": 0.37558454026278787,
      "learning_rate": 7.741147741147742e-06,
      "loss": 0.5065,
      "step": 635
    },
    {
      "epoch": 0.23305240014657383,
      "grad_norm": 0.36799601558593215,
      "learning_rate": 7.753357753357754e-06,
      "loss": 0.4925,
      "step": 636
    },
    {
      "epoch": 0.23341883473799926,
      "grad_norm": 0.41665320170099596,
      "learning_rate": 7.765567765567767e-06,
      "loss": 0.5087,
      "step": 637
    },
    {
      "epoch": 0.2337852693294247,
      "grad_norm": 0.35531777414948396,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.4723,
      "step": 638
    },
    {
      "epoch": 0.23415170392085013,
      "grad_norm": 0.42511796145241604,
      "learning_rate": 7.78998778998779e-06,
      "loss": 0.5014,
      "step": 639
    },
    {
      "epoch": 0.23451813851227557,
      "grad_norm": 0.34829872264702333,
      "learning_rate": 7.802197802197802e-06,
      "loss": 0.4869,
      "step": 640
    },
    {
      "epoch": 0.23488457310370098,
      "grad_norm": 0.35662444460667003,
      "learning_rate": 7.814407814407815e-06,
      "loss": 0.4764,
      "step": 641
    },
    {
      "epoch": 0.2352510076951264,
      "grad_norm": 0.3682144602222955,
      "learning_rate": 7.826617826617827e-06,
      "loss": 0.4753,
      "step": 642
    },
    {
      "epoch": 0.23561744228655185,
      "grad_norm": 0.3573845164424596,
      "learning_rate": 7.83882783882784e-06,
      "loss": 0.4826,
      "step": 643
    },
    {
      "epoch": 0.23598387687797728,
      "grad_norm": 0.3916292341960152,
      "learning_rate": 7.851037851037852e-06,
      "loss": 0.4889,
      "step": 644
    },
    {
      "epoch": 0.23635031146940272,
      "grad_norm": 0.39977541763417573,
      "learning_rate": 7.863247863247863e-06,
      "loss": 0.5158,
      "step": 645
    },
    {
      "epoch": 0.23671674606082815,
      "grad_norm": 0.3431613778112849,
      "learning_rate": 7.875457875457876e-06,
      "loss": 0.4742,
      "step": 646
    },
    {
      "epoch": 0.23708318065225356,
      "grad_norm": 0.344284014982128,
      "learning_rate": 7.887667887667888e-06,
      "loss": 0.4879,
      "step": 647
    },
    {
      "epoch": 0.237449615243679,
      "grad_norm": 0.3732359780607435,
      "learning_rate": 7.8998778998779e-06,
      "loss": 0.5043,
      "step": 648
    },
    {
      "epoch": 0.23781604983510443,
      "grad_norm": 0.4161490369497673,
      "learning_rate": 7.912087912087913e-06,
      "loss": 0.5039,
      "step": 649
    },
    {
      "epoch": 0.23818248442652987,
      "grad_norm": 0.3579353510648252,
      "learning_rate": 7.924297924297925e-06,
      "loss": 0.4798,
      "step": 650
    },
    {
      "epoch": 0.2385489190179553,
      "grad_norm": 0.38954983769672186,
      "learning_rate": 7.936507936507936e-06,
      "loss": 0.478,
      "step": 651
    },
    {
      "epoch": 0.23891535360938074,
      "grad_norm": 0.3458616709639291,
      "learning_rate": 7.948717948717949e-06,
      "loss": 0.4798,
      "step": 652
    },
    {
      "epoch": 0.23928178820080614,
      "grad_norm": 0.35102140891703565,
      "learning_rate": 7.960927960927961e-06,
      "loss": 0.514,
      "step": 653
    },
    {
      "epoch": 0.23964822279223158,
      "grad_norm": 0.3591503880044411,
      "learning_rate": 7.973137973137974e-06,
      "loss": 0.4992,
      "step": 654
    },
    {
      "epoch": 0.24001465738365702,
      "grad_norm": 0.3967349082448371,
      "learning_rate": 7.985347985347986e-06,
      "loss": 0.5282,
      "step": 655
    },
    {
      "epoch": 0.24038109197508245,
      "grad_norm": 0.3972632052247872,
      "learning_rate": 7.997557997557998e-06,
      "loss": 0.4973,
      "step": 656
    },
    {
      "epoch": 0.24074752656650789,
      "grad_norm": 0.4129382322260075,
      "learning_rate": 8.009768009768011e-06,
      "loss": 0.505,
      "step": 657
    },
    {
      "epoch": 0.24111396115793332,
      "grad_norm": 0.363644387530892,
      "learning_rate": 8.021978021978023e-06,
      "loss": 0.5011,
      "step": 658
    },
    {
      "epoch": 0.24148039574935873,
      "grad_norm": 0.3612807707581569,
      "learning_rate": 8.034188034188036e-06,
      "loss": 0.4773,
      "step": 659
    },
    {
      "epoch": 0.24184683034078416,
      "grad_norm": 0.3606222959885171,
      "learning_rate": 8.046398046398047e-06,
      "loss": 0.4728,
      "step": 660
    },
    {
      "epoch": 0.2422132649322096,
      "grad_norm": 0.4166867446696258,
      "learning_rate": 8.058608058608059e-06,
      "loss": 0.5122,
      "step": 661
    },
    {
      "epoch": 0.24257969952363503,
      "grad_norm": 0.35997948265290913,
      "learning_rate": 8.070818070818072e-06,
      "loss": 0.4786,
      "step": 662
    },
    {
      "epoch": 0.24294613411506047,
      "grad_norm": 0.3648000396162286,
      "learning_rate": 8.083028083028084e-06,
      "loss": 0.5013,
      "step": 663
    },
    {
      "epoch": 0.2433125687064859,
      "grad_norm": 0.5589790817732267,
      "learning_rate": 8.095238095238097e-06,
      "loss": 0.4823,
      "step": 664
    },
    {
      "epoch": 0.2436790032979113,
      "grad_norm": 0.3531213870859164,
      "learning_rate": 8.107448107448109e-06,
      "loss": 0.4786,
      "step": 665
    },
    {
      "epoch": 0.24404543788933675,
      "grad_norm": 0.3321121195713429,
      "learning_rate": 8.11965811965812e-06,
      "loss": 0.4483,
      "step": 666
    },
    {
      "epoch": 0.24441187248076218,
      "grad_norm": 0.3594657122842124,
      "learning_rate": 8.131868131868132e-06,
      "loss": 0.485,
      "step": 667
    },
    {
      "epoch": 0.24477830707218762,
      "grad_norm": 0.3606475078963179,
      "learning_rate": 8.144078144078145e-06,
      "loss": 0.5034,
      "step": 668
    },
    {
      "epoch": 0.24514474166361305,
      "grad_norm": 0.39644338930688117,
      "learning_rate": 8.156288156288157e-06,
      "loss": 0.5114,
      "step": 669
    },
    {
      "epoch": 0.2455111762550385,
      "grad_norm": 0.42269985805376,
      "learning_rate": 8.16849816849817e-06,
      "loss": 0.5024,
      "step": 670
    },
    {
      "epoch": 0.2458776108464639,
      "grad_norm": 0.4004947809343667,
      "learning_rate": 8.18070818070818e-06,
      "loss": 0.4632,
      "step": 671
    },
    {
      "epoch": 0.24624404543788933,
      "grad_norm": 0.43051496960594215,
      "learning_rate": 8.192918192918193e-06,
      "loss": 0.4805,
      "step": 672
    },
    {
      "epoch": 0.24661048002931477,
      "grad_norm": 0.3721450513719655,
      "learning_rate": 8.205128205128205e-06,
      "loss": 0.5068,
      "step": 673
    },
    {
      "epoch": 0.2469769146207402,
      "grad_norm": 0.38529575546881467,
      "learning_rate": 8.217338217338218e-06,
      "loss": 0.4779,
      "step": 674
    },
    {
      "epoch": 0.24734334921216564,
      "grad_norm": 0.3851940015795156,
      "learning_rate": 8.22954822954823e-06,
      "loss": 0.4618,
      "step": 675
    },
    {
      "epoch": 0.24770978380359107,
      "grad_norm": 0.3715535257349864,
      "learning_rate": 8.241758241758243e-06,
      "loss": 0.4884,
      "step": 676
    },
    {
      "epoch": 0.24807621839501648,
      "grad_norm": 0.416845665543893,
      "learning_rate": 8.253968253968254e-06,
      "loss": 0.5134,
      "step": 677
    },
    {
      "epoch": 0.24844265298644191,
      "grad_norm": 0.48214339792644967,
      "learning_rate": 8.266178266178266e-06,
      "loss": 0.49,
      "step": 678
    },
    {
      "epoch": 0.24880908757786735,
      "grad_norm": 0.3790994517245311,
      "learning_rate": 8.278388278388278e-06,
      "loss": 0.4864,
      "step": 679
    },
    {
      "epoch": 0.24917552216929278,
      "grad_norm": 0.41482797205951044,
      "learning_rate": 8.290598290598293e-06,
      "loss": 0.5188,
      "step": 680
    },
    {
      "epoch": 0.24954195676071822,
      "grad_norm": 0.38028962183518955,
      "learning_rate": 8.302808302808303e-06,
      "loss": 0.5271,
      "step": 681
    },
    {
      "epoch": 0.24990839135214366,
      "grad_norm": 0.6128326958569499,
      "learning_rate": 8.315018315018316e-06,
      "loss": 0.5001,
      "step": 682
    },
    {
      "epoch": 0.2502748259435691,
      "grad_norm": 0.3754517856010606,
      "learning_rate": 8.327228327228328e-06,
      "loss": 0.4811,
      "step": 683
    },
    {
      "epoch": 0.2506412605349945,
      "grad_norm": 0.36857242075936014,
      "learning_rate": 8.33943833943834e-06,
      "loss": 0.4661,
      "step": 684
    },
    {
      "epoch": 0.25100769512641996,
      "grad_norm": 0.4063883020031402,
      "learning_rate": 8.351648351648353e-06,
      "loss": 0.5019,
      "step": 685
    },
    {
      "epoch": 0.25137412971784534,
      "grad_norm": 0.3583732801907618,
      "learning_rate": 8.363858363858364e-06,
      "loss": 0.4833,
      "step": 686
    },
    {
      "epoch": 0.2517405643092708,
      "grad_norm": 0.37917347975307175,
      "learning_rate": 8.376068376068377e-06,
      "loss": 0.4983,
      "step": 687
    },
    {
      "epoch": 0.2521069989006962,
      "grad_norm": 0.3482370993644404,
      "learning_rate": 8.388278388278389e-06,
      "loss": 0.4785,
      "step": 688
    },
    {
      "epoch": 0.25247343349212165,
      "grad_norm": 0.3528517749312567,
      "learning_rate": 8.400488400488401e-06,
      "loss": 0.4902,
      "step": 689
    },
    {
      "epoch": 0.2528398680835471,
      "grad_norm": 0.36051978923504785,
      "learning_rate": 8.412698412698414e-06,
      "loss": 0.489,
      "step": 690
    },
    {
      "epoch": 0.2532063026749725,
      "grad_norm": 0.3663181285237452,
      "learning_rate": 8.424908424908426e-06,
      "loss": 0.4868,
      "step": 691
    },
    {
      "epoch": 0.25357273726639795,
      "grad_norm": 0.4472243714894805,
      "learning_rate": 8.437118437118437e-06,
      "loss": 0.4942,
      "step": 692
    },
    {
      "epoch": 0.2539391718578234,
      "grad_norm": 0.3743906567117739,
      "learning_rate": 8.44932844932845e-06,
      "loss": 0.4815,
      "step": 693
    },
    {
      "epoch": 0.2543056064492488,
      "grad_norm": 0.42324051755636166,
      "learning_rate": 8.461538461538462e-06,
      "loss": 0.4883,
      "step": 694
    },
    {
      "epoch": 0.25467204104067426,
      "grad_norm": 0.3970052185839475,
      "learning_rate": 8.473748473748475e-06,
      "loss": 0.5013,
      "step": 695
    },
    {
      "epoch": 0.2550384756320997,
      "grad_norm": 0.44168855647231314,
      "learning_rate": 8.485958485958487e-06,
      "loss": 0.4778,
      "step": 696
    },
    {
      "epoch": 0.25540491022352513,
      "grad_norm": 0.4264087954250368,
      "learning_rate": 8.498168498168498e-06,
      "loss": 0.495,
      "step": 697
    },
    {
      "epoch": 0.2557713448149505,
      "grad_norm": 0.38182211599928406,
      "learning_rate": 8.51037851037851e-06,
      "loss": 0.4767,
      "step": 698
    },
    {
      "epoch": 0.25613777940637594,
      "grad_norm": 0.5224704025654386,
      "learning_rate": 8.522588522588523e-06,
      "loss": 0.5091,
      "step": 699
    },
    {
      "epoch": 0.2565042139978014,
      "grad_norm": 0.47886160815455536,
      "learning_rate": 8.534798534798535e-06,
      "loss": 0.4981,
      "step": 700
    },
    {
      "epoch": 0.2568706485892268,
      "grad_norm": 0.42730875846937677,
      "learning_rate": 8.547008547008548e-06,
      "loss": 0.4777,
      "step": 701
    },
    {
      "epoch": 0.25723708318065225,
      "grad_norm": 0.39957244739975445,
      "learning_rate": 8.55921855921856e-06,
      "loss": 0.4911,
      "step": 702
    },
    {
      "epoch": 0.2576035177720777,
      "grad_norm": 0.5033821886433024,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.505,
      "step": 703
    },
    {
      "epoch": 0.2579699523635031,
      "grad_norm": 0.46373980119429103,
      "learning_rate": 8.583638583638583e-06,
      "loss": 0.4935,
      "step": 704
    },
    {
      "epoch": 0.25833638695492855,
      "grad_norm": 0.43587203673499986,
      "learning_rate": 8.595848595848598e-06,
      "loss": 0.4833,
      "step": 705
    },
    {
      "epoch": 0.258702821546354,
      "grad_norm": 0.3815639984552434,
      "learning_rate": 8.60805860805861e-06,
      "loss": 0.488,
      "step": 706
    },
    {
      "epoch": 0.2590692561377794,
      "grad_norm": 0.39061679265024074,
      "learning_rate": 8.62026862026862e-06,
      "loss": 0.4866,
      "step": 707
    },
    {
      "epoch": 0.25943569072920486,
      "grad_norm": 0.6166284286824331,
      "learning_rate": 8.632478632478633e-06,
      "loss": 0.5125,
      "step": 708
    },
    {
      "epoch": 0.2598021253206303,
      "grad_norm": 0.46001536865201403,
      "learning_rate": 8.644688644688646e-06,
      "loss": 0.4752,
      "step": 709
    },
    {
      "epoch": 0.2601685599120557,
      "grad_norm": 0.4028807379840428,
      "learning_rate": 8.656898656898658e-06,
      "loss": 0.5031,
      "step": 710
    },
    {
      "epoch": 0.2605349945034811,
      "grad_norm": 0.38130588917586233,
      "learning_rate": 8.66910866910867e-06,
      "loss": 0.5013,
      "step": 711
    },
    {
      "epoch": 0.26090142909490655,
      "grad_norm": 0.5077990918105902,
      "learning_rate": 8.681318681318681e-06,
      "loss": 0.504,
      "step": 712
    },
    {
      "epoch": 0.261267863686332,
      "grad_norm": 0.36416641891494683,
      "learning_rate": 8.693528693528694e-06,
      "loss": 0.4827,
      "step": 713
    },
    {
      "epoch": 0.2616342982777574,
      "grad_norm": 0.4077649660929022,
      "learning_rate": 8.705738705738706e-06,
      "loss": 0.4827,
      "step": 714
    },
    {
      "epoch": 0.26200073286918285,
      "grad_norm": 0.37862511528136267,
      "learning_rate": 8.717948717948719e-06,
      "loss": 0.4986,
      "step": 715
    },
    {
      "epoch": 0.2623671674606083,
      "grad_norm": 0.3637412236681666,
      "learning_rate": 8.730158730158731e-06,
      "loss": 0.494,
      "step": 716
    },
    {
      "epoch": 0.2627336020520337,
      "grad_norm": 0.35258327168038306,
      "learning_rate": 8.742368742368744e-06,
      "loss": 0.4846,
      "step": 717
    },
    {
      "epoch": 0.26310003664345916,
      "grad_norm": 0.4050326298891662,
      "learning_rate": 8.754578754578755e-06,
      "loss": 0.4949,
      "step": 718
    },
    {
      "epoch": 0.2634664712348846,
      "grad_norm": 0.38138503356723485,
      "learning_rate": 8.766788766788767e-06,
      "loss": 0.5085,
      "step": 719
    },
    {
      "epoch": 0.26383290582631,
      "grad_norm": 0.35975819724761976,
      "learning_rate": 8.77899877899878e-06,
      "loss": 0.489,
      "step": 720
    },
    {
      "epoch": 0.2641993404177354,
      "grad_norm": 0.37188720736494213,
      "learning_rate": 8.791208791208792e-06,
      "loss": 0.4987,
      "step": 721
    },
    {
      "epoch": 0.26456577500916084,
      "grad_norm": 0.3912738500585695,
      "learning_rate": 8.803418803418804e-06,
      "loss": 0.5063,
      "step": 722
    },
    {
      "epoch": 0.2649322096005863,
      "grad_norm": 0.34511630272337923,
      "learning_rate": 8.815628815628817e-06,
      "loss": 0.4689,
      "step": 723
    },
    {
      "epoch": 0.2652986441920117,
      "grad_norm": 0.40941443397388766,
      "learning_rate": 8.827838827838828e-06,
      "loss": 0.4777,
      "step": 724
    },
    {
      "epoch": 0.26566507878343715,
      "grad_norm": 0.3953743974697647,
      "learning_rate": 8.84004884004884e-06,
      "loss": 0.4919,
      "step": 725
    },
    {
      "epoch": 0.2660315133748626,
      "grad_norm": 0.3717861695817555,
      "learning_rate": 8.852258852258853e-06,
      "loss": 0.4877,
      "step": 726
    },
    {
      "epoch": 0.266397947966288,
      "grad_norm": 0.3695564315557365,
      "learning_rate": 8.864468864468865e-06,
      "loss": 0.5203,
      "step": 727
    },
    {
      "epoch": 0.26676438255771345,
      "grad_norm": 0.5311792271181744,
      "learning_rate": 8.876678876678878e-06,
      "loss": 0.4987,
      "step": 728
    },
    {
      "epoch": 0.2671308171491389,
      "grad_norm": 0.45987713273491854,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.4892,
      "step": 729
    },
    {
      "epoch": 0.2674972517405643,
      "grad_norm": 0.38942470628908815,
      "learning_rate": 8.9010989010989e-06,
      "loss": 0.4809,
      "step": 730
    },
    {
      "epoch": 0.26786368633198976,
      "grad_norm": 0.43706401165002884,
      "learning_rate": 8.913308913308915e-06,
      "loss": 0.4815,
      "step": 731
    },
    {
      "epoch": 0.2682301209234152,
      "grad_norm": 0.41187571049979005,
      "learning_rate": 8.925518925518927e-06,
      "loss": 0.4943,
      "step": 732
    },
    {
      "epoch": 0.2685965555148406,
      "grad_norm": 0.35534055221041067,
      "learning_rate": 8.937728937728938e-06,
      "loss": 0.4903,
      "step": 733
    },
    {
      "epoch": 0.268962990106266,
      "grad_norm": 0.39316260985050466,
      "learning_rate": 8.94993894993895e-06,
      "loss": 0.5121,
      "step": 734
    },
    {
      "epoch": 0.26932942469769144,
      "grad_norm": 0.40817465462423497,
      "learning_rate": 8.962148962148963e-06,
      "loss": 0.5123,
      "step": 735
    },
    {
      "epoch": 0.2696958592891169,
      "grad_norm": 0.3645037583598911,
      "learning_rate": 8.974358974358976e-06,
      "loss": 0.4667,
      "step": 736
    },
    {
      "epoch": 0.2700622938805423,
      "grad_norm": 0.33472277766146563,
      "learning_rate": 8.986568986568988e-06,
      "loss": 0.4827,
      "step": 737
    },
    {
      "epoch": 0.27042872847196775,
      "grad_norm": 0.3749787795357067,
      "learning_rate": 8.998778998779e-06,
      "loss": 0.4924,
      "step": 738
    },
    {
      "epoch": 0.2707951630633932,
      "grad_norm": 0.4207209364328427,
      "learning_rate": 9.010989010989011e-06,
      "loss": 0.503,
      "step": 739
    },
    {
      "epoch": 0.2711615976548186,
      "grad_norm": 0.3895521870207341,
      "learning_rate": 9.023199023199024e-06,
      "loss": 0.5214,
      "step": 740
    },
    {
      "epoch": 0.27152803224624406,
      "grad_norm": 0.3720728425291998,
      "learning_rate": 9.035409035409036e-06,
      "loss": 0.4871,
      "step": 741
    },
    {
      "epoch": 0.2718944668376695,
      "grad_norm": 0.3581647722413061,
      "learning_rate": 9.047619047619049e-06,
      "loss": 0.4812,
      "step": 742
    },
    {
      "epoch": 0.2722609014290949,
      "grad_norm": 0.4494516345388821,
      "learning_rate": 9.059829059829061e-06,
      "loss": 0.4964,
      "step": 743
    },
    {
      "epoch": 0.27262733602052036,
      "grad_norm": 0.3287827914560442,
      "learning_rate": 9.072039072039072e-06,
      "loss": 0.4792,
      "step": 744
    },
    {
      "epoch": 0.27299377061194574,
      "grad_norm": 0.3583864862033823,
      "learning_rate": 9.084249084249084e-06,
      "loss": 0.4626,
      "step": 745
    },
    {
      "epoch": 0.2733602052033712,
      "grad_norm": 0.3672474921296208,
      "learning_rate": 9.096459096459097e-06,
      "loss": 0.4926,
      "step": 746
    },
    {
      "epoch": 0.2737266397947966,
      "grad_norm": 0.4432180539800607,
      "learning_rate": 9.10866910866911e-06,
      "loss": 0.4955,
      "step": 747
    },
    {
      "epoch": 0.27409307438622205,
      "grad_norm": 0.3561670999380096,
      "learning_rate": 9.120879120879122e-06,
      "loss": 0.4982,
      "step": 748
    },
    {
      "epoch": 0.2744595089776475,
      "grad_norm": 0.35467915959810536,
      "learning_rate": 9.133089133089134e-06,
      "loss": 0.4768,
      "step": 749
    },
    {
      "epoch": 0.2748259435690729,
      "grad_norm": 0.37173776547366577,
      "learning_rate": 9.145299145299145e-06,
      "loss": 0.4928,
      "step": 750
    },
    {
      "epoch": 0.27519237816049835,
      "grad_norm": 0.38374110968556174,
      "learning_rate": 9.157509157509158e-06,
      "loss": 0.4898,
      "step": 751
    },
    {
      "epoch": 0.2755588127519238,
      "grad_norm": 0.3587086942313156,
      "learning_rate": 9.16971916971917e-06,
      "loss": 0.4776,
      "step": 752
    },
    {
      "epoch": 0.2759252473433492,
      "grad_norm": 0.36962841277631536,
      "learning_rate": 9.181929181929182e-06,
      "loss": 0.5047,
      "step": 753
    },
    {
      "epoch": 0.27629168193477466,
      "grad_norm": 0.3596551075603808,
      "learning_rate": 9.194139194139195e-06,
      "loss": 0.4824,
      "step": 754
    },
    {
      "epoch": 0.2766581165262001,
      "grad_norm": 0.3898787303968857,
      "learning_rate": 9.206349206349207e-06,
      "loss": 0.5053,
      "step": 755
    },
    {
      "epoch": 0.27702455111762553,
      "grad_norm": 0.3609157879391422,
      "learning_rate": 9.218559218559218e-06,
      "loss": 0.4992,
      "step": 756
    },
    {
      "epoch": 0.2773909857090509,
      "grad_norm": 0.45311838094446444,
      "learning_rate": 9.230769230769232e-06,
      "loss": 0.512,
      "step": 757
    },
    {
      "epoch": 0.27775742030047634,
      "grad_norm": 0.3660550216512867,
      "learning_rate": 9.242979242979245e-06,
      "loss": 0.4706,
      "step": 758
    },
    {
      "epoch": 0.2781238548919018,
      "grad_norm": 0.35131751304204967,
      "learning_rate": 9.255189255189256e-06,
      "loss": 0.4942,
      "step": 759
    },
    {
      "epoch": 0.2784902894833272,
      "grad_norm": 0.40997255037573044,
      "learning_rate": 9.267399267399268e-06,
      "loss": 0.5033,
      "step": 760
    },
    {
      "epoch": 0.27885672407475265,
      "grad_norm": 0.40864481585724405,
      "learning_rate": 9.27960927960928e-06,
      "loss": 0.4927,
      "step": 761
    },
    {
      "epoch": 0.2792231586661781,
      "grad_norm": 0.3467758055475284,
      "learning_rate": 9.291819291819293e-06,
      "loss": 0.4686,
      "step": 762
    },
    {
      "epoch": 0.2795895932576035,
      "grad_norm": 0.3885411346693868,
      "learning_rate": 9.304029304029305e-06,
      "loss": 0.4887,
      "step": 763
    },
    {
      "epoch": 0.27995602784902895,
      "grad_norm": 0.39495354815406547,
      "learning_rate": 9.316239316239318e-06,
      "loss": 0.4904,
      "step": 764
    },
    {
      "epoch": 0.2803224624404544,
      "grad_norm": 0.3736539742033064,
      "learning_rate": 9.328449328449329e-06,
      "loss": 0.4911,
      "step": 765
    },
    {
      "epoch": 0.2806888970318798,
      "grad_norm": 0.37845646114223475,
      "learning_rate": 9.340659340659341e-06,
      "loss": 0.485,
      "step": 766
    },
    {
      "epoch": 0.28105533162330526,
      "grad_norm": 0.3629830496925177,
      "learning_rate": 9.352869352869354e-06,
      "loss": 0.4737,
      "step": 767
    },
    {
      "epoch": 0.2814217662147307,
      "grad_norm": 0.403627925188817,
      "learning_rate": 9.365079365079366e-06,
      "loss": 0.4925,
      "step": 768
    },
    {
      "epoch": 0.2817882008061561,
      "grad_norm": 0.3970956223594606,
      "learning_rate": 9.377289377289379e-06,
      "loss": 0.4941,
      "step": 769
    },
    {
      "epoch": 0.2821546353975815,
      "grad_norm": 0.35528086824759786,
      "learning_rate": 9.389499389499391e-06,
      "loss": 0.4801,
      "step": 770
    },
    {
      "epoch": 0.28252106998900695,
      "grad_norm": 0.5497084354969882,
      "learning_rate": 9.401709401709402e-06,
      "loss": 0.4796,
      "step": 771
    },
    {
      "epoch": 0.2828875045804324,
      "grad_norm": 0.349820838691197,
      "learning_rate": 9.413919413919414e-06,
      "loss": 0.4961,
      "step": 772
    },
    {
      "epoch": 0.2832539391718578,
      "grad_norm": 0.433207752961731,
      "learning_rate": 9.426129426129427e-06,
      "loss": 0.4791,
      "step": 773
    },
    {
      "epoch": 0.28362037376328325,
      "grad_norm": 0.37400648248319857,
      "learning_rate": 9.43833943833944e-06,
      "loss": 0.4895,
      "step": 774
    },
    {
      "epoch": 0.2839868083547087,
      "grad_norm": 0.36291983515716425,
      "learning_rate": 9.450549450549452e-06,
      "loss": 0.4766,
      "step": 775
    },
    {
      "epoch": 0.2843532429461341,
      "grad_norm": 0.3925403165376891,
      "learning_rate": 9.462759462759462e-06,
      "loss": 0.49,
      "step": 776
    },
    {
      "epoch": 0.28471967753755956,
      "grad_norm": 0.3340704145164964,
      "learning_rate": 9.474969474969475e-06,
      "loss": 0.4832,
      "step": 777
    },
    {
      "epoch": 0.285086112128985,
      "grad_norm": 0.3791800058619776,
      "learning_rate": 9.487179487179487e-06,
      "loss": 0.496,
      "step": 778
    },
    {
      "epoch": 0.2854525467204104,
      "grad_norm": 0.35337793539834833,
      "learning_rate": 9.4993894993895e-06,
      "loss": 0.4988,
      "step": 779
    },
    {
      "epoch": 0.28581898131183586,
      "grad_norm": 0.4087444852381171,
      "learning_rate": 9.511599511599512e-06,
      "loss": 0.4926,
      "step": 780
    },
    {
      "epoch": 0.28618541590326124,
      "grad_norm": 0.3995732590237783,
      "learning_rate": 9.523809523809525e-06,
      "loss": 0.4867,
      "step": 781
    },
    {
      "epoch": 0.2865518504946867,
      "grad_norm": 0.36457220993789036,
      "learning_rate": 9.536019536019537e-06,
      "loss": 0.4965,
      "step": 782
    },
    {
      "epoch": 0.2869182850861121,
      "grad_norm": 0.4517496642975283,
      "learning_rate": 9.54822954822955e-06,
      "loss": 0.4943,
      "step": 783
    },
    {
      "epoch": 0.28728471967753755,
      "grad_norm": 0.46408379933086275,
      "learning_rate": 9.560439560439562e-06,
      "loss": 0.4989,
      "step": 784
    },
    {
      "epoch": 0.287651154268963,
      "grad_norm": 0.3735521600816869,
      "learning_rate": 9.572649572649575e-06,
      "loss": 0.4616,
      "step": 785
    },
    {
      "epoch": 0.2880175888603884,
      "grad_norm": 0.3687222244996573,
      "learning_rate": 9.584859584859585e-06,
      "loss": 0.4968,
      "step": 786
    },
    {
      "epoch": 0.28838402345181385,
      "grad_norm": 0.3699655603577364,
      "learning_rate": 9.597069597069598e-06,
      "loss": 0.4849,
      "step": 787
    },
    {
      "epoch": 0.2887504580432393,
      "grad_norm": 0.35011120803701096,
      "learning_rate": 9.60927960927961e-06,
      "loss": 0.4459,
      "step": 788
    },
    {
      "epoch": 0.2891168926346647,
      "grad_norm": 0.355231731201131,
      "learning_rate": 9.621489621489623e-06,
      "loss": 0.5089,
      "step": 789
    },
    {
      "epoch": 0.28948332722609016,
      "grad_norm": 0.3359836039449088,
      "learning_rate": 9.633699633699635e-06,
      "loss": 0.4817,
      "step": 790
    },
    {
      "epoch": 0.2898497618175156,
      "grad_norm": 0.3980277042880035,
      "learning_rate": 9.645909645909646e-06,
      "loss": 0.5068,
      "step": 791
    },
    {
      "epoch": 0.29021619640894103,
      "grad_norm": 0.37780491732049987,
      "learning_rate": 9.658119658119659e-06,
      "loss": 0.4873,
      "step": 792
    },
    {
      "epoch": 0.2905826310003664,
      "grad_norm": 0.3577855401421477,
      "learning_rate": 9.670329670329671e-06,
      "loss": 0.4891,
      "step": 793
    },
    {
      "epoch": 0.29094906559179184,
      "grad_norm": 0.5289375036136742,
      "learning_rate": 9.682539682539683e-06,
      "loss": 0.4873,
      "step": 794
    },
    {
      "epoch": 0.2913155001832173,
      "grad_norm": 0.42119567914572176,
      "learning_rate": 9.694749694749696e-06,
      "loss": 0.4932,
      "step": 795
    },
    {
      "epoch": 0.2916819347746427,
      "grad_norm": 0.3844797532102613,
      "learning_rate": 9.706959706959708e-06,
      "loss": 0.497,
      "step": 796
    },
    {
      "epoch": 0.29204836936606815,
      "grad_norm": 0.3262312567853422,
      "learning_rate": 9.719169719169719e-06,
      "loss": 0.4896,
      "step": 797
    },
    {
      "epoch": 0.2924148039574936,
      "grad_norm": 0.38371445416847405,
      "learning_rate": 9.731379731379732e-06,
      "loss": 0.5028,
      "step": 798
    },
    {
      "epoch": 0.292781238548919,
      "grad_norm": 0.36575915906200024,
      "learning_rate": 9.743589743589744e-06,
      "loss": 0.4766,
      "step": 799
    },
    {
      "epoch": 0.29314767314034446,
      "grad_norm": 0.33701876264919745,
      "learning_rate": 9.755799755799757e-06,
      "loss": 0.4763,
      "step": 800
    },
    {
      "epoch": 0.2935141077317699,
      "grad_norm": 0.3642072720305421,
      "learning_rate": 9.768009768009769e-06,
      "loss": 0.4943,
      "step": 801
    },
    {
      "epoch": 0.2938805423231953,
      "grad_norm": 0.35882585725145155,
      "learning_rate": 9.780219780219781e-06,
      "loss": 0.4855,
      "step": 802
    },
    {
      "epoch": 0.29424697691462076,
      "grad_norm": 0.43696949731560797,
      "learning_rate": 9.792429792429792e-06,
      "loss": 0.4953,
      "step": 803
    },
    {
      "epoch": 0.2946134115060462,
      "grad_norm": 0.36675545787963326,
      "learning_rate": 9.804639804639805e-06,
      "loss": 0.4732,
      "step": 804
    },
    {
      "epoch": 0.2949798460974716,
      "grad_norm": 0.37572561905878205,
      "learning_rate": 9.816849816849817e-06,
      "loss": 0.4882,
      "step": 805
    },
    {
      "epoch": 0.295346280688897,
      "grad_norm": 0.4380952896658236,
      "learning_rate": 9.82905982905983e-06,
      "loss": 0.4988,
      "step": 806
    },
    {
      "epoch": 0.29571271528032245,
      "grad_norm": 0.40911585717530374,
      "learning_rate": 9.841269841269842e-06,
      "loss": 0.4742,
      "step": 807
    },
    {
      "epoch": 0.2960791498717479,
      "grad_norm": 0.3526880859681006,
      "learning_rate": 9.853479853479855e-06,
      "loss": 0.4878,
      "step": 808
    },
    {
      "epoch": 0.2964455844631733,
      "grad_norm": 0.3394612889838655,
      "learning_rate": 9.865689865689867e-06,
      "loss": 0.478,
      "step": 809
    },
    {
      "epoch": 0.29681201905459875,
      "grad_norm": 0.4776008736559957,
      "learning_rate": 9.87789987789988e-06,
      "loss": 0.4779,
      "step": 810
    },
    {
      "epoch": 0.2971784536460242,
      "grad_norm": 0.3642881191246686,
      "learning_rate": 9.890109890109892e-06,
      "loss": 0.4863,
      "step": 811
    },
    {
      "epoch": 0.2975448882374496,
      "grad_norm": 0.43348608864602856,
      "learning_rate": 9.902319902319903e-06,
      "loss": 0.4972,
      "step": 812
    },
    {
      "epoch": 0.29791132282887506,
      "grad_norm": 0.3987639874738233,
      "learning_rate": 9.914529914529915e-06,
      "loss": 0.4853,
      "step": 813
    },
    {
      "epoch": 0.2982777574203005,
      "grad_norm": 0.390700657717214,
      "learning_rate": 9.926739926739928e-06,
      "loss": 0.4922,
      "step": 814
    },
    {
      "epoch": 0.29864419201172593,
      "grad_norm": 0.33282433792860544,
      "learning_rate": 9.93894993894994e-06,
      "loss": 0.4795,
      "step": 815
    },
    {
      "epoch": 0.29901062660315136,
      "grad_norm": 0.3987820312707686,
      "learning_rate": 9.951159951159953e-06,
      "loss": 0.4806,
      "step": 816
    },
    {
      "epoch": 0.29937706119457674,
      "grad_norm": 0.35434444035941337,
      "learning_rate": 9.963369963369965e-06,
      "loss": 0.4686,
      "step": 817
    },
    {
      "epoch": 0.2997434957860022,
      "grad_norm": 0.3289707024079982,
      "learning_rate": 9.975579975579976e-06,
      "loss": 0.475,
      "step": 818
    },
    {
      "epoch": 0.3001099303774276,
      "grad_norm": 0.4187630643770696,
      "learning_rate": 9.987789987789988e-06,
      "loss": 0.4938,
      "step": 819
    },
    {
      "epoch": 0.30047636496885305,
      "grad_norm": 0.41341405450193025,
      "learning_rate": 1e-05,
      "loss": 0.5042,
      "step": 820
    },
    {
      "epoch": 0.3008427995602785,
      "grad_norm": 0.41428663442956193,
      "learning_rate": 9.999999545493067e-06,
      "loss": 0.4809,
      "step": 821
    },
    {
      "epoch": 0.3012092341517039,
      "grad_norm": 0.40281789161147064,
      "learning_rate": 9.999998181972347e-06,
      "loss": 0.4757,
      "step": 822
    },
    {
      "epoch": 0.30157566874312935,
      "grad_norm": 0.4132418560404733,
      "learning_rate": 9.999995909438092e-06,
      "loss": 0.483,
      "step": 823
    },
    {
      "epoch": 0.3019421033345548,
      "grad_norm": 0.3671082530372566,
      "learning_rate": 9.999992727890711e-06,
      "loss": 0.4768,
      "step": 824
    },
    {
      "epoch": 0.3023085379259802,
      "grad_norm": 0.4068806598498716,
      "learning_rate": 9.999988637330786e-06,
      "loss": 0.5188,
      "step": 825
    },
    {
      "epoch": 0.30267497251740566,
      "grad_norm": 0.4125913032425474,
      "learning_rate": 9.999983637759059e-06,
      "loss": 0.4819,
      "step": 826
    },
    {
      "epoch": 0.3030414071088311,
      "grad_norm": 0.3510898305743237,
      "learning_rate": 9.999977729176437e-06,
      "loss": 0.4812,
      "step": 827
    },
    {
      "epoch": 0.30340784170025653,
      "grad_norm": 0.38228516649100863,
      "learning_rate": 9.999970911583999e-06,
      "loss": 0.507,
      "step": 828
    },
    {
      "epoch": 0.3037742762916819,
      "grad_norm": 0.47810526214652366,
      "learning_rate": 9.99996318498298e-06,
      "loss": 0.4945,
      "step": 829
    },
    {
      "epoch": 0.30414071088310735,
      "grad_norm": 0.39853861773491683,
      "learning_rate": 9.999954549374787e-06,
      "loss": 0.4728,
      "step": 830
    },
    {
      "epoch": 0.3045071454745328,
      "grad_norm": 0.3746307843577429,
      "learning_rate": 9.999945004760988e-06,
      "loss": 0.4721,
      "step": 831
    },
    {
      "epoch": 0.3048735800659582,
      "grad_norm": 0.36522110547759384,
      "learning_rate": 9.999934551143319e-06,
      "loss": 0.4809,
      "step": 832
    },
    {
      "epoch": 0.30524001465738365,
      "grad_norm": 0.4091304202367403,
      "learning_rate": 9.999923188523683e-06,
      "loss": 0.5176,
      "step": 833
    },
    {
      "epoch": 0.3056064492488091,
      "grad_norm": 0.3735651322247712,
      "learning_rate": 9.999910916904144e-06,
      "loss": 0.487,
      "step": 834
    },
    {
      "epoch": 0.3059728838402345,
      "grad_norm": 0.3407670981452426,
      "learning_rate": 9.999897736286932e-06,
      "loss": 0.5046,
      "step": 835
    },
    {
      "epoch": 0.30633931843165996,
      "grad_norm": 0.3358725268070786,
      "learning_rate": 9.999883646674445e-06,
      "loss": 0.4847,
      "step": 836
    },
    {
      "epoch": 0.3067057530230854,
      "grad_norm": 0.3723559188610394,
      "learning_rate": 9.999868648069243e-06,
      "loss": 0.4879,
      "step": 837
    },
    {
      "epoch": 0.3070721876145108,
      "grad_norm": 0.3970178481109056,
      "learning_rate": 9.999852740474054e-06,
      "loss": 0.4893,
      "step": 838
    },
    {
      "epoch": 0.30743862220593626,
      "grad_norm": 0.3864544531756429,
      "learning_rate": 9.999835923891768e-06,
      "loss": 0.4781,
      "step": 839
    },
    {
      "epoch": 0.3078050567973617,
      "grad_norm": 0.35629711566347455,
      "learning_rate": 9.999818198325444e-06,
      "loss": 0.4844,
      "step": 840
    },
    {
      "epoch": 0.3081714913887871,
      "grad_norm": 0.35823163049965295,
      "learning_rate": 9.999799563778307e-06,
      "loss": 0.4738,
      "step": 841
    },
    {
      "epoch": 0.3085379259802125,
      "grad_norm": 0.3616854816186216,
      "learning_rate": 9.99978002025374e-06,
      "loss": 0.4823,
      "step": 842
    },
    {
      "epoch": 0.30890436057163795,
      "grad_norm": 0.4473414044498849,
      "learning_rate": 9.999759567755298e-06,
      "loss": 0.4683,
      "step": 843
    },
    {
      "epoch": 0.3092707951630634,
      "grad_norm": 0.41734086620728544,
      "learning_rate": 9.9997382062867e-06,
      "loss": 0.5212,
      "step": 844
    },
    {
      "epoch": 0.3096372297544888,
      "grad_norm": 0.4077917620575579,
      "learning_rate": 9.999715935851829e-06,
      "loss": 0.4878,
      "step": 845
    },
    {
      "epoch": 0.31000366434591425,
      "grad_norm": 0.32630624328652214,
      "learning_rate": 9.999692756454734e-06,
      "loss": 0.4635,
      "step": 846
    },
    {
      "epoch": 0.3103700989373397,
      "grad_norm": 0.3506401031709525,
      "learning_rate": 9.999668668099628e-06,
      "loss": 0.4945,
      "step": 847
    },
    {
      "epoch": 0.3107365335287651,
      "grad_norm": 0.4078048364926696,
      "learning_rate": 9.999643670790894e-06,
      "loss": 0.4912,
      "step": 848
    },
    {
      "epoch": 0.31110296812019056,
      "grad_norm": 0.3736729823526688,
      "learning_rate": 9.999617764533071e-06,
      "loss": 0.4851,
      "step": 849
    },
    {
      "epoch": 0.311469402711616,
      "grad_norm": 0.3394517678904402,
      "learning_rate": 9.999590949330876e-06,
      "loss": 0.4896,
      "step": 850
    },
    {
      "epoch": 0.31183583730304143,
      "grad_norm": 0.3718268644495909,
      "learning_rate": 9.999563225189174e-06,
      "loss": 0.4819,
      "step": 851
    },
    {
      "epoch": 0.31220227189446687,
      "grad_norm": 0.40000431279409043,
      "learning_rate": 9.999534592113016e-06,
      "loss": 0.4912,
      "step": 852
    },
    {
      "epoch": 0.31256870648589224,
      "grad_norm": 0.3435751015618656,
      "learning_rate": 9.999505050107601e-06,
      "loss": 0.4634,
      "step": 853
    },
    {
      "epoch": 0.3129351410773177,
      "grad_norm": 0.3656552513580114,
      "learning_rate": 9.9994745991783e-06,
      "loss": 0.4777,
      "step": 854
    },
    {
      "epoch": 0.3133015756687431,
      "grad_norm": 0.34255993671414675,
      "learning_rate": 9.999443239330652e-06,
      "loss": 0.467,
      "step": 855
    },
    {
      "epoch": 0.31366801026016855,
      "grad_norm": 0.35574218835641785,
      "learning_rate": 9.999410970570358e-06,
      "loss": 0.4733,
      "step": 856
    },
    {
      "epoch": 0.314034444851594,
      "grad_norm": 0.35277335006259075,
      "learning_rate": 9.999377792903281e-06,
      "loss": 0.4899,
      "step": 857
    },
    {
      "epoch": 0.3144008794430194,
      "grad_norm": 0.3764557308782789,
      "learning_rate": 9.999343706335457e-06,
      "loss": 0.5074,
      "step": 858
    },
    {
      "epoch": 0.31476731403444486,
      "grad_norm": 0.3666469640245472,
      "learning_rate": 9.999308710873083e-06,
      "loss": 0.5038,
      "step": 859
    },
    {
      "epoch": 0.3151337486258703,
      "grad_norm": 0.3687292500827223,
      "learning_rate": 9.999272806522517e-06,
      "loss": 0.511,
      "step": 860
    },
    {
      "epoch": 0.3155001832172957,
      "grad_norm": 0.5329575045375692,
      "learning_rate": 9.99923599329029e-06,
      "loss": 0.5169,
      "step": 861
    },
    {
      "epoch": 0.31586661780872116,
      "grad_norm": 0.31506705870780954,
      "learning_rate": 9.999198271183094e-06,
      "loss": 0.4803,
      "step": 862
    },
    {
      "epoch": 0.3162330524001466,
      "grad_norm": 0.3393686220550401,
      "learning_rate": 9.999159640207787e-06,
      "loss": 0.5003,
      "step": 863
    },
    {
      "epoch": 0.316599486991572,
      "grad_norm": 0.3581028333717299,
      "learning_rate": 9.999120100371392e-06,
      "loss": 0.4814,
      "step": 864
    },
    {
      "epoch": 0.3169659215829974,
      "grad_norm": 0.392453271625103,
      "learning_rate": 9.999079651681096e-06,
      "loss": 0.4938,
      "step": 865
    },
    {
      "epoch": 0.31733235617442285,
      "grad_norm": 0.33913423993719477,
      "learning_rate": 9.999038294144254e-06,
      "loss": 0.4632,
      "step": 866
    },
    {
      "epoch": 0.3176987907658483,
      "grad_norm": 0.36874935063565545,
      "learning_rate": 9.998996027768387e-06,
      "loss": 0.4928,
      "step": 867
    },
    {
      "epoch": 0.3180652253572737,
      "grad_norm": 0.6706100622364325,
      "learning_rate": 9.998952852561176e-06,
      "loss": 0.4653,
      "step": 868
    },
    {
      "epoch": 0.31843165994869915,
      "grad_norm": 0.3822410602382159,
      "learning_rate": 9.998908768530473e-06,
      "loss": 0.4783,
      "step": 869
    },
    {
      "epoch": 0.3187980945401246,
      "grad_norm": 0.3838047436245271,
      "learning_rate": 9.99886377568429e-06,
      "loss": 0.4912,
      "step": 870
    },
    {
      "epoch": 0.31916452913155,
      "grad_norm": 0.3816427102232793,
      "learning_rate": 9.998817874030808e-06,
      "loss": 0.4916,
      "step": 871
    },
    {
      "epoch": 0.31953096372297546,
      "grad_norm": 0.3885893546145554,
      "learning_rate": 9.998771063578372e-06,
      "loss": 0.5033,
      "step": 872
    },
    {
      "epoch": 0.3198973983144009,
      "grad_norm": 0.3845807139166245,
      "learning_rate": 9.998723344335492e-06,
      "loss": 0.5011,
      "step": 873
    },
    {
      "epoch": 0.32026383290582633,
      "grad_norm": 0.36794308899379286,
      "learning_rate": 9.998674716310846e-06,
      "loss": 0.4889,
      "step": 874
    },
    {
      "epoch": 0.32063026749725176,
      "grad_norm": 0.4759354169976258,
      "learning_rate": 9.998625179513271e-06,
      "loss": 0.4696,
      "step": 875
    },
    {
      "epoch": 0.32099670208867714,
      "grad_norm": 0.419896426796065,
      "learning_rate": 9.998574733951775e-06,
      "loss": 0.4966,
      "step": 876
    },
    {
      "epoch": 0.3213631366801026,
      "grad_norm": 0.3639556773501901,
      "learning_rate": 9.998523379635527e-06,
      "loss": 0.492,
      "step": 877
    },
    {
      "epoch": 0.321729571271528,
      "grad_norm": 0.362497942951087,
      "learning_rate": 9.998471116573866e-06,
      "loss": 0.4755,
      "step": 878
    },
    {
      "epoch": 0.32209600586295345,
      "grad_norm": 0.35410598624293277,
      "learning_rate": 9.998417944776293e-06,
      "loss": 0.4803,
      "step": 879
    },
    {
      "epoch": 0.3224624404543789,
      "grad_norm": 0.34927140825113623,
      "learning_rate": 9.998363864252474e-06,
      "loss": 0.4789,
      "step": 880
    },
    {
      "epoch": 0.3228288750458043,
      "grad_norm": 0.376132795688513,
      "learning_rate": 9.998308875012242e-06,
      "loss": 0.5216,
      "step": 881
    },
    {
      "epoch": 0.32319530963722976,
      "grad_norm": 0.3308872228475439,
      "learning_rate": 9.998252977065594e-06,
      "loss": 0.4722,
      "step": 882
    },
    {
      "epoch": 0.3235617442286552,
      "grad_norm": 0.3914098934074977,
      "learning_rate": 9.99819617042269e-06,
      "loss": 0.4616,
      "step": 883
    },
    {
      "epoch": 0.3239281788200806,
      "grad_norm": 0.34365636457953597,
      "learning_rate": 9.99813845509386e-06,
      "loss": 0.4635,
      "step": 884
    },
    {
      "epoch": 0.32429461341150606,
      "grad_norm": 0.3749166195011908,
      "learning_rate": 9.998079831089598e-06,
      "loss": 0.4978,
      "step": 885
    },
    {
      "epoch": 0.3246610480029315,
      "grad_norm": 0.3356594240837966,
      "learning_rate": 9.998020298420559e-06,
      "loss": 0.4779,
      "step": 886
    },
    {
      "epoch": 0.32502748259435693,
      "grad_norm": 0.32157946505474994,
      "learning_rate": 9.997959857097567e-06,
      "loss": 0.4839,
      "step": 887
    },
    {
      "epoch": 0.3253939171857823,
      "grad_norm": 0.3461223345621298,
      "learning_rate": 9.997898507131611e-06,
      "loss": 0.4914,
      "step": 888
    },
    {
      "epoch": 0.32576035177720775,
      "grad_norm": 0.3759742146220676,
      "learning_rate": 9.997836248533845e-06,
      "loss": 0.4834,
      "step": 889
    },
    {
      "epoch": 0.3261267863686332,
      "grad_norm": 0.3403733516121136,
      "learning_rate": 9.997773081315587e-06,
      "loss": 0.4785,
      "step": 890
    },
    {
      "epoch": 0.3264932209600586,
      "grad_norm": 0.35435279168904416,
      "learning_rate": 9.997709005488322e-06,
      "loss": 0.4811,
      "step": 891
    },
    {
      "epoch": 0.32685965555148405,
      "grad_norm": 0.3825865498531042,
      "learning_rate": 9.997644021063698e-06,
      "loss": 0.5002,
      "step": 892
    },
    {
      "epoch": 0.3272260901429095,
      "grad_norm": 0.3409848940984805,
      "learning_rate": 9.997578128053528e-06,
      "loss": 0.4895,
      "step": 893
    },
    {
      "epoch": 0.3275925247343349,
      "grad_norm": 0.34214813462697935,
      "learning_rate": 9.997511326469797e-06,
      "loss": 0.4789,
      "step": 894
    },
    {
      "epoch": 0.32795895932576036,
      "grad_norm": 0.3842137698746825,
      "learning_rate": 9.997443616324645e-06,
      "loss": 0.4998,
      "step": 895
    },
    {
      "epoch": 0.3283253939171858,
      "grad_norm": 0.36877174216288766,
      "learning_rate": 9.997374997630381e-06,
      "loss": 0.4984,
      "step": 896
    },
    {
      "epoch": 0.32869182850861123,
      "grad_norm": 0.35120190414594427,
      "learning_rate": 9.997305470399483e-06,
      "loss": 0.5079,
      "step": 897
    },
    {
      "epoch": 0.32905826310003666,
      "grad_norm": 0.36094850842743403,
      "learning_rate": 9.99723503464459e-06,
      "loss": 0.4733,
      "step": 898
    },
    {
      "epoch": 0.3294246976914621,
      "grad_norm": 0.324887253937852,
      "learning_rate": 9.997163690378509e-06,
      "loss": 0.4609,
      "step": 899
    },
    {
      "epoch": 0.3297911322828875,
      "grad_norm": 0.3322606525530979,
      "learning_rate": 9.997091437614209e-06,
      "loss": 0.4737,
      "step": 900
    },
    {
      "epoch": 0.3301575668743129,
      "grad_norm": 0.35247156521090733,
      "learning_rate": 9.997018276364825e-06,
      "loss": 0.4674,
      "step": 901
    },
    {
      "epoch": 0.33052400146573835,
      "grad_norm": 1.3455019617141495,
      "learning_rate": 9.99694420664366e-06,
      "loss": 0.4748,
      "step": 902
    },
    {
      "epoch": 0.3308904360571638,
      "grad_norm": 0.3853801818222334,
      "learning_rate": 9.996869228464177e-06,
      "loss": 0.481,
      "step": 903
    },
    {
      "epoch": 0.3312568706485892,
      "grad_norm": 0.34434447042209737,
      "learning_rate": 9.99679334184001e-06,
      "loss": 0.4736,
      "step": 904
    },
    {
      "epoch": 0.33162330524001465,
      "grad_norm": 0.3673734429041379,
      "learning_rate": 9.996716546784956e-06,
      "loss": 0.4857,
      "step": 905
    },
    {
      "epoch": 0.3319897398314401,
      "grad_norm": 0.3667439218200377,
      "learning_rate": 9.996638843312974e-06,
      "loss": 0.4978,
      "step": 906
    },
    {
      "epoch": 0.3323561744228655,
      "grad_norm": 0.3681589766858387,
      "learning_rate": 9.996560231438193e-06,
      "loss": 0.4766,
      "step": 907
    },
    {
      "epoch": 0.33272260901429096,
      "grad_norm": 0.3438440418298864,
      "learning_rate": 9.996480711174904e-06,
      "loss": 0.5024,
      "step": 908
    },
    {
      "epoch": 0.3330890436057164,
      "grad_norm": 0.4069146877074087,
      "learning_rate": 9.996400282537562e-06,
      "loss": 0.5105,
      "step": 909
    },
    {
      "epoch": 0.33345547819714183,
      "grad_norm": 0.3507839851392909,
      "learning_rate": 9.996318945540792e-06,
      "loss": 0.4847,
      "step": 910
    },
    {
      "epoch": 0.33382191278856727,
      "grad_norm": 0.3611698377283544,
      "learning_rate": 9.99623670019938e-06,
      "loss": 0.4943,
      "step": 911
    },
    {
      "epoch": 0.33418834737999265,
      "grad_norm": 0.35450225996326323,
      "learning_rate": 9.996153546528277e-06,
      "loss": 0.4787,
      "step": 912
    },
    {
      "epoch": 0.3345547819714181,
      "grad_norm": 0.368356607924399,
      "learning_rate": 9.996069484542605e-06,
      "loss": 0.4794,
      "step": 913
    },
    {
      "epoch": 0.3349212165628435,
      "grad_norm": 0.33914084504444886,
      "learning_rate": 9.995984514257641e-06,
      "loss": 0.48,
      "step": 914
    },
    {
      "epoch": 0.33528765115426895,
      "grad_norm": 0.35190457498324956,
      "learning_rate": 9.995898635688838e-06,
      "loss": 0.4691,
      "step": 915
    },
    {
      "epoch": 0.3356540857456944,
      "grad_norm": 0.38114424881977627,
      "learning_rate": 9.995811848851807e-06,
      "loss": 0.4761,
      "step": 916
    },
    {
      "epoch": 0.3360205203371198,
      "grad_norm": 0.35434544761509706,
      "learning_rate": 9.995724153762325e-06,
      "loss": 0.4893,
      "step": 917
    },
    {
      "epoch": 0.33638695492854526,
      "grad_norm": 0.3828004074476827,
      "learning_rate": 9.995635550436336e-06,
      "loss": 0.4555,
      "step": 918
    },
    {
      "epoch": 0.3367533895199707,
      "grad_norm": 0.3520654255967265,
      "learning_rate": 9.995546038889948e-06,
      "loss": 0.5059,
      "step": 919
    },
    {
      "epoch": 0.3371198241113961,
      "grad_norm": 0.33601258562198283,
      "learning_rate": 9.995455619139435e-06,
      "loss": 0.4727,
      "step": 920
    },
    {
      "epoch": 0.33748625870282156,
      "grad_norm": 0.3784681734772082,
      "learning_rate": 9.995364291201236e-06,
      "loss": 0.4751,
      "step": 921
    },
    {
      "epoch": 0.337852693294247,
      "grad_norm": 0.3802149599784418,
      "learning_rate": 9.995272055091954e-06,
      "loss": 0.4898,
      "step": 922
    },
    {
      "epoch": 0.33821912788567243,
      "grad_norm": 0.3431137527200948,
      "learning_rate": 9.995178910828355e-06,
      "loss": 0.4669,
      "step": 923
    },
    {
      "epoch": 0.3385855624770978,
      "grad_norm": 0.38179228114556124,
      "learning_rate": 9.99508485842738e-06,
      "loss": 0.5057,
      "step": 924
    },
    {
      "epoch": 0.33895199706852325,
      "grad_norm": 0.37005687533561177,
      "learning_rate": 9.99498989790612e-06,
      "loss": 0.4759,
      "step": 925
    },
    {
      "epoch": 0.3393184316599487,
      "grad_norm": 0.35444211046243895,
      "learning_rate": 9.994894029281846e-06,
      "loss": 0.5055,
      "step": 926
    },
    {
      "epoch": 0.3396848662513741,
      "grad_norm": 0.3464329126697241,
      "learning_rate": 9.994797252571982e-06,
      "loss": 0.4711,
      "step": 927
    },
    {
      "epoch": 0.34005130084279955,
      "grad_norm": 0.3820348992955082,
      "learning_rate": 9.994699567794123e-06,
      "loss": 0.4694,
      "step": 928
    },
    {
      "epoch": 0.340417735434225,
      "grad_norm": 0.44604439155219,
      "learning_rate": 9.994600974966033e-06,
      "loss": 0.5175,
      "step": 929
    },
    {
      "epoch": 0.3407841700256504,
      "grad_norm": 0.3593560272791784,
      "learning_rate": 9.99450147410563e-06,
      "loss": 0.486,
      "step": 930
    },
    {
      "epoch": 0.34115060461707586,
      "grad_norm": 0.3354006609076205,
      "learning_rate": 9.994401065231008e-06,
      "loss": 0.4941,
      "step": 931
    },
    {
      "epoch": 0.3415170392085013,
      "grad_norm": 0.3679014998368014,
      "learning_rate": 9.994299748360418e-06,
      "loss": 0.4923,
      "step": 932
    },
    {
      "epoch": 0.34188347379992673,
      "grad_norm": 0.3877385107529982,
      "learning_rate": 9.994197523512286e-06,
      "loss": 0.5006,
      "step": 933
    },
    {
      "epoch": 0.34224990839135216,
      "grad_norm": 0.39911399129535463,
      "learning_rate": 9.994094390705189e-06,
      "loss": 0.5002,
      "step": 934
    },
    {
      "epoch": 0.3426163429827776,
      "grad_norm": 0.3629146727793879,
      "learning_rate": 9.993990349957883e-06,
      "loss": 0.4879,
      "step": 935
    },
    {
      "epoch": 0.342982777574203,
      "grad_norm": 0.3734861195383574,
      "learning_rate": 9.993885401289279e-06,
      "loss": 0.4833,
      "step": 936
    },
    {
      "epoch": 0.3433492121656284,
      "grad_norm": 0.40500449434177827,
      "learning_rate": 9.993779544718459e-06,
      "loss": 0.5167,
      "step": 937
    },
    {
      "epoch": 0.34371564675705385,
      "grad_norm": 0.3725428191162605,
      "learning_rate": 9.993672780264664e-06,
      "loss": 0.4935,
      "step": 938
    },
    {
      "epoch": 0.3440820813484793,
      "grad_norm": 0.35993882953501133,
      "learning_rate": 9.993565107947312e-06,
      "loss": 0.4854,
      "step": 939
    },
    {
      "epoch": 0.3444485159399047,
      "grad_norm": 0.31307294115466544,
      "learning_rate": 9.99345652778597e-06,
      "loss": 0.4539,
      "step": 940
    },
    {
      "epoch": 0.34481495053133016,
      "grad_norm": 0.3687207192019942,
      "learning_rate": 9.993347039800384e-06,
      "loss": 0.4835,
      "step": 941
    },
    {
      "epoch": 0.3451813851227556,
      "grad_norm": 0.3687348983124402,
      "learning_rate": 9.993236644010456e-06,
      "loss": 0.4788,
      "step": 942
    },
    {
      "epoch": 0.345547819714181,
      "grad_norm": 0.3752370972789874,
      "learning_rate": 9.993125340436258e-06,
      "loss": 0.4982,
      "step": 943
    },
    {
      "epoch": 0.34591425430560646,
      "grad_norm": 0.34854747798005864,
      "learning_rate": 9.993013129098024e-06,
      "loss": 0.4831,
      "step": 944
    },
    {
      "epoch": 0.3462806888970319,
      "grad_norm": 0.34168567861593807,
      "learning_rate": 9.992900010016155e-06,
      "loss": 0.4496,
      "step": 945
    },
    {
      "epoch": 0.34664712348845733,
      "grad_norm": 0.3593814471838252,
      "learning_rate": 9.992785983211214e-06,
      "loss": 0.4954,
      "step": 946
    },
    {
      "epoch": 0.34701355807988277,
      "grad_norm": 0.4288712028070973,
      "learning_rate": 9.992671048703936e-06,
      "loss": 0.4799,
      "step": 947
    },
    {
      "epoch": 0.34737999267130815,
      "grad_norm": 0.35999246038889054,
      "learning_rate": 9.992555206515213e-06,
      "loss": 0.4809,
      "step": 948
    },
    {
      "epoch": 0.3477464272627336,
      "grad_norm": 0.4316733518906925,
      "learning_rate": 9.992438456666108e-06,
      "loss": 0.4979,
      "step": 949
    },
    {
      "epoch": 0.348112861854159,
      "grad_norm": 0.3871260704983962,
      "learning_rate": 9.992320799177845e-06,
      "loss": 0.4832,
      "step": 950
    },
    {
      "epoch": 0.34847929644558445,
      "grad_norm": 0.32123335321102814,
      "learning_rate": 9.99220223407181e-06,
      "loss": 0.4839,
      "step": 951
    },
    {
      "epoch": 0.3488457310370099,
      "grad_norm": 0.3667405137149697,
      "learning_rate": 9.992082761369567e-06,
      "loss": 0.4804,
      "step": 952
    },
    {
      "epoch": 0.3492121656284353,
      "grad_norm": 0.3987619181263153,
      "learning_rate": 9.99196238109283e-06,
      "loss": 0.485,
      "step": 953
    },
    {
      "epoch": 0.34957860021986076,
      "grad_norm": 0.3534621117826914,
      "learning_rate": 9.991841093263488e-06,
      "loss": 0.4779,
      "step": 954
    },
    {
      "epoch": 0.3499450348112862,
      "grad_norm": 0.4070405311765395,
      "learning_rate": 9.99171889790359e-06,
      "loss": 0.4924,
      "step": 955
    },
    {
      "epoch": 0.35031146940271163,
      "grad_norm": 0.3486256168716824,
      "learning_rate": 9.991595795035352e-06,
      "loss": 0.4721,
      "step": 956
    },
    {
      "epoch": 0.35067790399413706,
      "grad_norm": 0.3680480335636397,
      "learning_rate": 9.991471784681153e-06,
      "loss": 0.4702,
      "step": 957
    },
    {
      "epoch": 0.3510443385855625,
      "grad_norm": 0.3510851100579493,
      "learning_rate": 9.99134686686354e-06,
      "loss": 0.478,
      "step": 958
    },
    {
      "epoch": 0.35141077317698793,
      "grad_norm": 0.393625715409865,
      "learning_rate": 9.991221041605224e-06,
      "loss": 0.5036,
      "step": 959
    },
    {
      "epoch": 0.3517772077684133,
      "grad_norm": 0.38036470207380113,
      "learning_rate": 9.991094308929077e-06,
      "loss": 0.4819,
      "step": 960
    },
    {
      "epoch": 0.35214364235983875,
      "grad_norm": 0.36485566551104,
      "learning_rate": 9.990966668858144e-06,
      "loss": 0.4998,
      "step": 961
    },
    {
      "epoch": 0.3525100769512642,
      "grad_norm": 0.4093034698049694,
      "learning_rate": 9.990838121415625e-06,
      "loss": 0.4877,
      "step": 962
    },
    {
      "epoch": 0.3528765115426896,
      "grad_norm": 0.504190530638233,
      "learning_rate": 9.990708666624895e-06,
      "loss": 0.483,
      "step": 963
    },
    {
      "epoch": 0.35324294613411505,
      "grad_norm": 0.4409161225317926,
      "learning_rate": 9.990578304509488e-06,
      "loss": 0.4961,
      "step": 964
    },
    {
      "epoch": 0.3536093807255405,
      "grad_norm": 0.4010009415313095,
      "learning_rate": 9.990447035093102e-06,
      "loss": 0.5165,
      "step": 965
    },
    {
      "epoch": 0.3539758153169659,
      "grad_norm": 0.39439260398387194,
      "learning_rate": 9.990314858399606e-06,
      "loss": 0.4916,
      "step": 966
    },
    {
      "epoch": 0.35434224990839136,
      "grad_norm": 0.34504116666633144,
      "learning_rate": 9.990181774453028e-06,
      "loss": 0.4696,
      "step": 967
    },
    {
      "epoch": 0.3547086844998168,
      "grad_norm": 0.36253960271934066,
      "learning_rate": 9.990047783277561e-06,
      "loss": 0.4911,
      "step": 968
    },
    {
      "epoch": 0.35507511909124223,
      "grad_norm": 0.34633988848873,
      "learning_rate": 9.989912884897568e-06,
      "loss": 0.4815,
      "step": 969
    },
    {
      "epoch": 0.35544155368266767,
      "grad_norm": 0.3506414598472753,
      "learning_rate": 9.989777079337572e-06,
      "loss": 0.4973,
      "step": 970
    },
    {
      "epoch": 0.3558079882740931,
      "grad_norm": 0.365926872395913,
      "learning_rate": 9.989640366622264e-06,
      "loss": 0.5045,
      "step": 971
    },
    {
      "epoch": 0.3561744228655185,
      "grad_norm": 0.3345001122585312,
      "learning_rate": 9.989502746776498e-06,
      "loss": 0.4676,
      "step": 972
    },
    {
      "epoch": 0.3565408574569439,
      "grad_norm": 0.3697013190452595,
      "learning_rate": 9.989364219825295e-06,
      "loss": 0.4972,
      "step": 973
    },
    {
      "epoch": 0.35690729204836935,
      "grad_norm": 0.32678158968878823,
      "learning_rate": 9.989224785793839e-06,
      "loss": 0.4872,
      "step": 974
    },
    {
      "epoch": 0.3572737266397948,
      "grad_norm": 0.3473479992615014,
      "learning_rate": 9.989084444707478e-06,
      "loss": 0.4892,
      "step": 975
    },
    {
      "epoch": 0.3576401612312202,
      "grad_norm": 0.39343444003885575,
      "learning_rate": 9.988943196591727e-06,
      "loss": 0.5009,
      "step": 976
    },
    {
      "epoch": 0.35800659582264566,
      "grad_norm": 0.3711730732644883,
      "learning_rate": 9.988801041472267e-06,
      "loss": 0.4826,
      "step": 977
    },
    {
      "epoch": 0.3583730304140711,
      "grad_norm": 0.3575021712642708,
      "learning_rate": 9.98865797937494e-06,
      "loss": 0.5113,
      "step": 978
    },
    {
      "epoch": 0.3587394650054965,
      "grad_norm": 0.3690560137982824,
      "learning_rate": 9.988514010325758e-06,
      "loss": 0.4734,
      "step": 979
    },
    {
      "epoch": 0.35910589959692196,
      "grad_norm": 0.3623031764413739,
      "learning_rate": 9.988369134350892e-06,
      "loss": 0.4618,
      "step": 980
    },
    {
      "epoch": 0.3594723341883474,
      "grad_norm": 0.3785376887382488,
      "learning_rate": 9.988223351476682e-06,
      "loss": 0.4734,
      "step": 981
    },
    {
      "epoch": 0.35983876877977283,
      "grad_norm": 0.38429886697783733,
      "learning_rate": 9.988076661729631e-06,
      "loss": 0.4869,
      "step": 982
    },
    {
      "epoch": 0.36020520337119827,
      "grad_norm": 0.3438129114586745,
      "learning_rate": 9.987929065136409e-06,
      "loss": 0.4888,
      "step": 983
    },
    {
      "epoch": 0.36057163796262365,
      "grad_norm": 0.41119983694477413,
      "learning_rate": 9.987780561723848e-06,
      "loss": 0.5032,
      "step": 984
    },
    {
      "epoch": 0.3609380725540491,
      "grad_norm": 0.40971534380347785,
      "learning_rate": 9.987631151518948e-06,
      "loss": 0.4825,
      "step": 985
    },
    {
      "epoch": 0.3613045071454745,
      "grad_norm": 0.33498425990963693,
      "learning_rate": 9.98748083454887e-06,
      "loss": 0.4836,
      "step": 986
    },
    {
      "epoch": 0.36167094173689995,
      "grad_norm": 0.36483006939732265,
      "learning_rate": 9.987329610840945e-06,
      "loss": 0.4853,
      "step": 987
    },
    {
      "epoch": 0.3620373763283254,
      "grad_norm": 0.44821305651972504,
      "learning_rate": 9.987177480422663e-06,
      "loss": 0.5082,
      "step": 988
    },
    {
      "epoch": 0.3624038109197508,
      "grad_norm": 0.40748429622565774,
      "learning_rate": 9.987024443321683e-06,
      "loss": 0.4854,
      "step": 989
    },
    {
      "epoch": 0.36277024551117626,
      "grad_norm": 0.8456231525664435,
      "learning_rate": 9.986870499565828e-06,
      "loss": 0.488,
      "step": 990
    },
    {
      "epoch": 0.3631366801026017,
      "grad_norm": 0.3246586779538845,
      "learning_rate": 9.986715649183084e-06,
      "loss": 0.4837,
      "step": 991
    },
    {
      "epoch": 0.36350311469402713,
      "grad_norm": 0.43812600406690017,
      "learning_rate": 9.986559892201606e-06,
      "loss": 0.4824,
      "step": 992
    },
    {
      "epoch": 0.36386954928545256,
      "grad_norm": 0.4216783391143744,
      "learning_rate": 9.986403228649707e-06,
      "loss": 0.4732,
      "step": 993
    },
    {
      "epoch": 0.364235983876878,
      "grad_norm": 0.33706852941562504,
      "learning_rate": 9.986245658555873e-06,
      "loss": 0.4656,
      "step": 994
    },
    {
      "epoch": 0.36460241846830344,
      "grad_norm": 0.37531451015966455,
      "learning_rate": 9.986087181948747e-06,
      "loss": 0.4884,
      "step": 995
    },
    {
      "epoch": 0.3649688530597288,
      "grad_norm": 0.3891886774940834,
      "learning_rate": 9.985927798857143e-06,
      "loss": 0.48,
      "step": 996
    },
    {
      "epoch": 0.36533528765115425,
      "grad_norm": 0.4053683138176225,
      "learning_rate": 9.985767509310035e-06,
      "loss": 0.4851,
      "step": 997
    },
    {
      "epoch": 0.3657017222425797,
      "grad_norm": 0.38328754036865603,
      "learning_rate": 9.985606313336567e-06,
      "loss": 0.4893,
      "step": 998
    },
    {
      "epoch": 0.3660681568340051,
      "grad_norm": 0.41172194490945674,
      "learning_rate": 9.985444210966044e-06,
      "loss": 0.5016,
      "step": 999
    },
    {
      "epoch": 0.36643459142543056,
      "grad_norm": 0.3360481810912452,
      "learning_rate": 9.985281202227936e-06,
      "loss": 0.4845,
      "step": 1000
    },
    {
      "epoch": 0.366801026016856,
      "grad_norm": 0.40582193304254915,
      "learning_rate": 9.985117287151877e-06,
      "loss": 0.4937,
      "step": 1001
    },
    {
      "epoch": 0.3671674606082814,
      "grad_norm": 0.436869882694962,
      "learning_rate": 9.98495246576767e-06,
      "loss": 0.5108,
      "step": 1002
    },
    {
      "epoch": 0.36753389519970686,
      "grad_norm": 0.3527754172261354,
      "learning_rate": 9.984786738105279e-06,
      "loss": 0.4885,
      "step": 1003
    },
    {
      "epoch": 0.3679003297911323,
      "grad_norm": 0.401622101088976,
      "learning_rate": 9.984620104194833e-06,
      "loss": 0.4856,
      "step": 1004
    },
    {
      "epoch": 0.36826676438255773,
      "grad_norm": 0.37594567089067965,
      "learning_rate": 9.984452564066627e-06,
      "loss": 0.4903,
      "step": 1005
    },
    {
      "epoch": 0.36863319897398317,
      "grad_norm": 0.36477652105001845,
      "learning_rate": 9.98428411775112e-06,
      "loss": 0.4851,
      "step": 1006
    },
    {
      "epoch": 0.36899963356540855,
      "grad_norm": 0.5147064689688183,
      "learning_rate": 9.984114765278938e-06,
      "loss": 0.4714,
      "step": 1007
    },
    {
      "epoch": 0.369366068156834,
      "grad_norm": 0.34595168327038955,
      "learning_rate": 9.983944506680867e-06,
      "loss": 0.4744,
      "step": 1008
    },
    {
      "epoch": 0.3697325027482594,
      "grad_norm": 0.35457167252881333,
      "learning_rate": 9.98377334198786e-06,
      "loss": 0.4993,
      "step": 1009
    },
    {
      "epoch": 0.37009893733968485,
      "grad_norm": 0.3435647774014055,
      "learning_rate": 9.983601271231039e-06,
      "loss": 0.5065,
      "step": 1010
    },
    {
      "epoch": 0.3704653719311103,
      "grad_norm": 0.37799259707959315,
      "learning_rate": 9.983428294441683e-06,
      "loss": 0.5021,
      "step": 1011
    },
    {
      "epoch": 0.3708318065225357,
      "grad_norm": 0.40199138744690077,
      "learning_rate": 9.983254411651242e-06,
      "loss": 0.4936,
      "step": 1012
    },
    {
      "epoch": 0.37119824111396116,
      "grad_norm": 0.41383589786742536,
      "learning_rate": 9.983079622891328e-06,
      "loss": 0.4852,
      "step": 1013
    },
    {
      "epoch": 0.3715646757053866,
      "grad_norm": 0.3080316747826409,
      "learning_rate": 9.982903928193717e-06,
      "loss": 0.4807,
      "step": 1014
    },
    {
      "epoch": 0.37193111029681203,
      "grad_norm": 0.47209962663893446,
      "learning_rate": 9.982727327590352e-06,
      "loss": 0.4765,
      "step": 1015
    },
    {
      "epoch": 0.37229754488823746,
      "grad_norm": 0.35280924766600585,
      "learning_rate": 9.982549821113338e-06,
      "loss": 0.4713,
      "step": 1016
    },
    {
      "epoch": 0.3726639794796629,
      "grad_norm": 0.3720604624657516,
      "learning_rate": 9.982371408794948e-06,
      "loss": 0.4655,
      "step": 1017
    },
    {
      "epoch": 0.37303041407108833,
      "grad_norm": 0.3474818806809959,
      "learning_rate": 9.982192090667618e-06,
      "loss": 0.4624,
      "step": 1018
    },
    {
      "epoch": 0.3733968486625137,
      "grad_norm": 0.4132741016341644,
      "learning_rate": 9.982011866763948e-06,
      "loss": 0.4962,
      "step": 1019
    },
    {
      "epoch": 0.37376328325393915,
      "grad_norm": 0.40564311452195156,
      "learning_rate": 9.9818307371167e-06,
      "loss": 0.4679,
      "step": 1020
    },
    {
      "epoch": 0.3741297178453646,
      "grad_norm": 0.3658883799435955,
      "learning_rate": 9.98164870175881e-06,
      "loss": 0.4986,
      "step": 1021
    },
    {
      "epoch": 0.37449615243679,
      "grad_norm": 0.408075461405066,
      "learning_rate": 9.981465760723368e-06,
      "loss": 0.4945,
      "step": 1022
    },
    {
      "epoch": 0.37486258702821545,
      "grad_norm": 0.3825136766601713,
      "learning_rate": 9.981281914043633e-06,
      "loss": 0.5,
      "step": 1023
    },
    {
      "epoch": 0.3752290216196409,
      "grad_norm": 0.3698835321423304,
      "learning_rate": 9.981097161753032e-06,
      "loss": 0.4844,
      "step": 1024
    },
    {
      "epoch": 0.3755954562110663,
      "grad_norm": 0.3134774028717853,
      "learning_rate": 9.980911503885151e-06,
      "loss": 0.4564,
      "step": 1025
    },
    {
      "epoch": 0.37596189080249176,
      "grad_norm": 0.34683735696491447,
      "learning_rate": 9.980724940473742e-06,
      "loss": 0.4887,
      "step": 1026
    },
    {
      "epoch": 0.3763283253939172,
      "grad_norm": 0.42631036783965415,
      "learning_rate": 9.980537471552728e-06,
      "loss": 0.4974,
      "step": 1027
    },
    {
      "epoch": 0.37669475998534263,
      "grad_norm": 0.4027042250239545,
      "learning_rate": 9.980349097156186e-06,
      "loss": 0.4606,
      "step": 1028
    },
    {
      "epoch": 0.37706119457676807,
      "grad_norm": 0.35571456551357533,
      "learning_rate": 9.980159817318365e-06,
      "loss": 0.4863,
      "step": 1029
    },
    {
      "epoch": 0.3774276291681935,
      "grad_norm": 0.441147610166631,
      "learning_rate": 9.979969632073678e-06,
      "loss": 0.5038,
      "step": 1030
    },
    {
      "epoch": 0.3777940637596189,
      "grad_norm": 0.37423926134828145,
      "learning_rate": 9.979778541456698e-06,
      "loss": 0.5044,
      "step": 1031
    },
    {
      "epoch": 0.3781604983510443,
      "grad_norm": 0.3780582341537142,
      "learning_rate": 9.979586545502168e-06,
      "loss": 0.5034,
      "step": 1032
    },
    {
      "epoch": 0.37852693294246975,
      "grad_norm": 0.394301491291551,
      "learning_rate": 9.979393644244992e-06,
      "loss": 0.4678,
      "step": 1033
    },
    {
      "epoch": 0.3788933675338952,
      "grad_norm": 0.4076314211223816,
      "learning_rate": 9.979199837720243e-06,
      "loss": 0.4625,
      "step": 1034
    },
    {
      "epoch": 0.3792598021253206,
      "grad_norm": 0.39352686575806894,
      "learning_rate": 9.979005125963153e-06,
      "loss": 0.4723,
      "step": 1035
    },
    {
      "epoch": 0.37962623671674606,
      "grad_norm": 0.3833557840721204,
      "learning_rate": 9.978809509009121e-06,
      "loss": 0.4919,
      "step": 1036
    },
    {
      "epoch": 0.3799926713081715,
      "grad_norm": 0.4088270109554436,
      "learning_rate": 9.978612986893712e-06,
      "loss": 0.4911,
      "step": 1037
    },
    {
      "epoch": 0.3803591058995969,
      "grad_norm": 0.4021340575776169,
      "learning_rate": 9.978415559652653e-06,
      "loss": 0.4765,
      "step": 1038
    },
    {
      "epoch": 0.38072554049102236,
      "grad_norm": 0.39616574333194543,
      "learning_rate": 9.978217227321837e-06,
      "loss": 0.4989,
      "step": 1039
    },
    {
      "epoch": 0.3810919750824478,
      "grad_norm": 0.5254572260517771,
      "learning_rate": 9.978017989937322e-06,
      "loss": 0.4954,
      "step": 1040
    },
    {
      "epoch": 0.38145840967387323,
      "grad_norm": 0.4465285621850101,
      "learning_rate": 9.977817847535329e-06,
      "loss": 0.4908,
      "step": 1041
    },
    {
      "epoch": 0.38182484426529867,
      "grad_norm": 0.4602323458838,
      "learning_rate": 9.977616800152248e-06,
      "loss": 0.4904,
      "step": 1042
    },
    {
      "epoch": 0.38219127885672405,
      "grad_norm": 1.645891342730488,
      "learning_rate": 9.977414847824625e-06,
      "loss": 0.4796,
      "step": 1043
    },
    {
      "epoch": 0.3825577134481495,
      "grad_norm": 0.39527049933206004,
      "learning_rate": 9.977211990589177e-06,
      "loss": 0.504,
      "step": 1044
    },
    {
      "epoch": 0.3829241480395749,
      "grad_norm": 0.3996282041469855,
      "learning_rate": 9.977008228482785e-06,
      "loss": 0.4616,
      "step": 1045
    },
    {
      "epoch": 0.38329058263100035,
      "grad_norm": 0.32653200264898874,
      "learning_rate": 9.976803561542494e-06,
      "loss": 0.4708,
      "step": 1046
    },
    {
      "epoch": 0.3836570172224258,
      "grad_norm": 0.3395662182163055,
      "learning_rate": 9.976597989805513e-06,
      "loss": 0.4648,
      "step": 1047
    },
    {
      "epoch": 0.3840234518138512,
      "grad_norm": 0.4088411913612648,
      "learning_rate": 9.976391513309212e-06,
      "loss": 0.4819,
      "step": 1048
    },
    {
      "epoch": 0.38438988640527666,
      "grad_norm": 0.30482004777383553,
      "learning_rate": 9.976184132091133e-06,
      "loss": 0.4721,
      "step": 1049
    },
    {
      "epoch": 0.3847563209967021,
      "grad_norm": 0.39170724217796427,
      "learning_rate": 9.975975846188976e-06,
      "loss": 0.4826,
      "step": 1050
    },
    {
      "epoch": 0.38512275558812753,
      "grad_norm": 0.36197178407812297,
      "learning_rate": 9.97576665564061e-06,
      "loss": 0.5028,
      "step": 1051
    },
    {
      "epoch": 0.38548919017955297,
      "grad_norm": 0.3548433331324174,
      "learning_rate": 9.975556560484067e-06,
      "loss": 0.4662,
      "step": 1052
    },
    {
      "epoch": 0.3858556247709784,
      "grad_norm": 0.37009406196861055,
      "learning_rate": 9.975345560757541e-06,
      "loss": 0.4598,
      "step": 1053
    },
    {
      "epoch": 0.38622205936240384,
      "grad_norm": 0.33719597456842726,
      "learning_rate": 9.975133656499392e-06,
      "loss": 0.4656,
      "step": 1054
    },
    {
      "epoch": 0.3865884939538292,
      "grad_norm": 0.37565856087790694,
      "learning_rate": 9.974920847748144e-06,
      "loss": 0.4662,
      "step": 1055
    },
    {
      "epoch": 0.38695492854525465,
      "grad_norm": 0.382811066384776,
      "learning_rate": 9.97470713454249e-06,
      "loss": 0.4818,
      "step": 1056
    },
    {
      "epoch": 0.3873213631366801,
      "grad_norm": 0.3611657439703506,
      "learning_rate": 9.97449251692128e-06,
      "loss": 0.4918,
      "step": 1057
    },
    {
      "epoch": 0.3876877977281055,
      "grad_norm": 0.36221723421861973,
      "learning_rate": 9.974276994923536e-06,
      "loss": 0.4792,
      "step": 1058
    },
    {
      "epoch": 0.38805423231953096,
      "grad_norm": 0.34054841915816636,
      "learning_rate": 9.974060568588436e-06,
      "loss": 0.478,
      "step": 1059
    },
    {
      "epoch": 0.3884206669109564,
      "grad_norm": 0.5714704377885451,
      "learning_rate": 9.973843237955328e-06,
      "loss": 0.4896,
      "step": 1060
    },
    {
      "epoch": 0.3887871015023818,
      "grad_norm": 0.3759172103800909,
      "learning_rate": 9.973625003063726e-06,
      "loss": 0.4623,
      "step": 1061
    },
    {
      "epoch": 0.38915353609380726,
      "grad_norm": 0.3789171521961233,
      "learning_rate": 9.9734058639533e-06,
      "loss": 0.4796,
      "step": 1062
    },
    {
      "epoch": 0.3895199706852327,
      "grad_norm": 0.35389425076215253,
      "learning_rate": 9.973185820663897e-06,
      "loss": 0.4611,
      "step": 1063
    },
    {
      "epoch": 0.38988640527665813,
      "grad_norm": 0.35366609318332143,
      "learning_rate": 9.97296487323552e-06,
      "loss": 0.4732,
      "step": 1064
    },
    {
      "epoch": 0.39025283986808357,
      "grad_norm": 0.3799934653794604,
      "learning_rate": 9.972743021708334e-06,
      "loss": 0.4769,
      "step": 1065
    },
    {
      "epoch": 0.390619274459509,
      "grad_norm": 0.3855830245728391,
      "learning_rate": 9.972520266122676e-06,
      "loss": 0.4919,
      "step": 1066
    },
    {
      "epoch": 0.3909857090509344,
      "grad_norm": 0.35672919770797595,
      "learning_rate": 9.972296606519041e-06,
      "loss": 0.4765,
      "step": 1067
    },
    {
      "epoch": 0.3913521436423598,
      "grad_norm": 0.36174806769199574,
      "learning_rate": 9.972072042938093e-06,
      "loss": 0.4573,
      "step": 1068
    },
    {
      "epoch": 0.39171857823378525,
      "grad_norm": 0.39919288258645724,
      "learning_rate": 9.971846575420656e-06,
      "loss": 0.4753,
      "step": 1069
    },
    {
      "epoch": 0.3920850128252107,
      "grad_norm": 0.40231279574678125,
      "learning_rate": 9.971620204007722e-06,
      "loss": 0.4956,
      "step": 1070
    },
    {
      "epoch": 0.3924514474166361,
      "grad_norm": 0.35698022521468203,
      "learning_rate": 9.971392928740448e-06,
      "loss": 0.4772,
      "step": 1071
    },
    {
      "epoch": 0.39281788200806156,
      "grad_norm": 0.34448888515208725,
      "learning_rate": 9.971164749660149e-06,
      "loss": 0.473,
      "step": 1072
    },
    {
      "epoch": 0.393184316599487,
      "grad_norm": 0.3495638562680776,
      "learning_rate": 9.970935666808313e-06,
      "loss": 0.476,
      "step": 1073
    },
    {
      "epoch": 0.39355075119091243,
      "grad_norm": 0.3519514994254813,
      "learning_rate": 9.970705680226583e-06,
      "loss": 0.4992,
      "step": 1074
    },
    {
      "epoch": 0.39391718578233786,
      "grad_norm": 0.34925624568549546,
      "learning_rate": 9.970474789956775e-06,
      "loss": 0.4899,
      "step": 1075
    },
    {
      "epoch": 0.3942836203737633,
      "grad_norm": 0.32780508212145393,
      "learning_rate": 9.970242996040865e-06,
      "loss": 0.4967,
      "step": 1076
    },
    {
      "epoch": 0.39465005496518873,
      "grad_norm": 0.3644446743670389,
      "learning_rate": 9.970010298520994e-06,
      "loss": 0.4807,
      "step": 1077
    },
    {
      "epoch": 0.39501648955661417,
      "grad_norm": 0.3445176962956108,
      "learning_rate": 9.969776697439463e-06,
      "loss": 0.473,
      "step": 1078
    },
    {
      "epoch": 0.39538292414803955,
      "grad_norm": 0.42522513135341444,
      "learning_rate": 9.969542192838748e-06,
      "loss": 0.4834,
      "step": 1079
    },
    {
      "epoch": 0.395749358739465,
      "grad_norm": 0.3059980747164909,
      "learning_rate": 9.969306784761476e-06,
      "loss": 0.496,
      "step": 1080
    },
    {
      "epoch": 0.3961157933308904,
      "grad_norm": 0.45334634614699637,
      "learning_rate": 9.96907047325045e-06,
      "loss": 0.4813,
      "step": 1081
    },
    {
      "epoch": 0.39648222792231586,
      "grad_norm": 0.3422777343722219,
      "learning_rate": 9.96883325834863e-06,
      "loss": 0.4828,
      "step": 1082
    },
    {
      "epoch": 0.3968486625137413,
      "grad_norm": 0.39947624643919044,
      "learning_rate": 9.968595140099142e-06,
      "loss": 0.4681,
      "step": 1083
    },
    {
      "epoch": 0.3972150971051667,
      "grad_norm": 0.337912600661978,
      "learning_rate": 9.968356118545277e-06,
      "loss": 0.4555,
      "step": 1084
    },
    {
      "epoch": 0.39758153169659216,
      "grad_norm": 0.38817561753742713,
      "learning_rate": 9.96811619373049e-06,
      "loss": 0.4514,
      "step": 1085
    },
    {
      "epoch": 0.3979479662880176,
      "grad_norm": 0.45831784516156837,
      "learning_rate": 9.9678753656984e-06,
      "loss": 0.4849,
      "step": 1086
    },
    {
      "epoch": 0.39831440087944303,
      "grad_norm": 0.3689841460252741,
      "learning_rate": 9.967633634492788e-06,
      "loss": 0.4875,
      "step": 1087
    },
    {
      "epoch": 0.39868083547086847,
      "grad_norm": 0.3541070222611498,
      "learning_rate": 9.967391000157605e-06,
      "loss": 0.4891,
      "step": 1088
    },
    {
      "epoch": 0.3990472700622939,
      "grad_norm": 0.35570780571298777,
      "learning_rate": 9.967147462736962e-06,
      "loss": 0.4709,
      "step": 1089
    },
    {
      "epoch": 0.39941370465371934,
      "grad_norm": 0.3469504191857469,
      "learning_rate": 9.966903022275131e-06,
      "loss": 0.4763,
      "step": 1090
    },
    {
      "epoch": 0.3997801392451447,
      "grad_norm": 0.37258958446758117,
      "learning_rate": 9.966657678816556e-06,
      "loss": 0.4668,
      "step": 1091
    },
    {
      "epoch": 0.40014657383657015,
      "grad_norm": 0.339973126668228,
      "learning_rate": 9.966411432405839e-06,
      "loss": 0.4812,
      "step": 1092
    },
    {
      "epoch": 0.4005130084279956,
      "grad_norm": 0.41028901322505856,
      "learning_rate": 9.96616428308775e-06,
      "loss": 0.4871,
      "step": 1093
    },
    {
      "epoch": 0.400879443019421,
      "grad_norm": 0.3723654312983103,
      "learning_rate": 9.96591623090722e-06,
      "loss": 0.5022,
      "step": 1094
    },
    {
      "epoch": 0.40124587761084646,
      "grad_norm": 0.4241639131276314,
      "learning_rate": 9.965667275909346e-06,
      "loss": 0.4983,
      "step": 1095
    },
    {
      "epoch": 0.4016123122022719,
      "grad_norm": 0.3472206684682854,
      "learning_rate": 9.96541741813939e-06,
      "loss": 0.465,
      "step": 1096
    },
    {
      "epoch": 0.40197874679369733,
      "grad_norm": 0.3361033844858255,
      "learning_rate": 9.965166657642773e-06,
      "loss": 0.4841,
      "step": 1097
    },
    {
      "epoch": 0.40234518138512276,
      "grad_norm": 0.3483194335756022,
      "learning_rate": 9.96491499446509e-06,
      "loss": 0.4864,
      "step": 1098
    },
    {
      "epoch": 0.4027116159765482,
      "grad_norm": 0.42898806391072597,
      "learning_rate": 9.964662428652088e-06,
      "loss": 0.4791,
      "step": 1099
    },
    {
      "epoch": 0.40307805056797363,
      "grad_norm": 0.48696842976425675,
      "learning_rate": 9.964408960249688e-06,
      "loss": 0.4711,
      "step": 1100
    },
    {
      "epoch": 0.40344448515939907,
      "grad_norm": 0.3380980027310082,
      "learning_rate": 9.964154589303969e-06,
      "loss": 0.4854,
      "step": 1101
    },
    {
      "epoch": 0.4038109197508245,
      "grad_norm": 0.40697425755428657,
      "learning_rate": 9.963899315861176e-06,
      "loss": 0.4874,
      "step": 1102
    },
    {
      "epoch": 0.4041773543422499,
      "grad_norm": 0.40250094402414405,
      "learning_rate": 9.963643139967722e-06,
      "loss": 0.4759,
      "step": 1103
    },
    {
      "epoch": 0.4045437889336753,
      "grad_norm": 0.38779182844877247,
      "learning_rate": 9.963386061670179e-06,
      "loss": 0.4946,
      "step": 1104
    },
    {
      "epoch": 0.40491022352510075,
      "grad_norm": 0.7891990101559702,
      "learning_rate": 9.963128081015282e-06,
      "loss": 0.4559,
      "step": 1105
    },
    {
      "epoch": 0.4052766581165262,
      "grad_norm": 0.3754368162939026,
      "learning_rate": 9.962869198049936e-06,
      "loss": 0.4928,
      "step": 1106
    },
    {
      "epoch": 0.4056430927079516,
      "grad_norm": 0.37114008314118874,
      "learning_rate": 9.962609412821205e-06,
      "loss": 0.4922,
      "step": 1107
    },
    {
      "epoch": 0.40600952729937706,
      "grad_norm": 0.36275778930597985,
      "learning_rate": 9.962348725376318e-06,
      "loss": 0.4777,
      "step": 1108
    },
    {
      "epoch": 0.4063759618908025,
      "grad_norm": 0.353617938537047,
      "learning_rate": 9.962087135762668e-06,
      "loss": 0.4734,
      "step": 1109
    },
    {
      "epoch": 0.40674239648222793,
      "grad_norm": 0.35198637699584967,
      "learning_rate": 9.961824644027816e-06,
      "loss": 0.4918,
      "step": 1110
    },
    {
      "epoch": 0.40710883107365337,
      "grad_norm": 0.4653823803519016,
      "learning_rate": 9.961561250219482e-06,
      "loss": 0.4966,
      "step": 1111
    },
    {
      "epoch": 0.4074752656650788,
      "grad_norm": 0.330015651967236,
      "learning_rate": 9.96129695438555e-06,
      "loss": 0.5011,
      "step": 1112
    },
    {
      "epoch": 0.40784170025650424,
      "grad_norm": 0.38994389788796036,
      "learning_rate": 9.961031756574074e-06,
      "loss": 0.498,
      "step": 1113
    },
    {
      "epoch": 0.40820813484792967,
      "grad_norm": 0.31573380814687385,
      "learning_rate": 9.960765656833263e-06,
      "loss": 0.4934,
      "step": 1114
    },
    {
      "epoch": 0.40857456943935505,
      "grad_norm": 0.44146336547192283,
      "learning_rate": 9.960498655211496e-06,
      "loss": 0.4878,
      "step": 1115
    },
    {
      "epoch": 0.4089410040307805,
      "grad_norm": 0.36426140728209433,
      "learning_rate": 9.960230751757318e-06,
      "loss": 0.4942,
      "step": 1116
    },
    {
      "epoch": 0.4093074386222059,
      "grad_norm": 0.3494118115133787,
      "learning_rate": 9.959961946519431e-06,
      "loss": 0.4791,
      "step": 1117
    },
    {
      "epoch": 0.40967387321363136,
      "grad_norm": 0.3431010275961274,
      "learning_rate": 9.959692239546705e-06,
      "loss": 0.4619,
      "step": 1118
    },
    {
      "epoch": 0.4100403078050568,
      "grad_norm": 0.3229448836278248,
      "learning_rate": 9.959421630888173e-06,
      "loss": 0.4596,
      "step": 1119
    },
    {
      "epoch": 0.4104067423964822,
      "grad_norm": 0.35525848698395673,
      "learning_rate": 9.959150120593035e-06,
      "loss": 0.4901,
      "step": 1120
    },
    {
      "epoch": 0.41077317698790766,
      "grad_norm": 0.3410630964419126,
      "learning_rate": 9.95887770871065e-06,
      "loss": 0.4647,
      "step": 1121
    },
    {
      "epoch": 0.4111396115793331,
      "grad_norm": 0.3207868010869608,
      "learning_rate": 9.958604395290543e-06,
      "loss": 0.475,
      "step": 1122
    },
    {
      "epoch": 0.41150604617075853,
      "grad_norm": 0.34293582851781906,
      "learning_rate": 9.958330180382405e-06,
      "loss": 0.4713,
      "step": 1123
    },
    {
      "epoch": 0.41187248076218397,
      "grad_norm": 0.3523541839328477,
      "learning_rate": 9.95805506403609e-06,
      "loss": 0.4862,
      "step": 1124
    },
    {
      "epoch": 0.4122389153536094,
      "grad_norm": 0.3213908199560734,
      "learning_rate": 9.95777904630161e-06,
      "loss": 0.4842,
      "step": 1125
    },
    {
      "epoch": 0.41260534994503484,
      "grad_norm": 0.32715448019729787,
      "learning_rate": 9.95750212722915e-06,
      "loss": 0.4748,
      "step": 1126
    },
    {
      "epoch": 0.4129717845364602,
      "grad_norm": 0.7135305410588745,
      "learning_rate": 9.957224306869053e-06,
      "loss": 0.486,
      "step": 1127
    },
    {
      "epoch": 0.41333821912788565,
      "grad_norm": 0.3643480628038832,
      "learning_rate": 9.956945585271829e-06,
      "loss": 0.4708,
      "step": 1128
    },
    {
      "epoch": 0.4137046537193111,
      "grad_norm": 0.3312808361593881,
      "learning_rate": 9.956665962488148e-06,
      "loss": 0.4762,
      "step": 1129
    },
    {
      "epoch": 0.4140710883107365,
      "grad_norm": 0.3468494190106635,
      "learning_rate": 9.956385438568848e-06,
      "loss": 0.4824,
      "step": 1130
    },
    {
      "epoch": 0.41443752290216196,
      "grad_norm": 0.354239744030121,
      "learning_rate": 9.956104013564928e-06,
      "loss": 0.4628,
      "step": 1131
    },
    {
      "epoch": 0.4148039574935874,
      "grad_norm": 0.34187546696242177,
      "learning_rate": 9.955821687527554e-06,
      "loss": 0.4781,
      "step": 1132
    },
    {
      "epoch": 0.41517039208501283,
      "grad_norm": 0.3930755908921338,
      "learning_rate": 9.95553846050805e-06,
      "loss": 0.4652,
      "step": 1133
    },
    {
      "epoch": 0.41553682667643826,
      "grad_norm": 0.3169142156652077,
      "learning_rate": 9.95525433255791e-06,
      "loss": 0.4816,
      "step": 1134
    },
    {
      "epoch": 0.4159032612678637,
      "grad_norm": 0.4050072544930659,
      "learning_rate": 9.95496930372879e-06,
      "loss": 0.4963,
      "step": 1135
    },
    {
      "epoch": 0.41626969585928913,
      "grad_norm": 0.39737237298408873,
      "learning_rate": 9.954683374072507e-06,
      "loss": 0.4774,
      "step": 1136
    },
    {
      "epoch": 0.41663613045071457,
      "grad_norm": 0.35337392110275034,
      "learning_rate": 9.954396543641044e-06,
      "loss": 0.4512,
      "step": 1137
    },
    {
      "epoch": 0.41700256504214,
      "grad_norm": 0.3627702062319581,
      "learning_rate": 9.95410881248655e-06,
      "loss": 0.4448,
      "step": 1138
    },
    {
      "epoch": 0.4173689996335654,
      "grad_norm": 0.4523537765127956,
      "learning_rate": 9.95382018066133e-06,
      "loss": 0.4883,
      "step": 1139
    },
    {
      "epoch": 0.4177354342249908,
      "grad_norm": 0.4051224203029738,
      "learning_rate": 9.953530648217864e-06,
      "loss": 0.4784,
      "step": 1140
    },
    {
      "epoch": 0.41810186881641626,
      "grad_norm": 0.3170054690408905,
      "learning_rate": 9.953240215208787e-06,
      "loss": 0.4713,
      "step": 1141
    },
    {
      "epoch": 0.4184683034078417,
      "grad_norm": 0.36297794379456705,
      "learning_rate": 9.952948881686902e-06,
      "loss": 0.4609,
      "step": 1142
    },
    {
      "epoch": 0.4188347379992671,
      "grad_norm": 0.380073637345379,
      "learning_rate": 9.952656647705171e-06,
      "loss": 0.4814,
      "step": 1143
    },
    {
      "epoch": 0.41920117259069256,
      "grad_norm": 0.38276474386512227,
      "learning_rate": 9.952363513316727e-06,
      "loss": 0.5023,
      "step": 1144
    },
    {
      "epoch": 0.419567607182118,
      "grad_norm": 0.3165309411272166,
      "learning_rate": 9.95206947857486e-06,
      "loss": 0.4932,
      "step": 1145
    },
    {
      "epoch": 0.41993404177354343,
      "grad_norm": 0.34708773402379095,
      "learning_rate": 9.951774543533027e-06,
      "loss": 0.4862,
      "step": 1146
    },
    {
      "epoch": 0.42030047636496887,
      "grad_norm": 0.3431218885266525,
      "learning_rate": 9.951478708244847e-06,
      "loss": 0.4514,
      "step": 1147
    },
    {
      "epoch": 0.4206669109563943,
      "grad_norm": 0.39357224160657256,
      "learning_rate": 9.951181972764106e-06,
      "loss": 0.4796,
      "step": 1148
    },
    {
      "epoch": 0.42103334554781974,
      "grad_norm": 0.3583405650770147,
      "learning_rate": 9.950884337144749e-06,
      "loss": 0.4567,
      "step": 1149
    },
    {
      "epoch": 0.4213997801392451,
      "grad_norm": 0.35972146277085854,
      "learning_rate": 9.950585801440889e-06,
      "loss": 0.4914,
      "step": 1150
    },
    {
      "epoch": 0.42176621473067055,
      "grad_norm": 0.33659884024441533,
      "learning_rate": 9.950286365706799e-06,
      "loss": 0.4581,
      "step": 1151
    },
    {
      "epoch": 0.422132649322096,
      "grad_norm": 0.35462483168975123,
      "learning_rate": 9.94998602999692e-06,
      "loss": 0.4788,
      "step": 1152
    },
    {
      "epoch": 0.4224990839135214,
      "grad_norm": 0.3883599127497516,
      "learning_rate": 9.949684794365848e-06,
      "loss": 0.4895,
      "step": 1153
    },
    {
      "epoch": 0.42286551850494686,
      "grad_norm": 0.3899883399775778,
      "learning_rate": 9.949382658868353e-06,
      "loss": 0.4558,
      "step": 1154
    },
    {
      "epoch": 0.4232319530963723,
      "grad_norm": 0.3217119171946446,
      "learning_rate": 9.949079623559366e-06,
      "loss": 0.4746,
      "step": 1155
    },
    {
      "epoch": 0.42359838768779773,
      "grad_norm": 0.3988700101314445,
      "learning_rate": 9.948775688493974e-06,
      "loss": 0.4545,
      "step": 1156
    },
    {
      "epoch": 0.42396482227922316,
      "grad_norm": 0.40500957054881825,
      "learning_rate": 9.94847085372744e-06,
      "loss": 0.4999,
      "step": 1157
    },
    {
      "epoch": 0.4243312568706486,
      "grad_norm": 0.31731314342594197,
      "learning_rate": 9.948165119315177e-06,
      "loss": 0.473,
      "step": 1158
    },
    {
      "epoch": 0.42469769146207403,
      "grad_norm": 0.32418659128401367,
      "learning_rate": 9.947858485312772e-06,
      "loss": 0.4739,
      "step": 1159
    },
    {
      "epoch": 0.42506412605349947,
      "grad_norm": 0.34582358545579145,
      "learning_rate": 9.94755095177597e-06,
      "loss": 0.4735,
      "step": 1160
    },
    {
      "epoch": 0.4254305606449249,
      "grad_norm": 0.43307356316484524,
      "learning_rate": 9.947242518760685e-06,
      "loss": 0.4921,
      "step": 1161
    },
    {
      "epoch": 0.4257969952363503,
      "grad_norm": 0.3486266115320226,
      "learning_rate": 9.946933186322988e-06,
      "loss": 0.4546,
      "step": 1162
    },
    {
      "epoch": 0.4261634298277757,
      "grad_norm": 0.3394947970065438,
      "learning_rate": 9.946622954519117e-06,
      "loss": 0.4626,
      "step": 1163
    },
    {
      "epoch": 0.42652986441920115,
      "grad_norm": 0.3599149780328244,
      "learning_rate": 9.946311823405471e-06,
      "loss": 0.4676,
      "step": 1164
    },
    {
      "epoch": 0.4268962990106266,
      "grad_norm": 0.36472894241638276,
      "learning_rate": 9.94599979303862e-06,
      "loss": 0.4739,
      "step": 1165
    },
    {
      "epoch": 0.427262733602052,
      "grad_norm": 0.37628019597073753,
      "learning_rate": 9.945686863475289e-06,
      "loss": 0.4886,
      "step": 1166
    },
    {
      "epoch": 0.42762916819347746,
      "grad_norm": 0.3545014604507153,
      "learning_rate": 9.945373034772365e-06,
      "loss": 0.5013,
      "step": 1167
    },
    {
      "epoch": 0.4279956027849029,
      "grad_norm": 0.4175272722875645,
      "learning_rate": 9.945058306986911e-06,
      "loss": 0.4957,
      "step": 1168
    },
    {
      "epoch": 0.42836203737632833,
      "grad_norm": 0.42098069222241924,
      "learning_rate": 9.94474268017614e-06,
      "loss": 0.4989,
      "step": 1169
    },
    {
      "epoch": 0.42872847196775377,
      "grad_norm": 0.3447273680136115,
      "learning_rate": 9.944426154397435e-06,
      "loss": 0.4601,
      "step": 1170
    },
    {
      "epoch": 0.4290949065591792,
      "grad_norm": 1.1326113526107584,
      "learning_rate": 9.944108729708342e-06,
      "loss": 0.4786,
      "step": 1171
    },
    {
      "epoch": 0.42946134115060464,
      "grad_norm": 0.34772556371987956,
      "learning_rate": 9.94379040616657e-06,
      "loss": 0.4883,
      "step": 1172
    },
    {
      "epoch": 0.42982777574203007,
      "grad_norm": 0.3948423270818809,
      "learning_rate": 9.94347118382999e-06,
      "loss": 0.4903,
      "step": 1173
    },
    {
      "epoch": 0.43019421033345545,
      "grad_norm": 0.35440843588833937,
      "learning_rate": 9.943151062756638e-06,
      "loss": 0.5032,
      "step": 1174
    },
    {
      "epoch": 0.4305606449248809,
      "grad_norm": 0.3619878490453968,
      "learning_rate": 9.942830043004713e-06,
      "loss": 0.4673,
      "step": 1175
    },
    {
      "epoch": 0.4309270795163063,
      "grad_norm": 0.45356302989058217,
      "learning_rate": 9.942508124632576e-06,
      "loss": 0.4856,
      "step": 1176
    },
    {
      "epoch": 0.43129351410773176,
      "grad_norm": 0.3319751716555285,
      "learning_rate": 9.942185307698754e-06,
      "loss": 0.4704,
      "step": 1177
    },
    {
      "epoch": 0.4316599486991572,
      "grad_norm": 0.3452824428786535,
      "learning_rate": 9.941861592261936e-06,
      "loss": 0.4547,
      "step": 1178
    },
    {
      "epoch": 0.4320263832905826,
      "grad_norm": 0.3370503554260688,
      "learning_rate": 9.941536978380974e-06,
      "loss": 0.4703,
      "step": 1179
    },
    {
      "epoch": 0.43239281788200806,
      "grad_norm": 0.4301376220802731,
      "learning_rate": 9.941211466114883e-06,
      "loss": 0.509,
      "step": 1180
    },
    {
      "epoch": 0.4327592524734335,
      "grad_norm": 0.3408627298020712,
      "learning_rate": 9.940885055522844e-06,
      "loss": 0.4671,
      "step": 1181
    },
    {
      "epoch": 0.43312568706485893,
      "grad_norm": 0.34781453079552377,
      "learning_rate": 9.940557746664197e-06,
      "loss": 0.4586,
      "step": 1182
    },
    {
      "epoch": 0.43349212165628437,
      "grad_norm": 0.35818059179812567,
      "learning_rate": 9.940229539598449e-06,
      "loss": 0.4835,
      "step": 1183
    },
    {
      "epoch": 0.4338585562477098,
      "grad_norm": 0.31911293285495834,
      "learning_rate": 9.93990043438527e-06,
      "loss": 0.4571,
      "step": 1184
    },
    {
      "epoch": 0.43422499083913524,
      "grad_norm": 0.3677158929926494,
      "learning_rate": 9.93957043108449e-06,
      "loss": 0.4988,
      "step": 1185
    },
    {
      "epoch": 0.4345914254305606,
      "grad_norm": 0.334939207568525,
      "learning_rate": 9.939239529756106e-06,
      "loss": 0.4721,
      "step": 1186
    },
    {
      "epoch": 0.43495786002198605,
      "grad_norm": 0.3908490604904947,
      "learning_rate": 9.938907730460277e-06,
      "loss": 0.5072,
      "step": 1187
    },
    {
      "epoch": 0.4353242946134115,
      "grad_norm": 0.32537713025496334,
      "learning_rate": 9.938575033257322e-06,
      "loss": 0.4665,
      "step": 1188
    },
    {
      "epoch": 0.4356907292048369,
      "grad_norm": 0.34846039124917877,
      "learning_rate": 9.93824143820773e-06,
      "loss": 0.4561,
      "step": 1189
    },
    {
      "epoch": 0.43605716379626236,
      "grad_norm": 0.3870258199788134,
      "learning_rate": 9.93790694537215e-06,
      "loss": 0.4779,
      "step": 1190
    },
    {
      "epoch": 0.4364235983876878,
      "grad_norm": 0.36271872458649185,
      "learning_rate": 9.93757155481139e-06,
      "loss": 0.4868,
      "step": 1191
    },
    {
      "epoch": 0.43679003297911323,
      "grad_norm": 0.3608732833454894,
      "learning_rate": 9.937235266586425e-06,
      "loss": 0.4781,
      "step": 1192
    },
    {
      "epoch": 0.43715646757053866,
      "grad_norm": 0.33167032300032284,
      "learning_rate": 9.936898080758397e-06,
      "loss": 0.4741,
      "step": 1193
    },
    {
      "epoch": 0.4375229021619641,
      "grad_norm": 0.3314901169021132,
      "learning_rate": 9.936559997388606e-06,
      "loss": 0.484,
      "step": 1194
    },
    {
      "epoch": 0.43788933675338954,
      "grad_norm": 0.32511006979914947,
      "learning_rate": 9.936221016538514e-06,
      "loss": 0.466,
      "step": 1195
    },
    {
      "epoch": 0.43825577134481497,
      "grad_norm": 0.3420042017591301,
      "learning_rate": 9.93588113826975e-06,
      "loss": 0.4928,
      "step": 1196
    },
    {
      "epoch": 0.4386222059362404,
      "grad_norm": 0.3426930792316042,
      "learning_rate": 9.935540362644106e-06,
      "loss": 0.4664,
      "step": 1197
    },
    {
      "epoch": 0.4389886405276658,
      "grad_norm": 0.3395389982704228,
      "learning_rate": 9.935198689723537e-06,
      "loss": 0.4698,
      "step": 1198
    },
    {
      "epoch": 0.4393550751190912,
      "grad_norm": 0.4014875533004529,
      "learning_rate": 9.934856119570156e-06,
      "loss": 0.4829,
      "step": 1199
    },
    {
      "epoch": 0.43972150971051666,
      "grad_norm": 0.34502641838329773,
      "learning_rate": 9.934512652246246e-06,
      "loss": 0.4797,
      "step": 1200
    },
    {
      "epoch": 0.4400879443019421,
      "grad_norm": 0.3387048950472487,
      "learning_rate": 9.93416828781425e-06,
      "loss": 0.4574,
      "step": 1201
    },
    {
      "epoch": 0.4404543788933675,
      "grad_norm": 0.3290046042608199,
      "learning_rate": 9.933823026336774e-06,
      "loss": 0.46,
      "step": 1202
    },
    {
      "epoch": 0.44082081348479296,
      "grad_norm": 0.340759906079425,
      "learning_rate": 9.933476867876588e-06,
      "loss": 0.5171,
      "step": 1203
    },
    {
      "epoch": 0.4411872480762184,
      "grad_norm": 0.37175701360462854,
      "learning_rate": 9.933129812496623e-06,
      "loss": 0.4648,
      "step": 1204
    },
    {
      "epoch": 0.44155368266764383,
      "grad_norm": 0.4077869305014768,
      "learning_rate": 9.932781860259977e-06,
      "loss": 0.4677,
      "step": 1205
    },
    {
      "epoch": 0.44192011725906927,
      "grad_norm": 0.369418177649091,
      "learning_rate": 9.932433011229907e-06,
      "loss": 0.5194,
      "step": 1206
    },
    {
      "epoch": 0.4422865518504947,
      "grad_norm": 0.3300479740575993,
      "learning_rate": 9.932083265469836e-06,
      "loss": 0.4884,
      "step": 1207
    },
    {
      "epoch": 0.44265298644192014,
      "grad_norm": 0.3328701983866472,
      "learning_rate": 9.931732623043347e-06,
      "loss": 0.4558,
      "step": 1208
    },
    {
      "epoch": 0.4430194210333456,
      "grad_norm": 0.5709707777017338,
      "learning_rate": 9.931381084014189e-06,
      "loss": 0.4879,
      "step": 1209
    },
    {
      "epoch": 0.44338585562477095,
      "grad_norm": 0.33197961263067144,
      "learning_rate": 9.931028648446273e-06,
      "loss": 0.4749,
      "step": 1210
    },
    {
      "epoch": 0.4437522902161964,
      "grad_norm": 0.3781891403024365,
      "learning_rate": 9.930675316403673e-06,
      "loss": 0.4578,
      "step": 1211
    },
    {
      "epoch": 0.4441187248076218,
      "grad_norm": 0.37182317822564387,
      "learning_rate": 9.930321087950624e-06,
      "loss": 0.4914,
      "step": 1212
    },
    {
      "epoch": 0.44448515939904726,
      "grad_norm": 0.3359649965269327,
      "learning_rate": 9.929965963151526e-06,
      "loss": 0.458,
      "step": 1213
    },
    {
      "epoch": 0.4448515939904727,
      "grad_norm": 0.33208202149921007,
      "learning_rate": 9.929609942070944e-06,
      "loss": 0.4711,
      "step": 1214
    },
    {
      "epoch": 0.44521802858189813,
      "grad_norm": 0.4080738549413266,
      "learning_rate": 9.929253024773603e-06,
      "loss": 0.5026,
      "step": 1215
    },
    {
      "epoch": 0.44558446317332356,
      "grad_norm": 0.3381761086344294,
      "learning_rate": 9.928895211324387e-06,
      "loss": 0.4707,
      "step": 1216
    },
    {
      "epoch": 0.445950897764749,
      "grad_norm": 0.35281619867563574,
      "learning_rate": 9.928536501788354e-06,
      "loss": 0.487,
      "step": 1217
    },
    {
      "epoch": 0.44631733235617443,
      "grad_norm": 0.32847097153867166,
      "learning_rate": 9.928176896230715e-06,
      "loss": 0.4719,
      "step": 1218
    },
    {
      "epoch": 0.44668376694759987,
      "grad_norm": 0.4086974889646117,
      "learning_rate": 9.927816394716847e-06,
      "loss": 0.4782,
      "step": 1219
    },
    {
      "epoch": 0.4470502015390253,
      "grad_norm": 0.3714450952349168,
      "learning_rate": 9.927454997312291e-06,
      "loss": 0.483,
      "step": 1220
    },
    {
      "epoch": 0.44741663613045074,
      "grad_norm": 0.37952250534011944,
      "learning_rate": 9.927092704082751e-06,
      "loss": 0.5079,
      "step": 1221
    },
    {
      "epoch": 0.4477830707218761,
      "grad_norm": 0.40105747466199804,
      "learning_rate": 9.926729515094092e-06,
      "loss": 0.4786,
      "step": 1222
    },
    {
      "epoch": 0.44814950531330155,
      "grad_norm": 0.34166543523656284,
      "learning_rate": 9.926365430412341e-06,
      "loss": 0.4839,
      "step": 1223
    },
    {
      "epoch": 0.448515939904727,
      "grad_norm": 0.37854108531238145,
      "learning_rate": 9.926000450103692e-06,
      "loss": 0.4855,
      "step": 1224
    },
    {
      "epoch": 0.4488823744961524,
      "grad_norm": 0.3441763230122492,
      "learning_rate": 9.925634574234499e-06,
      "loss": 0.4875,
      "step": 1225
    },
    {
      "epoch": 0.44924880908757786,
      "grad_norm": 0.3754107928910458,
      "learning_rate": 9.92526780287128e-06,
      "loss": 0.4847,
      "step": 1226
    },
    {
      "epoch": 0.4496152436790033,
      "grad_norm": 0.3785112220917919,
      "learning_rate": 9.924900136080712e-06,
      "loss": 0.4745,
      "step": 1227
    },
    {
      "epoch": 0.44998167827042873,
      "grad_norm": 0.35996188463239687,
      "learning_rate": 9.924531573929641e-06,
      "loss": 0.4824,
      "step": 1228
    },
    {
      "epoch": 0.45034811286185417,
      "grad_norm": 0.3408816147215836,
      "learning_rate": 9.92416211648507e-06,
      "loss": 0.4867,
      "step": 1229
    },
    {
      "epoch": 0.4507145474532796,
      "grad_norm": 0.35326542620342405,
      "learning_rate": 9.92379176381417e-06,
      "loss": 0.4751,
      "step": 1230
    },
    {
      "epoch": 0.45108098204470504,
      "grad_norm": 0.33955079018729967,
      "learning_rate": 9.923420515984272e-06,
      "loss": 0.4651,
      "step": 1231
    },
    {
      "epoch": 0.45144741663613047,
      "grad_norm": 0.37839691821828947,
      "learning_rate": 9.923048373062867e-06,
      "loss": 0.4991,
      "step": 1232
    },
    {
      "epoch": 0.4518138512275559,
      "grad_norm": 0.3570960226871995,
      "learning_rate": 9.922675335117615e-06,
      "loss": 0.4761,
      "step": 1233
    },
    {
      "epoch": 0.4521802858189813,
      "grad_norm": 0.3590975773905069,
      "learning_rate": 9.922301402216334e-06,
      "loss": 0.4943,
      "step": 1234
    },
    {
      "epoch": 0.4525467204104067,
      "grad_norm": 0.3534719817289141,
      "learning_rate": 9.921926574427004e-06,
      "loss": 0.4824,
      "step": 1235
    },
    {
      "epoch": 0.45291315500183216,
      "grad_norm": 0.35189153715304816,
      "learning_rate": 9.921550851817774e-06,
      "loss": 0.4817,
      "step": 1236
    },
    {
      "epoch": 0.4532795895932576,
      "grad_norm": 0.32316204403418164,
      "learning_rate": 9.921174234456947e-06,
      "loss": 0.4487,
      "step": 1237
    },
    {
      "epoch": 0.453646024184683,
      "grad_norm": 0.35746713573392175,
      "learning_rate": 9.920796722412995e-06,
      "loss": 0.4961,
      "step": 1238
    },
    {
      "epoch": 0.45401245877610846,
      "grad_norm": 0.32727274448843274,
      "learning_rate": 9.920418315754552e-06,
      "loss": 0.4816,
      "step": 1239
    },
    {
      "epoch": 0.4543788933675339,
      "grad_norm": 0.33918317730603703,
      "learning_rate": 9.920039014550413e-06,
      "loss": 0.4726,
      "step": 1240
    },
    {
      "epoch": 0.45474532795895933,
      "grad_norm": 0.32794972541789685,
      "learning_rate": 9.919658818869533e-06,
      "loss": 0.4645,
      "step": 1241
    },
    {
      "epoch": 0.45511176255038477,
      "grad_norm": 0.34315200251910605,
      "learning_rate": 9.919277728781036e-06,
      "loss": 0.4761,
      "step": 1242
    },
    {
      "epoch": 0.4554781971418102,
      "grad_norm": 0.32649973579682795,
      "learning_rate": 9.918895744354204e-06,
      "loss": 0.4846,
      "step": 1243
    },
    {
      "epoch": 0.45584463173323564,
      "grad_norm": 0.3335918281005749,
      "learning_rate": 9.918512865658485e-06,
      "loss": 0.4769,
      "step": 1244
    },
    {
      "epoch": 0.4562110663246611,
      "grad_norm": 0.3679626981627195,
      "learning_rate": 9.918129092763482e-06,
      "loss": 0.4848,
      "step": 1245
    },
    {
      "epoch": 0.45657750091608645,
      "grad_norm": 0.33660399277725206,
      "learning_rate": 9.917744425738971e-06,
      "loss": 0.4746,
      "step": 1246
    },
    {
      "epoch": 0.4569439355075119,
      "grad_norm": 0.35400820460225624,
      "learning_rate": 9.917358864654884e-06,
      "loss": 0.4925,
      "step": 1247
    },
    {
      "epoch": 0.4573103700989373,
      "grad_norm": 0.3231949741882881,
      "learning_rate": 9.916972409581316e-06,
      "loss": 0.4597,
      "step": 1248
    },
    {
      "epoch": 0.45767680469036276,
      "grad_norm": 0.3452493982483522,
      "learning_rate": 9.916585060588526e-06,
      "loss": 0.4782,
      "step": 1249
    },
    {
      "epoch": 0.4580432392817882,
      "grad_norm": 0.3250876858736982,
      "learning_rate": 9.916196817746936e-06,
      "loss": 0.4612,
      "step": 1250
    },
    {
      "epoch": 0.45840967387321363,
      "grad_norm": 0.35255197050759285,
      "learning_rate": 9.915807681127131e-06,
      "loss": 0.4725,
      "step": 1251
    },
    {
      "epoch": 0.45877610846463907,
      "grad_norm": 0.3485893236723856,
      "learning_rate": 9.915417650799855e-06,
      "loss": 0.506,
      "step": 1252
    },
    {
      "epoch": 0.4591425430560645,
      "grad_norm": 0.4149431567106927,
      "learning_rate": 9.915026726836016e-06,
      "loss": 0.4832,
      "step": 1253
    },
    {
      "epoch": 0.45950897764748994,
      "grad_norm": 0.3315439905771507,
      "learning_rate": 9.914634909306685e-06,
      "loss": 0.4872,
      "step": 1254
    },
    {
      "epoch": 0.45987541223891537,
      "grad_norm": 0.39109601505809216,
      "learning_rate": 9.914242198283099e-06,
      "loss": 0.4875,
      "step": 1255
    },
    {
      "epoch": 0.4602418468303408,
      "grad_norm": 0.3588948813187481,
      "learning_rate": 9.91384859383665e-06,
      "loss": 0.4793,
      "step": 1256
    },
    {
      "epoch": 0.46060828142176624,
      "grad_norm": 0.3576423330625272,
      "learning_rate": 9.913454096038898e-06,
      "loss": 0.4925,
      "step": 1257
    },
    {
      "epoch": 0.4609747160131916,
      "grad_norm": 0.36180694847889583,
      "learning_rate": 9.913058704961565e-06,
      "loss": 0.4871,
      "step": 1258
    },
    {
      "epoch": 0.46134115060461706,
      "grad_norm": 0.32071795773407186,
      "learning_rate": 9.912662420676532e-06,
      "loss": 0.4693,
      "step": 1259
    },
    {
      "epoch": 0.4617075851960425,
      "grad_norm": 0.35209462897630367,
      "learning_rate": 9.912265243255844e-06,
      "loss": 0.4738,
      "step": 1260
    },
    {
      "epoch": 0.4620740197874679,
      "grad_norm": 0.39414472337715223,
      "learning_rate": 9.911867172771711e-06,
      "loss": 0.4859,
      "step": 1261
    },
    {
      "epoch": 0.46244045437889336,
      "grad_norm": 0.36128661757803004,
      "learning_rate": 9.911468209296504e-06,
      "loss": 0.4803,
      "step": 1262
    },
    {
      "epoch": 0.4628068889703188,
      "grad_norm": 0.3597082564208622,
      "learning_rate": 9.911068352902754e-06,
      "loss": 0.4679,
      "step": 1263
    },
    {
      "epoch": 0.46317332356174423,
      "grad_norm": 0.36395789709589166,
      "learning_rate": 9.910667603663156e-06,
      "loss": 0.4773,
      "step": 1264
    },
    {
      "epoch": 0.46353975815316967,
      "grad_norm": 0.37056115545840174,
      "learning_rate": 9.910265961650567e-06,
      "loss": 0.4854,
      "step": 1265
    },
    {
      "epoch": 0.4639061927445951,
      "grad_norm": 0.3288128260681619,
      "learning_rate": 9.90986342693801e-06,
      "loss": 0.4854,
      "step": 1266
    },
    {
      "epoch": 0.46427262733602054,
      "grad_norm": 0.4485602582844438,
      "learning_rate": 9.90945999959866e-06,
      "loss": 0.4603,
      "step": 1267
    },
    {
      "epoch": 0.464639061927446,
      "grad_norm": 0.37786692924734416,
      "learning_rate": 9.90905567970587e-06,
      "loss": 0.4985,
      "step": 1268
    },
    {
      "epoch": 0.4650054965188714,
      "grad_norm": 0.3455407694351336,
      "learning_rate": 9.908650467333138e-06,
      "loss": 0.4932,
      "step": 1269
    },
    {
      "epoch": 0.4653719311102968,
      "grad_norm": 0.4379425309955117,
      "learning_rate": 9.90824436255414e-06,
      "loss": 0.4825,
      "step": 1270
    },
    {
      "epoch": 0.4657383657017222,
      "grad_norm": 0.35901042775843983,
      "learning_rate": 9.907837365442702e-06,
      "loss": 0.4916,
      "step": 1271
    },
    {
      "epoch": 0.46610480029314766,
      "grad_norm": 0.33814250021383363,
      "learning_rate": 9.907429476072819e-06,
      "loss": 0.4727,
      "step": 1272
    },
    {
      "epoch": 0.4664712348845731,
      "grad_norm": 0.3885170155061806,
      "learning_rate": 9.907020694518646e-06,
      "loss": 0.4806,
      "step": 1273
    },
    {
      "epoch": 0.46683766947599853,
      "grad_norm": 0.36065378505041423,
      "learning_rate": 9.906611020854501e-06,
      "loss": 0.4939,
      "step": 1274
    },
    {
      "epoch": 0.46720410406742396,
      "grad_norm": 0.3673881816810962,
      "learning_rate": 9.906200455154865e-06,
      "loss": 0.4738,
      "step": 1275
    },
    {
      "epoch": 0.4675705386588494,
      "grad_norm": 0.3212015490158451,
      "learning_rate": 9.905788997494377e-06,
      "loss": 0.4999,
      "step": 1276
    },
    {
      "epoch": 0.46793697325027483,
      "grad_norm": 0.36977756914845433,
      "learning_rate": 9.905376647947844e-06,
      "loss": 0.4608,
      "step": 1277
    },
    {
      "epoch": 0.46830340784170027,
      "grad_norm": 0.381963693374118,
      "learning_rate": 9.90496340659023e-06,
      "loss": 0.4707,
      "step": 1278
    },
    {
      "epoch": 0.4686698424331257,
      "grad_norm": 0.39165859482960963,
      "learning_rate": 9.904549273496666e-06,
      "loss": 0.485,
      "step": 1279
    },
    {
      "epoch": 0.46903627702455114,
      "grad_norm": 0.350455323182188,
      "learning_rate": 9.90413424874244e-06,
      "loss": 0.4875,
      "step": 1280
    },
    {
      "epoch": 0.4694027116159766,
      "grad_norm": 0.3458185964033527,
      "learning_rate": 9.903718332403007e-06,
      "loss": 0.4742,
      "step": 1281
    },
    {
      "epoch": 0.46976914620740196,
      "grad_norm": 0.5685224260163415,
      "learning_rate": 9.90330152455398e-06,
      "loss": 0.4934,
      "step": 1282
    },
    {
      "epoch": 0.4701355807988274,
      "grad_norm": 0.38636916318563325,
      "learning_rate": 9.902883825271138e-06,
      "loss": 0.4946,
      "step": 1283
    },
    {
      "epoch": 0.4705020153902528,
      "grad_norm": 0.33502897007198335,
      "learning_rate": 9.902465234630415e-06,
      "loss": 0.4578,
      "step": 1284
    },
    {
      "epoch": 0.47086844998167826,
      "grad_norm": 0.363625294761928,
      "learning_rate": 9.902045752707916e-06,
      "loss": 0.4793,
      "step": 1285
    },
    {
      "epoch": 0.4712348845731037,
      "grad_norm": 0.3811218942935005,
      "learning_rate": 9.901625379579903e-06,
      "loss": 0.4958,
      "step": 1286
    },
    {
      "epoch": 0.47160131916452913,
      "grad_norm": 0.3643614096653435,
      "learning_rate": 9.901204115322803e-06,
      "loss": 0.4765,
      "step": 1287
    },
    {
      "epoch": 0.47196775375595457,
      "grad_norm": 0.3459796953932444,
      "learning_rate": 9.9007819600132e-06,
      "loss": 0.4521,
      "step": 1288
    },
    {
      "epoch": 0.47233418834738,
      "grad_norm": 0.36980175346165184,
      "learning_rate": 9.900358913727843e-06,
      "loss": 0.4715,
      "step": 1289
    },
    {
      "epoch": 0.47270062293880544,
      "grad_norm": 0.3226644433481222,
      "learning_rate": 9.899934976543646e-06,
      "loss": 0.4837,
      "step": 1290
    },
    {
      "epoch": 0.47306705753023087,
      "grad_norm": 0.35725318482904045,
      "learning_rate": 9.89951014853768e-06,
      "loss": 0.475,
      "step": 1291
    },
    {
      "epoch": 0.4734334921216563,
      "grad_norm": 0.3354524927583145,
      "learning_rate": 9.899084429787178e-06,
      "loss": 0.4702,
      "step": 1292
    },
    {
      "epoch": 0.4737999267130817,
      "grad_norm": 0.3430719921572459,
      "learning_rate": 9.898657820369539e-06,
      "loss": 0.496,
      "step": 1293
    },
    {
      "epoch": 0.4741663613045071,
      "grad_norm": 0.39233591791377376,
      "learning_rate": 9.898230320362323e-06,
      "loss": 0.5066,
      "step": 1294
    },
    {
      "epoch": 0.47453279589593256,
      "grad_norm": 0.3591619333971561,
      "learning_rate": 9.897801929843248e-06,
      "loss": 0.4692,
      "step": 1295
    },
    {
      "epoch": 0.474899230487358,
      "grad_norm": 0.3352007330250629,
      "learning_rate": 9.8973726488902e-06,
      "loss": 0.46,
      "step": 1296
    },
    {
      "epoch": 0.47526566507878343,
      "grad_norm": 0.38885245248457234,
      "learning_rate": 9.896942477581221e-06,
      "loss": 0.4925,
      "step": 1297
    },
    {
      "epoch": 0.47563209967020886,
      "grad_norm": 0.3940918207341639,
      "learning_rate": 9.896511415994517e-06,
      "loss": 0.473,
      "step": 1298
    },
    {
      "epoch": 0.4759985342616343,
      "grad_norm": 0.29532923443288556,
      "learning_rate": 9.896079464208457e-06,
      "loss": 0.4538,
      "step": 1299
    },
    {
      "epoch": 0.47636496885305973,
      "grad_norm": 0.3307351369868918,
      "learning_rate": 9.89564662230157e-06,
      "loss": 0.483,
      "step": 1300
    },
    {
      "epoch": 0.47673140344448517,
      "grad_norm": 0.3726202039146688,
      "learning_rate": 9.895212890352551e-06,
      "loss": 0.5057,
      "step": 1301
    },
    {
      "epoch": 0.4770978380359106,
      "grad_norm": 0.3873952729150689,
      "learning_rate": 9.89477826844025e-06,
      "loss": 0.5065,
      "step": 1302
    },
    {
      "epoch": 0.47746427262733604,
      "grad_norm": 0.3262414689904565,
      "learning_rate": 9.894342756643684e-06,
      "loss": 0.4658,
      "step": 1303
    },
    {
      "epoch": 0.4778307072187615,
      "grad_norm": 0.3331892214958134,
      "learning_rate": 9.89390635504203e-06,
      "loss": 0.4942,
      "step": 1304
    },
    {
      "epoch": 0.47819714181018685,
      "grad_norm": 0.3661197701177818,
      "learning_rate": 9.893469063714628e-06,
      "loss": 0.4766,
      "step": 1305
    },
    {
      "epoch": 0.4785635764016123,
      "grad_norm": 0.33593596918646423,
      "learning_rate": 9.893030882740977e-06,
      "loss": 0.5109,
      "step": 1306
    },
    {
      "epoch": 0.4789300109930377,
      "grad_norm": 0.3535774928900838,
      "learning_rate": 9.89259181220074e-06,
      "loss": 0.4868,
      "step": 1307
    },
    {
      "epoch": 0.47929644558446316,
      "grad_norm": 0.3383313300572853,
      "learning_rate": 9.892151852173744e-06,
      "loss": 0.4735,
      "step": 1308
    },
    {
      "epoch": 0.4796628801758886,
      "grad_norm": 0.32594495102807364,
      "learning_rate": 9.891711002739971e-06,
      "loss": 0.472,
      "step": 1309
    },
    {
      "epoch": 0.48002931476731403,
      "grad_norm": 0.3584420674159499,
      "learning_rate": 9.891269263979571e-06,
      "loss": 0.5123,
      "step": 1310
    },
    {
      "epoch": 0.48039574935873947,
      "grad_norm": 0.3542612656252512,
      "learning_rate": 9.890826635972852e-06,
      "loss": 0.4659,
      "step": 1311
    },
    {
      "epoch": 0.4807621839501649,
      "grad_norm": 0.3136187351877026,
      "learning_rate": 9.890383118800287e-06,
      "loss": 0.4793,
      "step": 1312
    },
    {
      "epoch": 0.48112861854159034,
      "grad_norm": 0.35621059939208366,
      "learning_rate": 9.889938712542507e-06,
      "loss": 0.4978,
      "step": 1313
    },
    {
      "epoch": 0.48149505313301577,
      "grad_norm": 0.33292904988134675,
      "learning_rate": 9.889493417280305e-06,
      "loss": 0.4757,
      "step": 1314
    },
    {
      "epoch": 0.4818614877244412,
      "grad_norm": 0.32021234945352617,
      "learning_rate": 9.88904723309464e-06,
      "loss": 0.4883,
      "step": 1315
    },
    {
      "epoch": 0.48222792231586664,
      "grad_norm": 0.34253735751083286,
      "learning_rate": 9.888600160066627e-06,
      "loss": 0.4903,
      "step": 1316
    },
    {
      "epoch": 0.482594356907292,
      "grad_norm": 0.32118069982569925,
      "learning_rate": 9.888152198277548e-06,
      "loss": 0.4657,
      "step": 1317
    },
    {
      "epoch": 0.48296079149871746,
      "grad_norm": 0.3520152255464055,
      "learning_rate": 9.887703347808838e-06,
      "loss": 0.4646,
      "step": 1318
    },
    {
      "epoch": 0.4833272260901429,
      "grad_norm": 0.345720993092728,
      "learning_rate": 9.887253608742107e-06,
      "loss": 0.4474,
      "step": 1319
    },
    {
      "epoch": 0.4836936606815683,
      "grad_norm": 0.3639967714044352,
      "learning_rate": 9.886802981159114e-06,
      "loss": 0.4663,
      "step": 1320
    },
    {
      "epoch": 0.48406009527299376,
      "grad_norm": 0.307892684648314,
      "learning_rate": 9.886351465141785e-06,
      "loss": 0.4805,
      "step": 1321
    },
    {
      "epoch": 0.4844265298644192,
      "grad_norm": 0.35607033439800634,
      "learning_rate": 9.885899060772207e-06,
      "loss": 0.4625,
      "step": 1322
    },
    {
      "epoch": 0.48479296445584463,
      "grad_norm": 0.33953090368568634,
      "learning_rate": 9.88544576813263e-06,
      "loss": 0.4859,
      "step": 1323
    },
    {
      "epoch": 0.48515939904727007,
      "grad_norm": 0.33951654566687794,
      "learning_rate": 9.884991587305459e-06,
      "loss": 0.4804,
      "step": 1324
    },
    {
      "epoch": 0.4855258336386955,
      "grad_norm": 0.3488593689721262,
      "learning_rate": 9.884536518373271e-06,
      "loss": 0.5058,
      "step": 1325
    },
    {
      "epoch": 0.48589226823012094,
      "grad_norm": 0.3811725682052719,
      "learning_rate": 9.884080561418797e-06,
      "loss": 0.4849,
      "step": 1326
    },
    {
      "epoch": 0.4862587028215464,
      "grad_norm": 0.3483817422655987,
      "learning_rate": 9.883623716524929e-06,
      "loss": 0.4765,
      "step": 1327
    },
    {
      "epoch": 0.4866251374129718,
      "grad_norm": 0.3396127682251299,
      "learning_rate": 9.883165983774726e-06,
      "loss": 0.4567,
      "step": 1328
    },
    {
      "epoch": 0.4869915720043972,
      "grad_norm": 0.44358060209911765,
      "learning_rate": 9.882707363251403e-06,
      "loss": 0.4779,
      "step": 1329
    },
    {
      "epoch": 0.4873580065958226,
      "grad_norm": 0.33852183832150845,
      "learning_rate": 9.882247855038339e-06,
      "loss": 0.4717,
      "step": 1330
    },
    {
      "epoch": 0.48772444118724806,
      "grad_norm": 0.35501830423341385,
      "learning_rate": 9.881787459219074e-06,
      "loss": 0.469,
      "step": 1331
    },
    {
      "epoch": 0.4880908757786735,
      "grad_norm": 0.34533235700211157,
      "learning_rate": 9.881326175877309e-06,
      "loss": 0.4642,
      "step": 1332
    },
    {
      "epoch": 0.48845731037009893,
      "grad_norm": 0.3338235187992942,
      "learning_rate": 9.880864005096906e-06,
      "loss": 0.4794,
      "step": 1333
    },
    {
      "epoch": 0.48882374496152436,
      "grad_norm": 0.32441753066657564,
      "learning_rate": 9.88040094696189e-06,
      "loss": 0.4803,
      "step": 1334
    },
    {
      "epoch": 0.4891901795529498,
      "grad_norm": 0.3204468611541099,
      "learning_rate": 9.879937001556446e-06,
      "loss": 0.4554,
      "step": 1335
    },
    {
      "epoch": 0.48955661414437524,
      "grad_norm": 0.33302337130443616,
      "learning_rate": 9.87947216896492e-06,
      "loss": 0.4849,
      "step": 1336
    },
    {
      "epoch": 0.48992304873580067,
      "grad_norm": 0.3358955380734829,
      "learning_rate": 9.87900644927182e-06,
      "loss": 0.4517,
      "step": 1337
    },
    {
      "epoch": 0.4902894833272261,
      "grad_norm": 0.3204261490090355,
      "learning_rate": 9.878539842561816e-06,
      "loss": 0.4827,
      "step": 1338
    },
    {
      "epoch": 0.49065591791865154,
      "grad_norm": 0.33502562943401576,
      "learning_rate": 9.878072348919738e-06,
      "loss": 0.4804,
      "step": 1339
    },
    {
      "epoch": 0.491022352510077,
      "grad_norm": 0.3555987866924369,
      "learning_rate": 9.877603968430578e-06,
      "loss": 0.4447,
      "step": 1340
    },
    {
      "epoch": 0.49138878710150236,
      "grad_norm": 0.33208802263727527,
      "learning_rate": 9.877134701179487e-06,
      "loss": 0.4852,
      "step": 1341
    },
    {
      "epoch": 0.4917552216929278,
      "grad_norm": 0.3310292745969881,
      "learning_rate": 9.876664547251781e-06,
      "loss": 0.4571,
      "step": 1342
    },
    {
      "epoch": 0.4921216562843532,
      "grad_norm": 0.31612942152349477,
      "learning_rate": 9.876193506732934e-06,
      "loss": 0.4954,
      "step": 1343
    },
    {
      "epoch": 0.49248809087577866,
      "grad_norm": 0.33329131782734456,
      "learning_rate": 9.875721579708583e-06,
      "loss": 0.4814,
      "step": 1344
    },
    {
      "epoch": 0.4928545254672041,
      "grad_norm": 0.36130445988970694,
      "learning_rate": 9.875248766264527e-06,
      "loss": 0.4814,
      "step": 1345
    },
    {
      "epoch": 0.49322096005862953,
      "grad_norm": 0.34176269879361704,
      "learning_rate": 9.874775066486724e-06,
      "loss": 0.464,
      "step": 1346
    },
    {
      "epoch": 0.49358739465005497,
      "grad_norm": 0.3380411931673656,
      "learning_rate": 9.874300480461292e-06,
      "loss": 0.4531,
      "step": 1347
    },
    {
      "epoch": 0.4939538292414804,
      "grad_norm": 0.3423190568586275,
      "learning_rate": 9.873825008274514e-06,
      "loss": 0.4816,
      "step": 1348
    },
    {
      "epoch": 0.49432026383290584,
      "grad_norm": 0.34460903335734416,
      "learning_rate": 9.873348650012833e-06,
      "loss": 0.4984,
      "step": 1349
    },
    {
      "epoch": 0.4946866984243313,
      "grad_norm": 0.34305279386482984,
      "learning_rate": 9.87287140576285e-06,
      "loss": 0.444,
      "step": 1350
    },
    {
      "epoch": 0.4950531330157567,
      "grad_norm": 0.3338773776569606,
      "learning_rate": 9.872393275611329e-06,
      "loss": 0.485,
      "step": 1351
    },
    {
      "epoch": 0.49541956760718214,
      "grad_norm": 0.3182919418062384,
      "learning_rate": 9.871914259645198e-06,
      "loss": 0.4813,
      "step": 1352
    },
    {
      "epoch": 0.4957860021986075,
      "grad_norm": 0.8324591442698283,
      "learning_rate": 9.871434357951544e-06,
      "loss": 0.5037,
      "step": 1353
    },
    {
      "epoch": 0.49615243679003296,
      "grad_norm": 0.3014828100682889,
      "learning_rate": 9.87095357061761e-06,
      "loss": 0.463,
      "step": 1354
    },
    {
      "epoch": 0.4965188713814584,
      "grad_norm": 0.3295452846540232,
      "learning_rate": 9.870471897730807e-06,
      "loss": 0.5038,
      "step": 1355
    },
    {
      "epoch": 0.49688530597288383,
      "grad_norm": 0.33343475406260137,
      "learning_rate": 9.869989339378706e-06,
      "loss": 0.4673,
      "step": 1356
    },
    {
      "epoch": 0.49725174056430926,
      "grad_norm": 0.3311023302187062,
      "learning_rate": 9.869505895649036e-06,
      "loss": 0.4841,
      "step": 1357
    },
    {
      "epoch": 0.4976181751557347,
      "grad_norm": 0.32040137349919967,
      "learning_rate": 9.869021566629687e-06,
      "loss": 0.4611,
      "step": 1358
    },
    {
      "epoch": 0.49798460974716013,
      "grad_norm": 0.313742085972362,
      "learning_rate": 9.868536352408714e-06,
      "loss": 0.4672,
      "step": 1359
    },
    {
      "epoch": 0.49835104433858557,
      "grad_norm": 0.31450573826359157,
      "learning_rate": 9.868050253074328e-06,
      "loss": 0.454,
      "step": 1360
    },
    {
      "epoch": 0.498717478930011,
      "grad_norm": 0.3370637766348485,
      "learning_rate": 9.867563268714906e-06,
      "loss": 0.4673,
      "step": 1361
    },
    {
      "epoch": 0.49908391352143644,
      "grad_norm": 0.3558003380806816,
      "learning_rate": 9.86707539941898e-06,
      "loss": 0.4955,
      "step": 1362
    },
    {
      "epoch": 0.4994503481128619,
      "grad_norm": 0.3770894415632178,
      "learning_rate": 9.866586645275247e-06,
      "loss": 0.5265,
      "step": 1363
    },
    {
      "epoch": 0.4998167827042873,
      "grad_norm": 0.3524733120844753,
      "learning_rate": 9.866097006372565e-06,
      "loss": 0.4784,
      "step": 1364
    },
    {
      "epoch": 0.5001832172957127,
      "grad_norm": 0.32407616696208175,
      "learning_rate": 9.865606482799951e-06,
      "loss": 0.4894,
      "step": 1365
    },
    {
      "epoch": 0.5005496518871382,
      "grad_norm": 0.34809237474378,
      "learning_rate": 9.865115074646583e-06,
      "loss": 0.4907,
      "step": 1366
    },
    {
      "epoch": 0.5009160864785636,
      "grad_norm": 0.3849434485609519,
      "learning_rate": 9.8646227820018e-06,
      "loss": 0.4757,
      "step": 1367
    },
    {
      "epoch": 0.501282521069989,
      "grad_norm": 0.43884607084302596,
      "learning_rate": 9.864129604955105e-06,
      "loss": 0.4848,
      "step": 1368
    },
    {
      "epoch": 0.5016489556614144,
      "grad_norm": 0.3937016598642911,
      "learning_rate": 9.863635543596156e-06,
      "loss": 0.4875,
      "step": 1369
    },
    {
      "epoch": 0.5020153902528399,
      "grad_norm": 0.35333663378114355,
      "learning_rate": 9.863140598014776e-06,
      "loss": 0.4771,
      "step": 1370
    },
    {
      "epoch": 0.5023818248442653,
      "grad_norm": 0.44654343122784235,
      "learning_rate": 9.862644768300947e-06,
      "loss": 0.4836,
      "step": 1371
    },
    {
      "epoch": 0.5027482594356907,
      "grad_norm": 0.32883430281773324,
      "learning_rate": 9.862148054544812e-06,
      "loss": 0.4924,
      "step": 1372
    },
    {
      "epoch": 0.5031146940271162,
      "grad_norm": 0.35322613830451843,
      "learning_rate": 9.861650456836675e-06,
      "loss": 0.4833,
      "step": 1373
    },
    {
      "epoch": 0.5034811286185416,
      "grad_norm": 0.2936917325827801,
      "learning_rate": 9.861151975267002e-06,
      "loss": 0.4604,
      "step": 1374
    },
    {
      "epoch": 0.503847563209967,
      "grad_norm": 0.4193617407827907,
      "learning_rate": 9.860652609926417e-06,
      "loss": 0.4644,
      "step": 1375
    },
    {
      "epoch": 0.5042139978013924,
      "grad_norm": 0.339984425313873,
      "learning_rate": 9.860152360905705e-06,
      "loss": 0.4935,
      "step": 1376
    },
    {
      "epoch": 0.5045804323928179,
      "grad_norm": 0.38844812629998454,
      "learning_rate": 9.859651228295816e-06,
      "loss": 0.4934,
      "step": 1377
    },
    {
      "epoch": 0.5049468669842433,
      "grad_norm": 0.35504744741843175,
      "learning_rate": 9.859149212187855e-06,
      "loss": 0.4552,
      "step": 1378
    },
    {
      "epoch": 0.5053133015756688,
      "grad_norm": 0.366229325632918,
      "learning_rate": 9.858646312673088e-06,
      "loss": 0.472,
      "step": 1379
    },
    {
      "epoch": 0.5056797361670942,
      "grad_norm": 0.33722588736585213,
      "learning_rate": 9.858142529842947e-06,
      "loss": 0.4781,
      "step": 1380
    },
    {
      "epoch": 0.5060461707585197,
      "grad_norm": 0.3398218039454455,
      "learning_rate": 9.85763786378902e-06,
      "loss": 0.4499,
      "step": 1381
    },
    {
      "epoch": 0.506412605349945,
      "grad_norm": 0.33808367604097134,
      "learning_rate": 9.857132314603058e-06,
      "loss": 0.4979,
      "step": 1382
    },
    {
      "epoch": 0.5067790399413704,
      "grad_norm": 0.3419562883047194,
      "learning_rate": 9.856625882376967e-06,
      "loss": 0.4498,
      "step": 1383
    },
    {
      "epoch": 0.5071454745327959,
      "grad_norm": 0.3283681950158234,
      "learning_rate": 9.85611856720282e-06,
      "loss": 0.444,
      "step": 1384
    },
    {
      "epoch": 0.5075119091242213,
      "grad_norm": 0.34243892777717916,
      "learning_rate": 9.855610369172852e-06,
      "loss": 0.4759,
      "step": 1385
    },
    {
      "epoch": 0.5078783437156468,
      "grad_norm": 0.35041947670291895,
      "learning_rate": 9.855101288379452e-06,
      "loss": 0.4811,
      "step": 1386
    },
    {
      "epoch": 0.5082447783070722,
      "grad_norm": 0.3564053657659693,
      "learning_rate": 9.85459132491517e-06,
      "loss": 0.467,
      "step": 1387
    },
    {
      "epoch": 0.5086112128984976,
      "grad_norm": 0.3319844470251787,
      "learning_rate": 9.854080478872721e-06,
      "loss": 0.4784,
      "step": 1388
    },
    {
      "epoch": 0.508977647489923,
      "grad_norm": 0.33939000646244866,
      "learning_rate": 9.853568750344977e-06,
      "loss": 0.4666,
      "step": 1389
    },
    {
      "epoch": 0.5093440820813485,
      "grad_norm": 0.40657196460121753,
      "learning_rate": 9.853056139424974e-06,
      "loss": 0.469,
      "step": 1390
    },
    {
      "epoch": 0.5097105166727739,
      "grad_norm": 0.3639100517226516,
      "learning_rate": 9.852542646205905e-06,
      "loss": 0.4766,
      "step": 1391
    },
    {
      "epoch": 0.5100769512641994,
      "grad_norm": 0.34913602650679976,
      "learning_rate": 9.852028270781122e-06,
      "loss": 0.4945,
      "step": 1392
    },
    {
      "epoch": 0.5104433858556248,
      "grad_norm": 0.33881731118925434,
      "learning_rate": 9.851513013244144e-06,
      "loss": 0.4645,
      "step": 1393
    },
    {
      "epoch": 0.5108098204470503,
      "grad_norm": 0.32280205772907933,
      "learning_rate": 9.850996873688644e-06,
      "loss": 0.4886,
      "step": 1394
    },
    {
      "epoch": 0.5111762550384756,
      "grad_norm": 0.3452942150057405,
      "learning_rate": 9.850479852208458e-06,
      "loss": 0.4703,
      "step": 1395
    },
    {
      "epoch": 0.511542689629901,
      "grad_norm": 0.3549260253529213,
      "learning_rate": 9.849961948897582e-06,
      "loss": 0.5037,
      "step": 1396
    },
    {
      "epoch": 0.5119091242213265,
      "grad_norm": 0.3103173630272495,
      "learning_rate": 9.849443163850172e-06,
      "loss": 0.4823,
      "step": 1397
    },
    {
      "epoch": 0.5122755588127519,
      "grad_norm": 0.3686724190047589,
      "learning_rate": 9.848923497160543e-06,
      "loss": 0.4632,
      "step": 1398
    },
    {
      "epoch": 0.5126419934041774,
      "grad_norm": 0.36684287088707657,
      "learning_rate": 9.848402948923177e-06,
      "loss": 0.4777,
      "step": 1399
    },
    {
      "epoch": 0.5130084279956028,
      "grad_norm": 0.394594872123801,
      "learning_rate": 9.847881519232705e-06,
      "loss": 0.4709,
      "step": 1400
    },
    {
      "epoch": 0.5133748625870282,
      "grad_norm": 0.39443451906407634,
      "learning_rate": 9.847359208183927e-06,
      "loss": 0.4783,
      "step": 1401
    },
    {
      "epoch": 0.5137412971784536,
      "grad_norm": 0.3418417867570158,
      "learning_rate": 9.846836015871802e-06,
      "loss": 0.4648,
      "step": 1402
    },
    {
      "epoch": 0.5141077317698791,
      "grad_norm": 0.3301471391961097,
      "learning_rate": 9.846311942391445e-06,
      "loss": 0.4739,
      "step": 1403
    },
    {
      "epoch": 0.5144741663613045,
      "grad_norm": 0.36807988032794986,
      "learning_rate": 9.845786987838136e-06,
      "loss": 0.4637,
      "step": 1404
    },
    {
      "epoch": 0.51484060095273,
      "grad_norm": 0.40734117383251345,
      "learning_rate": 9.845261152307312e-06,
      "loss": 0.4749,
      "step": 1405
    },
    {
      "epoch": 0.5152070355441554,
      "grad_norm": 0.3556289142341691,
      "learning_rate": 9.844734435894575e-06,
      "loss": 0.4563,
      "step": 1406
    },
    {
      "epoch": 0.5155734701355807,
      "grad_norm": 0.40907534935822204,
      "learning_rate": 9.844206838695676e-06,
      "loss": 0.4738,
      "step": 1407
    },
    {
      "epoch": 0.5159399047270062,
      "grad_norm": 0.3671162376211414,
      "learning_rate": 9.843678360806542e-06,
      "loss": 0.4872,
      "step": 1408
    },
    {
      "epoch": 0.5163063393184316,
      "grad_norm": 0.338106040922876,
      "learning_rate": 9.843149002323245e-06,
      "loss": 0.4894,
      "step": 1409
    },
    {
      "epoch": 0.5166727739098571,
      "grad_norm": 0.33433018860057,
      "learning_rate": 9.842618763342029e-06,
      "loss": 0.4621,
      "step": 1410
    },
    {
      "epoch": 0.5170392085012825,
      "grad_norm": 0.33260607833621136,
      "learning_rate": 9.842087643959288e-06,
      "loss": 0.4515,
      "step": 1411
    },
    {
      "epoch": 0.517405643092708,
      "grad_norm": 0.33061697788173167,
      "learning_rate": 9.841555644271585e-06,
      "loss": 0.4943,
      "step": 1412
    },
    {
      "epoch": 0.5177720776841334,
      "grad_norm": 0.3120452083170936,
      "learning_rate": 9.841022764375638e-06,
      "loss": 0.4678,
      "step": 1413
    },
    {
      "epoch": 0.5181385122755588,
      "grad_norm": 0.3467604513075146,
      "learning_rate": 9.840489004368325e-06,
      "loss": 0.4605,
      "step": 1414
    },
    {
      "epoch": 0.5185049468669842,
      "grad_norm": 0.3601620863199198,
      "learning_rate": 9.839954364346687e-06,
      "loss": 0.4977,
      "step": 1415
    },
    {
      "epoch": 0.5188713814584097,
      "grad_norm": 0.3203328915367949,
      "learning_rate": 9.839418844407921e-06,
      "loss": 0.4826,
      "step": 1416
    },
    {
      "epoch": 0.5192378160498351,
      "grad_norm": 0.34829684343856293,
      "learning_rate": 9.838882444649387e-06,
      "loss": 0.4568,
      "step": 1417
    },
    {
      "epoch": 0.5196042506412606,
      "grad_norm": 0.34889071711666547,
      "learning_rate": 9.838345165168604e-06,
      "loss": 0.466,
      "step": 1418
    },
    {
      "epoch": 0.519970685232686,
      "grad_norm": 0.3237868067112904,
      "learning_rate": 9.83780700606325e-06,
      "loss": 0.4664,
      "step": 1419
    },
    {
      "epoch": 0.5203371198241113,
      "grad_norm": 0.3146521176713479,
      "learning_rate": 9.837267967431164e-06,
      "loss": 0.4593,
      "step": 1420
    },
    {
      "epoch": 0.5207035544155368,
      "grad_norm": 0.36386540544634405,
      "learning_rate": 9.836728049370345e-06,
      "loss": 0.5022,
      "step": 1421
    },
    {
      "epoch": 0.5210699890069622,
      "grad_norm": 0.36560976375365617,
      "learning_rate": 9.836187251978953e-06,
      "loss": 0.4819,
      "step": 1422
    },
    {
      "epoch": 0.5214364235983877,
      "grad_norm": 0.33553834096227986,
      "learning_rate": 9.835645575355304e-06,
      "loss": 0.4751,
      "step": 1423
    },
    {
      "epoch": 0.5218028581898131,
      "grad_norm": 0.33327749803241574,
      "learning_rate": 9.835103019597877e-06,
      "loss": 0.4764,
      "step": 1424
    },
    {
      "epoch": 0.5221692927812386,
      "grad_norm": 0.30473783008258853,
      "learning_rate": 9.834559584805313e-06,
      "loss": 0.473,
      "step": 1425
    },
    {
      "epoch": 0.522535727372664,
      "grad_norm": 0.37933714816657743,
      "learning_rate": 9.834015271076405e-06,
      "loss": 0.4602,
      "step": 1426
    },
    {
      "epoch": 0.5229021619640895,
      "grad_norm": 0.39667362252210747,
      "learning_rate": 9.833470078510116e-06,
      "loss": 0.4616,
      "step": 1427
    },
    {
      "epoch": 0.5232685965555148,
      "grad_norm": 0.38487427455126383,
      "learning_rate": 9.83292400720556e-06,
      "loss": 0.4777,
      "step": 1428
    },
    {
      "epoch": 0.5236350311469403,
      "grad_norm": 0.36112065039148483,
      "learning_rate": 9.832377057262015e-06,
      "loss": 0.479,
      "step": 1429
    },
    {
      "epoch": 0.5240014657383657,
      "grad_norm": 0.3216667435476645,
      "learning_rate": 9.831829228778917e-06,
      "loss": 0.4803,
      "step": 1430
    },
    {
      "epoch": 0.5243679003297911,
      "grad_norm": 0.387676195086693,
      "learning_rate": 9.831280521855866e-06,
      "loss": 0.4659,
      "step": 1431
    },
    {
      "epoch": 0.5247343349212166,
      "grad_norm": 0.36049308696950183,
      "learning_rate": 9.830730936592615e-06,
      "loss": 0.4952,
      "step": 1432
    },
    {
      "epoch": 0.525100769512642,
      "grad_norm": 0.3533970312690624,
      "learning_rate": 9.830180473089082e-06,
      "loss": 0.4933,
      "step": 1433
    },
    {
      "epoch": 0.5254672041040674,
      "grad_norm": 0.340800968071729,
      "learning_rate": 9.829629131445342e-06,
      "loss": 0.4959,
      "step": 1434
    },
    {
      "epoch": 0.5258336386954928,
      "grad_norm": 0.324745998712875,
      "learning_rate": 9.829076911761631e-06,
      "loss": 0.4644,
      "step": 1435
    },
    {
      "epoch": 0.5262000732869183,
      "grad_norm": 0.3399650955026245,
      "learning_rate": 9.828523814138344e-06,
      "loss": 0.4815,
      "step": 1436
    },
    {
      "epoch": 0.5265665078783437,
      "grad_norm": 0.3692959861273313,
      "learning_rate": 9.827969838676035e-06,
      "loss": 0.4932,
      "step": 1437
    },
    {
      "epoch": 0.5269329424697692,
      "grad_norm": 0.3225479561225344,
      "learning_rate": 9.827414985475419e-06,
      "loss": 0.4664,
      "step": 1438
    },
    {
      "epoch": 0.5272993770611946,
      "grad_norm": 0.40961493961960344,
      "learning_rate": 9.826859254637369e-06,
      "loss": 0.4933,
      "step": 1439
    },
    {
      "epoch": 0.52766581165262,
      "grad_norm": 0.3144870603143933,
      "learning_rate": 9.82630264626292e-06,
      "loss": 0.4773,
      "step": 1440
    },
    {
      "epoch": 0.5280322462440454,
      "grad_norm": 0.34932537257690305,
      "learning_rate": 9.825745160453264e-06,
      "loss": 0.4614,
      "step": 1441
    },
    {
      "epoch": 0.5283986808354708,
      "grad_norm": 0.35995609269121176,
      "learning_rate": 9.825186797309754e-06,
      "loss": 0.4935,
      "step": 1442
    },
    {
      "epoch": 0.5287651154268963,
      "grad_norm": 0.40434876613380516,
      "learning_rate": 9.824627556933898e-06,
      "loss": 0.4811,
      "step": 1443
    },
    {
      "epoch": 0.5291315500183217,
      "grad_norm": 0.3211884848768759,
      "learning_rate": 9.824067439427374e-06,
      "loss": 0.4773,
      "step": 1444
    },
    {
      "epoch": 0.5294979846097472,
      "grad_norm": 0.3285416270101821,
      "learning_rate": 9.82350644489201e-06,
      "loss": 0.4596,
      "step": 1445
    },
    {
      "epoch": 0.5298644192011726,
      "grad_norm": 0.36341787178401697,
      "learning_rate": 9.822944573429795e-06,
      "loss": 0.4694,
      "step": 1446
    },
    {
      "epoch": 0.530230853792598,
      "grad_norm": 0.42394153295739634,
      "learning_rate": 9.822381825142879e-06,
      "loss": 0.4924,
      "step": 1447
    },
    {
      "epoch": 0.5305972883840234,
      "grad_norm": 0.3317260942031331,
      "learning_rate": 9.821818200133574e-06,
      "loss": 0.4901,
      "step": 1448
    },
    {
      "epoch": 0.5309637229754489,
      "grad_norm": 0.3932753968999003,
      "learning_rate": 9.821253698504346e-06,
      "loss": 0.5041,
      "step": 1449
    },
    {
      "epoch": 0.5313301575668743,
      "grad_norm": 0.3995097424587736,
      "learning_rate": 9.820688320357823e-06,
      "loss": 0.5028,
      "step": 1450
    },
    {
      "epoch": 0.5316965921582998,
      "grad_norm": 0.3442771148754249,
      "learning_rate": 9.820122065796794e-06,
      "loss": 0.4664,
      "step": 1451
    },
    {
      "epoch": 0.5320630267497252,
      "grad_norm": 0.35646773473362936,
      "learning_rate": 9.819554934924205e-06,
      "loss": 0.4991,
      "step": 1452
    },
    {
      "epoch": 0.5324294613411507,
      "grad_norm": 0.35814549112994354,
      "learning_rate": 9.81898692784316e-06,
      "loss": 0.4921,
      "step": 1453
    },
    {
      "epoch": 0.532795895932576,
      "grad_norm": 0.36302622640026744,
      "learning_rate": 9.818418044656927e-06,
      "loss": 0.4893,
      "step": 1454
    },
    {
      "epoch": 0.5331623305240014,
      "grad_norm": 0.36208457859383386,
      "learning_rate": 9.81784828546893e-06,
      "loss": 0.4633,
      "step": 1455
    },
    {
      "epoch": 0.5335287651154269,
      "grad_norm": 0.3302817372564158,
      "learning_rate": 9.81727765038275e-06,
      "loss": 0.4817,
      "step": 1456
    },
    {
      "epoch": 0.5338951997068523,
      "grad_norm": 0.3438049045926076,
      "learning_rate": 9.816706139502134e-06,
      "loss": 0.486,
      "step": 1457
    },
    {
      "epoch": 0.5342616342982778,
      "grad_norm": 0.3344928407373188,
      "learning_rate": 9.816133752930983e-06,
      "loss": 0.4643,
      "step": 1458
    },
    {
      "epoch": 0.5346280688897032,
      "grad_norm": 0.30617021944209827,
      "learning_rate": 9.815560490773356e-06,
      "loss": 0.4742,
      "step": 1459
    },
    {
      "epoch": 0.5349945034811286,
      "grad_norm": 0.32282009176672644,
      "learning_rate": 9.814986353133477e-06,
      "loss": 0.4815,
      "step": 1460
    },
    {
      "epoch": 0.535360938072554,
      "grad_norm": 0.34387337896025416,
      "learning_rate": 9.814411340115723e-06,
      "loss": 0.4779,
      "step": 1461
    },
    {
      "epoch": 0.5357273726639795,
      "grad_norm": 0.35722546479423223,
      "learning_rate": 9.813835451824636e-06,
      "loss": 0.4564,
      "step": 1462
    },
    {
      "epoch": 0.5360938072554049,
      "grad_norm": 0.3315893374384504,
      "learning_rate": 9.813258688364911e-06,
      "loss": 0.4994,
      "step": 1463
    },
    {
      "epoch": 0.5364602418468304,
      "grad_norm": 0.3219504025004336,
      "learning_rate": 9.812681049841409e-06,
      "loss": 0.4655,
      "step": 1464
    },
    {
      "epoch": 0.5368266764382558,
      "grad_norm": 0.34061752661111266,
      "learning_rate": 9.812102536359142e-06,
      "loss": 0.4508,
      "step": 1465
    },
    {
      "epoch": 0.5371931110296811,
      "grad_norm": 0.3554757923077559,
      "learning_rate": 9.811523148023287e-06,
      "loss": 0.4691,
      "step": 1466
    },
    {
      "epoch": 0.5375595456211066,
      "grad_norm": 0.322486090667523,
      "learning_rate": 9.81094288493918e-06,
      "loss": 0.4764,
      "step": 1467
    },
    {
      "epoch": 0.537925980212532,
      "grad_norm": 0.3363414914051058,
      "learning_rate": 9.810361747212313e-06,
      "loss": 0.4748,
      "step": 1468
    },
    {
      "epoch": 0.5382924148039575,
      "grad_norm": 0.3257785750622576,
      "learning_rate": 9.80977973494834e-06,
      "loss": 0.5147,
      "step": 1469
    },
    {
      "epoch": 0.5386588493953829,
      "grad_norm": 0.3678763545265607,
      "learning_rate": 9.80919684825307e-06,
      "loss": 0.4591,
      "step": 1470
    },
    {
      "epoch": 0.5390252839868084,
      "grad_norm": 0.33920147384629934,
      "learning_rate": 9.808613087232473e-06,
      "loss": 0.4702,
      "step": 1471
    },
    {
      "epoch": 0.5393917185782338,
      "grad_norm": 0.3976997776139131,
      "learning_rate": 9.808028451992681e-06,
      "loss": 0.4778,
      "step": 1472
    },
    {
      "epoch": 0.5397581531696592,
      "grad_norm": 0.36442384221196217,
      "learning_rate": 9.80744294263998e-06,
      "loss": 0.469,
      "step": 1473
    },
    {
      "epoch": 0.5401245877610846,
      "grad_norm": 0.31316244043314456,
      "learning_rate": 9.806856559280819e-06,
      "loss": 0.4562,
      "step": 1474
    },
    {
      "epoch": 0.5404910223525101,
      "grad_norm": 0.35976667457597644,
      "learning_rate": 9.806269302021805e-06,
      "loss": 0.4726,
      "step": 1475
    },
    {
      "epoch": 0.5408574569439355,
      "grad_norm": 0.31604461770248315,
      "learning_rate": 9.8056811709697e-06,
      "loss": 0.4633,
      "step": 1476
    },
    {
      "epoch": 0.541223891535361,
      "grad_norm": 0.32482422740889894,
      "learning_rate": 9.80509216623143e-06,
      "loss": 0.4652,
      "step": 1477
    },
    {
      "epoch": 0.5415903261267864,
      "grad_norm": 0.33388756905682826,
      "learning_rate": 9.804502287914078e-06,
      "loss": 0.4495,
      "step": 1478
    },
    {
      "epoch": 0.5419567607182117,
      "grad_norm": 0.33543750563255886,
      "learning_rate": 9.803911536124882e-06,
      "loss": 0.4745,
      "step": 1479
    },
    {
      "epoch": 0.5423231953096372,
      "grad_norm": 0.32791215836055343,
      "learning_rate": 9.803319910971248e-06,
      "loss": 0.4931,
      "step": 1480
    },
    {
      "epoch": 0.5426896299010626,
      "grad_norm": 0.3175028654968894,
      "learning_rate": 9.80272741256073e-06,
      "loss": 0.4665,
      "step": 1481
    },
    {
      "epoch": 0.5430560644924881,
      "grad_norm": 0.3603893732068546,
      "learning_rate": 9.802134041001048e-06,
      "loss": 0.4786,
      "step": 1482
    },
    {
      "epoch": 0.5434224990839135,
      "grad_norm": 0.33696988481964685,
      "learning_rate": 9.801539796400078e-06,
      "loss": 0.4688,
      "step": 1483
    },
    {
      "epoch": 0.543788933675339,
      "grad_norm": 0.3215033995215954,
      "learning_rate": 9.800944678865857e-06,
      "loss": 0.4604,
      "step": 1484
    },
    {
      "epoch": 0.5441553682667644,
      "grad_norm": 0.3440589297237535,
      "learning_rate": 9.800348688506577e-06,
      "loss": 0.4762,
      "step": 1485
    },
    {
      "epoch": 0.5445218028581899,
      "grad_norm": 0.3092618618193074,
      "learning_rate": 9.799751825430592e-06,
      "loss": 0.457,
      "step": 1486
    },
    {
      "epoch": 0.5448882374496152,
      "grad_norm": 0.34456614007611114,
      "learning_rate": 9.799154089746412e-06,
      "loss": 0.4869,
      "step": 1487
    },
    {
      "epoch": 0.5452546720410407,
      "grad_norm": 0.34262315121097625,
      "learning_rate": 9.798555481562708e-06,
      "loss": 0.4967,
      "step": 1488
    },
    {
      "epoch": 0.5456211066324661,
      "grad_norm": 0.33608943562171195,
      "learning_rate": 9.79795600098831e-06,
      "loss": 0.4909,
      "step": 1489
    },
    {
      "epoch": 0.5459875412238915,
      "grad_norm": 0.3265100682583629,
      "learning_rate": 9.7973556481322e-06,
      "loss": 0.4881,
      "step": 1490
    },
    {
      "epoch": 0.546353975815317,
      "grad_norm": 0.37734218891914356,
      "learning_rate": 9.796754423103531e-06,
      "loss": 0.4945,
      "step": 1491
    },
    {
      "epoch": 0.5467204104067424,
      "grad_norm": 0.31159082275764516,
      "learning_rate": 9.796152326011604e-06,
      "loss": 0.4751,
      "step": 1492
    },
    {
      "epoch": 0.5470868449981678,
      "grad_norm": 0.3025404018668238,
      "learning_rate": 9.79554935696588e-06,
      "loss": 0.4614,
      "step": 1493
    },
    {
      "epoch": 0.5474532795895932,
      "grad_norm": 0.3293488949426941,
      "learning_rate": 9.794945516075982e-06,
      "loss": 0.4864,
      "step": 1494
    },
    {
      "epoch": 0.5478197141810187,
      "grad_norm": 0.33547295287984435,
      "learning_rate": 9.794340803451692e-06,
      "loss": 0.4744,
      "step": 1495
    },
    {
      "epoch": 0.5481861487724441,
      "grad_norm": 0.3745670350245115,
      "learning_rate": 9.793735219202945e-06,
      "loss": 0.4885,
      "step": 1496
    },
    {
      "epoch": 0.5485525833638696,
      "grad_norm": 0.3107770786017927,
      "learning_rate": 9.793128763439841e-06,
      "loss": 0.4967,
      "step": 1497
    },
    {
      "epoch": 0.548919017955295,
      "grad_norm": 0.340188220414171,
      "learning_rate": 9.792521436272633e-06,
      "loss": 0.4647,
      "step": 1498
    },
    {
      "epoch": 0.5492854525467205,
      "grad_norm": 0.3734130097339171,
      "learning_rate": 9.791913237811736e-06,
      "loss": 0.4737,
      "step": 1499
    },
    {
      "epoch": 0.5496518871381458,
      "grad_norm": 0.3310092946665534,
      "learning_rate": 9.79130416816772e-06,
      "loss": 0.4856,
      "step": 1500
    },
    {
      "epoch": 0.5500183217295713,
      "grad_norm": 0.38684256884192914,
      "learning_rate": 9.79069422745132e-06,
      "loss": 0.4921,
      "step": 1501
    },
    {
      "epoch": 0.5503847563209967,
      "grad_norm": 0.37821903666307793,
      "learning_rate": 9.790083415773422e-06,
      "loss": 0.5121,
      "step": 1502
    },
    {
      "epoch": 0.5507511909124221,
      "grad_norm": 0.3052359192332031,
      "learning_rate": 9.789471733245073e-06,
      "loss": 0.4787,
      "step": 1503
    },
    {
      "epoch": 0.5511176255038476,
      "grad_norm": 0.3117852678000023,
      "learning_rate": 9.788859179977478e-06,
      "loss": 0.464,
      "step": 1504
    },
    {
      "epoch": 0.551484060095273,
      "grad_norm": 0.346244494993302,
      "learning_rate": 9.788245756082002e-06,
      "loss": 0.4504,
      "step": 1505
    },
    {
      "epoch": 0.5518504946866984,
      "grad_norm": 0.34792018623243165,
      "learning_rate": 9.787631461670169e-06,
      "loss": 0.4766,
      "step": 1506
    },
    {
      "epoch": 0.5522169292781238,
      "grad_norm": 0.3162910685936409,
      "learning_rate": 9.787016296853657e-06,
      "loss": 0.4536,
      "step": 1507
    },
    {
      "epoch": 0.5525833638695493,
      "grad_norm": 0.36220228617401234,
      "learning_rate": 9.786400261744304e-06,
      "loss": 0.4668,
      "step": 1508
    },
    {
      "epoch": 0.5529497984609747,
      "grad_norm": 0.3524866060270971,
      "learning_rate": 9.785783356454108e-06,
      "loss": 0.4743,
      "step": 1509
    },
    {
      "epoch": 0.5533162330524002,
      "grad_norm": 0.3481955710418153,
      "learning_rate": 9.785165581095225e-06,
      "loss": 0.4653,
      "step": 1510
    },
    {
      "epoch": 0.5536826676438256,
      "grad_norm": 0.28939313660954685,
      "learning_rate": 9.784546935779969e-06,
      "loss": 0.4715,
      "step": 1511
    },
    {
      "epoch": 0.5540491022352511,
      "grad_norm": 0.3691823473039044,
      "learning_rate": 9.783927420620809e-06,
      "loss": 0.4976,
      "step": 1512
    },
    {
      "epoch": 0.5544155368266764,
      "grad_norm": 0.33464755490286585,
      "learning_rate": 9.783307035730375e-06,
      "loss": 0.4783,
      "step": 1513
    },
    {
      "epoch": 0.5547819714181018,
      "grad_norm": 0.3511980469847747,
      "learning_rate": 9.782685781221455e-06,
      "loss": 0.4597,
      "step": 1514
    },
    {
      "epoch": 0.5551484060095273,
      "grad_norm": 0.3305089464101922,
      "learning_rate": 9.782063657206995e-06,
      "loss": 0.4722,
      "step": 1515
    },
    {
      "epoch": 0.5555148406009527,
      "grad_norm": 0.3120066624152192,
      "learning_rate": 9.781440663800099e-06,
      "loss": 0.4592,
      "step": 1516
    },
    {
      "epoch": 0.5558812751923782,
      "grad_norm": 0.35199503516060054,
      "learning_rate": 9.780816801114029e-06,
      "loss": 0.4815,
      "step": 1517
    },
    {
      "epoch": 0.5562477097838036,
      "grad_norm": 0.37679420160259847,
      "learning_rate": 9.780192069262204e-06,
      "loss": 0.4943,
      "step": 1518
    },
    {
      "epoch": 0.556614144375229,
      "grad_norm": 0.3539407472252221,
      "learning_rate": 9.779566468358205e-06,
      "loss": 0.4462,
      "step": 1519
    },
    {
      "epoch": 0.5569805789666544,
      "grad_norm": 0.3067693058650467,
      "learning_rate": 9.778939998515764e-06,
      "loss": 0.4583,
      "step": 1520
    },
    {
      "epoch": 0.5573470135580799,
      "grad_norm": 0.3455374178050988,
      "learning_rate": 9.778312659848776e-06,
      "loss": 0.4689,
      "step": 1521
    },
    {
      "epoch": 0.5577134481495053,
      "grad_norm": 0.3317073915085135,
      "learning_rate": 9.777684452471296e-06,
      "loss": 0.4729,
      "step": 1522
    },
    {
      "epoch": 0.5580798827409308,
      "grad_norm": 0.322483499769926,
      "learning_rate": 9.77705537649753e-06,
      "loss": 0.4415,
      "step": 1523
    },
    {
      "epoch": 0.5584463173323562,
      "grad_norm": 0.33981185224376326,
      "learning_rate": 9.776425432041847e-06,
      "loss": 0.4638,
      "step": 1524
    },
    {
      "epoch": 0.5588127519237817,
      "grad_norm": 0.3180445026756105,
      "learning_rate": 9.775794619218773e-06,
      "loss": 0.4761,
      "step": 1525
    },
    {
      "epoch": 0.559179186515207,
      "grad_norm": 0.3458391029305091,
      "learning_rate": 9.775162938142991e-06,
      "loss": 0.4879,
      "step": 1526
    },
    {
      "epoch": 0.5595456211066324,
      "grad_norm": 0.32454133649935557,
      "learning_rate": 9.774530388929343e-06,
      "loss": 0.4628,
      "step": 1527
    },
    {
      "epoch": 0.5599120556980579,
      "grad_norm": 0.33151018923438935,
      "learning_rate": 9.77389697169283e-06,
      "loss": 0.4815,
      "step": 1528
    },
    {
      "epoch": 0.5602784902894833,
      "grad_norm": 0.4017160512617154,
      "learning_rate": 9.773262686548604e-06,
      "loss": 0.4708,
      "step": 1529
    },
    {
      "epoch": 0.5606449248809088,
      "grad_norm": 0.36010405104783244,
      "learning_rate": 9.772627533611982e-06,
      "loss": 0.4734,
      "step": 1530
    },
    {
      "epoch": 0.5610113594723342,
      "grad_norm": 0.3334963959551884,
      "learning_rate": 9.77199151299844e-06,
      "loss": 0.446,
      "step": 1531
    },
    {
      "epoch": 0.5613777940637596,
      "grad_norm": 0.33454273732973205,
      "learning_rate": 9.771354624823605e-06,
      "loss": 0.4983,
      "step": 1532
    },
    {
      "epoch": 0.561744228655185,
      "grad_norm": 0.3176673362718567,
      "learning_rate": 9.770716869203265e-06,
      "loss": 0.4558,
      "step": 1533
    },
    {
      "epoch": 0.5621106632466105,
      "grad_norm": 0.3188713912169947,
      "learning_rate": 9.770078246253367e-06,
      "loss": 0.4778,
      "step": 1534
    },
    {
      "epoch": 0.5624770978380359,
      "grad_norm": 0.3540731123816726,
      "learning_rate": 9.769438756090012e-06,
      "loss": 0.4623,
      "step": 1535
    },
    {
      "epoch": 0.5628435324294614,
      "grad_norm": 0.328734194005767,
      "learning_rate": 9.768798398829466e-06,
      "loss": 0.479,
      "step": 1536
    },
    {
      "epoch": 0.5632099670208868,
      "grad_norm": 0.30324124504925704,
      "learning_rate": 9.768157174588144e-06,
      "loss": 0.4841,
      "step": 1537
    },
    {
      "epoch": 0.5635764016123122,
      "grad_norm": 0.4029062978867119,
      "learning_rate": 9.767515083482622e-06,
      "loss": 0.4758,
      "step": 1538
    },
    {
      "epoch": 0.5639428362037376,
      "grad_norm": 0.3964843623644931,
      "learning_rate": 9.766872125629634e-06,
      "loss": 0.4863,
      "step": 1539
    },
    {
      "epoch": 0.564309270795163,
      "grad_norm": 0.3336997405853562,
      "learning_rate": 9.766228301146074e-06,
      "loss": 0.4676,
      "step": 1540
    },
    {
      "epoch": 0.5646757053865885,
      "grad_norm": 0.32198687306082846,
      "learning_rate": 9.76558361014899e-06,
      "loss": 0.4854,
      "step": 1541
    },
    {
      "epoch": 0.5650421399780139,
      "grad_norm": 0.3625551889672219,
      "learning_rate": 9.764938052755586e-06,
      "loss": 0.5018,
      "step": 1542
    },
    {
      "epoch": 0.5654085745694394,
      "grad_norm": 0.2999427004034081,
      "learning_rate": 9.76429162908323e-06,
      "loss": 0.4736,
      "step": 1543
    },
    {
      "epoch": 0.5657750091608648,
      "grad_norm": 0.3423832414581499,
      "learning_rate": 9.763644339249441e-06,
      "loss": 0.4788,
      "step": 1544
    },
    {
      "epoch": 0.5661414437522903,
      "grad_norm": 0.33452938223614603,
      "learning_rate": 9.7629961833719e-06,
      "loss": 0.4823,
      "step": 1545
    },
    {
      "epoch": 0.5665078783437156,
      "grad_norm": 0.31364299615066155,
      "learning_rate": 9.762347161568441e-06,
      "loss": 0.4788,
      "step": 1546
    },
    {
      "epoch": 0.5668743129351411,
      "grad_norm": 0.4412575370901754,
      "learning_rate": 9.76169727395706e-06,
      "loss": 0.4639,
      "step": 1547
    },
    {
      "epoch": 0.5672407475265665,
      "grad_norm": 0.3052797302280054,
      "learning_rate": 9.761046520655907e-06,
      "loss": 0.49,
      "step": 1548
    },
    {
      "epoch": 0.567607182117992,
      "grad_norm": 0.35124677300301144,
      "learning_rate": 9.760394901783294e-06,
      "loss": 0.4588,
      "step": 1549
    },
    {
      "epoch": 0.5679736167094174,
      "grad_norm": 0.4578722856103231,
      "learning_rate": 9.759742417457683e-06,
      "loss": 0.4606,
      "step": 1550
    },
    {
      "epoch": 0.5683400513008428,
      "grad_norm": 0.3109924811679939,
      "learning_rate": 9.7590890677977e-06,
      "loss": 0.4611,
      "step": 1551
    },
    {
      "epoch": 0.5687064858922682,
      "grad_norm": 0.325401322326984,
      "learning_rate": 9.758434852922124e-06,
      "loss": 0.4792,
      "step": 1552
    },
    {
      "epoch": 0.5690729204836936,
      "grad_norm": 0.34755520811198803,
      "learning_rate": 9.757779772949895e-06,
      "loss": 0.4698,
      "step": 1553
    },
    {
      "epoch": 0.5694393550751191,
      "grad_norm": 0.28607959910021,
      "learning_rate": 9.757123828000106e-06,
      "loss": 0.4594,
      "step": 1554
    },
    {
      "epoch": 0.5698057896665445,
      "grad_norm": 0.32572181925370347,
      "learning_rate": 9.756467018192013e-06,
      "loss": 0.4574,
      "step": 1555
    },
    {
      "epoch": 0.57017222425797,
      "grad_norm": 0.32714197008097834,
      "learning_rate": 9.755809343645021e-06,
      "loss": 0.4854,
      "step": 1556
    },
    {
      "epoch": 0.5705386588493954,
      "grad_norm": 0.3394595554677765,
      "learning_rate": 9.755150804478704e-06,
      "loss": 0.4908,
      "step": 1557
    },
    {
      "epoch": 0.5709050934408209,
      "grad_norm": 0.3339268608802838,
      "learning_rate": 9.754491400812779e-06,
      "loss": 0.4743,
      "step": 1558
    },
    {
      "epoch": 0.5712715280322462,
      "grad_norm": 0.35339941481087195,
      "learning_rate": 9.75383113276713e-06,
      "loss": 0.4814,
      "step": 1559
    },
    {
      "epoch": 0.5716379626236717,
      "grad_norm": 0.34837209886783954,
      "learning_rate": 9.753170000461799e-06,
      "loss": 0.4685,
      "step": 1560
    },
    {
      "epoch": 0.5720043972150971,
      "grad_norm": 0.3959239352547852,
      "learning_rate": 9.752508004016976e-06,
      "loss": 0.484,
      "step": 1561
    },
    {
      "epoch": 0.5723708318065225,
      "grad_norm": 0.34465422827765646,
      "learning_rate": 9.751845143553018e-06,
      "loss": 0.486,
      "step": 1562
    },
    {
      "epoch": 0.572737266397948,
      "grad_norm": 0.3013010145047974,
      "learning_rate": 9.751181419190435e-06,
      "loss": 0.4886,
      "step": 1563
    },
    {
      "epoch": 0.5731037009893734,
      "grad_norm": 0.3038750608645962,
      "learning_rate": 9.75051683104989e-06,
      "loss": 0.4329,
      "step": 1564
    },
    {
      "epoch": 0.5734701355807988,
      "grad_norm": 0.3637126302430395,
      "learning_rate": 9.74985137925221e-06,
      "loss": 0.4674,
      "step": 1565
    },
    {
      "epoch": 0.5738365701722242,
      "grad_norm": 0.3274367979021695,
      "learning_rate": 9.749185063918375e-06,
      "loss": 0.4841,
      "step": 1566
    },
    {
      "epoch": 0.5742030047636497,
      "grad_norm": 0.38639152534347115,
      "learning_rate": 9.748517885169525e-06,
      "loss": 0.4829,
      "step": 1567
    },
    {
      "epoch": 0.5745694393550751,
      "grad_norm": 0.3242767201659219,
      "learning_rate": 9.74784984312695e-06,
      "loss": 0.445,
      "step": 1568
    },
    {
      "epoch": 0.5749358739465006,
      "grad_norm": 0.3627110602038673,
      "learning_rate": 9.747180937912109e-06,
      "loss": 0.4714,
      "step": 1569
    },
    {
      "epoch": 0.575302308537926,
      "grad_norm": 0.30889099162928774,
      "learning_rate": 9.746511169646604e-06,
      "loss": 0.4749,
      "step": 1570
    },
    {
      "epoch": 0.5756687431293515,
      "grad_norm": 0.3476997696414378,
      "learning_rate": 9.745840538452204e-06,
      "loss": 0.4572,
      "step": 1571
    },
    {
      "epoch": 0.5760351777207768,
      "grad_norm": 0.36800644146155564,
      "learning_rate": 9.745169044450834e-06,
      "loss": 0.4848,
      "step": 1572
    },
    {
      "epoch": 0.5764016123122023,
      "grad_norm": 0.32504734850670763,
      "learning_rate": 9.744496687764568e-06,
      "loss": 0.4669,
      "step": 1573
    },
    {
      "epoch": 0.5767680469036277,
      "grad_norm": 0.3132229444891331,
      "learning_rate": 9.743823468515647e-06,
      "loss": 0.4518,
      "step": 1574
    },
    {
      "epoch": 0.5771344814950531,
      "grad_norm": 0.3821369455078288,
      "learning_rate": 9.743149386826461e-06,
      "loss": 0.4749,
      "step": 1575
    },
    {
      "epoch": 0.5775009160864786,
      "grad_norm": 0.40049549598956263,
      "learning_rate": 9.742474442819561e-06,
      "loss": 0.5176,
      "step": 1576
    },
    {
      "epoch": 0.577867350677904,
      "grad_norm": 0.32994783322717924,
      "learning_rate": 9.741798636617654e-06,
      "loss": 0.4775,
      "step": 1577
    },
    {
      "epoch": 0.5782337852693294,
      "grad_norm": 0.3286963939407201,
      "learning_rate": 9.741121968343605e-06,
      "loss": 0.4883,
      "step": 1578
    },
    {
      "epoch": 0.5786002198607548,
      "grad_norm": 0.4098500139537914,
      "learning_rate": 9.74044443812043e-06,
      "loss": 0.4755,
      "step": 1579
    },
    {
      "epoch": 0.5789666544521803,
      "grad_norm": 0.3330280014529633,
      "learning_rate": 9.739766046071311e-06,
      "loss": 0.4682,
      "step": 1580
    },
    {
      "epoch": 0.5793330890436057,
      "grad_norm": 0.379369609854492,
      "learning_rate": 9.739086792319577e-06,
      "loss": 0.4764,
      "step": 1581
    },
    {
      "epoch": 0.5796995236350312,
      "grad_norm": 0.31667737702179444,
      "learning_rate": 9.738406676988721e-06,
      "loss": 0.4779,
      "step": 1582
    },
    {
      "epoch": 0.5800659582264566,
      "grad_norm": 0.336981781926578,
      "learning_rate": 9.737725700202388e-06,
      "loss": 0.4802,
      "step": 1583
    },
    {
      "epoch": 0.5804323928178821,
      "grad_norm": 0.32914347080100426,
      "learning_rate": 9.737043862084383e-06,
      "loss": 0.4852,
      "step": 1584
    },
    {
      "epoch": 0.5807988274093074,
      "grad_norm": 0.31406672781745154,
      "learning_rate": 9.736361162758665e-06,
      "loss": 0.4822,
      "step": 1585
    },
    {
      "epoch": 0.5811652620007328,
      "grad_norm": 0.5428379102104804,
      "learning_rate": 9.735677602349352e-06,
      "loss": 0.4792,
      "step": 1586
    },
    {
      "epoch": 0.5815316965921583,
      "grad_norm": 0.3561708419766296,
      "learning_rate": 9.734993180980715e-06,
      "loss": 0.4489,
      "step": 1587
    },
    {
      "epoch": 0.5818981311835837,
      "grad_norm": 0.31710152434676864,
      "learning_rate": 9.734307898777187e-06,
      "loss": 0.4655,
      "step": 1588
    },
    {
      "epoch": 0.5822645657750092,
      "grad_norm": 0.3902881546893,
      "learning_rate": 9.733621755863349e-06,
      "loss": 0.4528,
      "step": 1589
    },
    {
      "epoch": 0.5826310003664346,
      "grad_norm": 0.34037343631928796,
      "learning_rate": 9.73293475236395e-06,
      "loss": 0.4774,
      "step": 1590
    },
    {
      "epoch": 0.58299743495786,
      "grad_norm": 0.38556538289734893,
      "learning_rate": 9.732246888403885e-06,
      "loss": 0.4911,
      "step": 1591
    },
    {
      "epoch": 0.5833638695492854,
      "grad_norm": 0.28728541801123847,
      "learning_rate": 9.73155816410821e-06,
      "loss": 0.4592,
      "step": 1592
    },
    {
      "epoch": 0.5837303041407109,
      "grad_norm": 0.3073137903119383,
      "learning_rate": 9.730868579602138e-06,
      "loss": 0.4901,
      "step": 1593
    },
    {
      "epoch": 0.5840967387321363,
      "grad_norm": 0.3416779698337659,
      "learning_rate": 9.730178135011036e-06,
      "loss": 0.4644,
      "step": 1594
    },
    {
      "epoch": 0.5844631733235618,
      "grad_norm": 0.3598605822779923,
      "learning_rate": 9.729486830460433e-06,
      "loss": 0.4495,
      "step": 1595
    },
    {
      "epoch": 0.5848296079149872,
      "grad_norm": 0.3610329511714763,
      "learning_rate": 9.728794666076004e-06,
      "loss": 0.4927,
      "step": 1596
    },
    {
      "epoch": 0.5851960425064126,
      "grad_norm": 0.35077379628825867,
      "learning_rate": 9.72810164198359e-06,
      "loss": 0.4635,
      "step": 1597
    },
    {
      "epoch": 0.585562477097838,
      "grad_norm": 0.34413400987063736,
      "learning_rate": 9.727407758309184e-06,
      "loss": 0.461,
      "step": 1598
    },
    {
      "epoch": 0.5859289116892634,
      "grad_norm": 0.3672968606532333,
      "learning_rate": 9.726713015178934e-06,
      "loss": 0.4608,
      "step": 1599
    },
    {
      "epoch": 0.5862953462806889,
      "grad_norm": 1.143864842967924,
      "learning_rate": 9.726017412719151e-06,
      "loss": 0.4912,
      "step": 1600
    },
    {
      "epoch": 0.5866617808721143,
      "grad_norm": 0.3172043720703272,
      "learning_rate": 9.725320951056292e-06,
      "loss": 0.4724,
      "step": 1601
    },
    {
      "epoch": 0.5870282154635398,
      "grad_norm": 0.3745452270940885,
      "learning_rate": 9.72462363031698e-06,
      "loss": 0.4801,
      "step": 1602
    },
    {
      "epoch": 0.5873946500549652,
      "grad_norm": 0.3558927257873667,
      "learning_rate": 9.723925450627988e-06,
      "loss": 0.4951,
      "step": 1603
    },
    {
      "epoch": 0.5877610846463907,
      "grad_norm": 0.3262572897366275,
      "learning_rate": 9.723226412116247e-06,
      "loss": 0.4707,
      "step": 1604
    },
    {
      "epoch": 0.588127519237816,
      "grad_norm": 0.34564481060441543,
      "learning_rate": 9.722526514908843e-06,
      "loss": 0.4778,
      "step": 1605
    },
    {
      "epoch": 0.5884939538292415,
      "grad_norm": 0.310235834257553,
      "learning_rate": 9.721825759133022e-06,
      "loss": 0.4571,
      "step": 1606
    },
    {
      "epoch": 0.5888603884206669,
      "grad_norm": 0.31798293518563736,
      "learning_rate": 9.721124144916182e-06,
      "loss": 0.4778,
      "step": 1607
    },
    {
      "epoch": 0.5892268230120924,
      "grad_norm": 0.3241005325576626,
      "learning_rate": 9.720421672385875e-06,
      "loss": 0.4751,
      "step": 1608
    },
    {
      "epoch": 0.5895932576035178,
      "grad_norm": 0.323645765316664,
      "learning_rate": 9.71971834166982e-06,
      "loss": 0.4938,
      "step": 1609
    },
    {
      "epoch": 0.5899596921949432,
      "grad_norm": 0.32561787952342863,
      "learning_rate": 9.719014152895879e-06,
      "loss": 0.4839,
      "step": 1610
    },
    {
      "epoch": 0.5903261267863686,
      "grad_norm": 0.39252783841609945,
      "learning_rate": 9.718309106192074e-06,
      "loss": 0.4813,
      "step": 1611
    },
    {
      "epoch": 0.590692561377794,
      "grad_norm": 0.3468524121071481,
      "learning_rate": 9.71760320168659e-06,
      "loss": 0.4597,
      "step": 1612
    },
    {
      "epoch": 0.5910589959692195,
      "grad_norm": 0.3445494013593732,
      "learning_rate": 9.716896439507759e-06,
      "loss": 0.462,
      "step": 1613
    },
    {
      "epoch": 0.5914254305606449,
      "grad_norm": 0.3409851115960381,
      "learning_rate": 9.716188819784071e-06,
      "loss": 0.4776,
      "step": 1614
    },
    {
      "epoch": 0.5917918651520704,
      "grad_norm": 0.34129550633678774,
      "learning_rate": 9.715480342644177e-06,
      "loss": 0.4944,
      "step": 1615
    },
    {
      "epoch": 0.5921582997434958,
      "grad_norm": 0.39109292477241886,
      "learning_rate": 9.714771008216876e-06,
      "loss": 0.4823,
      "step": 1616
    },
    {
      "epoch": 0.5925247343349213,
      "grad_norm": 0.2990594341024583,
      "learning_rate": 9.714060816631132e-06,
      "loss": 0.467,
      "step": 1617
    },
    {
      "epoch": 0.5928911689263466,
      "grad_norm": 0.3256573375389477,
      "learning_rate": 9.713349768016055e-06,
      "loss": 0.4813,
      "step": 1618
    },
    {
      "epoch": 0.5932576035177721,
      "grad_norm": 0.3562052691956579,
      "learning_rate": 9.712637862500918e-06,
      "loss": 0.4766,
      "step": 1619
    },
    {
      "epoch": 0.5936240381091975,
      "grad_norm": 0.3169981827214232,
      "learning_rate": 9.711925100215147e-06,
      "loss": 0.4525,
      "step": 1620
    },
    {
      "epoch": 0.5939904727006229,
      "grad_norm": 0.3089837089068,
      "learning_rate": 9.711211481288323e-06,
      "loss": 0.4367,
      "step": 1621
    },
    {
      "epoch": 0.5943569072920484,
      "grad_norm": 0.44735205929318056,
      "learning_rate": 9.710497005850185e-06,
      "loss": 0.478,
      "step": 1622
    },
    {
      "epoch": 0.5947233418834738,
      "grad_norm": 0.30722691935567736,
      "learning_rate": 9.709781674030628e-06,
      "loss": 0.4683,
      "step": 1623
    },
    {
      "epoch": 0.5950897764748992,
      "grad_norm": 0.3279764363991997,
      "learning_rate": 9.7090654859597e-06,
      "loss": 0.472,
      "step": 1624
    },
    {
      "epoch": 0.5954562110663246,
      "grad_norm": 0.34494697562994237,
      "learning_rate": 9.708348441767606e-06,
      "loss": 0.4729,
      "step": 1625
    },
    {
      "epoch": 0.5958226456577501,
      "grad_norm": 0.357006470705126,
      "learning_rate": 9.707630541584705e-06,
      "loss": 0.4779,
      "step": 1626
    },
    {
      "epoch": 0.5961890802491755,
      "grad_norm": 0.3213369959982843,
      "learning_rate": 9.706911785541515e-06,
      "loss": 0.448,
      "step": 1627
    },
    {
      "epoch": 0.596555514840601,
      "grad_norm": 0.33979283513125325,
      "learning_rate": 9.706192173768708e-06,
      "loss": 0.4713,
      "step": 1628
    },
    {
      "epoch": 0.5969219494320264,
      "grad_norm": 0.3906098593861048,
      "learning_rate": 9.705471706397112e-06,
      "loss": 0.4892,
      "step": 1629
    },
    {
      "epoch": 0.5972883840234519,
      "grad_norm": 0.303730348593113,
      "learning_rate": 9.704750383557707e-06,
      "loss": 0.472,
      "step": 1630
    },
    {
      "epoch": 0.5976548186148772,
      "grad_norm": 0.33263583974801997,
      "learning_rate": 9.704028205381634e-06,
      "loss": 0.4643,
      "step": 1631
    },
    {
      "epoch": 0.5980212532063027,
      "grad_norm": 0.342026022924897,
      "learning_rate": 9.703305172000187e-06,
      "loss": 0.4942,
      "step": 1632
    },
    {
      "epoch": 0.5983876877977281,
      "grad_norm": 0.40131948255928274,
      "learning_rate": 9.702581283544813e-06,
      "loss": 0.4841,
      "step": 1633
    },
    {
      "epoch": 0.5987541223891535,
      "grad_norm": 0.3428047891930018,
      "learning_rate": 9.701856540147123e-06,
      "loss": 0.4707,
      "step": 1634
    },
    {
      "epoch": 0.599120556980579,
      "grad_norm": 0.3032419170530265,
      "learning_rate": 9.70113094193887e-06,
      "loss": 0.4683,
      "step": 1635
    },
    {
      "epoch": 0.5994869915720044,
      "grad_norm": 0.33068242014388377,
      "learning_rate": 9.700404489051974e-06,
      "loss": 0.4759,
      "step": 1636
    },
    {
      "epoch": 0.5998534261634298,
      "grad_norm": 0.31861193773013397,
      "learning_rate": 9.699677181618506e-06,
      "loss": 0.4622,
      "step": 1637
    },
    {
      "epoch": 0.6002198607548552,
      "grad_norm": 0.33493351700507645,
      "learning_rate": 9.69894901977069e-06,
      "loss": 0.4825,
      "step": 1638
    },
    {
      "epoch": 0.6005862953462807,
      "grad_norm": 0.31451865503560866,
      "learning_rate": 9.69822000364091e-06,
      "loss": 0.4422,
      "step": 1639
    },
    {
      "epoch": 0.6009527299377061,
      "grad_norm": 0.31857203727991945,
      "learning_rate": 9.697490133361703e-06,
      "loss": 0.4904,
      "step": 1640
    },
    {
      "epoch": 0.6013191645291316,
      "grad_norm": 0.28866763479192276,
      "learning_rate": 9.696759409065759e-06,
      "loss": 0.4535,
      "step": 1641
    },
    {
      "epoch": 0.601685599120557,
      "grad_norm": 0.33155926088170257,
      "learning_rate": 9.69602783088593e-06,
      "loss": 0.4824,
      "step": 1642
    },
    {
      "epoch": 0.6020520337119825,
      "grad_norm": 0.3213654233318701,
      "learning_rate": 9.695295398955215e-06,
      "loss": 0.4739,
      "step": 1643
    },
    {
      "epoch": 0.6024184683034078,
      "grad_norm": 0.34527795659158467,
      "learning_rate": 9.694562113406776e-06,
      "loss": 0.4619,
      "step": 1644
    },
    {
      "epoch": 0.6027849028948332,
      "grad_norm": 0.3197121854682555,
      "learning_rate": 9.693827974373925e-06,
      "loss": 0.4898,
      "step": 1645
    },
    {
      "epoch": 0.6031513374862587,
      "grad_norm": 0.3495937581403054,
      "learning_rate": 9.693092981990126e-06,
      "loss": 0.4981,
      "step": 1646
    },
    {
      "epoch": 0.6035177720776841,
      "grad_norm": 0.31624165800041776,
      "learning_rate": 9.69235713638901e-06,
      "loss": 0.4779,
      "step": 1647
    },
    {
      "epoch": 0.6038842066691096,
      "grad_norm": 0.3071581033421603,
      "learning_rate": 9.69162043770435e-06,
      "loss": 0.4655,
      "step": 1648
    },
    {
      "epoch": 0.604250641260535,
      "grad_norm": 0.33839029777726315,
      "learning_rate": 9.690882886070083e-06,
      "loss": 0.4837,
      "step": 1649
    },
    {
      "epoch": 0.6046170758519605,
      "grad_norm": 0.2957953548717626,
      "learning_rate": 9.690144481620295e-06,
      "loss": 0.464,
      "step": 1650
    },
    {
      "epoch": 0.6049835104433858,
      "grad_norm": 0.35346484026632125,
      "learning_rate": 9.689405224489235e-06,
      "loss": 0.4656,
      "step": 1651
    },
    {
      "epoch": 0.6053499450348113,
      "grad_norm": 0.6018454475439221,
      "learning_rate": 9.688665114811298e-06,
      "loss": 0.4986,
      "step": 1652
    },
    {
      "epoch": 0.6057163796262367,
      "grad_norm": 0.3391792721197751,
      "learning_rate": 9.687924152721038e-06,
      "loss": 0.4559,
      "step": 1653
    },
    {
      "epoch": 0.6060828142176622,
      "grad_norm": 0.3211915617229444,
      "learning_rate": 9.687182338353166e-06,
      "loss": 0.46,
      "step": 1654
    },
    {
      "epoch": 0.6064492488090876,
      "grad_norm": 0.3528964250328152,
      "learning_rate": 9.686439671842543e-06,
      "loss": 0.4843,
      "step": 1655
    },
    {
      "epoch": 0.6068156834005131,
      "grad_norm": 0.3374957655590205,
      "learning_rate": 9.685696153324192e-06,
      "loss": 0.4846,
      "step": 1656
    },
    {
      "epoch": 0.6071821179919384,
      "grad_norm": 0.32746512619377705,
      "learning_rate": 9.684951782933282e-06,
      "loss": 0.4701,
      "step": 1657
    },
    {
      "epoch": 0.6075485525833638,
      "grad_norm": 0.2954707329210968,
      "learning_rate": 9.684206560805145e-06,
      "loss": 0.4802,
      "step": 1658
    },
    {
      "epoch": 0.6079149871747893,
      "grad_norm": 0.32410565253886886,
      "learning_rate": 9.683460487075263e-06,
      "loss": 0.4714,
      "step": 1659
    },
    {
      "epoch": 0.6082814217662147,
      "grad_norm": 0.2955418638005018,
      "learning_rate": 9.682713561879275e-06,
      "loss": 0.4712,
      "step": 1660
    },
    {
      "epoch": 0.6086478563576402,
      "grad_norm": 0.3065021634819279,
      "learning_rate": 9.681965785352974e-06,
      "loss": 0.4739,
      "step": 1661
    },
    {
      "epoch": 0.6090142909490656,
      "grad_norm": 0.35178236451461414,
      "learning_rate": 9.681217157632308e-06,
      "loss": 0.4692,
      "step": 1662
    },
    {
      "epoch": 0.609380725540491,
      "grad_norm": 0.36444782665054043,
      "learning_rate": 9.680467678853378e-06,
      "loss": 0.4757,
      "step": 1663
    },
    {
      "epoch": 0.6097471601319164,
      "grad_norm": 0.34204378590197365,
      "learning_rate": 9.679717349152444e-06,
      "loss": 0.5005,
      "step": 1664
    },
    {
      "epoch": 0.6101135947233419,
      "grad_norm": 0.34066861246810093,
      "learning_rate": 9.678966168665915e-06,
      "loss": 0.4704,
      "step": 1665
    },
    {
      "epoch": 0.6104800293147673,
      "grad_norm": 0.3051829761732453,
      "learning_rate": 9.678214137530361e-06,
      "loss": 0.4652,
      "step": 1666
    },
    {
      "epoch": 0.6108464639061928,
      "grad_norm": 0.35924224483972805,
      "learning_rate": 9.6774612558825e-06,
      "loss": 0.4636,
      "step": 1667
    },
    {
      "epoch": 0.6112128984976182,
      "grad_norm": 0.37944092581330824,
      "learning_rate": 9.67670752385921e-06,
      "loss": 0.475,
      "step": 1668
    },
    {
      "epoch": 0.6115793330890436,
      "grad_norm": 0.3022127247017098,
      "learning_rate": 9.675952941597522e-06,
      "loss": 0.4675,
      "step": 1669
    },
    {
      "epoch": 0.611945767680469,
      "grad_norm": 0.2971984951532198,
      "learning_rate": 9.675197509234619e-06,
      "loss": 0.4704,
      "step": 1670
    },
    {
      "epoch": 0.6123122022718944,
      "grad_norm": 0.2948680634289387,
      "learning_rate": 9.674441226907842e-06,
      "loss": 0.4728,
      "step": 1671
    },
    {
      "epoch": 0.6126786368633199,
      "grad_norm": 0.3253002343723525,
      "learning_rate": 9.673684094754686e-06,
      "loss": 0.4597,
      "step": 1672
    },
    {
      "epoch": 0.6130450714547453,
      "grad_norm": 0.3382744956752839,
      "learning_rate": 9.6729261129128e-06,
      "loss": 0.4941,
      "step": 1673
    },
    {
      "epoch": 0.6134115060461708,
      "grad_norm": 0.7207546341774002,
      "learning_rate": 9.672167281519984e-06,
      "loss": 0.4833,
      "step": 1674
    },
    {
      "epoch": 0.6137779406375962,
      "grad_norm": 0.3408969916386793,
      "learning_rate": 9.671407600714197e-06,
      "loss": 0.4532,
      "step": 1675
    },
    {
      "epoch": 0.6141443752290217,
      "grad_norm": 0.3443734685014367,
      "learning_rate": 9.670647070633554e-06,
      "loss": 0.4791,
      "step": 1676
    },
    {
      "epoch": 0.614510809820447,
      "grad_norm": 0.34325486985609893,
      "learning_rate": 9.669885691416318e-06,
      "loss": 0.4749,
      "step": 1677
    },
    {
      "epoch": 0.6148772444118725,
      "grad_norm": 0.3482662514319337,
      "learning_rate": 9.669123463200914e-06,
      "loss": 0.4843,
      "step": 1678
    },
    {
      "epoch": 0.6152436790032979,
      "grad_norm": 0.3203080018798903,
      "learning_rate": 9.668360386125911e-06,
      "loss": 0.447,
      "step": 1679
    },
    {
      "epoch": 0.6156101135947234,
      "grad_norm": 0.3378302656807861,
      "learning_rate": 9.667596460330045e-06,
      "loss": 0.4775,
      "step": 1680
    },
    {
      "epoch": 0.6159765481861488,
      "grad_norm": 0.32128558873787194,
      "learning_rate": 9.666831685952194e-06,
      "loss": 0.4617,
      "step": 1681
    },
    {
      "epoch": 0.6163429827775742,
      "grad_norm": 0.31569355594556603,
      "learning_rate": 9.6660660631314e-06,
      "loss": 0.4643,
      "step": 1682
    },
    {
      "epoch": 0.6167094173689996,
      "grad_norm": 0.35276081949232563,
      "learning_rate": 9.665299592006854e-06,
      "loss": 0.4806,
      "step": 1683
    },
    {
      "epoch": 0.617075851960425,
      "grad_norm": 0.3299689907624397,
      "learning_rate": 9.664532272717902e-06,
      "loss": 0.4698,
      "step": 1684
    },
    {
      "epoch": 0.6174422865518505,
      "grad_norm": 0.3032861625559838,
      "learning_rate": 9.663764105404047e-06,
      "loss": 0.4489,
      "step": 1685
    },
    {
      "epoch": 0.6178087211432759,
      "grad_norm": 0.6236114574464505,
      "learning_rate": 9.662995090204942e-06,
      "loss": 0.4725,
      "step": 1686
    },
    {
      "epoch": 0.6181751557347014,
      "grad_norm": 0.3375843070337673,
      "learning_rate": 9.662225227260395e-06,
      "loss": 0.472,
      "step": 1687
    },
    {
      "epoch": 0.6185415903261268,
      "grad_norm": 0.34681644628434216,
      "learning_rate": 9.661454516710374e-06,
      "loss": 0.4695,
      "step": 1688
    },
    {
      "epoch": 0.6189080249175523,
      "grad_norm": 0.32714354618075964,
      "learning_rate": 9.66068295869499e-06,
      "loss": 0.4878,
      "step": 1689
    },
    {
      "epoch": 0.6192744595089776,
      "grad_norm": 0.3277708120165392,
      "learning_rate": 9.65991055335452e-06,
      "loss": 0.4759,
      "step": 1690
    },
    {
      "epoch": 0.6196408941004031,
      "grad_norm": 0.3152551628509673,
      "learning_rate": 9.659137300829384e-06,
      "loss": 0.4892,
      "step": 1691
    },
    {
      "epoch": 0.6200073286918285,
      "grad_norm": 0.3818398794482309,
      "learning_rate": 9.658363201260167e-06,
      "loss": 0.4823,
      "step": 1692
    },
    {
      "epoch": 0.6203737632832539,
      "grad_norm": 0.32941106098636397,
      "learning_rate": 9.657588254787598e-06,
      "loss": 0.4867,
      "step": 1693
    },
    {
      "epoch": 0.6207401978746794,
      "grad_norm": 0.3208284770017711,
      "learning_rate": 9.656812461552568e-06,
      "loss": 0.473,
      "step": 1694
    },
    {
      "epoch": 0.6211066324661048,
      "grad_norm": 0.33284401964453264,
      "learning_rate": 9.656035821696118e-06,
      "loss": 0.4641,
      "step": 1695
    },
    {
      "epoch": 0.6214730670575302,
      "grad_norm": 0.33401240781070024,
      "learning_rate": 9.655258335359438e-06,
      "loss": 0.4619,
      "step": 1696
    },
    {
      "epoch": 0.6218395016489556,
      "grad_norm": 0.3471136423889752,
      "learning_rate": 9.654480002683884e-06,
      "loss": 0.4757,
      "step": 1697
    },
    {
      "epoch": 0.6222059362403811,
      "grad_norm": 0.35074976664112023,
      "learning_rate": 9.653700823810955e-06,
      "loss": 0.4639,
      "step": 1698
    },
    {
      "epoch": 0.6225723708318065,
      "grad_norm": 0.36061265184421065,
      "learning_rate": 9.652920798882307e-06,
      "loss": 0.4783,
      "step": 1699
    },
    {
      "epoch": 0.622938805423232,
      "grad_norm": 0.34464723522477536,
      "learning_rate": 9.652139928039757e-06,
      "loss": 0.4765,
      "step": 1700
    },
    {
      "epoch": 0.6233052400146574,
      "grad_norm": 0.3085503188942665,
      "learning_rate": 9.651358211425262e-06,
      "loss": 0.4617,
      "step": 1701
    },
    {
      "epoch": 0.6236716746060829,
      "grad_norm": 0.3413458711999711,
      "learning_rate": 9.650575649180944e-06,
      "loss": 0.4621,
      "step": 1702
    },
    {
      "epoch": 0.6240381091975082,
      "grad_norm": 0.32356339883104773,
      "learning_rate": 9.649792241449074e-06,
      "loss": 0.4715,
      "step": 1703
    },
    {
      "epoch": 0.6244045437889337,
      "grad_norm": 0.3094131529243528,
      "learning_rate": 9.649007988372081e-06,
      "loss": 0.4724,
      "step": 1704
    },
    {
      "epoch": 0.6247709783803591,
      "grad_norm": 0.3495962412861437,
      "learning_rate": 9.648222890092538e-06,
      "loss": 0.4783,
      "step": 1705
    },
    {
      "epoch": 0.6251374129717845,
      "grad_norm": 0.32772877533556166,
      "learning_rate": 9.647436946753184e-06,
      "loss": 0.4729,
      "step": 1706
    },
    {
      "epoch": 0.62550384756321,
      "grad_norm": 0.3933618681027318,
      "learning_rate": 9.646650158496902e-06,
      "loss": 0.4733,
      "step": 1707
    },
    {
      "epoch": 0.6258702821546354,
      "grad_norm": 0.33395743596083666,
      "learning_rate": 9.645862525466734e-06,
      "loss": 0.4743,
      "step": 1708
    },
    {
      "epoch": 0.6262367167460609,
      "grad_norm": 0.41587477177223403,
      "learning_rate": 9.645074047805873e-06,
      "loss": 0.4531,
      "step": 1709
    },
    {
      "epoch": 0.6266031513374862,
      "grad_norm": 0.401970019661568,
      "learning_rate": 9.644284725657667e-06,
      "loss": 0.4714,
      "step": 1710
    },
    {
      "epoch": 0.6269695859289117,
      "grad_norm": 0.3575208626001208,
      "learning_rate": 9.643494559165616e-06,
      "loss": 0.4732,
      "step": 1711
    },
    {
      "epoch": 0.6273360205203371,
      "grad_norm": 0.3239371657913646,
      "learning_rate": 9.642703548473376e-06,
      "loss": 0.4678,
      "step": 1712
    },
    {
      "epoch": 0.6277024551117626,
      "grad_norm": 0.35393094125570485,
      "learning_rate": 9.641911693724754e-06,
      "loss": 0.4736,
      "step": 1713
    },
    {
      "epoch": 0.628068889703188,
      "grad_norm": 0.36503025646284354,
      "learning_rate": 9.641118995063712e-06,
      "loss": 0.4657,
      "step": 1714
    },
    {
      "epoch": 0.6284353242946135,
      "grad_norm": 0.33576248902277017,
      "learning_rate": 9.640325452634364e-06,
      "loss": 0.4924,
      "step": 1715
    },
    {
      "epoch": 0.6288017588860388,
      "grad_norm": 0.3316936716845354,
      "learning_rate": 9.639531066580979e-06,
      "loss": 0.4782,
      "step": 1716
    },
    {
      "epoch": 0.6291681934774642,
      "grad_norm": 0.294948407143815,
      "learning_rate": 9.638735837047976e-06,
      "loss": 0.4911,
      "step": 1717
    },
    {
      "epoch": 0.6295346280688897,
      "grad_norm": 0.320620253223771,
      "learning_rate": 9.637939764179935e-06,
      "loss": 0.4648,
      "step": 1718
    },
    {
      "epoch": 0.6299010626603151,
      "grad_norm": 0.3320150463462027,
      "learning_rate": 9.63714284812158e-06,
      "loss": 0.4712,
      "step": 1719
    },
    {
      "epoch": 0.6302674972517406,
      "grad_norm": 0.34079426997436096,
      "learning_rate": 9.636345089017795e-06,
      "loss": 0.4775,
      "step": 1720
    },
    {
      "epoch": 0.630633931843166,
      "grad_norm": 0.33131214419138594,
      "learning_rate": 9.635546487013611e-06,
      "loss": 0.4526,
      "step": 1721
    },
    {
      "epoch": 0.6310003664345915,
      "grad_norm": 0.3160381748444647,
      "learning_rate": 9.634747042254223e-06,
      "loss": 0.449,
      "step": 1722
    },
    {
      "epoch": 0.6313668010260168,
      "grad_norm": 0.29322671345127665,
      "learning_rate": 9.633946754884963e-06,
      "loss": 0.452,
      "step": 1723
    },
    {
      "epoch": 0.6317332356174423,
      "grad_norm": 0.35370498171013215,
      "learning_rate": 9.633145625051335e-06,
      "loss": 0.466,
      "step": 1724
    },
    {
      "epoch": 0.6320996702088677,
      "grad_norm": 0.31780859269024725,
      "learning_rate": 9.63234365289898e-06,
      "loss": 0.4618,
      "step": 1725
    },
    {
      "epoch": 0.6324661048002932,
      "grad_norm": 0.34473274011130256,
      "learning_rate": 9.6315408385737e-06,
      "loss": 0.469,
      "step": 1726
    },
    {
      "epoch": 0.6328325393917186,
      "grad_norm": 0.3123718090827457,
      "learning_rate": 9.630737182221453e-06,
      "loss": 0.4688,
      "step": 1727
    },
    {
      "epoch": 0.633198973983144,
      "grad_norm": 0.3572795400972479,
      "learning_rate": 9.629932683988341e-06,
      "loss": 0.4621,
      "step": 1728
    },
    {
      "epoch": 0.6335654085745694,
      "grad_norm": 0.3202416447053963,
      "learning_rate": 9.629127344020625e-06,
      "loss": 0.4636,
      "step": 1729
    },
    {
      "epoch": 0.6339318431659948,
      "grad_norm": 0.3355753148406141,
      "learning_rate": 9.62832116246472e-06,
      "loss": 0.467,
      "step": 1730
    },
    {
      "epoch": 0.6342982777574203,
      "grad_norm": 0.3715214313155459,
      "learning_rate": 9.627514139467189e-06,
      "loss": 0.4684,
      "step": 1731
    },
    {
      "epoch": 0.6346647123488457,
      "grad_norm": 0.32613224120691897,
      "learning_rate": 9.626706275174754e-06,
      "loss": 0.4568,
      "step": 1732
    },
    {
      "epoch": 0.6350311469402712,
      "grad_norm": 0.3303372815942467,
      "learning_rate": 9.625897569734287e-06,
      "loss": 0.4767,
      "step": 1733
    },
    {
      "epoch": 0.6353975815316966,
      "grad_norm": 0.30929270945016024,
      "learning_rate": 9.62508802329281e-06,
      "loss": 0.4491,
      "step": 1734
    },
    {
      "epoch": 0.635764016123122,
      "grad_norm": 0.3505737302019091,
      "learning_rate": 9.624277635997503e-06,
      "loss": 0.4659,
      "step": 1735
    },
    {
      "epoch": 0.6361304507145474,
      "grad_norm": 0.3244648587368659,
      "learning_rate": 9.623466407995697e-06,
      "loss": 0.4855,
      "step": 1736
    },
    {
      "epoch": 0.6364968853059729,
      "grad_norm": 0.3248590179435871,
      "learning_rate": 9.622654339434873e-06,
      "loss": 0.4737,
      "step": 1737
    },
    {
      "epoch": 0.6368633198973983,
      "grad_norm": 0.32688459546925186,
      "learning_rate": 9.62184143046267e-06,
      "loss": 0.4524,
      "step": 1738
    },
    {
      "epoch": 0.6372297544888238,
      "grad_norm": 0.7151932375161782,
      "learning_rate": 9.621027681226875e-06,
      "loss": 0.4624,
      "step": 1739
    },
    {
      "epoch": 0.6375961890802492,
      "grad_norm": 0.32279052435246,
      "learning_rate": 9.620213091875431e-06,
      "loss": 0.4823,
      "step": 1740
    },
    {
      "epoch": 0.6379626236716746,
      "grad_norm": 0.3413225078062629,
      "learning_rate": 9.619397662556434e-06,
      "loss": 0.5026,
      "step": 1741
    },
    {
      "epoch": 0.6383290582631,
      "grad_norm": 0.36991896443661926,
      "learning_rate": 9.61858139341813e-06,
      "loss": 0.4678,
      "step": 1742
    },
    {
      "epoch": 0.6386954928545254,
      "grad_norm": 0.36317438379634925,
      "learning_rate": 9.617764284608918e-06,
      "loss": 0.477,
      "step": 1743
    },
    {
      "epoch": 0.6390619274459509,
      "grad_norm": 0.32460361416113576,
      "learning_rate": 9.61694633627735e-06,
      "loss": 0.4633,
      "step": 1744
    },
    {
      "epoch": 0.6394283620373763,
      "grad_norm": 0.3833172545789272,
      "learning_rate": 9.616127548572137e-06,
      "loss": 0.5007,
      "step": 1745
    },
    {
      "epoch": 0.6397947966288018,
      "grad_norm": 0.34049790965107485,
      "learning_rate": 9.61530792164213e-06,
      "loss": 0.4798,
      "step": 1746
    },
    {
      "epoch": 0.6401612312202272,
      "grad_norm": 0.3165551263490173,
      "learning_rate": 9.614487455636341e-06,
      "loss": 0.475,
      "step": 1747
    },
    {
      "epoch": 0.6405276658116527,
      "grad_norm": 0.3390189824817185,
      "learning_rate": 9.613666150703935e-06,
      "loss": 0.4441,
      "step": 1748
    },
    {
      "epoch": 0.640894100403078,
      "grad_norm": 0.2792102030979026,
      "learning_rate": 9.612844006994226e-06,
      "loss": 0.4399,
      "step": 1749
    },
    {
      "epoch": 0.6412605349945035,
      "grad_norm": 0.3432369245720703,
      "learning_rate": 9.612021024656685e-06,
      "loss": 0.4544,
      "step": 1750
    },
    {
      "epoch": 0.6416269695859289,
      "grad_norm": 0.3658477481575573,
      "learning_rate": 9.611197203840927e-06,
      "loss": 0.4768,
      "step": 1751
    },
    {
      "epoch": 0.6419934041773543,
      "grad_norm": 0.3562858976587348,
      "learning_rate": 9.610372544696731e-06,
      "loss": 0.4778,
      "step": 1752
    },
    {
      "epoch": 0.6423598387687798,
      "grad_norm": 0.3402737604619502,
      "learning_rate": 9.609547047374018e-06,
      "loss": 0.4671,
      "step": 1753
    },
    {
      "epoch": 0.6427262733602052,
      "grad_norm": 0.4036063544353359,
      "learning_rate": 9.608720712022866e-06,
      "loss": 0.4798,
      "step": 1754
    },
    {
      "epoch": 0.6430927079516306,
      "grad_norm": 0.3170258959724051,
      "learning_rate": 9.607893538793506e-06,
      "loss": 0.4627,
      "step": 1755
    },
    {
      "epoch": 0.643459142543056,
      "grad_norm": 0.4264795346965528,
      "learning_rate": 9.607065527836324e-06,
      "loss": 0.4719,
      "step": 1756
    },
    {
      "epoch": 0.6438255771344815,
      "grad_norm": 0.3398503147537644,
      "learning_rate": 9.606236679301848e-06,
      "loss": 0.4437,
      "step": 1757
    },
    {
      "epoch": 0.6441920117259069,
      "grad_norm": 0.3518824729766209,
      "learning_rate": 9.605406993340768e-06,
      "loss": 0.4578,
      "step": 1758
    },
    {
      "epoch": 0.6445584463173324,
      "grad_norm": 0.30246522875072257,
      "learning_rate": 9.604576470103923e-06,
      "loss": 0.4953,
      "step": 1759
    },
    {
      "epoch": 0.6449248809087578,
      "grad_norm": 0.3484439152117294,
      "learning_rate": 9.603745109742306e-06,
      "loss": 0.4876,
      "step": 1760
    },
    {
      "epoch": 0.6452913155001833,
      "grad_norm": 0.3387112537115552,
      "learning_rate": 9.60291291240706e-06,
      "loss": 0.5049,
      "step": 1761
    },
    {
      "epoch": 0.6456577500916086,
      "grad_norm": 0.32592599678256173,
      "learning_rate": 9.60207987824948e-06,
      "loss": 0.4576,
      "step": 1762
    },
    {
      "epoch": 0.6460241846830341,
      "grad_norm": 0.321861772413566,
      "learning_rate": 9.601246007421013e-06,
      "loss": 0.4707,
      "step": 1763
    },
    {
      "epoch": 0.6463906192744595,
      "grad_norm": 0.36497752845658515,
      "learning_rate": 9.60041130007326e-06,
      "loss": 0.4761,
      "step": 1764
    },
    {
      "epoch": 0.6467570538658849,
      "grad_norm": 0.3346990934385534,
      "learning_rate": 9.599575756357974e-06,
      "loss": 0.4697,
      "step": 1765
    },
    {
      "epoch": 0.6471234884573104,
      "grad_norm": 0.36919787082975103,
      "learning_rate": 9.598739376427058e-06,
      "loss": 0.4781,
      "step": 1766
    },
    {
      "epoch": 0.6474899230487358,
      "grad_norm": 0.33481023411900396,
      "learning_rate": 9.597902160432569e-06,
      "loss": 0.4606,
      "step": 1767
    },
    {
      "epoch": 0.6478563576401613,
      "grad_norm": 0.3158990841882706,
      "learning_rate": 9.597064108526715e-06,
      "loss": 0.4816,
      "step": 1768
    },
    {
      "epoch": 0.6482227922315866,
      "grad_norm": 0.3520710061210504,
      "learning_rate": 9.596225220861855e-06,
      "loss": 0.4848,
      "step": 1769
    },
    {
      "epoch": 0.6485892268230121,
      "grad_norm": 0.3553550930528338,
      "learning_rate": 9.595385497590503e-06,
      "loss": 0.4632,
      "step": 1770
    },
    {
      "epoch": 0.6489556614144375,
      "grad_norm": 0.35195085562925643,
      "learning_rate": 9.59454493886532e-06,
      "loss": 0.4644,
      "step": 1771
    },
    {
      "epoch": 0.649322096005863,
      "grad_norm": 0.3231801115508992,
      "learning_rate": 9.593703544839126e-06,
      "loss": 0.4697,
      "step": 1772
    },
    {
      "epoch": 0.6496885305972884,
      "grad_norm": 0.32734811740866493,
      "learning_rate": 9.592861315664886e-06,
      "loss": 0.4541,
      "step": 1773
    },
    {
      "epoch": 0.6500549651887139,
      "grad_norm": 0.36962735648884604,
      "learning_rate": 9.59201825149572e-06,
      "loss": 0.4701,
      "step": 1774
    },
    {
      "epoch": 0.6504213997801392,
      "grad_norm": 0.3521190683270996,
      "learning_rate": 9.591174352484902e-06,
      "loss": 0.4821,
      "step": 1775
    },
    {
      "epoch": 0.6507878343715646,
      "grad_norm": 0.3944880333853217,
      "learning_rate": 9.59032961878585e-06,
      "loss": 0.4779,
      "step": 1776
    },
    {
      "epoch": 0.6511542689629901,
      "grad_norm": 0.33863043975460627,
      "learning_rate": 9.589484050552142e-06,
      "loss": 0.474,
      "step": 1777
    },
    {
      "epoch": 0.6515207035544155,
      "grad_norm": 0.3433076936951028,
      "learning_rate": 9.588637647937507e-06,
      "loss": 0.4798,
      "step": 1778
    },
    {
      "epoch": 0.651887138145841,
      "grad_norm": 0.326269412321057,
      "learning_rate": 9.587790411095819e-06,
      "loss": 0.4851,
      "step": 1779
    },
    {
      "epoch": 0.6522535727372664,
      "grad_norm": 0.3446941928059998,
      "learning_rate": 9.58694234018111e-06,
      "loss": 0.4909,
      "step": 1780
    },
    {
      "epoch": 0.6526200073286919,
      "grad_norm": 0.37401131544342314,
      "learning_rate": 9.586093435347561e-06,
      "loss": 0.4775,
      "step": 1781
    },
    {
      "epoch": 0.6529864419201172,
      "grad_norm": 0.35202333557857196,
      "learning_rate": 9.585243696749506e-06,
      "loss": 0.4688,
      "step": 1782
    },
    {
      "epoch": 0.6533528765115427,
      "grad_norm": 0.33320758065961403,
      "learning_rate": 9.584393124541431e-06,
      "loss": 0.4842,
      "step": 1783
    },
    {
      "epoch": 0.6537193111029681,
      "grad_norm": 0.3405147569879103,
      "learning_rate": 9.58354171887797e-06,
      "loss": 0.4753,
      "step": 1784
    },
    {
      "epoch": 0.6540857456943936,
      "grad_norm": 0.36670467822862707,
      "learning_rate": 9.582689479913912e-06,
      "loss": 0.4679,
      "step": 1785
    },
    {
      "epoch": 0.654452180285819,
      "grad_norm": 0.3176103548604317,
      "learning_rate": 9.581836407804196e-06,
      "loss": 0.477,
      "step": 1786
    },
    {
      "epoch": 0.6548186148772445,
      "grad_norm": 0.3585693711456387,
      "learning_rate": 9.580982502703913e-06,
      "loss": 0.4761,
      "step": 1787
    },
    {
      "epoch": 0.6551850494686698,
      "grad_norm": 0.5805923306804135,
      "learning_rate": 9.580127764768305e-06,
      "loss": 0.4757,
      "step": 1788
    },
    {
      "epoch": 0.6555514840600952,
      "grad_norm": 0.3596547078544456,
      "learning_rate": 9.579272194152767e-06,
      "loss": 0.4696,
      "step": 1789
    },
    {
      "epoch": 0.6559179186515207,
      "grad_norm": 0.3186798850401182,
      "learning_rate": 9.578415791012843e-06,
      "loss": 0.4785,
      "step": 1790
    },
    {
      "epoch": 0.6562843532429461,
      "grad_norm": 0.35644369258105596,
      "learning_rate": 9.57755855550423e-06,
      "loss": 0.4974,
      "step": 1791
    },
    {
      "epoch": 0.6566507878343716,
      "grad_norm": 0.3438294799990214,
      "learning_rate": 9.576700487782775e-06,
      "loss": 0.4603,
      "step": 1792
    },
    {
      "epoch": 0.657017222425797,
      "grad_norm": 0.3290932360057555,
      "learning_rate": 9.57584158800448e-06,
      "loss": 0.4819,
      "step": 1793
    },
    {
      "epoch": 0.6573836570172225,
      "grad_norm": 0.3294473566722038,
      "learning_rate": 9.57498185632549e-06,
      "loss": 0.472,
      "step": 1794
    },
    {
      "epoch": 0.6577500916086478,
      "grad_norm": 0.3085784688897355,
      "learning_rate": 9.57412129290211e-06,
      "loss": 0.4582,
      "step": 1795
    },
    {
      "epoch": 0.6581165262000733,
      "grad_norm": 0.33754328982946663,
      "learning_rate": 9.573259897890794e-06,
      "loss": 0.4726,
      "step": 1796
    },
    {
      "epoch": 0.6584829607914987,
      "grad_norm": 0.33134231063660163,
      "learning_rate": 9.572397671448144e-06,
      "loss": 0.449,
      "step": 1797
    },
    {
      "epoch": 0.6588493953829242,
      "grad_norm": 0.3340208562023715,
      "learning_rate": 9.571534613730915e-06,
      "loss": 0.4947,
      "step": 1798
    },
    {
      "epoch": 0.6592158299743496,
      "grad_norm": 0.38579168866994784,
      "learning_rate": 9.570670724896013e-06,
      "loss": 0.4903,
      "step": 1799
    },
    {
      "epoch": 0.659582264565775,
      "grad_norm": 0.31639158219634667,
      "learning_rate": 9.5698060051005e-06,
      "loss": 0.4648,
      "step": 1800
    },
    {
      "epoch": 0.6599486991572004,
      "grad_norm": 0.3267720857878399,
      "learning_rate": 9.568940454501578e-06,
      "loss": 0.4517,
      "step": 1801
    },
    {
      "epoch": 0.6603151337486258,
      "grad_norm": 0.32453018315055066,
      "learning_rate": 9.568074073256608e-06,
      "loss": 0.4703,
      "step": 1802
    },
    {
      "epoch": 0.6606815683400513,
      "grad_norm": 0.30501596081870974,
      "learning_rate": 9.567206861523105e-06,
      "loss": 0.4694,
      "step": 1803
    },
    {
      "epoch": 0.6610480029314767,
      "grad_norm": 0.336272684743686,
      "learning_rate": 9.566338819458726e-06,
      "loss": 0.4721,
      "step": 1804
    },
    {
      "epoch": 0.6614144375229022,
      "grad_norm": 0.3446780654560093,
      "learning_rate": 9.565469947221282e-06,
      "loss": 0.4928,
      "step": 1805
    },
    {
      "epoch": 0.6617808721143276,
      "grad_norm": 0.32235354784819303,
      "learning_rate": 9.564600244968742e-06,
      "loss": 0.4757,
      "step": 1806
    },
    {
      "epoch": 0.6621473067057531,
      "grad_norm": 0.363730986311918,
      "learning_rate": 9.563729712859216e-06,
      "loss": 0.5048,
      "step": 1807
    },
    {
      "epoch": 0.6625137412971784,
      "grad_norm": 0.38373994026184055,
      "learning_rate": 9.562858351050972e-06,
      "loss": 0.4685,
      "step": 1808
    },
    {
      "epoch": 0.6628801758886039,
      "grad_norm": 0.2962188193806418,
      "learning_rate": 9.561986159702424e-06,
      "loss": 0.4634,
      "step": 1809
    },
    {
      "epoch": 0.6632466104800293,
      "grad_norm": 0.3165163552841888,
      "learning_rate": 9.561113138972138e-06,
      "loss": 0.4789,
      "step": 1810
    },
    {
      "epoch": 0.6636130450714548,
      "grad_norm": 0.3644006598459135,
      "learning_rate": 9.560239289018833e-06,
      "loss": 0.4878,
      "step": 1811
    },
    {
      "epoch": 0.6639794796628802,
      "grad_norm": 0.3328568021988627,
      "learning_rate": 9.559364610001379e-06,
      "loss": 0.4864,
      "step": 1812
    },
    {
      "epoch": 0.6643459142543056,
      "grad_norm": 0.3347004532323089,
      "learning_rate": 9.558489102078792e-06,
      "loss": 0.4724,
      "step": 1813
    },
    {
      "epoch": 0.664712348845731,
      "grad_norm": 0.3257831396514747,
      "learning_rate": 9.557612765410243e-06,
      "loss": 0.4373,
      "step": 1814
    },
    {
      "epoch": 0.6650787834371564,
      "grad_norm": 0.3073266146597233,
      "learning_rate": 9.556735600155053e-06,
      "loss": 0.4672,
      "step": 1815
    },
    {
      "epoch": 0.6654452180285819,
      "grad_norm": 0.3390814923914724,
      "learning_rate": 9.555857606472692e-06,
      "loss": 0.4894,
      "step": 1816
    },
    {
      "epoch": 0.6658116526200073,
      "grad_norm": 0.31935425972048787,
      "learning_rate": 9.554978784522784e-06,
      "loss": 0.4558,
      "step": 1817
    },
    {
      "epoch": 0.6661780872114328,
      "grad_norm": 1.204762640707574,
      "learning_rate": 9.554099134465097e-06,
      "loss": 0.4832,
      "step": 1818
    },
    {
      "epoch": 0.6665445218028582,
      "grad_norm": 0.3399023886183897,
      "learning_rate": 9.553218656459558e-06,
      "loss": 0.4931,
      "step": 1819
    },
    {
      "epoch": 0.6669109563942837,
      "grad_norm": 0.35590111430646254,
      "learning_rate": 9.552337350666239e-06,
      "loss": 0.4703,
      "step": 1820
    },
    {
      "epoch": 0.667277390985709,
      "grad_norm": 0.3567277834720204,
      "learning_rate": 9.551455217245362e-06,
      "loss": 0.4639,
      "step": 1821
    },
    {
      "epoch": 0.6676438255771345,
      "grad_norm": 0.317875464211911,
      "learning_rate": 9.550572256357305e-06,
      "loss": 0.4791,
      "step": 1822
    },
    {
      "epoch": 0.6680102601685599,
      "grad_norm": 0.30362243503963404,
      "learning_rate": 9.54968846816259e-06,
      "loss": 0.4679,
      "step": 1823
    },
    {
      "epoch": 0.6683766947599853,
      "grad_norm": 0.3151481940218672,
      "learning_rate": 9.548803852821891e-06,
      "loss": 0.4804,
      "step": 1824
    },
    {
      "epoch": 0.6687431293514108,
      "grad_norm": 0.2998200858332865,
      "learning_rate": 9.547918410496037e-06,
      "loss": 0.4545,
      "step": 1825
    },
    {
      "epoch": 0.6691095639428362,
      "grad_norm": 0.3347200903609805,
      "learning_rate": 9.547032141346002e-06,
      "loss": 0.4635,
      "step": 1826
    },
    {
      "epoch": 0.6694759985342617,
      "grad_norm": 0.36408603024072994,
      "learning_rate": 9.546145045532911e-06,
      "loss": 0.4806,
      "step": 1827
    },
    {
      "epoch": 0.669842433125687,
      "grad_norm": 0.30121586521351557,
      "learning_rate": 9.545257123218043e-06,
      "loss": 0.4604,
      "step": 1828
    },
    {
      "epoch": 0.6702088677171125,
      "grad_norm": 0.3303862237015951,
      "learning_rate": 9.544368374562825e-06,
      "loss": 0.4699,
      "step": 1829
    },
    {
      "epoch": 0.6705753023085379,
      "grad_norm": 0.31456276085042817,
      "learning_rate": 9.54347879972883e-06,
      "loss": 0.4598,
      "step": 1830
    },
    {
      "epoch": 0.6709417368999634,
      "grad_norm": 0.3387598474538159,
      "learning_rate": 9.542588398877787e-06,
      "loss": 0.4805,
      "step": 1831
    },
    {
      "epoch": 0.6713081714913888,
      "grad_norm": 0.33683604963104574,
      "learning_rate": 9.541697172171577e-06,
      "loss": 0.4624,
      "step": 1832
    },
    {
      "epoch": 0.6716746060828143,
      "grad_norm": 0.3224074649789024,
      "learning_rate": 9.540805119772223e-06,
      "loss": 0.462,
      "step": 1833
    },
    {
      "epoch": 0.6720410406742396,
      "grad_norm": 0.3167708911376573,
      "learning_rate": 9.539912241841904e-06,
      "loss": 0.4467,
      "step": 1834
    },
    {
      "epoch": 0.6724074752656651,
      "grad_norm": 0.3084435561992738,
      "learning_rate": 9.539018538542949e-06,
      "loss": 0.4938,
      "step": 1835
    },
    {
      "epoch": 0.6727739098570905,
      "grad_norm": 0.35401325037224596,
      "learning_rate": 9.538124010037832e-06,
      "loss": 0.4884,
      "step": 1836
    },
    {
      "epoch": 0.6731403444485159,
      "grad_norm": 0.5637540659484231,
      "learning_rate": 9.537228656489187e-06,
      "loss": 0.4766,
      "step": 1837
    },
    {
      "epoch": 0.6735067790399414,
      "grad_norm": 0.31080190213251285,
      "learning_rate": 9.536332478059784e-06,
      "loss": 0.475,
      "step": 1838
    },
    {
      "epoch": 0.6738732136313668,
      "grad_norm": 0.31640232836594645,
      "learning_rate": 9.535435474912559e-06,
      "loss": 0.4859,
      "step": 1839
    },
    {
      "epoch": 0.6742396482227923,
      "grad_norm": 0.33166440030165695,
      "learning_rate": 9.534537647210582e-06,
      "loss": 0.4481,
      "step": 1840
    },
    {
      "epoch": 0.6746060828142176,
      "grad_norm": 0.3504380260957974,
      "learning_rate": 9.533638995117085e-06,
      "loss": 0.4774,
      "step": 1841
    },
    {
      "epoch": 0.6749725174056431,
      "grad_norm": 0.3132055978630974,
      "learning_rate": 9.532739518795443e-06,
      "loss": 0.4857,
      "step": 1842
    },
    {
      "epoch": 0.6753389519970685,
      "grad_norm": 0.3194892474408759,
      "learning_rate": 9.531839218409186e-06,
      "loss": 0.4626,
      "step": 1843
    },
    {
      "epoch": 0.675705386588494,
      "grad_norm": 0.2926799723207414,
      "learning_rate": 9.530938094121992e-06,
      "loss": 0.4681,
      "step": 1844
    },
    {
      "epoch": 0.6760718211799194,
      "grad_norm": 0.3365994277788381,
      "learning_rate": 9.530036146097683e-06,
      "loss": 0.4649,
      "step": 1845
    },
    {
      "epoch": 0.6764382557713449,
      "grad_norm": 0.32180039815963807,
      "learning_rate": 9.52913337450024e-06,
      "loss": 0.4632,
      "step": 1846
    },
    {
      "epoch": 0.6768046903627702,
      "grad_norm": 0.3513484498742536,
      "learning_rate": 9.528229779493787e-06,
      "loss": 0.4877,
      "step": 1847
    },
    {
      "epoch": 0.6771711249541956,
      "grad_norm": 0.3460131479794343,
      "learning_rate": 9.5273253612426e-06,
      "loss": 0.4557,
      "step": 1848
    },
    {
      "epoch": 0.6775375595456211,
      "grad_norm": 0.33370736665438455,
      "learning_rate": 9.526420119911109e-06,
      "loss": 0.4568,
      "step": 1849
    },
    {
      "epoch": 0.6779039941370465,
      "grad_norm": 0.37346805802315114,
      "learning_rate": 9.525514055663882e-06,
      "loss": 0.4717,
      "step": 1850
    },
    {
      "epoch": 0.678270428728472,
      "grad_norm": 0.42951004832767165,
      "learning_rate": 9.52460716866565e-06,
      "loss": 0.4567,
      "step": 1851
    },
    {
      "epoch": 0.6786368633198974,
      "grad_norm": 0.3225999261266418,
      "learning_rate": 9.523699459081285e-06,
      "loss": 0.4672,
      "step": 1852
    },
    {
      "epoch": 0.6790032979113229,
      "grad_norm": 0.3461582413432567,
      "learning_rate": 9.522790927075811e-06,
      "loss": 0.4765,
      "step": 1853
    },
    {
      "epoch": 0.6793697325027482,
      "grad_norm": 0.44274060927709685,
      "learning_rate": 9.521881572814403e-06,
      "loss": 0.465,
      "step": 1854
    },
    {
      "epoch": 0.6797361670941737,
      "grad_norm": 0.3658660227879599,
      "learning_rate": 9.520971396462383e-06,
      "loss": 0.4851,
      "step": 1855
    },
    {
      "epoch": 0.6801026016855991,
      "grad_norm": 0.34690354973920745,
      "learning_rate": 9.520060398185226e-06,
      "loss": 0.466,
      "step": 1856
    },
    {
      "epoch": 0.6804690362770246,
      "grad_norm": 0.30874861985558755,
      "learning_rate": 9.519148578148552e-06,
      "loss": 0.4387,
      "step": 1857
    },
    {
      "epoch": 0.68083547086845,
      "grad_norm": 0.3496965092314128,
      "learning_rate": 9.51823593651813e-06,
      "loss": 0.4784,
      "step": 1858
    },
    {
      "epoch": 0.6812019054598755,
      "grad_norm": 0.3461764194095652,
      "learning_rate": 9.517322473459884e-06,
      "loss": 0.4722,
      "step": 1859
    },
    {
      "epoch": 0.6815683400513008,
      "grad_norm": 0.3109409585574862,
      "learning_rate": 9.516408189139883e-06,
      "loss": 0.469,
      "step": 1860
    },
    {
      "epoch": 0.6819347746427262,
      "grad_norm": 0.6658409289630646,
      "learning_rate": 9.515493083724348e-06,
      "loss": 0.4755,
      "step": 1861
    },
    {
      "epoch": 0.6823012092341517,
      "grad_norm": 0.45589834745662716,
      "learning_rate": 9.514577157379647e-06,
      "loss": 0.4802,
      "step": 1862
    },
    {
      "epoch": 0.6826676438255771,
      "grad_norm": 0.3106013798914157,
      "learning_rate": 9.513660410272296e-06,
      "loss": 0.4704,
      "step": 1863
    },
    {
      "epoch": 0.6830340784170026,
      "grad_norm": 0.3435477689175926,
      "learning_rate": 9.512742842568964e-06,
      "loss": 0.489,
      "step": 1864
    },
    {
      "epoch": 0.683400513008428,
      "grad_norm": 0.3242511077751796,
      "learning_rate": 9.511824454436466e-06,
      "loss": 0.4835,
      "step": 1865
    },
    {
      "epoch": 0.6837669475998535,
      "grad_norm": 0.3195586775943994,
      "learning_rate": 9.510905246041771e-06,
      "loss": 0.4662,
      "step": 1866
    },
    {
      "epoch": 0.6841333821912788,
      "grad_norm": 0.3551423128289305,
      "learning_rate": 9.509985217551989e-06,
      "loss": 0.4559,
      "step": 1867
    },
    {
      "epoch": 0.6844998167827043,
      "grad_norm": 0.3053951343468664,
      "learning_rate": 9.509064369134387e-06,
      "loss": 0.4711,
      "step": 1868
    },
    {
      "epoch": 0.6848662513741297,
      "grad_norm": 0.2967128562586158,
      "learning_rate": 9.508142700956376e-06,
      "loss": 0.4592,
      "step": 1869
    },
    {
      "epoch": 0.6852326859655552,
      "grad_norm": 0.3585925016752028,
      "learning_rate": 9.507220213185517e-06,
      "loss": 0.4704,
      "step": 1870
    },
    {
      "epoch": 0.6855991205569806,
      "grad_norm": 0.34875148168165326,
      "learning_rate": 9.506296905989523e-06,
      "loss": 0.4843,
      "step": 1871
    },
    {
      "epoch": 0.685965555148406,
      "grad_norm": 0.31765235672701414,
      "learning_rate": 9.505372779536255e-06,
      "loss": 0.4725,
      "step": 1872
    },
    {
      "epoch": 0.6863319897398314,
      "grad_norm": 0.33934213564749904,
      "learning_rate": 9.504447833993717e-06,
      "loss": 0.4921,
      "step": 1873
    },
    {
      "epoch": 0.6866984243312568,
      "grad_norm": 0.35007945658718553,
      "learning_rate": 9.503522069530072e-06,
      "loss": 0.4491,
      "step": 1874
    },
    {
      "epoch": 0.6870648589226823,
      "grad_norm": 0.3386595212176214,
      "learning_rate": 9.502595486313623e-06,
      "loss": 0.4543,
      "step": 1875
    },
    {
      "epoch": 0.6874312935141077,
      "grad_norm": 0.3162771473191873,
      "learning_rate": 9.501668084512827e-06,
      "loss": 0.483,
      "step": 1876
    },
    {
      "epoch": 0.6877977281055332,
      "grad_norm": 0.3108375614325068,
      "learning_rate": 9.500739864296287e-06,
      "loss": 0.4831,
      "step": 1877
    },
    {
      "epoch": 0.6881641626969586,
      "grad_norm": 0.31189051477356383,
      "learning_rate": 9.499810825832757e-06,
      "loss": 0.4792,
      "step": 1878
    },
    {
      "epoch": 0.6885305972883841,
      "grad_norm": 0.29458385023558664,
      "learning_rate": 9.498880969291138e-06,
      "loss": 0.4602,
      "step": 1879
    },
    {
      "epoch": 0.6888970318798094,
      "grad_norm": 0.32716668185945247,
      "learning_rate": 9.497950294840481e-06,
      "loss": 0.4703,
      "step": 1880
    },
    {
      "epoch": 0.6892634664712349,
      "grad_norm": 0.3361998029612432,
      "learning_rate": 9.497018802649986e-06,
      "loss": 0.4826,
      "step": 1881
    },
    {
      "epoch": 0.6896299010626603,
      "grad_norm": 0.30605591272721844,
      "learning_rate": 9.496086492889001e-06,
      "loss": 0.4552,
      "step": 1882
    },
    {
      "epoch": 0.6899963356540857,
      "grad_norm": 0.30952808988783337,
      "learning_rate": 9.495153365727019e-06,
      "loss": 0.4782,
      "step": 1883
    },
    {
      "epoch": 0.6903627702455112,
      "grad_norm": 0.32525586981793103,
      "learning_rate": 9.494219421333688e-06,
      "loss": 0.4706,
      "step": 1884
    },
    {
      "epoch": 0.6907292048369366,
      "grad_norm": 0.3564000847628441,
      "learning_rate": 9.493284659878802e-06,
      "loss": 0.487,
      "step": 1885
    },
    {
      "epoch": 0.691095639428362,
      "grad_norm": 0.3504989098197117,
      "learning_rate": 9.492349081532305e-06,
      "loss": 0.4955,
      "step": 1886
    },
    {
      "epoch": 0.6914620740197874,
      "grad_norm": 0.33160393015795075,
      "learning_rate": 9.49141268646428e-06,
      "loss": 0.4736,
      "step": 1887
    },
    {
      "epoch": 0.6918285086112129,
      "grad_norm": 0.30797447152288643,
      "learning_rate": 9.490475474844976e-06,
      "loss": 0.4638,
      "step": 1888
    },
    {
      "epoch": 0.6921949432026383,
      "grad_norm": 0.3311620710579034,
      "learning_rate": 9.489537446844775e-06,
      "loss": 0.4845,
      "step": 1889
    },
    {
      "epoch": 0.6925613777940638,
      "grad_norm": 0.33861534468440124,
      "learning_rate": 9.488598602634213e-06,
      "loss": 0.4478,
      "step": 1890
    },
    {
      "epoch": 0.6929278123854892,
      "grad_norm": 0.337113178658717,
      "learning_rate": 9.487658942383975e-06,
      "loss": 0.4807,
      "step": 1891
    },
    {
      "epoch": 0.6932942469769147,
      "grad_norm": 0.3143262937347115,
      "learning_rate": 9.486718466264898e-06,
      "loss": 0.4582,
      "step": 1892
    },
    {
      "epoch": 0.69366068156834,
      "grad_norm": 0.32430539586944224,
      "learning_rate": 9.485777174447957e-06,
      "loss": 0.4556,
      "step": 1893
    },
    {
      "epoch": 0.6940271161597655,
      "grad_norm": 0.3237789201457776,
      "learning_rate": 9.484835067104285e-06,
      "loss": 0.4869,
      "step": 1894
    },
    {
      "epoch": 0.6943935507511909,
      "grad_norm": 0.34453769080308305,
      "learning_rate": 9.483892144405157e-06,
      "loss": 0.4688,
      "step": 1895
    },
    {
      "epoch": 0.6947599853426163,
      "grad_norm": 0.29636656060740485,
      "learning_rate": 9.482948406522003e-06,
      "loss": 0.4842,
      "step": 1896
    },
    {
      "epoch": 0.6951264199340418,
      "grad_norm": 0.34705784496451136,
      "learning_rate": 9.482003853626396e-06,
      "loss": 0.4611,
      "step": 1897
    },
    {
      "epoch": 0.6954928545254672,
      "grad_norm": 0.32188358172299264,
      "learning_rate": 9.481058485890055e-06,
      "loss": 0.4585,
      "step": 1898
    },
    {
      "epoch": 0.6958592891168927,
      "grad_norm": 0.32376563221808025,
      "learning_rate": 9.480112303484855e-06,
      "loss": 0.4535,
      "step": 1899
    },
    {
      "epoch": 0.696225723708318,
      "grad_norm": 0.32529251864224346,
      "learning_rate": 9.479165306582811e-06,
      "loss": 0.4711,
      "step": 1900
    },
    {
      "epoch": 0.6965921582997435,
      "grad_norm": 0.3412896622589412,
      "learning_rate": 9.478217495356092e-06,
      "loss": 0.4653,
      "step": 1901
    },
    {
      "epoch": 0.6969585928911689,
      "grad_norm": 0.3214465266466881,
      "learning_rate": 9.477268869977012e-06,
      "loss": 0.4582,
      "step": 1902
    },
    {
      "epoch": 0.6973250274825944,
      "grad_norm": 0.38915660280889663,
      "learning_rate": 9.476319430618033e-06,
      "loss": 0.4663,
      "step": 1903
    },
    {
      "epoch": 0.6976914620740198,
      "grad_norm": 0.31365499597612423,
      "learning_rate": 9.475369177451765e-06,
      "loss": 0.4635,
      "step": 1904
    },
    {
      "epoch": 0.6980578966654453,
      "grad_norm": 0.34827552142138474,
      "learning_rate": 9.47441811065097e-06,
      "loss": 0.456,
      "step": 1905
    },
    {
      "epoch": 0.6984243312568706,
      "grad_norm": 0.31950015070719184,
      "learning_rate": 9.473466230388552e-06,
      "loss": 0.4537,
      "step": 1906
    },
    {
      "epoch": 0.698790765848296,
      "grad_norm": 0.3746704328158076,
      "learning_rate": 9.472513536837566e-06,
      "loss": 0.4954,
      "step": 1907
    },
    {
      "epoch": 0.6991572004397215,
      "grad_norm": 0.3292956477317344,
      "learning_rate": 9.471560030171213e-06,
      "loss": 0.4757,
      "step": 1908
    },
    {
      "epoch": 0.6995236350311469,
      "grad_norm": 0.3171526582714549,
      "learning_rate": 9.470605710562845e-06,
      "loss": 0.4644,
      "step": 1909
    },
    {
      "epoch": 0.6998900696225724,
      "grad_norm": 0.33313503483697654,
      "learning_rate": 9.46965057818596e-06,
      "loss": 0.4477,
      "step": 1910
    },
    {
      "epoch": 0.7002565042139978,
      "grad_norm": 0.33658415230972877,
      "learning_rate": 9.468694633214204e-06,
      "loss": 0.4477,
      "step": 1911
    },
    {
      "epoch": 0.7006229388054233,
      "grad_norm": 0.3178483733641845,
      "learning_rate": 9.467737875821368e-06,
      "loss": 0.4841,
      "step": 1912
    },
    {
      "epoch": 0.7009893733968486,
      "grad_norm": 0.35260401590179097,
      "learning_rate": 9.466780306181396e-06,
      "loss": 0.4872,
      "step": 1913
    },
    {
      "epoch": 0.7013558079882741,
      "grad_norm": 0.35797573951026274,
      "learning_rate": 9.465821924468375e-06,
      "loss": 0.4772,
      "step": 1914
    },
    {
      "epoch": 0.7017222425796995,
      "grad_norm": 0.3368761025271847,
      "learning_rate": 9.464862730856542e-06,
      "loss": 0.4666,
      "step": 1915
    },
    {
      "epoch": 0.702088677171125,
      "grad_norm": 0.31561187744282293,
      "learning_rate": 9.46390272552028e-06,
      "loss": 0.4818,
      "step": 1916
    },
    {
      "epoch": 0.7024551117625504,
      "grad_norm": 0.34320589036349625,
      "learning_rate": 9.462941908634122e-06,
      "loss": 0.4912,
      "step": 1917
    },
    {
      "epoch": 0.7028215463539759,
      "grad_norm": 0.34400272059044246,
      "learning_rate": 9.461980280372748e-06,
      "loss": 0.4632,
      "step": 1918
    },
    {
      "epoch": 0.7031879809454012,
      "grad_norm": 0.30129643661794997,
      "learning_rate": 9.461017840910982e-06,
      "loss": 0.4464,
      "step": 1919
    },
    {
      "epoch": 0.7035544155368266,
      "grad_norm": 0.3497595571319319,
      "learning_rate": 9.460054590423801e-06,
      "loss": 0.4871,
      "step": 1920
    },
    {
      "epoch": 0.7039208501282521,
      "grad_norm": 0.31612241488919823,
      "learning_rate": 9.459090529086325e-06,
      "loss": 0.4711,
      "step": 1921
    },
    {
      "epoch": 0.7042872847196775,
      "grad_norm": 0.32755714326367463,
      "learning_rate": 9.458125657073823e-06,
      "loss": 0.4734,
      "step": 1922
    },
    {
      "epoch": 0.704653719311103,
      "grad_norm": 0.35104021574188987,
      "learning_rate": 9.457159974561712e-06,
      "loss": 0.4746,
      "step": 1923
    },
    {
      "epoch": 0.7050201539025284,
      "grad_norm": 0.32829051588189273,
      "learning_rate": 9.456193481725555e-06,
      "loss": 0.4697,
      "step": 1924
    },
    {
      "epoch": 0.7053865884939539,
      "grad_norm": 0.3758424973956133,
      "learning_rate": 9.455226178741065e-06,
      "loss": 0.4968,
      "step": 1925
    },
    {
      "epoch": 0.7057530230853792,
      "grad_norm": 0.32661237134508697,
      "learning_rate": 9.4542580657841e-06,
      "loss": 0.4681,
      "step": 1926
    },
    {
      "epoch": 0.7061194576768047,
      "grad_norm": 0.3192960288780993,
      "learning_rate": 9.453289143030662e-06,
      "loss": 0.4808,
      "step": 1927
    },
    {
      "epoch": 0.7064858922682301,
      "grad_norm": 0.3342102445599269,
      "learning_rate": 9.452319410656907e-06,
      "loss": 0.4636,
      "step": 1928
    },
    {
      "epoch": 0.7068523268596556,
      "grad_norm": 0.37288863310649095,
      "learning_rate": 9.451348868839135e-06,
      "loss": 0.487,
      "step": 1929
    },
    {
      "epoch": 0.707218761451081,
      "grad_norm": 0.34170315846983107,
      "learning_rate": 9.45037751775379e-06,
      "loss": 0.4736,
      "step": 1930
    },
    {
      "epoch": 0.7075851960425064,
      "grad_norm": 0.33404467804816557,
      "learning_rate": 9.449405357577473e-06,
      "loss": 0.4789,
      "step": 1931
    },
    {
      "epoch": 0.7079516306339318,
      "grad_norm": 0.36591193858962073,
      "learning_rate": 9.44843238848692e-06,
      "loss": 0.4821,
      "step": 1932
    },
    {
      "epoch": 0.7083180652253572,
      "grad_norm": 0.3312194811054758,
      "learning_rate": 9.447458610659019e-06,
      "loss": 0.4662,
      "step": 1933
    },
    {
      "epoch": 0.7086844998167827,
      "grad_norm": 0.49552853354401016,
      "learning_rate": 9.44648402427081e-06,
      "loss": 0.5088,
      "step": 1934
    },
    {
      "epoch": 0.7090509344082081,
      "grad_norm": 0.3142242955129153,
      "learning_rate": 9.445508629499472e-06,
      "loss": 0.4807,
      "step": 1935
    },
    {
      "epoch": 0.7094173689996336,
      "grad_norm": 0.327364875070892,
      "learning_rate": 9.444532426522334e-06,
      "loss": 0.4348,
      "step": 1936
    },
    {
      "epoch": 0.709783803591059,
      "grad_norm": 0.3502728308904058,
      "learning_rate": 9.443555415516875e-06,
      "loss": 0.4915,
      "step": 1937
    },
    {
      "epoch": 0.7101502381824845,
      "grad_norm": 0.3802738704125049,
      "learning_rate": 9.442577596660717e-06,
      "loss": 0.4811,
      "step": 1938
    },
    {
      "epoch": 0.7105166727739098,
      "grad_norm": 0.3157966683916068,
      "learning_rate": 9.44159897013163e-06,
      "loss": 0.4623,
      "step": 1939
    },
    {
      "epoch": 0.7108831073653353,
      "grad_norm": 0.33180868536571856,
      "learning_rate": 9.440619536107531e-06,
      "loss": 0.4706,
      "step": 1940
    },
    {
      "epoch": 0.7112495419567607,
      "grad_norm": 0.332876914786784,
      "learning_rate": 9.439639294766485e-06,
      "loss": 0.4659,
      "step": 1941
    },
    {
      "epoch": 0.7116159765481862,
      "grad_norm": 0.3669401490669464,
      "learning_rate": 9.4386582462867e-06,
      "loss": 0.4875,
      "step": 1942
    },
    {
      "epoch": 0.7119824111396116,
      "grad_norm": 0.3130221746943202,
      "learning_rate": 9.437676390846537e-06,
      "loss": 0.4815,
      "step": 1943
    },
    {
      "epoch": 0.712348845731037,
      "grad_norm": 0.3251447601974065,
      "learning_rate": 9.436693728624497e-06,
      "loss": 0.4654,
      "step": 1944
    },
    {
      "epoch": 0.7127152803224625,
      "grad_norm": 0.30336329880223495,
      "learning_rate": 9.435710259799234e-06,
      "loss": 0.4652,
      "step": 1945
    },
    {
      "epoch": 0.7130817149138878,
      "grad_norm": 0.3924599708036331,
      "learning_rate": 9.43472598454954e-06,
      "loss": 0.4851,
      "step": 1946
    },
    {
      "epoch": 0.7134481495053133,
      "grad_norm": 0.35424248252304325,
      "learning_rate": 9.433740903054363e-06,
      "loss": 0.4426,
      "step": 1947
    },
    {
      "epoch": 0.7138145840967387,
      "grad_norm": 0.363700105462963,
      "learning_rate": 9.432755015492794e-06,
      "loss": 0.4955,
      "step": 1948
    },
    {
      "epoch": 0.7141810186881642,
      "grad_norm": 0.31572620862500944,
      "learning_rate": 9.431768322044069e-06,
      "loss": 0.4688,
      "step": 1949
    },
    {
      "epoch": 0.7145474532795896,
      "grad_norm": 0.3186151419117438,
      "learning_rate": 9.430780822887568e-06,
      "loss": 0.4555,
      "step": 1950
    },
    {
      "epoch": 0.7149138878710151,
      "grad_norm": 0.3376249878556975,
      "learning_rate": 9.429792518202826e-06,
      "loss": 0.4432,
      "step": 1951
    },
    {
      "epoch": 0.7152803224624404,
      "grad_norm": 0.31646333815786803,
      "learning_rate": 9.428803408169521e-06,
      "loss": 0.4611,
      "step": 1952
    },
    {
      "epoch": 0.7156467570538659,
      "grad_norm": 0.3553713101955225,
      "learning_rate": 9.42781349296747e-06,
      "loss": 0.4782,
      "step": 1953
    },
    {
      "epoch": 0.7160131916452913,
      "grad_norm": 0.3184395075231948,
      "learning_rate": 9.426822772776645e-06,
      "loss": 0.4657,
      "step": 1954
    },
    {
      "epoch": 0.7163796262367167,
      "grad_norm": 0.3122900338457371,
      "learning_rate": 9.425831247777163e-06,
      "loss": 0.4627,
      "step": 1955
    },
    {
      "epoch": 0.7167460608281422,
      "grad_norm": 0.3633570469871141,
      "learning_rate": 9.424838918149285e-06,
      "loss": 0.4976,
      "step": 1956
    },
    {
      "epoch": 0.7171124954195676,
      "grad_norm": 0.33769100491293774,
      "learning_rate": 9.42384578407342e-06,
      "loss": 0.4779,
      "step": 1957
    },
    {
      "epoch": 0.717478930010993,
      "grad_norm": 0.3204424334724945,
      "learning_rate": 9.42285184573012e-06,
      "loss": 0.4676,
      "step": 1958
    },
    {
      "epoch": 0.7178453646024184,
      "grad_norm": 0.3258942233948997,
      "learning_rate": 9.421857103300087e-06,
      "loss": 0.4616,
      "step": 1959
    },
    {
      "epoch": 0.7182117991938439,
      "grad_norm": 0.36144335594200033,
      "learning_rate": 9.42086155696417e-06,
      "loss": 0.4684,
      "step": 1960
    },
    {
      "epoch": 0.7185782337852693,
      "grad_norm": 0.34099908087020664,
      "learning_rate": 9.419865206903361e-06,
      "loss": 0.4508,
      "step": 1961
    },
    {
      "epoch": 0.7189446683766948,
      "grad_norm": 0.3309238862048949,
      "learning_rate": 9.418868053298797e-06,
      "loss": 0.4666,
      "step": 1962
    },
    {
      "epoch": 0.7193111029681202,
      "grad_norm": 0.32546449643359804,
      "learning_rate": 9.417870096331764e-06,
      "loss": 0.4778,
      "step": 1963
    },
    {
      "epoch": 0.7196775375595457,
      "grad_norm": 0.3681744142915993,
      "learning_rate": 9.416871336183698e-06,
      "loss": 0.4748,
      "step": 1964
    },
    {
      "epoch": 0.720043972150971,
      "grad_norm": 0.3001835438912081,
      "learning_rate": 9.41587177303617e-06,
      "loss": 0.4695,
      "step": 1965
    },
    {
      "epoch": 0.7204104067423965,
      "grad_norm": 0.3068492457100541,
      "learning_rate": 9.414871407070906e-06,
      "loss": 0.4344,
      "step": 1966
    },
    {
      "epoch": 0.7207768413338219,
      "grad_norm": 0.40150150254957184,
      "learning_rate": 9.413870238469778e-06,
      "loss": 0.4715,
      "step": 1967
    },
    {
      "epoch": 0.7211432759252473,
      "grad_norm": 0.3487879090479508,
      "learning_rate": 9.412868267414797e-06,
      "loss": 0.4852,
      "step": 1968
    },
    {
      "epoch": 0.7215097105166728,
      "grad_norm": 0.3178657805042771,
      "learning_rate": 9.411865494088124e-06,
      "loss": 0.4575,
      "step": 1969
    },
    {
      "epoch": 0.7218761451080982,
      "grad_norm": 0.31008562303991544,
      "learning_rate": 9.41086191867207e-06,
      "loss": 0.4754,
      "step": 1970
    },
    {
      "epoch": 0.7222425796995237,
      "grad_norm": 0.3454728392211863,
      "learning_rate": 9.409857541349086e-06,
      "loss": 0.4668,
      "step": 1971
    },
    {
      "epoch": 0.722609014290949,
      "grad_norm": 0.3161308397266021,
      "learning_rate": 9.408852362301768e-06,
      "loss": 0.4749,
      "step": 1972
    },
    {
      "epoch": 0.7229754488823745,
      "grad_norm": 0.35250186416567547,
      "learning_rate": 9.407846381712862e-06,
      "loss": 0.4723,
      "step": 1973
    },
    {
      "epoch": 0.7233418834737999,
      "grad_norm": 0.3360799337604492,
      "learning_rate": 9.406839599765261e-06,
      "loss": 0.477,
      "step": 1974
    },
    {
      "epoch": 0.7237083180652254,
      "grad_norm": 0.3034950187398148,
      "learning_rate": 9.405832016641997e-06,
      "loss": 0.4829,
      "step": 1975
    },
    {
      "epoch": 0.7240747526566508,
      "grad_norm": 0.3414659795363816,
      "learning_rate": 9.404823632526253e-06,
      "loss": 0.4855,
      "step": 1976
    },
    {
      "epoch": 0.7244411872480763,
      "grad_norm": 0.3473351355357309,
      "learning_rate": 9.403814447601354e-06,
      "loss": 0.4609,
      "step": 1977
    },
    {
      "epoch": 0.7248076218395016,
      "grad_norm": 0.29160284045527163,
      "learning_rate": 9.402804462050776e-06,
      "loss": 0.4802,
      "step": 1978
    },
    {
      "epoch": 0.725174056430927,
      "grad_norm": 0.2952061057538384,
      "learning_rate": 9.401793676058135e-06,
      "loss": 0.4643,
      "step": 1979
    },
    {
      "epoch": 0.7255404910223525,
      "grad_norm": 0.3107861657585485,
      "learning_rate": 9.400782089807194e-06,
      "loss": 0.4666,
      "step": 1980
    },
    {
      "epoch": 0.7259069256137779,
      "grad_norm": 0.34643324015723087,
      "learning_rate": 9.399769703481865e-06,
      "loss": 0.4627,
      "step": 1981
    },
    {
      "epoch": 0.7262733602052034,
      "grad_norm": 0.305979661717283,
      "learning_rate": 9.3987565172662e-06,
      "loss": 0.4644,
      "step": 1982
    },
    {
      "epoch": 0.7266397947966288,
      "grad_norm": 0.32093668218694754,
      "learning_rate": 9.397742531344401e-06,
      "loss": 0.4632,
      "step": 1983
    },
    {
      "epoch": 0.7270062293880543,
      "grad_norm": 0.3112650496930872,
      "learning_rate": 9.396727745900811e-06,
      "loss": 0.4851,
      "step": 1984
    },
    {
      "epoch": 0.7273726639794796,
      "grad_norm": 1.027993342626995,
      "learning_rate": 9.395712161119924e-06,
      "loss": 0.474,
      "step": 1985
    },
    {
      "epoch": 0.7277390985709051,
      "grad_norm": 0.30334098796540615,
      "learning_rate": 9.394695777186373e-06,
      "loss": 0.4653,
      "step": 1986
    },
    {
      "epoch": 0.7281055331623305,
      "grad_norm": 0.2928219834587544,
      "learning_rate": 9.39367859428494e-06,
      "loss": 0.4772,
      "step": 1987
    },
    {
      "epoch": 0.728471967753756,
      "grad_norm": 0.32523367391407,
      "learning_rate": 9.392660612600555e-06,
      "loss": 0.4849,
      "step": 1988
    },
    {
      "epoch": 0.7288384023451814,
      "grad_norm": 0.30891410054656865,
      "learning_rate": 9.391641832318286e-06,
      "loss": 0.4777,
      "step": 1989
    },
    {
      "epoch": 0.7292048369366069,
      "grad_norm": 0.2952772969771446,
      "learning_rate": 9.390622253623353e-06,
      "loss": 0.4833,
      "step": 1990
    },
    {
      "epoch": 0.7295712715280323,
      "grad_norm": 0.33054094655339983,
      "learning_rate": 9.389601876701115e-06,
      "loss": 0.4777,
      "step": 1991
    },
    {
      "epoch": 0.7299377061194576,
      "grad_norm": 0.2993352726618424,
      "learning_rate": 9.388580701737082e-06,
      "loss": 0.4471,
      "step": 1992
    },
    {
      "epoch": 0.7303041407108831,
      "grad_norm": 0.3103230802952866,
      "learning_rate": 9.387558728916905e-06,
      "loss": 0.4504,
      "step": 1993
    },
    {
      "epoch": 0.7306705753023085,
      "grad_norm": 0.29740405381817164,
      "learning_rate": 9.386535958426382e-06,
      "loss": 0.4663,
      "step": 1994
    },
    {
      "epoch": 0.731037009893734,
      "grad_norm": 0.3317342667221009,
      "learning_rate": 9.385512390451456e-06,
      "loss": 0.4858,
      "step": 1995
    },
    {
      "epoch": 0.7314034444851594,
      "grad_norm": 0.2895033890165353,
      "learning_rate": 9.384488025178214e-06,
      "loss": 0.4819,
      "step": 1996
    },
    {
      "epoch": 0.7317698790765849,
      "grad_norm": 0.4076103188284932,
      "learning_rate": 9.383462862792887e-06,
      "loss": 0.48,
      "step": 1997
    },
    {
      "epoch": 0.7321363136680102,
      "grad_norm": 0.3174216442078818,
      "learning_rate": 9.382436903481856e-06,
      "loss": 0.4808,
      "step": 1998
    },
    {
      "epoch": 0.7325027482594357,
      "grad_norm": 0.3924769359594308,
      "learning_rate": 9.38141014743164e-06,
      "loss": 0.4546,
      "step": 1999
    },
    {
      "epoch": 0.7328691828508611,
      "grad_norm": 0.32518470115814135,
      "learning_rate": 9.380382594828906e-06,
      "loss": 0.4744,
      "step": 2000
    }
  ],
  "logging_steps": 1,
  "max_steps": 8187,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 712109932609536.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}