{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 24849,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0012072920439454304,
      "grad_norm": 9.275238037109375,
      "learning_rate": 4.5e-06,
      "loss": 2.9788,
      "step": 10
    },
    {
      "epoch": 0.002414584087890861,
      "grad_norm": 9.452595710754395,
      "learning_rate": 9.5e-06,
      "loss": 2.8308,
      "step": 20
    },
    {
      "epoch": 0.003621876131836291,
      "grad_norm": 6.564192295074463,
      "learning_rate": 1.45e-05,
      "loss": 2.5594,
      "step": 30
    },
    {
      "epoch": 0.004829168175781722,
      "grad_norm": 8.191930770874023,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 2.2765,
      "step": 40
    },
    {
      "epoch": 0.006036460219727152,
      "grad_norm": 7.330264568328857,
      "learning_rate": 2.45e-05,
      "loss": 1.9964,
      "step": 50
    },
    {
      "epoch": 0.007243752263672582,
      "grad_norm": 4.648830890655518,
      "learning_rate": 2.95e-05,
      "loss": 1.9472,
      "step": 60
    },
    {
      "epoch": 0.008451044307618012,
      "grad_norm": 7.797567844390869,
      "learning_rate": 3.45e-05,
      "loss": 2.2882,
      "step": 70
    },
    {
      "epoch": 0.009658336351563443,
      "grad_norm": 8.396490097045898,
      "learning_rate": 3.9500000000000005e-05,
      "loss": 1.9277,
      "step": 80
    },
    {
      "epoch": 0.010865628395508874,
      "grad_norm": 7.616096019744873,
      "learning_rate": 4.4500000000000004e-05,
      "loss": 2.0172,
      "step": 90
    },
    {
      "epoch": 0.012072920439454304,
      "grad_norm": 6.0004096031188965,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 2.1916,
      "step": 100
    },
    {
      "epoch": 0.013280212483399735,
      "grad_norm": 8.90979290008545,
      "learning_rate": 4.998181744716959e-05,
      "loss": 1.7532,
      "step": 110
    },
    {
      "epoch": 0.014487504527345164,
      "grad_norm": 10.78310775756836,
      "learning_rate": 4.9961614610691346e-05,
      "loss": 2.0114,
      "step": 120
    },
    {
      "epoch": 0.015694796571290595,
      "grad_norm": 10.907149314880371,
      "learning_rate": 4.9941411774213104e-05,
      "loss": 1.6906,
      "step": 130
    },
    {
      "epoch": 0.016902088615236024,
      "grad_norm": 12.938363075256348,
      "learning_rate": 4.9921208937734856e-05,
      "loss": 1.6514,
      "step": 140
    },
    {
      "epoch": 0.018109380659181457,
      "grad_norm": 11.261585235595703,
      "learning_rate": 4.990100610125662e-05,
      "loss": 1.9814,
      "step": 150
    },
    {
      "epoch": 0.019316672703126887,
      "grad_norm": 58.045230865478516,
      "learning_rate": 4.988080326477837e-05,
      "loss": 1.2393,
      "step": 160
    },
    {
      "epoch": 0.020523964747072316,
      "grad_norm": 13.091780662536621,
      "learning_rate": 4.986060042830014e-05,
      "loss": 1.6619,
      "step": 170
    },
    {
      "epoch": 0.02173125679101775,
      "grad_norm": 7.509953498840332,
      "learning_rate": 4.9840397591821896e-05,
      "loss": 1.5073,
      "step": 180
    },
    {
      "epoch": 0.022938548834963178,
      "grad_norm": 12.29320240020752,
      "learning_rate": 4.9820194755343655e-05,
      "loss": 1.7218,
      "step": 190
    },
    {
      "epoch": 0.024145840878908607,
      "grad_norm": 12.777913093566895,
      "learning_rate": 4.979999191886541e-05,
      "loss": 1.7257,
      "step": 200
    },
    {
      "epoch": 0.025353132922854037,
      "grad_norm": 6.258416652679443,
      "learning_rate": 4.9779789082387165e-05,
      "loss": 1.7946,
      "step": 210
    },
    {
      "epoch": 0.02656042496679947,
      "grad_norm": 24.06138801574707,
      "learning_rate": 4.975958624590893e-05,
      "loss": 1.9596,
      "step": 220
    },
    {
      "epoch": 0.0277677170107449,
      "grad_norm": 11.42760944366455,
      "learning_rate": 4.973938340943068e-05,
      "loss": 1.6279,
      "step": 230
    },
    {
      "epoch": 0.028975009054690328,
      "grad_norm": 8.943782806396484,
      "learning_rate": 4.971918057295245e-05,
      "loss": 1.5058,
      "step": 240
    },
    {
      "epoch": 0.03018230109863576,
      "grad_norm": 7.511715888977051,
      "learning_rate": 4.9698977736474206e-05,
      "loss": 1.6097,
      "step": 250
    },
    {
      "epoch": 0.03138959314258119,
      "grad_norm": 16.66579818725586,
      "learning_rate": 4.9678774899995964e-05,
      "loss": 1.7869,
      "step": 260
    },
    {
      "epoch": 0.03259688518652662,
      "grad_norm": 13.319332122802734,
      "learning_rate": 4.965857206351772e-05,
      "loss": 1.6985,
      "step": 270
    },
    {
      "epoch": 0.03380417723047205,
      "grad_norm": 7.227231502532959,
      "learning_rate": 4.9638369227039474e-05,
      "loss": 1.4694,
      "step": 280
    },
    {
      "epoch": 0.03501146927441748,
      "grad_norm": 8.204809188842773,
      "learning_rate": 4.961816639056124e-05,
      "loss": 1.6566,
      "step": 290
    },
    {
      "epoch": 0.036218761318362915,
      "grad_norm": 16.91630744934082,
      "learning_rate": 4.959796355408299e-05,
      "loss": 2.0499,
      "step": 300
    },
    {
      "epoch": 0.037426053362308344,
      "grad_norm": 10.307326316833496,
      "learning_rate": 4.9577760717604757e-05,
      "loss": 1.7973,
      "step": 310
    },
    {
      "epoch": 0.03863334540625377,
      "grad_norm": 7.013141632080078,
      "learning_rate": 4.9557557881126515e-05,
      "loss": 1.7598,
      "step": 320
    },
    {
      "epoch": 0.0398406374501992,
      "grad_norm": 11.12799072265625,
      "learning_rate": 4.9537355044648274e-05,
      "loss": 1.3294,
      "step": 330
    },
    {
      "epoch": 0.04104792949414463,
      "grad_norm": 18.232412338256836,
      "learning_rate": 4.951715220817003e-05,
      "loss": 1.4147,
      "step": 340
    },
    {
      "epoch": 0.04225522153809006,
      "grad_norm": 12.806726455688477,
      "learning_rate": 4.9496949371691784e-05,
      "loss": 1.6231,
      "step": 350
    },
    {
      "epoch": 0.0434625135820355,
      "grad_norm": 13.561619758605957,
      "learning_rate": 4.947674653521355e-05,
      "loss": 1.0687,
      "step": 360
    },
    {
      "epoch": 0.04466980562598093,
      "grad_norm": 12.685995101928711,
      "learning_rate": 4.94565436987353e-05,
      "loss": 1.7345,
      "step": 370
    },
    {
      "epoch": 0.045877097669926356,
      "grad_norm": 4.854790210723877,
      "learning_rate": 4.9436340862257066e-05,
      "loss": 1.6417,
      "step": 380
    },
    {
      "epoch": 0.047084389713871785,
      "grad_norm": 19.98137092590332,
      "learning_rate": 4.9416138025778824e-05,
      "loss": 1.4927,
      "step": 390
    },
    {
      "epoch": 0.048291681757817215,
      "grad_norm": 15.927617073059082,
      "learning_rate": 4.939593518930058e-05,
      "loss": 1.3949,
      "step": 400
    },
    {
      "epoch": 0.049498973801762644,
      "grad_norm": 18.98552894592285,
      "learning_rate": 4.937573235282234e-05,
      "loss": 1.7391,
      "step": 410
    },
    {
      "epoch": 0.05070626584570807,
      "grad_norm": 22.160247802734375,
      "learning_rate": 4.935552951634409e-05,
      "loss": 1.8821,
      "step": 420
    },
    {
      "epoch": 0.05191355788965351,
      "grad_norm": 28.589061737060547,
      "learning_rate": 4.933532667986586e-05,
      "loss": 1.5805,
      "step": 430
    },
    {
      "epoch": 0.05312084993359894,
      "grad_norm": 12.758599281311035,
      "learning_rate": 4.931512384338761e-05,
      "loss": 1.8922,
      "step": 440
    },
    {
      "epoch": 0.05432814197754437,
      "grad_norm": 8.21611213684082,
      "learning_rate": 4.9294921006909375e-05,
      "loss": 1.5596,
      "step": 450
    },
    {
      "epoch": 0.0555354340214898,
      "grad_norm": 6.6640777587890625,
      "learning_rate": 4.9274718170431134e-05,
      "loss": 1.3691,
      "step": 460
    },
    {
      "epoch": 0.05674272606543523,
      "grad_norm": 14.22506046295166,
      "learning_rate": 4.9254515333952885e-05,
      "loss": 1.4684,
      "step": 470
    },
    {
      "epoch": 0.057950018109380656,
      "grad_norm": 10.40058708190918,
      "learning_rate": 4.923431249747465e-05,
      "loss": 1.6429,
      "step": 480
    },
    {
      "epoch": 0.05915731015332609,
      "grad_norm": 10.339471817016602,
      "learning_rate": 4.92141096609964e-05,
      "loss": 1.2098,
      "step": 490
    },
    {
      "epoch": 0.06036460219727152,
      "grad_norm": 17.218908309936523,
      "learning_rate": 4.919390682451817e-05,
      "loss": 1.4122,
      "step": 500
    },
    {
      "epoch": 0.06157189424121695,
      "grad_norm": 7.813619136810303,
      "learning_rate": 4.917370398803992e-05,
      "loss": 1.839,
      "step": 510
    },
    {
      "epoch": 0.06277918628516238,
      "grad_norm": 8.563981056213379,
      "learning_rate": 4.9153501151561685e-05,
      "loss": 1.5228,
      "step": 520
    },
    {
      "epoch": 0.06398647832910781,
      "grad_norm": 13.853601455688477,
      "learning_rate": 4.913329831508344e-05,
      "loss": 1.6589,
      "step": 530
    },
    {
      "epoch": 0.06519377037305324,
      "grad_norm": 13.146242141723633,
      "learning_rate": 4.9113095478605195e-05,
      "loss": 1.413,
      "step": 540
    },
    {
      "epoch": 0.06640106241699867,
      "grad_norm": 11.832603454589844,
      "learning_rate": 4.909289264212696e-05,
      "loss": 1.3251,
      "step": 550
    },
    {
      "epoch": 0.0676083544609441,
      "grad_norm": 10.441021919250488,
      "learning_rate": 4.907268980564871e-05,
      "loss": 1.2825,
      "step": 560
    },
    {
      "epoch": 0.06881564650488953,
      "grad_norm": 20.174440383911133,
      "learning_rate": 4.905248696917048e-05,
      "loss": 1.0532,
      "step": 570
    },
    {
      "epoch": 0.07002293854883496,
      "grad_norm": 28.26873016357422,
      "learning_rate": 4.903228413269223e-05,
      "loss": 1.3134,
      "step": 580
    },
    {
      "epoch": 0.0712302305927804,
      "grad_norm": 11.23534870147705,
      "learning_rate": 4.9012081296213994e-05,
      "loss": 1.4265,
      "step": 590
    },
    {
      "epoch": 0.07243752263672583,
      "grad_norm": 14.088513374328613,
      "learning_rate": 4.899187845973575e-05,
      "loss": 1.7516,
      "step": 600
    },
    {
      "epoch": 0.07364481468067126,
      "grad_norm": 12.843744277954102,
      "learning_rate": 4.8971675623257504e-05,
      "loss": 1.5273,
      "step": 610
    },
    {
      "epoch": 0.07485210672461669,
      "grad_norm": 8.675108909606934,
      "learning_rate": 4.895147278677927e-05,
      "loss": 1.6928,
      "step": 620
    },
    {
      "epoch": 0.07605939876856212,
      "grad_norm": 10.954988479614258,
      "learning_rate": 4.893126995030102e-05,
      "loss": 1.4604,
      "step": 630
    },
    {
      "epoch": 0.07726669081250755,
      "grad_norm": 22.95113182067871,
      "learning_rate": 4.8911067113822786e-05,
      "loss": 1.5711,
      "step": 640
    },
    {
      "epoch": 0.07847398285645298,
      "grad_norm": 10.261492729187012,
      "learning_rate": 4.889086427734454e-05,
      "loss": 1.6281,
      "step": 650
    },
    {
      "epoch": 0.0796812749003984,
      "grad_norm": 14.267106056213379,
      "learning_rate": 4.88706614408663e-05,
      "loss": 1.6191,
      "step": 660
    },
    {
      "epoch": 0.08088856694434383,
      "grad_norm": 7.991502285003662,
      "learning_rate": 4.885045860438806e-05,
      "loss": 1.3924,
      "step": 670
    },
    {
      "epoch": 0.08209585898828926,
      "grad_norm": 9.072630882263184,
      "learning_rate": 4.883025576790981e-05,
      "loss": 1.7118,
      "step": 680
    },
    {
      "epoch": 0.08330315103223469,
      "grad_norm": 7.452462673187256,
      "learning_rate": 4.881005293143158e-05,
      "loss": 1.149,
      "step": 690
    },
    {
      "epoch": 0.08451044307618012,
      "grad_norm": 15.651378631591797,
      "learning_rate": 4.878985009495333e-05,
      "loss": 1.4979,
      "step": 700
    },
    {
      "epoch": 0.08571773512012555,
      "grad_norm": 15.411689758300781,
      "learning_rate": 4.8769647258475096e-05,
      "loss": 1.4243,
      "step": 710
    },
    {
      "epoch": 0.086925027164071,
      "grad_norm": 12.783531188964844,
      "learning_rate": 4.874944442199685e-05,
      "loss": 1.3898,
      "step": 720
    },
    {
      "epoch": 0.08813231920801642,
      "grad_norm": 18.151527404785156,
      "learning_rate": 4.8729241585518606e-05,
      "loss": 1.3177,
      "step": 730
    },
    {
      "epoch": 0.08933961125196185,
      "grad_norm": 11.814696311950684,
      "learning_rate": 4.870903874904037e-05,
      "loss": 1.2741,
      "step": 740
    },
    {
      "epoch": 0.09054690329590728,
      "grad_norm": 11.758604049682617,
      "learning_rate": 4.868883591256212e-05,
      "loss": 1.4683,
      "step": 750
    },
    {
      "epoch": 0.09175419533985271,
      "grad_norm": 18.662107467651367,
      "learning_rate": 4.866863307608389e-05,
      "loss": 1.4551,
      "step": 760
    },
    {
      "epoch": 0.09296148738379814,
      "grad_norm": 15.011588096618652,
      "learning_rate": 4.864843023960564e-05,
      "loss": 1.1149,
      "step": 770
    },
    {
      "epoch": 0.09416877942774357,
      "grad_norm": 14.521595001220703,
      "learning_rate": 4.8628227403127405e-05,
      "loss": 1.1625,
      "step": 780
    },
    {
      "epoch": 0.095376071471689,
      "grad_norm": 19.445602416992188,
      "learning_rate": 4.8608024566649157e-05,
      "loss": 1.2544,
      "step": 790
    },
    {
      "epoch": 0.09658336351563443,
      "grad_norm": 20.03233528137207,
      "learning_rate": 4.8587821730170915e-05,
      "loss": 1.8591,
      "step": 800
    },
    {
      "epoch": 0.09779065555957986,
      "grad_norm": 18.21596336364746,
      "learning_rate": 4.856761889369268e-05,
      "loss": 1.6532,
      "step": 810
    },
    {
      "epoch": 0.09899794760352529,
      "grad_norm": 20.792818069458008,
      "learning_rate": 4.854741605721443e-05,
      "loss": 1.8679,
      "step": 820
    },
    {
      "epoch": 0.10020523964747072,
      "grad_norm": 11.928009986877441,
      "learning_rate": 4.85272132207362e-05,
      "loss": 1.1994,
      "step": 830
    },
    {
      "epoch": 0.10141253169141615,
      "grad_norm": 25.11162757873535,
      "learning_rate": 4.850701038425795e-05,
      "loss": 1.6699,
      "step": 840
    },
    {
      "epoch": 0.10261982373536159,
      "grad_norm": 8.17197036743164,
      "learning_rate": 4.8486807547779714e-05,
      "loss": 1.1168,
      "step": 850
    },
    {
      "epoch": 0.10382711577930702,
      "grad_norm": 12.91263198852539,
      "learning_rate": 4.8466604711301466e-05,
      "loss": 1.3457,
      "step": 860
    },
    {
      "epoch": 0.10503440782325245,
      "grad_norm": 6.471170425415039,
      "learning_rate": 4.8446401874823224e-05,
      "loss": 1.6162,
      "step": 870
    },
    {
      "epoch": 0.10624169986719788,
      "grad_norm": 9.942479133605957,
      "learning_rate": 4.842619903834499e-05,
      "loss": 1.4697,
      "step": 880
    },
    {
      "epoch": 0.10744899191114331,
      "grad_norm": 10.31945514678955,
      "learning_rate": 4.840599620186674e-05,
      "loss": 1.8475,
      "step": 890
    },
    {
      "epoch": 0.10865628395508874,
      "grad_norm": 13.239459991455078,
      "learning_rate": 4.8385793365388507e-05,
      "loss": 1.5302,
      "step": 900
    },
    {
      "epoch": 0.10986357599903417,
      "grad_norm": 14.270588874816895,
      "learning_rate": 4.836559052891026e-05,
      "loss": 1.5908,
      "step": 910
    },
    {
      "epoch": 0.1110708680429796,
      "grad_norm": 11.940027236938477,
      "learning_rate": 4.8345387692432024e-05,
      "loss": 1.5045,
      "step": 920
    },
    {
      "epoch": 0.11227816008692502,
      "grad_norm": 6.636003017425537,
      "learning_rate": 4.8325184855953775e-05,
      "loss": 1.4601,
      "step": 930
    },
    {
      "epoch": 0.11348545213087045,
      "grad_norm": 6.393827438354492,
      "learning_rate": 4.8304982019475534e-05,
      "loss": 1.1863,
      "step": 940
    },
    {
      "epoch": 0.11469274417481588,
      "grad_norm": 16.41211700439453,
      "learning_rate": 4.82847791829973e-05,
      "loss": 1.4751,
      "step": 950
    },
    {
      "epoch": 0.11590003621876131,
      "grad_norm": 12.556045532226562,
      "learning_rate": 4.826457634651905e-05,
      "loss": 1.4615,
      "step": 960
    },
    {
      "epoch": 0.11710732826270674,
      "grad_norm": 9.441774368286133,
      "learning_rate": 4.8244373510040816e-05,
      "loss": 1.3514,
      "step": 970
    },
    {
      "epoch": 0.11831462030665219,
      "grad_norm": 10.880135536193848,
      "learning_rate": 4.822417067356257e-05,
      "loss": 1.3681,
      "step": 980
    },
    {
      "epoch": 0.11952191235059761,
      "grad_norm": 9.910809516906738,
      "learning_rate": 4.820396783708433e-05,
      "loss": 1.4222,
      "step": 990
    },
    {
      "epoch": 0.12072920439454304,
      "grad_norm": 15.677468299865723,
      "learning_rate": 4.8183765000606085e-05,
      "loss": 1.518,
      "step": 1000
    },
    {
      "epoch": 0.12193649643848847,
      "grad_norm": 4.044098854064941,
      "learning_rate": 4.816356216412784e-05,
      "loss": 1.6859,
      "step": 1010
    },
    {
      "epoch": 0.1231437884824339,
      "grad_norm": 14.098213195800781,
      "learning_rate": 4.814335932764961e-05,
      "loss": 1.1927,
      "step": 1020
    },
    {
      "epoch": 0.12435108052637933,
      "grad_norm": 7.806868553161621,
      "learning_rate": 4.812315649117136e-05,
      "loss": 1.3333,
      "step": 1030
    },
    {
      "epoch": 0.12555837257032476,
      "grad_norm": 15.979628562927246,
      "learning_rate": 4.8102953654693125e-05,
      "loss": 1.4089,
      "step": 1040
    },
    {
      "epoch": 0.1267656646142702,
      "grad_norm": 10.027960777282715,
      "learning_rate": 4.808275081821488e-05,
      "loss": 1.1557,
      "step": 1050
    },
    {
      "epoch": 0.12797295665821562,
      "grad_norm": 7.701732158660889,
      "learning_rate": 4.8062547981736635e-05,
      "loss": 1.279,
      "step": 1060
    },
    {
      "epoch": 0.12918024870216105,
      "grad_norm": 16.805604934692383,
      "learning_rate": 4.8042345145258394e-05,
      "loss": 1.1216,
      "step": 1070
    },
    {
      "epoch": 0.13038754074610648,
      "grad_norm": 17.213546752929688,
      "learning_rate": 4.802214230878015e-05,
      "loss": 1.155,
      "step": 1080
    },
    {
      "epoch": 0.1315948327900519,
      "grad_norm": 8.207999229431152,
      "learning_rate": 4.800193947230192e-05,
      "loss": 1.6443,
      "step": 1090
    },
    {
      "epoch": 0.13280212483399734,
      "grad_norm": 17.410259246826172,
      "learning_rate": 4.798173663582367e-05,
      "loss": 1.5332,
      "step": 1100
    },
    {
      "epoch": 0.13400941687794277,
      "grad_norm": 10.78391170501709,
      "learning_rate": 4.7961533799345435e-05,
      "loss": 1.3603,
      "step": 1110
    },
    {
      "epoch": 0.1352167089218882,
      "grad_norm": 14.313050270080566,
      "learning_rate": 4.7941330962867186e-05,
      "loss": 1.5066,
      "step": 1120
    },
    {
      "epoch": 0.13642400096583363,
      "grad_norm": 8.3862886428833,
      "learning_rate": 4.7921128126388945e-05,
      "loss": 1.3182,
      "step": 1130
    },
    {
      "epoch": 0.13763129300977905,
      "grad_norm": 6.555320739746094,
      "learning_rate": 4.79009252899107e-05,
      "loss": 1.6927,
      "step": 1140
    },
    {
      "epoch": 0.13883858505372448,
      "grad_norm": 5.530238151550293,
      "learning_rate": 4.788072245343246e-05,
      "loss": 1.4342,
      "step": 1150
    },
    {
      "epoch": 0.1400458770976699,
      "grad_norm": 13.069168090820312,
      "learning_rate": 4.786051961695423e-05,
      "loss": 1.635,
      "step": 1160
    },
    {
      "epoch": 0.14125316914161537,
      "grad_norm": 14.852108001708984,
      "learning_rate": 4.784031678047598e-05,
      "loss": 1.358,
      "step": 1170
    },
    {
      "epoch": 0.1424604611855608,
      "grad_norm": 16.23404312133789,
      "learning_rate": 4.7820113943997744e-05,
      "loss": 1.5927,
      "step": 1180
    },
    {
      "epoch": 0.14366775322950623,
      "grad_norm": 10.260706901550293,
      "learning_rate": 4.7799911107519496e-05,
      "loss": 1.6218,
      "step": 1190
    },
    {
      "epoch": 0.14487504527345166,
      "grad_norm": 18.92452049255371,
      "learning_rate": 4.7779708271041254e-05,
      "loss": 1.3995,
      "step": 1200
    },
    {
      "epoch": 0.1460823373173971,
      "grad_norm": 13.07711410522461,
      "learning_rate": 4.775950543456301e-05,
      "loss": 1.1995,
      "step": 1210
    },
    {
      "epoch": 0.14728962936134252,
      "grad_norm": 18.864797592163086,
      "learning_rate": 4.773930259808477e-05,
      "loss": 1.4071,
      "step": 1220
    },
    {
      "epoch": 0.14849692140528795,
      "grad_norm": 8.001008987426758,
      "learning_rate": 4.7719099761606536e-05,
      "loss": 1.1935,
      "step": 1230
    },
    {
      "epoch": 0.14970421344923338,
      "grad_norm": 10.160765647888184,
      "learning_rate": 4.769889692512829e-05,
      "loss": 1.5043,
      "step": 1240
    },
    {
      "epoch": 0.1509115054931788,
      "grad_norm": 24.56706428527832,
      "learning_rate": 4.767869408865005e-05,
      "loss": 1.0508,
      "step": 1250
    },
    {
      "epoch": 0.15211879753712423,
      "grad_norm": 7.233272075653076,
      "learning_rate": 4.7658491252171805e-05,
      "loss": 1.0304,
      "step": 1260
    },
    {
      "epoch": 0.15332608958106966,
      "grad_norm": 10.575057029724121,
      "learning_rate": 4.763828841569356e-05,
      "loss": 1.8129,
      "step": 1270
    },
    {
      "epoch": 0.1545333816250151,
      "grad_norm": 6.6973466873168945,
      "learning_rate": 4.761808557921533e-05,
      "loss": 1.7691,
      "step": 1280
    },
    {
      "epoch": 0.15574067366896052,
      "grad_norm": 21.352863311767578,
      "learning_rate": 4.759788274273708e-05,
      "loss": 1.2366,
      "step": 1290
    },
    {
      "epoch": 0.15694796571290595,
      "grad_norm": 11.174928665161133,
      "learning_rate": 4.7577679906258846e-05,
      "loss": 1.2173,
      "step": 1300
    },
    {
      "epoch": 0.15815525775685138,
      "grad_norm": 14.561117172241211,
      "learning_rate": 4.75574770697806e-05,
      "loss": 1.3908,
      "step": 1310
    },
    {
      "epoch": 0.1593625498007968,
      "grad_norm": 13.982307434082031,
      "learning_rate": 4.7537274233302356e-05,
      "loss": 1.3084,
      "step": 1320
    },
    {
      "epoch": 0.16056984184474224,
      "grad_norm": 12.887901306152344,
      "learning_rate": 4.7517071396824114e-05,
      "loss": 1.342,
      "step": 1330
    },
    {
      "epoch": 0.16177713388868767,
      "grad_norm": 8.075447082519531,
      "learning_rate": 4.749686856034587e-05,
      "loss": 1.326,
      "step": 1340
    },
    {
      "epoch": 0.1629844259326331,
      "grad_norm": 5.228328227996826,
      "learning_rate": 4.747666572386764e-05,
      "loss": 0.9845,
      "step": 1350
    },
    {
      "epoch": 0.16419171797657853,
      "grad_norm": 8.118184089660645,
      "learning_rate": 4.745646288738939e-05,
      "loss": 1.2093,
      "step": 1360
    },
    {
      "epoch": 0.16539901002052396,
      "grad_norm": 13.997716903686523,
      "learning_rate": 4.7436260050911155e-05,
      "loss": 1.603,
      "step": 1370
    },
    {
      "epoch": 0.16660630206446939,
      "grad_norm": 3.0456855297088623,
      "learning_rate": 4.7416057214432907e-05,
      "loss": 1.1583,
      "step": 1380
    },
    {
      "epoch": 0.16781359410841482,
      "grad_norm": 19.701936721801758,
      "learning_rate": 4.7395854377954665e-05,
      "loss": 1.79,
      "step": 1390
    },
    {
      "epoch": 0.16902088615236024,
      "grad_norm": 6.100048065185547,
      "learning_rate": 4.7375651541476424e-05,
      "loss": 1.4285,
      "step": 1400
    },
    {
      "epoch": 0.17022817819630567,
      "grad_norm": 32.51299285888672,
      "learning_rate": 4.735544870499818e-05,
      "loss": 1.007,
      "step": 1410
    },
    {
      "epoch": 0.1714354702402511,
      "grad_norm": 13.89777946472168,
      "learning_rate": 4.733524586851995e-05,
      "loss": 1.0662,
      "step": 1420
    },
    {
      "epoch": 0.17264276228419656,
      "grad_norm": 8.625432014465332,
      "learning_rate": 4.73150430320417e-05,
      "loss": 1.2265,
      "step": 1430
    },
    {
      "epoch": 0.173850054328142,
      "grad_norm": 12.52755355834961,
      "learning_rate": 4.7294840195563464e-05,
      "loss": 1.3175,
      "step": 1440
    },
    {
      "epoch": 0.17505734637208742,
      "grad_norm": 7.011743545532227,
      "learning_rate": 4.7274637359085216e-05,
      "loss": 1.3613,
      "step": 1450
    },
    {
      "epoch": 0.17626463841603285,
      "grad_norm": 27.03123664855957,
      "learning_rate": 4.7254434522606974e-05,
      "loss": 1.2327,
      "step": 1460
    },
    {
      "epoch": 0.17747193045997828,
      "grad_norm": 19.251615524291992,
      "learning_rate": 4.723423168612873e-05,
      "loss": 1.3975,
      "step": 1470
    },
    {
      "epoch": 0.1786792225039237,
      "grad_norm": 10.123787879943848,
      "learning_rate": 4.721402884965049e-05,
      "loss": 1.7086,
      "step": 1480
    },
    {
      "epoch": 0.17988651454786914,
      "grad_norm": 4.793344497680664,
      "learning_rate": 4.7193826013172257e-05,
      "loss": 0.9422,
      "step": 1490
    },
    {
      "epoch": 0.18109380659181457,
      "grad_norm": 15.687713623046875,
      "learning_rate": 4.717362317669401e-05,
      "loss": 1.1628,
      "step": 1500
    },
    {
      "epoch": 0.18230109863576,
      "grad_norm": 13.548020362854004,
      "learning_rate": 4.7153420340215774e-05,
      "loss": 1.2458,
      "step": 1510
    },
    {
      "epoch": 0.18350839067970542,
      "grad_norm": 14.213974952697754,
      "learning_rate": 4.7133217503737525e-05,
      "loss": 0.801,
      "step": 1520
    },
    {
      "epoch": 0.18471568272365085,
      "grad_norm": 21.768686294555664,
      "learning_rate": 4.7113014667259284e-05,
      "loss": 1.2402,
      "step": 1530
    },
    {
      "epoch": 0.18592297476759628,
      "grad_norm": 4.675285816192627,
      "learning_rate": 4.709281183078104e-05,
      "loss": 1.2738,
      "step": 1540
    },
    {
      "epoch": 0.1871302668115417,
      "grad_norm": 13.430849075317383,
      "learning_rate": 4.70726089943028e-05,
      "loss": 1.6566,
      "step": 1550
    },
    {
      "epoch": 0.18833755885548714,
      "grad_norm": 10.261524200439453,
      "learning_rate": 4.7052406157824566e-05,
      "loss": 1.4868,
      "step": 1560
    },
    {
      "epoch": 0.18954485089943257,
      "grad_norm": 26.612010955810547,
      "learning_rate": 4.703220332134632e-05,
      "loss": 1.3324,
      "step": 1570
    },
    {
      "epoch": 0.190752142943378,
      "grad_norm": 16.320634841918945,
      "learning_rate": 4.701200048486808e-05,
      "loss": 1.1528,
      "step": 1580
    },
    {
      "epoch": 0.19195943498732343,
      "grad_norm": 7.375401020050049,
      "learning_rate": 4.6991797648389835e-05,
      "loss": 1.7783,
      "step": 1590
    },
    {
      "epoch": 0.19316672703126886,
      "grad_norm": 12.007126808166504,
      "learning_rate": 4.697159481191159e-05,
      "loss": 1.2573,
      "step": 1600
    },
    {
      "epoch": 0.1943740190752143,
      "grad_norm": 10.93433952331543,
      "learning_rate": 4.695139197543335e-05,
      "loss": 1.3532,
      "step": 1610
    },
    {
      "epoch": 0.19558131111915972,
      "grad_norm": 20.017738342285156,
      "learning_rate": 4.693118913895511e-05,
      "loss": 1.3978,
      "step": 1620
    },
    {
      "epoch": 0.19678860316310515,
      "grad_norm": 7.2689738273620605,
      "learning_rate": 4.6910986302476875e-05,
      "loss": 1.4389,
      "step": 1630
    },
    {
      "epoch": 0.19799589520705058,
      "grad_norm": 7.010345935821533,
      "learning_rate": 4.689078346599863e-05,
      "loss": 1.5128,
      "step": 1640
    },
    {
      "epoch": 0.199203187250996,
      "grad_norm": 7.650420665740967,
      "learning_rate": 4.6870580629520385e-05,
      "loss": 1.1954,
      "step": 1650
    },
    {
      "epoch": 0.20041047929494143,
      "grad_norm": 34.956077575683594,
      "learning_rate": 4.6850377793042144e-05,
      "loss": 1.2292,
      "step": 1660
    },
    {
      "epoch": 0.20161777133888686,
      "grad_norm": 7.855569362640381,
      "learning_rate": 4.68301749565639e-05,
      "loss": 1.1295,
      "step": 1670
    },
    {
      "epoch": 0.2028250633828323,
      "grad_norm": 25.676971435546875,
      "learning_rate": 4.680997212008566e-05,
      "loss": 1.4634,
      "step": 1680
    },
    {
      "epoch": 0.20403235542677775,
      "grad_norm": 4.123684406280518,
      "learning_rate": 4.678976928360742e-05,
      "loss": 1.7712,
      "step": 1690
    },
    {
      "epoch": 0.20523964747072318,
      "grad_norm": 10.01121711730957,
      "learning_rate": 4.6769566447129185e-05,
      "loss": 1.3331,
      "step": 1700
    },
    {
      "epoch": 0.2064469395146686,
      "grad_norm": 5.998635292053223,
      "learning_rate": 4.6749363610650936e-05,
      "loss": 1.7638,
      "step": 1710
    },
    {
      "epoch": 0.20765423155861404,
      "grad_norm": 16.880443572998047,
      "learning_rate": 4.6729160774172695e-05,
      "loss": 1.4302,
      "step": 1720
    },
    {
      "epoch": 0.20886152360255947,
      "grad_norm": 7.825206279754639,
      "learning_rate": 4.670895793769445e-05,
      "loss": 1.4297,
      "step": 1730
    },
    {
      "epoch": 0.2100688156465049,
      "grad_norm": 10.868940353393555,
      "learning_rate": 4.668875510121621e-05,
      "loss": 1.3953,
      "step": 1740
    },
    {
      "epoch": 0.21127610769045033,
      "grad_norm": 9.225337028503418,
      "learning_rate": 4.666855226473797e-05,
      "loss": 1.0227,
      "step": 1750
    },
    {
      "epoch": 0.21248339973439576,
      "grad_norm": 4.306819438934326,
      "learning_rate": 4.664834942825973e-05,
      "loss": 0.9752,
      "step": 1760
    },
    {
      "epoch": 0.21369069177834119,
      "grad_norm": 6.5991692543029785,
      "learning_rate": 4.6628146591781494e-05,
      "loss": 1.2568,
      "step": 1770
    },
    {
      "epoch": 0.21489798382228661,
      "grad_norm": 10.09531307220459,
      "learning_rate": 4.6607943755303246e-05,
      "loss": 1.2129,
      "step": 1780
    },
    {
      "epoch": 0.21610527586623204,
      "grad_norm": 60.62458801269531,
      "learning_rate": 4.6587740918825004e-05,
      "loss": 1.4717,
      "step": 1790
    },
    {
      "epoch": 0.21731256791017747,
      "grad_norm": 37.9005126953125,
      "learning_rate": 4.656753808234676e-05,
      "loss": 1.6294,
      "step": 1800
    },
    {
      "epoch": 0.2185198599541229,
      "grad_norm": 8.195784568786621,
      "learning_rate": 4.654733524586852e-05,
      "loss": 1.4433,
      "step": 1810
    },
    {
      "epoch": 0.21972715199806833,
      "grad_norm": 9.735424995422363,
      "learning_rate": 4.652713240939028e-05,
      "loss": 1.1082,
      "step": 1820
    },
    {
      "epoch": 0.22093444404201376,
      "grad_norm": 12.799370765686035,
      "learning_rate": 4.650692957291204e-05,
      "loss": 1.152,
      "step": 1830
    },
    {
      "epoch": 0.2221417360859592,
      "grad_norm": 17.47879409790039,
      "learning_rate": 4.64867267364338e-05,
      "loss": 1.2594,
      "step": 1840
    },
    {
      "epoch": 0.22334902812990462,
      "grad_norm": 16.908063888549805,
      "learning_rate": 4.6466523899955555e-05,
      "loss": 1.4419,
      "step": 1850
    },
    {
      "epoch": 0.22455632017385005,
      "grad_norm": 11.247140884399414,
      "learning_rate": 4.6446321063477313e-05,
      "loss": 1.5153,
      "step": 1860
    },
    {
      "epoch": 0.22576361221779548,
      "grad_norm": 13.506928443908691,
      "learning_rate": 4.642611822699907e-05,
      "loss": 1.527,
      "step": 1870
    },
    {
      "epoch": 0.2269709042617409,
      "grad_norm": 20.46377182006836,
      "learning_rate": 4.640591539052083e-05,
      "loss": 1.3863,
      "step": 1880
    },
    {
      "epoch": 0.22817819630568634,
      "grad_norm": 21.828020095825195,
      "learning_rate": 4.638571255404259e-05,
      "loss": 1.1371,
      "step": 1890
    },
    {
      "epoch": 0.22938548834963177,
      "grad_norm": 4.66279411315918,
      "learning_rate": 4.636550971756435e-05,
      "loss": 1.2135,
      "step": 1900
    },
    {
      "epoch": 0.2305927803935772,
      "grad_norm": 9.881929397583008,
      "learning_rate": 4.6345306881086106e-05,
      "loss": 1.3616,
      "step": 1910
    },
    {
      "epoch": 0.23180007243752263,
      "grad_norm": 15.277690887451172,
      "learning_rate": 4.6325104044607864e-05,
      "loss": 1.09,
      "step": 1920
    },
    {
      "epoch": 0.23300736448146805,
      "grad_norm": 21.580551147460938,
      "learning_rate": 4.630490120812962e-05,
      "loss": 1.2027,
      "step": 1930
    },
    {
      "epoch": 0.23421465652541348,
      "grad_norm": 28.54668617248535,
      "learning_rate": 4.628469837165138e-05,
      "loss": 1.378,
      "step": 1940
    },
    {
      "epoch": 0.23542194856935894,
      "grad_norm": 27.845537185668945,
      "learning_rate": 4.626449553517314e-05,
      "loss": 1.7166,
      "step": 1950
    },
    {
      "epoch": 0.23662924061330437,
      "grad_norm": 7.06273078918457,
      "learning_rate": 4.62442926986949e-05,
      "loss": 1.2312,
      "step": 1960
    },
    {
      "epoch": 0.2378365326572498,
      "grad_norm": 9.48945426940918,
      "learning_rate": 4.622408986221666e-05,
      "loss": 1.3555,
      "step": 1970
    },
    {
      "epoch": 0.23904382470119523,
      "grad_norm": 13.238455772399902,
      "learning_rate": 4.6203887025738415e-05,
      "loss": 1.5681,
      "step": 1980
    },
    {
      "epoch": 0.24025111674514066,
      "grad_norm": 19.26911735534668,
      "learning_rate": 4.6183684189260174e-05,
      "loss": 1.4444,
      "step": 1990
    },
    {
      "epoch": 0.2414584087890861,
      "grad_norm": 14.012389183044434,
      "learning_rate": 4.616348135278193e-05,
      "loss": 1.1536,
      "step": 2000
    },
    {
      "epoch": 0.24266570083303152,
      "grad_norm": 12.465118408203125,
      "learning_rate": 4.614327851630369e-05,
      "loss": 1.4052,
      "step": 2010
    },
    {
      "epoch": 0.24387299287697695,
      "grad_norm": 12.004776000976562,
      "learning_rate": 4.612307567982545e-05,
      "loss": 1.2122,
      "step": 2020
    },
    {
      "epoch": 0.24508028492092238,
      "grad_norm": 10.17117977142334,
      "learning_rate": 4.610287284334721e-05,
      "loss": 1.9346,
      "step": 2030
    },
    {
      "epoch": 0.2462875769648678,
      "grad_norm": 10.781609535217285,
      "learning_rate": 4.6082670006868966e-05,
      "loss": 1.0924,
      "step": 2040
    },
    {
      "epoch": 0.24749486900881323,
      "grad_norm": 24.12435531616211,
      "learning_rate": 4.6062467170390724e-05,
      "loss": 1.4508,
      "step": 2050
    },
    {
      "epoch": 0.24870216105275866,
      "grad_norm": 15.126108169555664,
      "learning_rate": 4.604226433391248e-05,
      "loss": 1.3745,
      "step": 2060
    },
    {
      "epoch": 0.2499094530967041,
      "grad_norm": 9.983051300048828,
      "learning_rate": 4.602206149743424e-05,
      "loss": 1.4068,
      "step": 2070
    },
    {
      "epoch": 0.2511167451406495,
      "grad_norm": 13.977287292480469,
      "learning_rate": 4.6001858660956e-05,
      "loss": 1.2594,
      "step": 2080
    },
    {
      "epoch": 0.25232403718459495,
      "grad_norm": 11.70623779296875,
      "learning_rate": 4.598165582447776e-05,
      "loss": 1.2123,
      "step": 2090
    },
    {
      "epoch": 0.2535313292285404,
      "grad_norm": 5.455778121948242,
      "learning_rate": 4.596145298799952e-05,
      "loss": 1.3749,
      "step": 2100
    },
    {
      "epoch": 0.2547386212724858,
      "grad_norm": 10.889780044555664,
      "learning_rate": 4.5941250151521275e-05,
      "loss": 1.0532,
      "step": 2110
    },
    {
      "epoch": 0.25594591331643124,
      "grad_norm": 13.141742706298828,
      "learning_rate": 4.5921047315043034e-05,
      "loss": 1.4611,
      "step": 2120
    },
    {
      "epoch": 0.25715320536037667,
      "grad_norm": 10.340712547302246,
      "learning_rate": 4.590084447856479e-05,
      "loss": 1.2444,
      "step": 2130
    },
    {
      "epoch": 0.2583604974043221,
      "grad_norm": 5.682750225067139,
      "learning_rate": 4.588064164208655e-05,
      "loss": 1.5581,
      "step": 2140
    },
    {
      "epoch": 0.2595677894482675,
      "grad_norm": 9.827606201171875,
      "learning_rate": 4.586043880560831e-05,
      "loss": 1.1955,
      "step": 2150
    },
    {
      "epoch": 0.26077508149221296,
      "grad_norm": 10.144163131713867,
      "learning_rate": 4.584023596913007e-05,
      "loss": 1.4239,
      "step": 2160
    },
    {
      "epoch": 0.2619823735361584,
      "grad_norm": 3.9006004333496094,
      "learning_rate": 4.5820033132651826e-05,
      "loss": 1.1671,
      "step": 2170
    },
    {
      "epoch": 0.2631896655801038,
      "grad_norm": 9.541402816772461,
      "learning_rate": 4.5799830296173585e-05,
      "loss": 1.295,
      "step": 2180
    },
    {
      "epoch": 0.26439695762404924,
      "grad_norm": 12.119535446166992,
      "learning_rate": 4.577962745969534e-05,
      "loss": 0.9746,
      "step": 2190
    },
    {
      "epoch": 0.2656042496679947,
      "grad_norm": 7.668707370758057,
      "learning_rate": 4.57594246232171e-05,
      "loss": 1.107,
      "step": 2200
    },
    {
      "epoch": 0.2668115417119401,
      "grad_norm": 17.13339614868164,
      "learning_rate": 4.573922178673886e-05,
      "loss": 1.1998,
      "step": 2210
    },
    {
      "epoch": 0.26801883375588553,
      "grad_norm": 23.124265670776367,
      "learning_rate": 4.571901895026062e-05,
      "loss": 1.6125,
      "step": 2220
    },
    {
      "epoch": 0.26922612579983096,
      "grad_norm": 8.596065521240234,
      "learning_rate": 4.569881611378238e-05,
      "loss": 1.1312,
      "step": 2230
    },
    {
      "epoch": 0.2704334178437764,
      "grad_norm": 12.597529411315918,
      "learning_rate": 4.5678613277304135e-05,
      "loss": 1.3432,
      "step": 2240
    },
    {
      "epoch": 0.2716407098877218,
      "grad_norm": 5.108142852783203,
      "learning_rate": 4.5658410440825894e-05,
      "loss": 0.9722,
      "step": 2250
    },
    {
      "epoch": 0.27284800193166725,
      "grad_norm": 21.639436721801758,
      "learning_rate": 4.563820760434765e-05,
      "loss": 1.5108,
      "step": 2260
    },
    {
      "epoch": 0.2740552939756127,
      "grad_norm": 16.49915313720703,
      "learning_rate": 4.561800476786941e-05,
      "loss": 0.9613,
      "step": 2270
    },
    {
      "epoch": 0.2752625860195581,
      "grad_norm": 6.012657165527344,
      "learning_rate": 4.559780193139117e-05,
      "loss": 0.8367,
      "step": 2280
    },
    {
      "epoch": 0.27646987806350354,
      "grad_norm": 46.227638244628906,
      "learning_rate": 4.557759909491293e-05,
      "loss": 1.3591,
      "step": 2290
    },
    {
      "epoch": 0.27767717010744897,
      "grad_norm": 15.463981628417969,
      "learning_rate": 4.5557396258434686e-05,
      "loss": 1.141,
      "step": 2300
    },
    {
      "epoch": 0.2788844621513944,
      "grad_norm": 23.341672897338867,
      "learning_rate": 4.5537193421956445e-05,
      "loss": 1.4015,
      "step": 2310
    },
    {
      "epoch": 0.2800917541953398,
      "grad_norm": 9.3795804977417,
      "learning_rate": 4.55169905854782e-05,
      "loss": 1.0225,
      "step": 2320
    },
    {
      "epoch": 0.2812990462392853,
      "grad_norm": 5.406980037689209,
      "learning_rate": 4.549678774899996e-05,
      "loss": 1.1794,
      "step": 2330
    },
    {
      "epoch": 0.28250633828323074,
      "grad_norm": 55.37160110473633,
      "learning_rate": 4.547658491252172e-05,
      "loss": 1.7329,
      "step": 2340
    },
    {
      "epoch": 0.28371363032717617,
      "grad_norm": 11.730996131896973,
      "learning_rate": 4.545638207604348e-05,
      "loss": 1.3531,
      "step": 2350
    },
    {
      "epoch": 0.2849209223711216,
      "grad_norm": 26.810592651367188,
      "learning_rate": 4.543617923956524e-05,
      "loss": 1.0099,
      "step": 2360
    },
    {
      "epoch": 0.28612821441506703,
      "grad_norm": 5.320415019989014,
      "learning_rate": 4.5415976403086996e-05,
      "loss": 0.9144,
      "step": 2370
    },
    {
      "epoch": 0.28733550645901246,
      "grad_norm": 9.783456802368164,
      "learning_rate": 4.5395773566608754e-05,
      "loss": 1.2418,
      "step": 2380
    },
    {
      "epoch": 0.2885427985029579,
      "grad_norm": 13.23823070526123,
      "learning_rate": 4.537557073013051e-05,
      "loss": 1.7033,
      "step": 2390
    },
    {
      "epoch": 0.2897500905469033,
      "grad_norm": 21.210071563720703,
      "learning_rate": 4.535536789365227e-05,
      "loss": 1.2961,
      "step": 2400
    },
    {
      "epoch": 0.29095738259084875,
      "grad_norm": 10.58446979522705,
      "learning_rate": 4.533516505717403e-05,
      "loss": 1.6142,
      "step": 2410
    },
    {
      "epoch": 0.2921646746347942,
      "grad_norm": 13.622685432434082,
      "learning_rate": 4.531496222069579e-05,
      "loss": 1.61,
      "step": 2420
    },
    {
      "epoch": 0.2933719666787396,
      "grad_norm": 12.963950157165527,
      "learning_rate": 4.5294759384217546e-05,
      "loss": 1.2118,
      "step": 2430
    },
    {
      "epoch": 0.29457925872268503,
      "grad_norm": 14.19684886932373,
      "learning_rate": 4.5274556547739305e-05,
      "loss": 1.1937,
      "step": 2440
    },
    {
      "epoch": 0.29578655076663046,
      "grad_norm": 14.39604377746582,
      "learning_rate": 4.5254353711261063e-05,
      "loss": 1.3376,
      "step": 2450
    },
    {
      "epoch": 0.2969938428105759,
      "grad_norm": 4.026514053344727,
      "learning_rate": 4.523415087478282e-05,
      "loss": 1.3256,
      "step": 2460
    },
    {
      "epoch": 0.2982011348545213,
      "grad_norm": 27.728099822998047,
      "learning_rate": 4.521394803830458e-05,
      "loss": 1.4356,
      "step": 2470
    },
    {
      "epoch": 0.29940842689846675,
      "grad_norm": 9.098320960998535,
      "learning_rate": 4.519374520182634e-05,
      "loss": 1.1717,
      "step": 2480
    },
    {
      "epoch": 0.3006157189424122,
      "grad_norm": 12.053803443908691,
      "learning_rate": 4.51735423653481e-05,
      "loss": 1.0712,
      "step": 2490
    },
    {
      "epoch": 0.3018230109863576,
      "grad_norm": 5.71108865737915,
      "learning_rate": 4.5153339528869856e-05,
      "loss": 1.1777,
      "step": 2500
    },
    {
      "epoch": 0.30303030303030304,
      "grad_norm": 17.753007888793945,
      "learning_rate": 4.5133136692391614e-05,
      "loss": 0.927,
      "step": 2510
    },
    {
      "epoch": 0.30423759507424847,
      "grad_norm": 26.85407829284668,
      "learning_rate": 4.511293385591337e-05,
      "loss": 1.6259,
      "step": 2520
    },
    {
      "epoch": 0.3054448871181939,
      "grad_norm": 10.358158111572266,
      "learning_rate": 4.509273101943513e-05,
      "loss": 1.3677,
      "step": 2530
    },
    {
      "epoch": 0.3066521791621393,
      "grad_norm": 21.72525405883789,
      "learning_rate": 4.507252818295689e-05,
      "loss": 1.4709,
      "step": 2540
    },
    {
      "epoch": 0.30785947120608476,
      "grad_norm": 4.551169395446777,
      "learning_rate": 4.505232534647865e-05,
      "loss": 1.0104,
      "step": 2550
    },
    {
      "epoch": 0.3090667632500302,
      "grad_norm": 22.30368423461914,
      "learning_rate": 4.503212251000041e-05,
      "loss": 0.9084,
      "step": 2560
    },
    {
      "epoch": 0.3102740552939756,
      "grad_norm": 20.115114212036133,
      "learning_rate": 4.5011919673522165e-05,
      "loss": 0.8655,
      "step": 2570
    },
    {
      "epoch": 0.31148134733792104,
      "grad_norm": 13.660867691040039,
      "learning_rate": 4.4991716837043924e-05,
      "loss": 1.7534,
      "step": 2580
    },
    {
      "epoch": 0.3126886393818665,
      "grad_norm": 18.70029640197754,
      "learning_rate": 4.497151400056568e-05,
      "loss": 1.3612,
      "step": 2590
    },
    {
      "epoch": 0.3138959314258119,
      "grad_norm": 7.6045918464660645,
      "learning_rate": 4.495131116408744e-05,
      "loss": 1.2088,
      "step": 2600
    },
    {
      "epoch": 0.31510322346975733,
      "grad_norm": 24.201080322265625,
      "learning_rate": 4.49311083276092e-05,
      "loss": 1.4048,
      "step": 2610
    },
    {
      "epoch": 0.31631051551370276,
      "grad_norm": 17.77806854248047,
      "learning_rate": 4.491090549113096e-05,
      "loss": 1.2488,
      "step": 2620
    },
    {
      "epoch": 0.3175178075576482,
      "grad_norm": 13.561559677124023,
      "learning_rate": 4.4890702654652716e-05,
      "loss": 1.4018,
      "step": 2630
    },
    {
      "epoch": 0.3187250996015936,
      "grad_norm": 20.884668350219727,
      "learning_rate": 4.4870499818174474e-05,
      "loss": 1.2615,
      "step": 2640
    },
    {
      "epoch": 0.31993239164553905,
      "grad_norm": 13.134846687316895,
      "learning_rate": 4.485029698169623e-05,
      "loss": 1.5482,
      "step": 2650
    },
    {
      "epoch": 0.3211396836894845,
      "grad_norm": 13.651280403137207,
      "learning_rate": 4.483009414521799e-05,
      "loss": 1.2608,
      "step": 2660
    },
    {
      "epoch": 0.3223469757334299,
      "grad_norm": 8.194324493408203,
      "learning_rate": 4.480989130873975e-05,
      "loss": 1.5103,
      "step": 2670
    },
    {
      "epoch": 0.32355426777737534,
      "grad_norm": 14.692306518554688,
      "learning_rate": 4.478968847226151e-05,
      "loss": 1.1334,
      "step": 2680
    },
    {
      "epoch": 0.32476155982132077,
      "grad_norm": 12.189238548278809,
      "learning_rate": 4.476948563578327e-05,
      "loss": 1.5735,
      "step": 2690
    },
    {
      "epoch": 0.3259688518652662,
      "grad_norm": 9.967313766479492,
      "learning_rate": 4.4749282799305025e-05,
      "loss": 1.1465,
      "step": 2700
    },
    {
      "epoch": 0.3271761439092116,
      "grad_norm": 12.82974624633789,
      "learning_rate": 4.4729079962826784e-05,
      "loss": 1.1387,
      "step": 2710
    },
    {
      "epoch": 0.32838343595315705,
      "grad_norm": 7.670962810516357,
      "learning_rate": 4.470887712634854e-05,
      "loss": 1.0769,
      "step": 2720
    },
    {
      "epoch": 0.3295907279971025,
      "grad_norm": 7.622762203216553,
      "learning_rate": 4.46886742898703e-05,
      "loss": 1.3424,
      "step": 2730
    },
    {
      "epoch": 0.3307980200410479,
      "grad_norm": 16.582876205444336,
      "learning_rate": 4.466847145339206e-05,
      "loss": 0.9649,
      "step": 2740
    },
    {
      "epoch": 0.33200531208499334,
      "grad_norm": 23.246932983398438,
      "learning_rate": 4.464826861691382e-05,
      "loss": 1.6987,
      "step": 2750
    },
    {
      "epoch": 0.33321260412893877,
      "grad_norm": 6.82628870010376,
      "learning_rate": 4.4628065780435576e-05,
      "loss": 1.4019,
      "step": 2760
    },
    {
      "epoch": 0.3344198961728842,
      "grad_norm": 19.110023498535156,
      "learning_rate": 4.4607862943957335e-05,
      "loss": 1.2315,
      "step": 2770
    },
    {
      "epoch": 0.33562718821682963,
      "grad_norm": 8.397043228149414,
      "learning_rate": 4.458766010747909e-05,
      "loss": 1.431,
      "step": 2780
    },
    {
      "epoch": 0.33683448026077506,
      "grad_norm": 14.006730079650879,
      "learning_rate": 4.456745727100085e-05,
      "loss": 1.1916,
      "step": 2790
    },
    {
      "epoch": 0.3380417723047205,
      "grad_norm": 14.785843849182129,
      "learning_rate": 4.454725443452261e-05,
      "loss": 1.2145,
      "step": 2800
    },
    {
      "epoch": 0.3392490643486659,
      "grad_norm": 27.793302536010742,
      "learning_rate": 4.452705159804437e-05,
      "loss": 0.9969,
      "step": 2810
    },
    {
      "epoch": 0.34045635639261135,
      "grad_norm": 18.20464324951172,
      "learning_rate": 4.450684876156613e-05,
      "loss": 1.0573,
      "step": 2820
    },
    {
      "epoch": 0.3416636484365568,
      "grad_norm": 16.448253631591797,
      "learning_rate": 4.4486645925087885e-05,
      "loss": 1.4611,
      "step": 2830
    },
    {
      "epoch": 0.3428709404805022,
      "grad_norm": 15.40036678314209,
      "learning_rate": 4.4466443088609644e-05,
      "loss": 1.3228,
      "step": 2840
    },
    {
      "epoch": 0.3440782325244477,
      "grad_norm": 18.153051376342773,
      "learning_rate": 4.44462402521314e-05,
      "loss": 1.2209,
      "step": 2850
    },
    {
      "epoch": 0.3452855245683931,
      "grad_norm": 7.2596116065979,
      "learning_rate": 4.442603741565316e-05,
      "loss": 1.2455,
      "step": 2860
    },
    {
      "epoch": 0.34649281661233855,
      "grad_norm": 11.477952003479004,
      "learning_rate": 4.440583457917492e-05,
      "loss": 1.2597,
      "step": 2870
    },
    {
      "epoch": 0.347700108656284,
      "grad_norm": 6.482874870300293,
      "learning_rate": 4.438563174269668e-05,
      "loss": 0.8658,
      "step": 2880
    },
    {
      "epoch": 0.3489074007002294,
      "grad_norm": 35.54882049560547,
      "learning_rate": 4.4365428906218436e-05,
      "loss": 1.0959,
      "step": 2890
    },
    {
      "epoch": 0.35011469274417484,
      "grad_norm": 15.930017471313477,
      "learning_rate": 4.4345226069740195e-05,
      "loss": 1.2369,
      "step": 2900
    },
    {
      "epoch": 0.35132198478812027,
      "grad_norm": 41.58671188354492,
      "learning_rate": 4.432502323326195e-05,
      "loss": 1.3944,
      "step": 2910
    },
    {
      "epoch": 0.3525292768320657,
      "grad_norm": 23.988384246826172,
      "learning_rate": 4.4304820396783705e-05,
      "loss": 1.1431,
      "step": 2920
    },
    {
      "epoch": 0.3537365688760111,
      "grad_norm": 30.049528121948242,
      "learning_rate": 4.428461756030547e-05,
      "loss": 1.4105,
      "step": 2930
    },
    {
      "epoch": 0.35494386091995656,
      "grad_norm": 12.288514137268066,
      "learning_rate": 4.426441472382723e-05,
      "loss": 1.3671,
      "step": 2940
    },
    {
      "epoch": 0.356151152963902,
      "grad_norm": 21.20829963684082,
      "learning_rate": 4.424421188734899e-05,
      "loss": 1.1026,
      "step": 2950
    },
    {
      "epoch": 0.3573584450078474,
      "grad_norm": 18.07496452331543,
      "learning_rate": 4.4224009050870746e-05,
      "loss": 1.3093,
      "step": 2960
    },
    {
      "epoch": 0.35856573705179284,
      "grad_norm": 27.410478591918945,
      "learning_rate": 4.4203806214392504e-05,
      "loss": 1.603,
      "step": 2970
    },
    {
      "epoch": 0.3597730290957383,
      "grad_norm": 28.87078857421875,
      "learning_rate": 4.418360337791426e-05,
      "loss": 1.0971,
      "step": 2980
    },
    {
      "epoch": 0.3609803211396837,
      "grad_norm": 14.704648971557617,
      "learning_rate": 4.4163400541436014e-05,
      "loss": 1.3089,
      "step": 2990
    },
    {
      "epoch": 0.36218761318362913,
      "grad_norm": 14.575715065002441,
      "learning_rate": 4.414319770495778e-05,
      "loss": 1.6346,
      "step": 3000
    },
    {
      "epoch": 0.36339490522757456,
      "grad_norm": 12.183820724487305,
      "learning_rate": 4.412299486847954e-05,
      "loss": 1.3566,
      "step": 3010
    },
    {
      "epoch": 0.36460219727152,
      "grad_norm": 16.289274215698242,
      "learning_rate": 4.4102792032001297e-05,
      "loss": 1.5175,
      "step": 3020
    },
    {
      "epoch": 0.3658094893154654,
      "grad_norm": 5.464224815368652,
      "learning_rate": 4.4082589195523055e-05,
      "loss": 0.9785,
      "step": 3030
    },
    {
      "epoch": 0.36701678135941085,
      "grad_norm": 10.966645240783691,
      "learning_rate": 4.4062386359044813e-05,
      "loss": 1.4365,
      "step": 3040
    },
    {
      "epoch": 0.3682240734033563,
      "grad_norm": 9.651285171508789,
      "learning_rate": 4.404218352256657e-05,
      "loss": 1.6014,
      "step": 3050
    },
    {
      "epoch": 0.3694313654473017,
      "grad_norm": 17.764057159423828,
      "learning_rate": 4.4021980686088324e-05,
      "loss": 1.4559,
      "step": 3060
    },
    {
      "epoch": 0.37063865749124714,
      "grad_norm": 6.9669270515441895,
      "learning_rate": 4.400177784961009e-05,
      "loss": 0.9335,
      "step": 3070
    },
    {
      "epoch": 0.37184594953519257,
      "grad_norm": 16.282068252563477,
      "learning_rate": 4.398157501313185e-05,
      "loss": 1.0428,
      "step": 3080
    },
    {
      "epoch": 0.373053241579138,
      "grad_norm": 14.208762168884277,
      "learning_rate": 4.3961372176653606e-05,
      "loss": 1.2186,
      "step": 3090
    },
    {
      "epoch": 0.3742605336230834,
      "grad_norm": 23.55340003967285,
      "learning_rate": 4.3941169340175364e-05,
      "loss": 1.2596,
      "step": 3100
    },
    {
      "epoch": 0.37546782566702885,
      "grad_norm": 5.546447277069092,
      "learning_rate": 4.392096650369712e-05,
      "loss": 0.8901,
      "step": 3110
    },
    {
      "epoch": 0.3766751177109743,
      "grad_norm": 22.360876083374023,
      "learning_rate": 4.390076366721888e-05,
      "loss": 1.4893,
      "step": 3120
    },
    {
      "epoch": 0.3778824097549197,
      "grad_norm": 7.788496971130371,
      "learning_rate": 4.388056083074063e-05,
      "loss": 1.2426,
      "step": 3130
    },
    {
      "epoch": 0.37908970179886514,
      "grad_norm": 8.953337669372559,
      "learning_rate": 4.38603579942624e-05,
      "loss": 0.9917,
      "step": 3140
    },
    {
      "epoch": 0.38029699384281057,
      "grad_norm": 34.04292297363281,
      "learning_rate": 4.384015515778416e-05,
      "loss": 1.4174,
      "step": 3150
    },
    {
      "epoch": 0.381504285886756,
      "grad_norm": 10.96490478515625,
      "learning_rate": 4.3819952321305915e-05,
      "loss": 1.3436,
      "step": 3160
    },
    {
      "epoch": 0.38271157793070143,
      "grad_norm": 6.29488468170166,
      "learning_rate": 4.3799749484827674e-05,
      "loss": 1.6587,
      "step": 3170
    },
    {
      "epoch": 0.38391886997464686,
      "grad_norm": 9.935296058654785,
      "learning_rate": 4.377954664834943e-05,
      "loss": 0.9923,
      "step": 3180
    },
    {
      "epoch": 0.3851261620185923,
      "grad_norm": 5.370090007781982,
      "learning_rate": 4.375934381187119e-05,
      "loss": 1.1853,
      "step": 3190
    },
    {
      "epoch": 0.3863334540625377,
      "grad_norm": 34.48350143432617,
      "learning_rate": 4.373914097539294e-05,
      "loss": 1.1889,
      "step": 3200
    },
    {
      "epoch": 0.38754074610648315,
      "grad_norm": 7.0832343101501465,
      "learning_rate": 4.371893813891471e-05,
      "loss": 1.2137,
      "step": 3210
    },
    {
      "epoch": 0.3887480381504286,
      "grad_norm": 14.58543872833252,
      "learning_rate": 4.3698735302436466e-05,
      "loss": 1.1857,
      "step": 3220
    },
    {
      "epoch": 0.389955330194374,
      "grad_norm": 10.111153602600098,
      "learning_rate": 4.3678532465958224e-05,
      "loss": 1.0776,
      "step": 3230
    },
    {
      "epoch": 0.39116262223831944,
      "grad_norm": 29.29326057434082,
      "learning_rate": 4.365832962947998e-05,
      "loss": 1.1193,
      "step": 3240
    },
    {
      "epoch": 0.39236991428226486,
      "grad_norm": 9.426629066467285,
      "learning_rate": 4.3638126793001735e-05,
      "loss": 1.1566,
      "step": 3250
    },
    {
      "epoch": 0.3935772063262103,
      "grad_norm": 18.605775833129883,
      "learning_rate": 4.36179239565235e-05,
      "loss": 1.2413,
      "step": 3260
    },
    {
      "epoch": 0.3947844983701557,
      "grad_norm": 14.978432655334473,
      "learning_rate": 4.359772112004525e-05,
      "loss": 1.6638,
      "step": 3270
    },
    {
      "epoch": 0.39599179041410115,
      "grad_norm": 11.577205657958984,
      "learning_rate": 4.357751828356702e-05,
      "loss": 1.5006,
      "step": 3280
    },
    {
      "epoch": 0.3971990824580466,
      "grad_norm": 11.350707054138184,
      "learning_rate": 4.3557315447088775e-05,
      "loss": 1.3257,
      "step": 3290
    },
    {
      "epoch": 0.398406374501992,
      "grad_norm": 12.5741548538208,
      "learning_rate": 4.3537112610610534e-05,
      "loss": 1.3168,
      "step": 3300
    },
    {
      "epoch": 0.39961366654593744,
      "grad_norm": 8.999153137207031,
      "learning_rate": 4.351690977413229e-05,
      "loss": 1.1731,
      "step": 3310
    },
    {
      "epoch": 0.40082095858988287,
      "grad_norm": 8.121323585510254,
      "learning_rate": 4.3496706937654044e-05,
      "loss": 1.0765,
      "step": 3320
    },
    {
      "epoch": 0.4020282506338283,
      "grad_norm": 9.641090393066406,
      "learning_rate": 4.347650410117581e-05,
      "loss": 1.0762,
      "step": 3330
    },
    {
      "epoch": 0.40323554267777373,
      "grad_norm": 17.546192169189453,
      "learning_rate": 4.345630126469756e-05,
      "loss": 2.0923,
      "step": 3340
    },
    {
      "epoch": 0.40444283472171916,
      "grad_norm": 36.9171142578125,
      "learning_rate": 4.3436098428219326e-05,
      "loss": 1.3879,
      "step": 3350
    },
    {
      "epoch": 0.4056501267656646,
      "grad_norm": 11.159759521484375,
      "learning_rate": 4.3415895591741085e-05,
      "loss": 1.0467,
      "step": 3360
    },
    {
      "epoch": 0.40685741880961007,
      "grad_norm": 18.590736389160156,
      "learning_rate": 4.339569275526284e-05,
      "loss": 1.5621,
      "step": 3370
    },
    {
      "epoch": 0.4080647108535555,
      "grad_norm": 6.224109172821045,
      "learning_rate": 4.33754899187846e-05,
      "loss": 1.2587,
      "step": 3380
    },
    {
      "epoch": 0.40927200289750093,
      "grad_norm": 2.5112533569335938,
      "learning_rate": 4.335528708230635e-05,
      "loss": 0.8123,
      "step": 3390
    },
    {
      "epoch": 0.41047929494144636,
      "grad_norm": 19.40147590637207,
      "learning_rate": 4.333508424582812e-05,
      "loss": 1.1782,
      "step": 3400
    },
    {
      "epoch": 0.4116865869853918,
      "grad_norm": 22.796876907348633,
      "learning_rate": 4.331488140934987e-05,
      "loss": 1.3902,
      "step": 3410
    },
    {
      "epoch": 0.4128938790293372,
      "grad_norm": 6.7625532150268555,
      "learning_rate": 4.3294678572871636e-05,
      "loss": 1.0607,
      "step": 3420
    },
    {
      "epoch": 0.41410117107328265,
      "grad_norm": 10.557928085327148,
      "learning_rate": 4.3274475736393394e-05,
      "loss": 0.9897,
      "step": 3430
    },
    {
      "epoch": 0.4153084631172281,
      "grad_norm": 5.220957279205322,
      "learning_rate": 4.325427289991515e-05,
      "loss": 0.9378,
      "step": 3440
    },
    {
      "epoch": 0.4165157551611735,
      "grad_norm": 24.048385620117188,
      "learning_rate": 4.323407006343691e-05,
      "loss": 1.1559,
      "step": 3450
    },
    {
      "epoch": 0.41772304720511894,
      "grad_norm": 19.955415725708008,
      "learning_rate": 4.321386722695866e-05,
      "loss": 1.1639,
      "step": 3460
    },
    {
      "epoch": 0.41893033924906437,
      "grad_norm": 6.036765098571777,
      "learning_rate": 4.319366439048043e-05,
      "loss": 1.1289,
      "step": 3470
    },
    {
      "epoch": 0.4201376312930098,
      "grad_norm": 11.4838285446167,
      "learning_rate": 4.317346155400218e-05,
      "loss": 1.33,
      "step": 3480
    },
    {
      "epoch": 0.4213449233369552,
      "grad_norm": 11.677000045776367,
      "learning_rate": 4.3153258717523945e-05,
      "loss": 0.8747,
      "step": 3490
    },
    {
      "epoch": 0.42255221538090065,
      "grad_norm": 10.989776611328125,
      "learning_rate": 4.31330558810457e-05,
      "loss": 0.7294,
      "step": 3500
    },
    {
      "epoch": 0.4237595074248461,
      "grad_norm": 5.889261245727539,
      "learning_rate": 4.3112853044567455e-05,
      "loss": 0.9546,
      "step": 3510
    },
    {
      "epoch": 0.4249667994687915,
      "grad_norm": 9.179536819458008,
      "learning_rate": 4.309265020808922e-05,
      "loss": 0.9694,
      "step": 3520
    },
    {
      "epoch": 0.42617409151273694,
      "grad_norm": 15.909310340881348,
      "learning_rate": 4.307244737161097e-05,
      "loss": 1.5434,
      "step": 3530
    },
    {
      "epoch": 0.42738138355668237,
      "grad_norm": 11.908605575561523,
      "learning_rate": 4.305224453513274e-05,
      "loss": 1.6282,
      "step": 3540
    },
    {
      "epoch": 0.4285886756006278,
      "grad_norm": 14.551942825317383,
      "learning_rate": 4.303204169865449e-05,
      "loss": 1.3085,
      "step": 3550
    },
    {
      "epoch": 0.42979596764457323,
      "grad_norm": 4.719174385070801,
      "learning_rate": 4.3011838862176254e-05,
      "loss": 0.9076,
      "step": 3560
    },
    {
      "epoch": 0.43100325968851866,
      "grad_norm": 6.232408046722412,
      "learning_rate": 4.299163602569801e-05,
      "loss": 1.2265,
      "step": 3570
    },
    {
      "epoch": 0.4322105517324641,
      "grad_norm": 8.250030517578125,
      "learning_rate": 4.2971433189219764e-05,
      "loss": 1.2458,
      "step": 3580
    },
    {
      "epoch": 0.4334178437764095,
      "grad_norm": 16.61985969543457,
      "learning_rate": 4.295123035274153e-05,
      "loss": 1.2946,
      "step": 3590
    },
    {
      "epoch": 0.43462513582035495,
      "grad_norm": 16.908926010131836,
      "learning_rate": 4.293102751626328e-05,
      "loss": 1.5077,
      "step": 3600
    },
    {
      "epoch": 0.4358324278643004,
      "grad_norm": 34.204994201660156,
      "learning_rate": 4.2910824679785047e-05,
      "loss": 1.8074,
      "step": 3610
    },
    {
      "epoch": 0.4370397199082458,
      "grad_norm": 6.709868431091309,
      "learning_rate": 4.28906218433068e-05,
      "loss": 1.1231,
      "step": 3620
    },
    {
      "epoch": 0.43824701195219123,
      "grad_norm": 16.2998104095459,
      "learning_rate": 4.2870419006828563e-05,
      "loss": 1.1228,
      "step": 3630
    },
    {
      "epoch": 0.43945430399613666,
      "grad_norm": 6.007637977600098,
      "learning_rate": 4.285021617035032e-05,
      "loss": 1.1415,
      "step": 3640
    },
    {
      "epoch": 0.4406615960400821,
      "grad_norm": 2.5978376865386963,
      "learning_rate": 4.2830013333872074e-05,
      "loss": 1.0991,
      "step": 3650
    },
    {
      "epoch": 0.4418688880840275,
      "grad_norm": 12.730132102966309,
      "learning_rate": 4.280981049739384e-05,
      "loss": 1.3363,
      "step": 3660
    },
    {
      "epoch": 0.44307618012797295,
      "grad_norm": 6.902708053588867,
      "learning_rate": 4.278960766091559e-05,
      "loss": 1.4343,
      "step": 3670
    },
    {
      "epoch": 0.4442834721719184,
      "grad_norm": 17.429271697998047,
      "learning_rate": 4.2769404824437356e-05,
      "loss": 1.0775,
      "step": 3680
    },
    {
      "epoch": 0.4454907642158638,
      "grad_norm": 8.31417465209961,
      "learning_rate": 4.274920198795911e-05,
      "loss": 1.4973,
      "step": 3690
    },
    {
      "epoch": 0.44669805625980924,
      "grad_norm": 13.621916770935059,
      "learning_rate": 4.272899915148087e-05,
      "loss": 0.8602,
      "step": 3700
    },
    {
      "epoch": 0.44790534830375467,
      "grad_norm": 21.736705780029297,
      "learning_rate": 4.270879631500263e-05,
      "loss": 1.0396,
      "step": 3710
    },
    {
      "epoch": 0.4491126403477001,
      "grad_norm": 10.338930130004883,
      "learning_rate": 4.268859347852438e-05,
      "loss": 1.7329,
      "step": 3720
    },
    {
      "epoch": 0.4503199323916455,
      "grad_norm": 3.248728036880493,
      "learning_rate": 4.266839064204615e-05,
      "loss": 1.1748,
      "step": 3730
    },
    {
      "epoch": 0.45152722443559096,
      "grad_norm": 12.833627700805664,
      "learning_rate": 4.26481878055679e-05,
      "loss": 1.3436,
      "step": 3740
    },
    {
      "epoch": 0.4527345164795364,
      "grad_norm": 20.054931640625,
      "learning_rate": 4.2627984969089665e-05,
      "loss": 1.1273,
      "step": 3750
    },
    {
      "epoch": 0.4539418085234818,
      "grad_norm": 7.546112537384033,
      "learning_rate": 4.260778213261142e-05,
      "loss": 1.0332,
      "step": 3760
    },
    {
      "epoch": 0.45514910056742724,
      "grad_norm": 20.43124008178711,
      "learning_rate": 4.258757929613318e-05,
      "loss": 1.2536,
      "step": 3770
    },
    {
      "epoch": 0.4563563926113727,
      "grad_norm": 9.08917236328125,
      "learning_rate": 4.256737645965494e-05,
      "loss": 1.1458,
      "step": 3780
    },
    {
      "epoch": 0.4575636846553181,
      "grad_norm": 13.538667678833008,
      "learning_rate": 4.254717362317669e-05,
      "loss": 1.0353,
      "step": 3790
    },
    {
      "epoch": 0.45877097669926353,
      "grad_norm": 31.642333984375,
      "learning_rate": 4.252697078669846e-05,
      "loss": 1.0665,
      "step": 3800
    },
    {
      "epoch": 0.45997826874320896,
      "grad_norm": 8.742785453796387,
      "learning_rate": 4.250676795022021e-05,
      "loss": 0.9125,
      "step": 3810
    },
    {
      "epoch": 0.4611855607871544,
      "grad_norm": 38.56406784057617,
      "learning_rate": 4.2486565113741975e-05,
      "loss": 1.3749,
      "step": 3820
    },
    {
      "epoch": 0.4623928528310998,
      "grad_norm": 9.55203628540039,
      "learning_rate": 4.246636227726373e-05,
      "loss": 1.1529,
      "step": 3830
    },
    {
      "epoch": 0.46360014487504525,
      "grad_norm": 6.766389846801758,
      "learning_rate": 4.2446159440785485e-05,
      "loss": 0.8763,
      "step": 3840
    },
    {
      "epoch": 0.4648074369189907,
      "grad_norm": 12.11893081665039,
      "learning_rate": 4.242595660430725e-05,
      "loss": 0.9689,
      "step": 3850
    },
    {
      "epoch": 0.4660147289629361,
      "grad_norm": 5.715039253234863,
      "learning_rate": 4.2405753767829e-05,
      "loss": 1.0864,
      "step": 3860
    },
    {
      "epoch": 0.46722202100688154,
      "grad_norm": 13.815186500549316,
      "learning_rate": 4.238555093135077e-05,
      "loss": 1.3198,
      "step": 3870
    },
    {
      "epoch": 0.46842931305082697,
      "grad_norm": 9.635967254638672,
      "learning_rate": 4.236534809487252e-05,
      "loss": 1.1601,
      "step": 3880
    },
    {
      "epoch": 0.46963660509477245,
      "grad_norm": 19.660606384277344,
      "learning_rate": 4.2345145258394284e-05,
      "loss": 1.4532,
      "step": 3890
    },
    {
      "epoch": 0.4708438971387179,
      "grad_norm": 25.331987380981445,
      "learning_rate": 4.232494242191604e-05,
      "loss": 1.2724,
      "step": 3900
    },
    {
      "epoch": 0.4720511891826633,
      "grad_norm": 25.698617935180664,
      "learning_rate": 4.2304739585437794e-05,
      "loss": 1.2142,
      "step": 3910
    },
    {
      "epoch": 0.47325848122660874,
      "grad_norm": 6.644265651702881,
      "learning_rate": 4.228453674895956e-05,
      "loss": 1.0777,
      "step": 3920
    },
    {
      "epoch": 0.47446577327055417,
      "grad_norm": 13.186749458312988,
      "learning_rate": 4.226433391248131e-05,
      "loss": 1.2584,
      "step": 3930
    },
    {
      "epoch": 0.4756730653144996,
      "grad_norm": 7.920766353607178,
      "learning_rate": 4.2244131076003076e-05,
      "loss": 1.1861,
      "step": 3940
    },
    {
      "epoch": 0.47688035735844503,
      "grad_norm": 12.21504020690918,
      "learning_rate": 4.222392823952483e-05,
      "loss": 1.6591,
      "step": 3950
    },
    {
      "epoch": 0.47808764940239046,
      "grad_norm": 6.293022632598877,
      "learning_rate": 4.220372540304659e-05,
      "loss": 0.8833,
      "step": 3960
    },
    {
      "epoch": 0.4792949414463359,
      "grad_norm": 9.152581214904785,
      "learning_rate": 4.218352256656835e-05,
      "loss": 1.1766,
      "step": 3970
    },
    {
      "epoch": 0.4805022334902813,
      "grad_norm": 20.313894271850586,
      "learning_rate": 4.21633197300901e-05,
      "loss": 1.3263,
      "step": 3980
    },
    {
      "epoch": 0.48170952553422675,
      "grad_norm": 13.554831504821777,
      "learning_rate": 4.214311689361187e-05,
      "loss": 1.0694,
      "step": 3990
    },
    {
      "epoch": 0.4829168175781722,
      "grad_norm": 18.37723731994629,
      "learning_rate": 4.212291405713362e-05,
      "loss": 1.1514,
      "step": 4000
    },
    {
      "epoch": 0.4841241096221176,
      "grad_norm": 6.752192497253418,
      "learning_rate": 4.2102711220655386e-05,
      "loss": 1.4614,
      "step": 4010
    },
    {
      "epoch": 0.48533140166606303,
      "grad_norm": 14.30095386505127,
      "learning_rate": 4.208250838417714e-05,
      "loss": 0.9033,
      "step": 4020
    },
    {
      "epoch": 0.48653869371000846,
      "grad_norm": 3.7858872413635254,
      "learning_rate": 4.20623055476989e-05,
      "loss": 1.0653,
      "step": 4030
    },
    {
      "epoch": 0.4877459857539539,
      "grad_norm": 21.5765380859375,
      "learning_rate": 4.204210271122066e-05,
      "loss": 1.2044,
      "step": 4040
    },
    {
      "epoch": 0.4889532777978993,
      "grad_norm": 11.055548667907715,
      "learning_rate": 4.202189987474241e-05,
      "loss": 1.4867,
      "step": 4050
    },
    {
      "epoch": 0.49016056984184475,
      "grad_norm": 15.855735778808594,
      "learning_rate": 4.200169703826418e-05,
      "loss": 1.2091,
      "step": 4060
    },
    {
      "epoch": 0.4913678618857902,
      "grad_norm": 15.279037475585938,
      "learning_rate": 4.198149420178593e-05,
      "loss": 0.8559,
      "step": 4070
    },
    {
      "epoch": 0.4925751539297356,
      "grad_norm": 12.265058517456055,
      "learning_rate": 4.1961291365307695e-05,
      "loss": 1.5232,
      "step": 4080
    },
    {
      "epoch": 0.49378244597368104,
      "grad_norm": 20.34554672241211,
      "learning_rate": 4.1941088528829447e-05,
      "loss": 1.4562,
      "step": 4090
    },
    {
      "epoch": 0.49498973801762647,
      "grad_norm": 10.31032657623291,
      "learning_rate": 4.1920885692351205e-05,
      "loss": 1.397,
      "step": 4100
    },
    {
      "epoch": 0.4961970300615719,
      "grad_norm": 11.90217399597168,
      "learning_rate": 4.190068285587297e-05,
      "loss": 1.0369,
      "step": 4110
    },
    {
      "epoch": 0.4974043221055173,
      "grad_norm": 10.753695487976074,
      "learning_rate": 4.188048001939472e-05,
      "loss": 1.2174,
      "step": 4120
    },
    {
      "epoch": 0.49861161414946276,
      "grad_norm": 14.939640045166016,
      "learning_rate": 4.186027718291649e-05,
      "loss": 1.4672,
      "step": 4130
    },
    {
      "epoch": 0.4998189061934082,
      "grad_norm": 13.929787635803223,
      "learning_rate": 4.184007434643824e-05,
      "loss": 1.4669,
      "step": 4140
    },
    {
      "epoch": 0.5010261982373536,
      "grad_norm": 6.982569694519043,
      "learning_rate": 4.1819871509960004e-05,
      "loss": 0.8377,
      "step": 4150
    },
    {
      "epoch": 0.502233490281299,
      "grad_norm": 2.7284746170043945,
      "learning_rate": 4.1799668673481756e-05,
      "loss": 1.1901,
      "step": 4160
    },
    {
      "epoch": 0.5034407823252445,
      "grad_norm": 4.2180562019348145,
      "learning_rate": 4.1779465837003514e-05,
      "loss": 0.7566,
      "step": 4170
    },
    {
      "epoch": 0.5046480743691899,
      "grad_norm": 37.08050537109375,
      "learning_rate": 4.175926300052528e-05,
      "loss": 1.7459,
      "step": 4180
    },
    {
      "epoch": 0.5058553664131353,
      "grad_norm": 10.269936561584473,
      "learning_rate": 4.173906016404703e-05,
      "loss": 1.0708,
      "step": 4190
    },
    {
      "epoch": 0.5070626584570808,
      "grad_norm": 8.216901779174805,
      "learning_rate": 4.1718857327568797e-05,
      "loss": 0.901,
      "step": 4200
    },
    {
      "epoch": 0.5082699505010262,
      "grad_norm": 4.374532222747803,
      "learning_rate": 4.169865449109055e-05,
      "loss": 1.7503,
      "step": 4210
    },
    {
      "epoch": 0.5094772425449716,
      "grad_norm": 16.887733459472656,
      "learning_rate": 4.1678451654612314e-05,
      "loss": 0.915,
      "step": 4220
    },
    {
      "epoch": 0.510684534588917,
      "grad_norm": 6.850905418395996,
      "learning_rate": 4.1658248818134065e-05,
      "loss": 1.0194,
      "step": 4230
    },
    {
      "epoch": 0.5118918266328625,
      "grad_norm": 14.611712455749512,
      "learning_rate": 4.1638045981655824e-05,
      "loss": 0.9452,
      "step": 4240
    },
    {
      "epoch": 0.5130991186768079,
      "grad_norm": 3.197889804840088,
      "learning_rate": 4.161784314517759e-05,
      "loss": 1.272,
      "step": 4250
    },
    {
      "epoch": 0.5143064107207533,
      "grad_norm": 41.11612319946289,
      "learning_rate": 4.159764030869934e-05,
      "loss": 1.8195,
      "step": 4260
    },
    {
      "epoch": 0.5155137027646988,
      "grad_norm": 10.140461921691895,
      "learning_rate": 4.1577437472221106e-05,
      "loss": 1.2543,
      "step": 4270
    },
    {
      "epoch": 0.5167209948086442,
      "grad_norm": 7.760611057281494,
      "learning_rate": 4.155723463574286e-05,
      "loss": 1.216,
      "step": 4280
    },
    {
      "epoch": 0.5179282868525896,
      "grad_norm": 21.814470291137695,
      "learning_rate": 4.153703179926462e-05,
      "loss": 1.0094,
      "step": 4290
    },
    {
      "epoch": 0.519135578896535,
      "grad_norm": 12.858716011047363,
      "learning_rate": 4.1516828962786375e-05,
      "loss": 1.167,
      "step": 4300
    },
    {
      "epoch": 0.5203428709404805,
      "grad_norm": 10.170001983642578,
      "learning_rate": 4.149662612630813e-05,
      "loss": 0.7568,
      "step": 4310
    },
    {
      "epoch": 0.5215501629844259,
      "grad_norm": 10.542889595031738,
      "learning_rate": 4.14764232898299e-05,
      "loss": 1.0747,
      "step": 4320
    },
    {
      "epoch": 0.5227574550283713,
      "grad_norm": 11.128294944763184,
      "learning_rate": 4.145622045335165e-05,
      "loss": 0.71,
      "step": 4330
    },
    {
      "epoch": 0.5239647470723168,
      "grad_norm": 5.704838275909424,
      "learning_rate": 4.1436017616873415e-05,
      "loss": 1.2181,
      "step": 4340
    },
    {
      "epoch": 0.5251720391162622,
      "grad_norm": 11.039192199707031,
      "learning_rate": 4.141581478039517e-05,
      "loss": 0.9673,
      "step": 4350
    },
    {
      "epoch": 0.5263793311602076,
      "grad_norm": 13.896295547485352,
      "learning_rate": 4.1395611943916925e-05,
      "loss": 0.8528,
      "step": 4360
    },
    {
      "epoch": 0.5275866232041531,
      "grad_norm": 5.884769916534424,
      "learning_rate": 4.1375409107438684e-05,
      "loss": 1.0762,
      "step": 4370
    },
    {
      "epoch": 0.5287939152480985,
      "grad_norm": 12.80189037322998,
      "learning_rate": 4.135520627096044e-05,
      "loss": 1.7327,
      "step": 4380
    },
    {
      "epoch": 0.5300012072920439,
      "grad_norm": 13.67371940612793,
      "learning_rate": 4.133500343448221e-05,
      "loss": 1.2214,
      "step": 4390
    },
    {
      "epoch": 0.5312084993359893,
      "grad_norm": 22.67274284362793,
      "learning_rate": 4.131480059800396e-05,
      "loss": 1.3201,
      "step": 4400
    },
    {
      "epoch": 0.5324157913799348,
      "grad_norm": 22.11157989501953,
      "learning_rate": 4.1294597761525725e-05,
      "loss": 1.1297,
      "step": 4410
    },
    {
      "epoch": 0.5336230834238802,
      "grad_norm": 11.211080551147461,
      "learning_rate": 4.1274394925047476e-05,
      "loss": 1.5195,
      "step": 4420
    },
    {
      "epoch": 0.5348303754678256,
      "grad_norm": 11.708219528198242,
      "learning_rate": 4.1254192088569235e-05,
      "loss": 1.4617,
      "step": 4430
    },
    {
      "epoch": 0.5360376675117711,
      "grad_norm": 17.998167037963867,
      "learning_rate": 4.123398925209099e-05,
      "loss": 1.7386,
      "step": 4440
    },
    {
      "epoch": 0.5372449595557165,
      "grad_norm": 9.646733283996582,
      "learning_rate": 4.121378641561275e-05,
      "loss": 1.2318,
      "step": 4450
    },
    {
      "epoch": 0.5384522515996619,
      "grad_norm": 5.035294055938721,
      "learning_rate": 4.119358357913452e-05,
      "loss": 1.1864,
      "step": 4460
    },
    {
      "epoch": 0.5396595436436074,
      "grad_norm": 11.207514762878418,
      "learning_rate": 4.117338074265627e-05,
      "loss": 1.5049,
      "step": 4470
    },
    {
      "epoch": 0.5408668356875528,
      "grad_norm": 8.468974113464355,
      "learning_rate": 4.1153177906178034e-05,
      "loss": 1.2689,
      "step": 4480
    },
    {
      "epoch": 0.5420741277314982,
      "grad_norm": 6.934230804443359,
      "learning_rate": 4.1132975069699786e-05,
      "loss": 1.0554,
      "step": 4490
    },
    {
      "epoch": 0.5432814197754436,
      "grad_norm": 3.8387107849121094,
      "learning_rate": 4.1112772233221544e-05,
      "loss": 1.1357,
      "step": 4500
    },
    {
      "epoch": 0.5444887118193891,
      "grad_norm": 18.947919845581055,
      "learning_rate": 4.10925693967433e-05,
      "loss": 0.946,
      "step": 4510
    },
    {
      "epoch": 0.5456960038633345,
      "grad_norm": 13.828507423400879,
      "learning_rate": 4.107236656026506e-05,
      "loss": 1.3849,
      "step": 4520
    },
    {
      "epoch": 0.5469032959072799,
      "grad_norm": 12.078511238098145,
      "learning_rate": 4.1052163723786826e-05,
      "loss": 1.32,
      "step": 4530
    },
    {
      "epoch": 0.5481105879512254,
      "grad_norm": 8.533878326416016,
      "learning_rate": 4.103196088730858e-05,
      "loss": 1.1433,
      "step": 4540
    },
    {
      "epoch": 0.5493178799951708,
      "grad_norm": 23.233829498291016,
      "learning_rate": 4.101175805083034e-05,
      "loss": 1.146,
      "step": 4550
    },
    {
      "epoch": 0.5505251720391162,
      "grad_norm": 66.20478820800781,
      "learning_rate": 4.0991555214352095e-05,
      "loss": 1.2825,
      "step": 4560
    },
    {
      "epoch": 0.5517324640830616,
      "grad_norm": 6.977896690368652,
      "learning_rate": 4.097135237787385e-05,
      "loss": 0.9652,
      "step": 4570
    },
    {
      "epoch": 0.5529397561270071,
      "grad_norm": 15.212750434875488,
      "learning_rate": 4.095114954139561e-05,
      "loss": 1.1131,
      "step": 4580
    },
    {
      "epoch": 0.5541470481709525,
      "grad_norm": 4.527120590209961,
      "learning_rate": 4.093094670491737e-05,
      "loss": 1.1177,
      "step": 4590
    },
    {
      "epoch": 0.5553543402148979,
      "grad_norm": 5.057064533233643,
      "learning_rate": 4.0910743868439136e-05,
      "loss": 0.8529,
      "step": 4600
    },
    {
      "epoch": 0.5565616322588434,
      "grad_norm": 6.267819881439209,
      "learning_rate": 4.089054103196089e-05,
      "loss": 1.2458,
      "step": 4610
    },
    {
      "epoch": 0.5577689243027888,
      "grad_norm": 9.89231014251709,
      "learning_rate": 4.087033819548265e-05,
      "loss": 1.2677,
      "step": 4620
    },
    {
      "epoch": 0.5589762163467342,
      "grad_norm": 16.48421287536621,
      "learning_rate": 4.0850135359004404e-05,
      "loss": 1.2853,
      "step": 4630
    },
    {
      "epoch": 0.5601835083906797,
      "grad_norm": 6.470500469207764,
      "learning_rate": 4.082993252252616e-05,
      "loss": 1.1743,
      "step": 4640
    },
    {
      "epoch": 0.5613908004346251,
      "grad_norm": 13.790871620178223,
      "learning_rate": 4.080972968604792e-05,
      "loss": 1.117,
      "step": 4650
    },
    {
      "epoch": 0.5625980924785706,
      "grad_norm": 10.966059684753418,
      "learning_rate": 4.078952684956968e-05,
      "loss": 1.05,
      "step": 4660
    },
    {
      "epoch": 0.563805384522516,
      "grad_norm": 9.249923706054688,
      "learning_rate": 4.0769324013091445e-05,
      "loss": 1.283,
      "step": 4670
    },
    {
      "epoch": 0.5650126765664615,
      "grad_norm": 14.90842342376709,
      "learning_rate": 4.0749121176613197e-05,
      "loss": 1.0513,
      "step": 4680
    },
    {
      "epoch": 0.5662199686104069,
      "grad_norm": 4.99460506439209,
      "learning_rate": 4.0728918340134955e-05,
      "loss": 1.2027,
      "step": 4690
    },
    {
      "epoch": 0.5674272606543523,
      "grad_norm": 4.3763909339904785,
      "learning_rate": 4.0708715503656714e-05,
      "loss": 1.4831,
      "step": 4700
    },
    {
      "epoch": 0.5686345526982978,
      "grad_norm": 11.867714881896973,
      "learning_rate": 4.068851266717847e-05,
      "loss": 1.5867,
      "step": 4710
    },
    {
      "epoch": 0.5698418447422432,
      "grad_norm": 20.496326446533203,
      "learning_rate": 4.066830983070023e-05,
      "loss": 0.9757,
      "step": 4720
    },
    {
      "epoch": 0.5710491367861886,
      "grad_norm": 12.915321350097656,
      "learning_rate": 4.064810699422199e-05,
      "loss": 1.1551,
      "step": 4730
    },
    {
      "epoch": 0.5722564288301341,
      "grad_norm": 8.523058891296387,
      "learning_rate": 4.0627904157743754e-05,
      "loss": 1.0768,
      "step": 4740
    },
    {
      "epoch": 0.5734637208740795,
      "grad_norm": 6.861374855041504,
      "learning_rate": 4.0607701321265506e-05,
      "loss": 1.0506,
      "step": 4750
    },
    {
      "epoch": 0.5746710129180249,
      "grad_norm": 9.177238464355469,
      "learning_rate": 4.0587498484787264e-05,
      "loss": 0.9946,
      "step": 4760
    },
    {
      "epoch": 0.5758783049619703,
      "grad_norm": 14.017538070678711,
      "learning_rate": 4.056729564830902e-05,
      "loss": 1.6594,
      "step": 4770
    },
    {
      "epoch": 0.5770855970059158,
      "grad_norm": 18.890010833740234,
      "learning_rate": 4.054709281183078e-05,
      "loss": 1.5251,
      "step": 4780
    },
    {
      "epoch": 0.5782928890498612,
      "grad_norm": 7.3946099281311035,
      "learning_rate": 4.052688997535254e-05,
      "loss": 0.9987,
      "step": 4790
    },
    {
      "epoch": 0.5795001810938066,
      "grad_norm": 12.184159278869629,
      "learning_rate": 4.05066871388743e-05,
      "loss": 1.5328,
      "step": 4800
    },
    {
      "epoch": 0.5807074731377521,
      "grad_norm": 12.907468795776367,
      "learning_rate": 4.0486484302396064e-05,
      "loss": 1.6197,
      "step": 4810
    },
    {
      "epoch": 0.5819147651816975,
      "grad_norm": 14.874164581298828,
      "learning_rate": 4.0466281465917815e-05,
      "loss": 1.4526,
      "step": 4820
    },
    {
      "epoch": 0.5831220572256429,
      "grad_norm": 21.70797348022461,
      "learning_rate": 4.0446078629439574e-05,
      "loss": 1.1287,
      "step": 4830
    },
    {
      "epoch": 0.5843293492695883,
      "grad_norm": 10.595853805541992,
      "learning_rate": 4.042587579296133e-05,
      "loss": 1.0983,
      "step": 4840
    },
    {
      "epoch": 0.5855366413135338,
      "grad_norm": 15.052566528320312,
      "learning_rate": 4.040567295648309e-05,
      "loss": 1.1492,
      "step": 4850
    },
    {
      "epoch": 0.5867439333574792,
      "grad_norm": 7.051451683044434,
      "learning_rate": 4.038547012000485e-05,
      "loss": 1.7443,
      "step": 4860
    },
    {
      "epoch": 0.5879512254014246,
      "grad_norm": 10.008295059204102,
      "learning_rate": 4.036526728352661e-05,
      "loss": 1.2583,
      "step": 4870
    },
    {
      "epoch": 0.5891585174453701,
      "grad_norm": 21.09678077697754,
      "learning_rate": 4.034506444704837e-05,
      "loss": 1.4203,
      "step": 4880
    },
    {
      "epoch": 0.5903658094893155,
      "grad_norm": 7.052413463592529,
      "learning_rate": 4.0324861610570125e-05,
      "loss": 1.0363,
      "step": 4890
    },
    {
      "epoch": 0.5915731015332609,
      "grad_norm": 10.709598541259766,
      "learning_rate": 4.030465877409188e-05,
      "loss": 1.1428,
      "step": 4900
    },
    {
      "epoch": 0.5927803935772064,
      "grad_norm": 44.44905471801758,
      "learning_rate": 4.028445593761364e-05,
      "loss": 1.0179,
      "step": 4910
    },
    {
      "epoch": 0.5939876856211518,
      "grad_norm": 9.740785598754883,
      "learning_rate": 4.02642531011354e-05,
      "loss": 1.1429,
      "step": 4920
    },
    {
      "epoch": 0.5951949776650972,
      "grad_norm": 8.214555740356445,
      "learning_rate": 4.024405026465716e-05,
      "loss": 1.0286,
      "step": 4930
    },
    {
      "epoch": 0.5964022697090426,
      "grad_norm": 13.791067123413086,
      "learning_rate": 4.022384742817892e-05,
      "loss": 1.4941,
      "step": 4940
    },
    {
      "epoch": 0.5976095617529881,
      "grad_norm": 17.87343978881836,
      "learning_rate": 4.0203644591700675e-05,
      "loss": 1.0789,
      "step": 4950
    },
    {
      "epoch": 0.5988168537969335,
      "grad_norm": 22.551776885986328,
      "learning_rate": 4.0183441755222434e-05,
      "loss": 1.2638,
      "step": 4960
    },
    {
      "epoch": 0.6000241458408789,
      "grad_norm": 22.09368896484375,
      "learning_rate": 4.016323891874419e-05,
      "loss": 1.0276,
      "step": 4970
    },
    {
      "epoch": 0.6012314378848244,
      "grad_norm": 33.659698486328125,
      "learning_rate": 4.014303608226595e-05,
      "loss": 0.9479,
      "step": 4980
    },
    {
      "epoch": 0.6024387299287698,
      "grad_norm": 11.301599502563477,
      "learning_rate": 4.012283324578771e-05,
      "loss": 0.8701,
      "step": 4990
    },
    {
      "epoch": 0.6036460219727152,
      "grad_norm": 13.657859802246094,
      "learning_rate": 4.010263040930947e-05,
      "loss": 1.0516,
      "step": 5000
    },
    {
      "epoch": 0.6048533140166606,
      "grad_norm": 21.950868606567383,
      "learning_rate": 4.0082427572831226e-05,
      "loss": 0.9351,
      "step": 5010
    },
    {
      "epoch": 0.6060606060606061,
      "grad_norm": 15.882399559020996,
      "learning_rate": 4.0062224736352985e-05,
      "loss": 1.2399,
      "step": 5020
    },
    {
      "epoch": 0.6072678981045515,
      "grad_norm": 4.625049114227295,
      "learning_rate": 4.004202189987474e-05,
      "loss": 0.7845,
      "step": 5030
    },
    {
      "epoch": 0.6084751901484969,
      "grad_norm": 22.891393661499023,
      "learning_rate": 4.00218190633965e-05,
      "loss": 1.3215,
      "step": 5040
    },
    {
      "epoch": 0.6096824821924424,
      "grad_norm": 9.824638366699219,
      "learning_rate": 4.000161622691826e-05,
      "loss": 1.4625,
      "step": 5050
    },
    {
      "epoch": 0.6108897742363878,
      "grad_norm": 5.578610897064209,
      "learning_rate": 3.998141339044002e-05,
      "loss": 0.6758,
      "step": 5060
    },
    {
      "epoch": 0.6120970662803332,
      "grad_norm": 22.71108627319336,
      "learning_rate": 3.996121055396178e-05,
      "loss": 1.3916,
      "step": 5070
    },
    {
      "epoch": 0.6133043583242787,
      "grad_norm": 15.669069290161133,
      "learning_rate": 3.9941007717483536e-05,
      "loss": 1.35,
      "step": 5080
    },
    {
      "epoch": 0.6145116503682241,
      "grad_norm": 3.0980358123779297,
      "learning_rate": 3.9920804881005294e-05,
      "loss": 0.7032,
      "step": 5090
    },
    {
      "epoch": 0.6157189424121695,
      "grad_norm": 33.77893829345703,
      "learning_rate": 3.990060204452705e-05,
      "loss": 1.0446,
      "step": 5100
    },
    {
      "epoch": 0.6169262344561149,
      "grad_norm": 12.6209077835083,
      "learning_rate": 3.988039920804881e-05,
      "loss": 1.2496,
      "step": 5110
    },
    {
      "epoch": 0.6181335265000604,
      "grad_norm": 5.3566718101501465,
      "learning_rate": 3.986019637157057e-05,
      "loss": 0.9733,
      "step": 5120
    },
    {
      "epoch": 0.6193408185440058,
      "grad_norm": 13.809636116027832,
      "learning_rate": 3.983999353509233e-05,
      "loss": 1.0173,
      "step": 5130
    },
    {
      "epoch": 0.6205481105879512,
      "grad_norm": 9.455422401428223,
      "learning_rate": 3.981979069861409e-05,
      "loss": 1.072,
      "step": 5140
    },
    {
      "epoch": 0.6217554026318967,
      "grad_norm": 18.3873348236084,
      "learning_rate": 3.9799587862135845e-05,
      "loss": 1.2023,
      "step": 5150
    },
    {
      "epoch": 0.6229626946758421,
      "grad_norm": 8.8668794631958,
      "learning_rate": 3.97793850256576e-05,
      "loss": 1.1322,
      "step": 5160
    },
    {
      "epoch": 0.6241699867197875,
      "grad_norm": 7.455340385437012,
      "learning_rate": 3.975918218917936e-05,
      "loss": 1.2354,
      "step": 5170
    },
    {
      "epoch": 0.625377278763733,
      "grad_norm": 12.75782585144043,
      "learning_rate": 3.973897935270112e-05,
      "loss": 1.2552,
      "step": 5180
    },
    {
      "epoch": 0.6265845708076784,
      "grad_norm": 50.620765686035156,
      "learning_rate": 3.971877651622288e-05,
      "loss": 1.4724,
      "step": 5190
    },
    {
      "epoch": 0.6277918628516238,
      "grad_norm": 12.42019271850586,
      "learning_rate": 3.969857367974464e-05,
      "loss": 1.1321,
      "step": 5200
    },
    {
      "epoch": 0.6289991548955692,
      "grad_norm": 8.6299467086792,
      "learning_rate": 3.96783708432664e-05,
      "loss": 1.2936,
      "step": 5210
    },
    {
      "epoch": 0.6302064469395147,
      "grad_norm": 11.467175483703613,
      "learning_rate": 3.9658168006788154e-05,
      "loss": 1.0749,
      "step": 5220
    },
    {
      "epoch": 0.6314137389834601,
      "grad_norm": 7.028493881225586,
      "learning_rate": 3.963796517030991e-05,
      "loss": 1.1539,
      "step": 5230
    },
    {
      "epoch": 0.6326210310274055,
      "grad_norm": 3.649244546890259,
      "learning_rate": 3.961776233383167e-05,
      "loss": 0.9853,
      "step": 5240
    },
    {
      "epoch": 0.633828323071351,
      "grad_norm": 15.779903411865234,
      "learning_rate": 3.959755949735343e-05,
      "loss": 1.1718,
      "step": 5250
    },
    {
      "epoch": 0.6350356151152964,
      "grad_norm": 6.363460540771484,
      "learning_rate": 3.957735666087519e-05,
      "loss": 0.9464,
      "step": 5260
    },
    {
      "epoch": 0.6362429071592418,
      "grad_norm": 7.587771892547607,
      "learning_rate": 3.9557153824396947e-05,
      "loss": 1.2583,
      "step": 5270
    },
    {
      "epoch": 0.6374501992031872,
      "grad_norm": 17.928760528564453,
      "learning_rate": 3.9536950987918705e-05,
      "loss": 1.0755,
      "step": 5280
    },
    {
      "epoch": 0.6386574912471327,
      "grad_norm": 17.554765701293945,
      "learning_rate": 3.9516748151440464e-05,
      "loss": 1.3151,
      "step": 5290
    },
    {
      "epoch": 0.6398647832910781,
      "grad_norm": 7.157963275909424,
      "learning_rate": 3.949654531496222e-05,
      "loss": 1.227,
      "step": 5300
    },
    {
      "epoch": 0.6410720753350235,
      "grad_norm": 16.035985946655273,
      "learning_rate": 3.947634247848398e-05,
      "loss": 1.5345,
      "step": 5310
    },
    {
      "epoch": 0.642279367378969,
      "grad_norm": 19.32333755493164,
      "learning_rate": 3.945613964200574e-05,
      "loss": 1.2805,
      "step": 5320
    },
    {
      "epoch": 0.6434866594229144,
      "grad_norm": 16.213775634765625,
      "learning_rate": 3.94359368055275e-05,
      "loss": 0.8571,
      "step": 5330
    },
    {
      "epoch": 0.6446939514668598,
      "grad_norm": 12.026822090148926,
      "learning_rate": 3.9415733969049256e-05,
      "loss": 1.345,
      "step": 5340
    },
    {
      "epoch": 0.6459012435108052,
      "grad_norm": 22.680526733398438,
      "learning_rate": 3.9395531132571014e-05,
      "loss": 1.2917,
      "step": 5350
    },
    {
      "epoch": 0.6471085355547507,
      "grad_norm": 6.5746965408325195,
      "learning_rate": 3.937532829609277e-05,
      "loss": 1.5016,
      "step": 5360
    },
    {
      "epoch": 0.6483158275986961,
      "grad_norm": 14.723548889160156,
      "learning_rate": 3.935512545961453e-05,
      "loss": 1.0902,
      "step": 5370
    },
    {
      "epoch": 0.6495231196426415,
      "grad_norm": 11.986374855041504,
      "learning_rate": 3.933492262313629e-05,
      "loss": 0.8413,
      "step": 5380
    },
    {
      "epoch": 0.650730411686587,
      "grad_norm": 11.90639591217041,
      "learning_rate": 3.931471978665805e-05,
      "loss": 1.2744,
      "step": 5390
    },
    {
      "epoch": 0.6519377037305324,
      "grad_norm": 5.55607271194458,
      "learning_rate": 3.929451695017981e-05,
      "loss": 1.2054,
      "step": 5400
    },
    {
      "epoch": 0.6531449957744778,
      "grad_norm": 14.516633987426758,
      "learning_rate": 3.9274314113701565e-05,
      "loss": 1.2284,
      "step": 5410
    },
    {
      "epoch": 0.6543522878184233,
      "grad_norm": 6.279738426208496,
      "learning_rate": 3.9254111277223324e-05,
      "loss": 1.1143,
      "step": 5420
    },
    {
      "epoch": 0.6555595798623687,
      "grad_norm": 7.707493305206299,
      "learning_rate": 3.923390844074508e-05,
      "loss": 1.0882,
      "step": 5430
    },
    {
      "epoch": 0.6567668719063141,
      "grad_norm": 7.619891166687012,
      "learning_rate": 3.921370560426684e-05,
      "loss": 0.9117,
      "step": 5440
    },
    {
      "epoch": 0.6579741639502595,
      "grad_norm": 9.258544921875,
      "learning_rate": 3.91935027677886e-05,
      "loss": 1.4582,
      "step": 5450
    },
    {
      "epoch": 0.659181455994205,
      "grad_norm": 14.241188049316406,
      "learning_rate": 3.917329993131036e-05,
      "loss": 1.1351,
      "step": 5460
    },
    {
      "epoch": 0.6603887480381504,
      "grad_norm": 13.899798393249512,
      "learning_rate": 3.9153097094832116e-05,
      "loss": 1.016,
      "step": 5470
    },
    {
      "epoch": 0.6615960400820958,
      "grad_norm": 12.051464080810547,
      "learning_rate": 3.9132894258353875e-05,
      "loss": 1.0737,
      "step": 5480
    },
    {
      "epoch": 0.6628033321260413,
      "grad_norm": 11.040992736816406,
      "learning_rate": 3.911269142187563e-05,
      "loss": 0.7658,
      "step": 5490
    },
    {
      "epoch": 0.6640106241699867,
      "grad_norm": 8.852428436279297,
      "learning_rate": 3.909248858539739e-05,
      "loss": 1.4789,
      "step": 5500
    },
    {
      "epoch": 0.6652179162139321,
      "grad_norm": 10.25948429107666,
      "learning_rate": 3.907228574891915e-05,
      "loss": 1.2037,
      "step": 5510
    },
    {
      "epoch": 0.6664252082578775,
      "grad_norm": 7.592520713806152,
      "learning_rate": 3.905208291244091e-05,
      "loss": 1.678,
      "step": 5520
    },
    {
      "epoch": 0.667632500301823,
      "grad_norm": 5.732512474060059,
      "learning_rate": 3.903188007596267e-05,
      "loss": 1.0924,
      "step": 5530
    },
    {
      "epoch": 0.6688397923457684,
      "grad_norm": 45.35914993286133,
      "learning_rate": 3.9011677239484425e-05,
      "loss": 1.3868,
      "step": 5540
    },
    {
      "epoch": 0.6700470843897138,
      "grad_norm": 7.487490177154541,
      "learning_rate": 3.8991474403006184e-05,
      "loss": 1.5142,
      "step": 5550
    },
    {
      "epoch": 0.6712543764336593,
      "grad_norm": 14.404407501220703,
      "learning_rate": 3.897127156652794e-05,
      "loss": 0.9689,
      "step": 5560
    },
    {
      "epoch": 0.6724616684776047,
      "grad_norm": 8.324043273925781,
      "learning_rate": 3.89510687300497e-05,
      "loss": 1.2228,
      "step": 5570
    },
    {
      "epoch": 0.6736689605215501,
      "grad_norm": 15.029932975769043,
      "learning_rate": 3.893086589357146e-05,
      "loss": 1.2684,
      "step": 5580
    },
    {
      "epoch": 0.6748762525654955,
      "grad_norm": 12.878973960876465,
      "learning_rate": 3.891066305709322e-05,
      "loss": 0.8494,
      "step": 5590
    },
    {
      "epoch": 0.676083544609441,
      "grad_norm": 4.496845245361328,
      "learning_rate": 3.8890460220614976e-05,
      "loss": 1.1794,
      "step": 5600
    },
    {
      "epoch": 0.6772908366533864,
      "grad_norm": 16.195331573486328,
      "learning_rate": 3.8870257384136735e-05,
      "loss": 1.2776,
      "step": 5610
    },
    {
      "epoch": 0.6784981286973318,
      "grad_norm": 10.16224193572998,
      "learning_rate": 3.885005454765849e-05,
      "loss": 1.3508,
      "step": 5620
    },
    {
      "epoch": 0.6797054207412773,
      "grad_norm": 5.4379730224609375,
      "learning_rate": 3.882985171118025e-05,
      "loss": 1.3249,
      "step": 5630
    },
    {
      "epoch": 0.6809127127852227,
      "grad_norm": 22.010841369628906,
      "learning_rate": 3.880964887470201e-05,
      "loss": 1.4536,
      "step": 5640
    },
    {
      "epoch": 0.6821200048291681,
      "grad_norm": 3.157125473022461,
      "learning_rate": 3.878944603822377e-05,
      "loss": 1.2514,
      "step": 5650
    },
    {
      "epoch": 0.6833272968731136,
      "grad_norm": 17.443885803222656,
      "learning_rate": 3.876924320174553e-05,
      "loss": 1.2468,
      "step": 5660
    },
    {
      "epoch": 0.684534588917059,
      "grad_norm": 15.788519859313965,
      "learning_rate": 3.8749040365267286e-05,
      "loss": 1.763,
      "step": 5670
    },
    {
      "epoch": 0.6857418809610044,
      "grad_norm": 20.593984603881836,
      "learning_rate": 3.8728837528789044e-05,
      "loss": 1.0428,
      "step": 5680
    },
    {
      "epoch": 0.6869491730049498,
      "grad_norm": 9.900472640991211,
      "learning_rate": 3.87086346923108e-05,
      "loss": 1.4254,
      "step": 5690
    },
    {
      "epoch": 0.6881564650488954,
      "grad_norm": 15.87972354888916,
      "learning_rate": 3.868843185583256e-05,
      "loss": 1.0274,
      "step": 5700
    },
    {
      "epoch": 0.6893637570928408,
      "grad_norm": 16.773414611816406,
      "learning_rate": 3.866822901935432e-05,
      "loss": 1.0705,
      "step": 5710
    },
    {
      "epoch": 0.6905710491367862,
      "grad_norm": 24.48480224609375,
      "learning_rate": 3.864802618287608e-05,
      "loss": 0.84,
      "step": 5720
    },
    {
      "epoch": 0.6917783411807317,
      "grad_norm": 9.862621307373047,
      "learning_rate": 3.8627823346397836e-05,
      "loss": 1.3057,
      "step": 5730
    },
    {
      "epoch": 0.6929856332246771,
      "grad_norm": 30.978071212768555,
      "learning_rate": 3.8607620509919595e-05,
      "loss": 1.6665,
      "step": 5740
    },
    {
      "epoch": 0.6941929252686225,
      "grad_norm": 11.588111877441406,
      "learning_rate": 3.8587417673441353e-05,
      "loss": 1.472,
      "step": 5750
    },
    {
      "epoch": 0.695400217312568,
      "grad_norm": 22.008153915405273,
      "learning_rate": 3.856721483696311e-05,
      "loss": 1.5328,
      "step": 5760
    },
    {
      "epoch": 0.6966075093565134,
      "grad_norm": 12.862824440002441,
      "learning_rate": 3.854701200048487e-05,
      "loss": 0.868,
      "step": 5770
    },
    {
      "epoch": 0.6978148014004588,
      "grad_norm": 10.184123992919922,
      "learning_rate": 3.852680916400663e-05,
      "loss": 0.9351,
      "step": 5780
    },
    {
      "epoch": 0.6990220934444042,
      "grad_norm": 7.214652061462402,
      "learning_rate": 3.850660632752839e-05,
      "loss": 1.2079,
      "step": 5790
    },
    {
      "epoch": 0.7002293854883497,
      "grad_norm": 2.9722847938537598,
      "learning_rate": 3.8486403491050146e-05,
      "loss": 1.0512,
      "step": 5800
    },
    {
      "epoch": 0.7014366775322951,
      "grad_norm": 8.610760688781738,
      "learning_rate": 3.8466200654571904e-05,
      "loss": 1.0169,
      "step": 5810
    },
    {
      "epoch": 0.7026439695762405,
      "grad_norm": 30.522092819213867,
      "learning_rate": 3.844599781809366e-05,
      "loss": 1.1286,
      "step": 5820
    },
    {
      "epoch": 0.703851261620186,
      "grad_norm": 16.12820816040039,
      "learning_rate": 3.842579498161542e-05,
      "loss": 0.8967,
      "step": 5830
    },
    {
      "epoch": 0.7050585536641314,
      "grad_norm": 16.85987663269043,
      "learning_rate": 3.840559214513718e-05,
      "loss": 1.3307,
      "step": 5840
    },
    {
      "epoch": 0.7062658457080768,
      "grad_norm": 18.246917724609375,
      "learning_rate": 3.838538930865894e-05,
      "loss": 1.5731,
      "step": 5850
    },
    {
      "epoch": 0.7074731377520223,
      "grad_norm": 3.69657564163208,
      "learning_rate": 3.83651864721807e-05,
      "loss": 1.3564,
      "step": 5860
    },
    {
      "epoch": 0.7086804297959677,
      "grad_norm": 13.367045402526855,
      "learning_rate": 3.8344983635702455e-05,
      "loss": 1.2901,
      "step": 5870
    },
    {
      "epoch": 0.7098877218399131,
      "grad_norm": 15.154350280761719,
      "learning_rate": 3.8324780799224214e-05,
      "loss": 0.8989,
      "step": 5880
    },
    {
      "epoch": 0.7110950138838585,
      "grad_norm": 11.028938293457031,
      "learning_rate": 3.830457796274597e-05,
      "loss": 1.3365,
      "step": 5890
    },
    {
      "epoch": 0.712302305927804,
      "grad_norm": 14.09022045135498,
      "learning_rate": 3.828437512626773e-05,
      "loss": 1.1615,
      "step": 5900
    },
    {
      "epoch": 0.7135095979717494,
      "grad_norm": 11.353936195373535,
      "learning_rate": 3.826417228978949e-05,
      "loss": 1.1285,
      "step": 5910
    },
    {
      "epoch": 0.7147168900156948,
      "grad_norm": 12.883540153503418,
      "learning_rate": 3.824396945331125e-05,
      "loss": 1.1309,
      "step": 5920
    },
    {
      "epoch": 0.7159241820596403,
      "grad_norm": 25.671188354492188,
      "learning_rate": 3.8223766616833006e-05,
      "loss": 0.9608,
      "step": 5930
    },
    {
      "epoch": 0.7171314741035857,
      "grad_norm": 8.088144302368164,
      "learning_rate": 3.8203563780354764e-05,
      "loss": 0.8094,
      "step": 5940
    },
    {
      "epoch": 0.7183387661475311,
      "grad_norm": 17.223508834838867,
      "learning_rate": 3.818336094387652e-05,
      "loss": 0.9107,
      "step": 5950
    },
    {
      "epoch": 0.7195460581914765,
      "grad_norm": 17.494342803955078,
      "learning_rate": 3.816315810739828e-05,
      "loss": 0.8527,
      "step": 5960
    },
    {
      "epoch": 0.720753350235422,
      "grad_norm": 8.442413330078125,
      "learning_rate": 3.814295527092004e-05,
      "loss": 1.1342,
      "step": 5970
    },
    {
      "epoch": 0.7219606422793674,
      "grad_norm": 16.660480499267578,
      "learning_rate": 3.81227524344418e-05,
      "loss": 0.8708,
      "step": 5980
    },
    {
      "epoch": 0.7231679343233128,
      "grad_norm": 15.465290069580078,
      "learning_rate": 3.810254959796356e-05,
      "loss": 1.4811,
      "step": 5990
    },
    {
      "epoch": 0.7243752263672583,
      "grad_norm": 24.823471069335938,
      "learning_rate": 3.8082346761485315e-05,
      "loss": 1.0563,
      "step": 6000
    },
    {
      "epoch": 0.7255825184112037,
      "grad_norm": 18.63069725036621,
      "learning_rate": 3.8062143925007074e-05,
      "loss": 1.197,
      "step": 6010
    },
    {
      "epoch": 0.7267898104551491,
      "grad_norm": 13.026390075683594,
      "learning_rate": 3.804194108852883e-05,
      "loss": 1.4025,
      "step": 6020
    },
    {
      "epoch": 0.7279971024990946,
      "grad_norm": 18.132537841796875,
      "learning_rate": 3.8021738252050584e-05,
      "loss": 1.3149,
      "step": 6030
    },
    {
      "epoch": 0.72920439454304,
      "grad_norm": 15.85509204864502,
      "learning_rate": 3.800153541557235e-05,
      "loss": 1.3619,
      "step": 6040
    },
    {
      "epoch": 0.7304116865869854,
      "grad_norm": 22.019498825073242,
      "learning_rate": 3.798133257909411e-05,
      "loss": 0.8123,
      "step": 6050
    },
    {
      "epoch": 0.7316189786309308,
      "grad_norm": 10.342978477478027,
      "learning_rate": 3.7961129742615866e-05,
      "loss": 1.311,
      "step": 6060
    },
    {
      "epoch": 0.7328262706748763,
      "grad_norm": 5.243163108825684,
      "learning_rate": 3.7940926906137625e-05,
      "loss": 1.4002,
      "step": 6070
    },
    {
      "epoch": 0.7340335627188217,
      "grad_norm": 19.03634262084961,
      "learning_rate": 3.792072406965938e-05,
      "loss": 1.275,
      "step": 6080
    },
    {
      "epoch": 0.7352408547627671,
      "grad_norm": 12.920637130737305,
      "learning_rate": 3.790052123318114e-05,
      "loss": 1.0692,
      "step": 6090
    },
    {
      "epoch": 0.7364481468067126,
      "grad_norm": 20.97330093383789,
      "learning_rate": 3.788031839670289e-05,
      "loss": 1.2747,
      "step": 6100
    },
    {
      "epoch": 0.737655438850658,
      "grad_norm": 9.79830265045166,
      "learning_rate": 3.786011556022466e-05,
      "loss": 0.774,
      "step": 6110
    },
    {
      "epoch": 0.7388627308946034,
      "grad_norm": 15.466949462890625,
      "learning_rate": 3.783991272374642e-05,
      "loss": 0.8515,
      "step": 6120
    },
    {
      "epoch": 0.7400700229385488,
      "grad_norm": 5.718857765197754,
      "learning_rate": 3.7819709887268175e-05,
      "loss": 1.0088,
      "step": 6130
    },
    {
      "epoch": 0.7412773149824943,
      "grad_norm": 18.900920867919922,
      "learning_rate": 3.7799507050789934e-05,
      "loss": 1.1782,
      "step": 6140
    },
    {
      "epoch": 0.7424846070264397,
      "grad_norm": 20.686071395874023,
      "learning_rate": 3.777930421431169e-05,
      "loss": 1.0686,
      "step": 6150
    },
    {
      "epoch": 0.7436918990703851,
      "grad_norm": 25.060802459716797,
      "learning_rate": 3.775910137783345e-05,
      "loss": 1.2997,
      "step": 6160
    },
    {
      "epoch": 0.7448991911143306,
      "grad_norm": 16.664579391479492,
      "learning_rate": 3.77388985413552e-05,
      "loss": 1.2535,
      "step": 6170
    },
    {
      "epoch": 0.746106483158276,
      "grad_norm": 9.081557273864746,
      "learning_rate": 3.771869570487697e-05,
      "loss": 1.0879,
      "step": 6180
    },
    {
      "epoch": 0.7473137752022214,
      "grad_norm": 6.086925506591797,
      "learning_rate": 3.7698492868398726e-05,
      "loss": 1.1765,
      "step": 6190
    },
    {
      "epoch": 0.7485210672461668,
      "grad_norm": 18.266639709472656,
      "learning_rate": 3.7678290031920485e-05,
      "loss": 1.079,
      "step": 6200
    },
    {
      "epoch": 0.7497283592901123,
      "grad_norm": 6.707126617431641,
      "learning_rate": 3.765808719544224e-05,
      "loss": 0.8651,
      "step": 6210
    },
    {
      "epoch": 0.7509356513340577,
      "grad_norm": 7.5502753257751465,
      "learning_rate": 3.7637884358964e-05,
      "loss": 1.4936,
      "step": 6220
    },
    {
      "epoch": 0.7521429433780031,
      "grad_norm": 8.294759750366211,
      "learning_rate": 3.761768152248576e-05,
      "loss": 1.3242,
      "step": 6230
    },
    {
      "epoch": 0.7533502354219486,
      "grad_norm": 12.217183113098145,
      "learning_rate": 3.759747868600751e-05,
      "loss": 1.4978,
      "step": 6240
    },
    {
      "epoch": 0.754557527465894,
      "grad_norm": 15.602599143981934,
      "learning_rate": 3.757727584952928e-05,
      "loss": 1.1139,
      "step": 6250
    },
    {
      "epoch": 0.7557648195098394,
      "grad_norm": 30.808263778686523,
      "learning_rate": 3.7557073013051036e-05,
      "loss": 1.0699,
      "step": 6260
    },
    {
      "epoch": 0.7569721115537849,
      "grad_norm": 17.670814514160156,
      "learning_rate": 3.7536870176572794e-05,
      "loss": 1.1453,
      "step": 6270
    },
    {
      "epoch": 0.7581794035977303,
      "grad_norm": 8.318049430847168,
      "learning_rate": 3.751666734009455e-05,
      "loss": 1.2287,
      "step": 6280
    },
    {
      "epoch": 0.7593866956416757,
      "grad_norm": 13.859355926513672,
      "learning_rate": 3.7496464503616304e-05,
      "loss": 0.8042,
      "step": 6290
    },
    {
      "epoch": 0.7605939876856211,
      "grad_norm": 14.095185279846191,
      "learning_rate": 3.747626166713807e-05,
      "loss": 1.163,
      "step": 6300
    },
    {
      "epoch": 0.7618012797295666,
      "grad_norm": 10.636726379394531,
      "learning_rate": 3.745605883065982e-05,
      "loss": 0.9425,
      "step": 6310
    },
    {
      "epoch": 0.763008571773512,
      "grad_norm": 13.010544776916504,
      "learning_rate": 3.7435855994181586e-05,
      "loss": 1.1415,
      "step": 6320
    },
    {
      "epoch": 0.7642158638174574,
      "grad_norm": 13.287128448486328,
      "learning_rate": 3.7415653157703345e-05,
      "loss": 1.0001,
      "step": 6330
    },
    {
      "epoch": 0.7654231558614029,
      "grad_norm": 26.043094635009766,
      "learning_rate": 3.7395450321225103e-05,
      "loss": 1.2829,
      "step": 6340
    },
    {
      "epoch": 0.7666304479053483,
      "grad_norm": 6.012359619140625,
      "learning_rate": 3.737524748474686e-05,
      "loss": 1.1179,
      "step": 6350
    },
    {
      "epoch": 0.7678377399492937,
      "grad_norm": 12.80908203125,
      "learning_rate": 3.7355044648268614e-05,
      "loss": 0.7993,
      "step": 6360
    },
    {
      "epoch": 0.7690450319932391,
      "grad_norm": 3.56353759765625,
      "learning_rate": 3.733484181179038e-05,
      "loss": 0.7913,
      "step": 6370
    },
    {
      "epoch": 0.7702523240371846,
      "grad_norm": 9.892097473144531,
      "learning_rate": 3.731463897531214e-05,
      "loss": 1.003,
      "step": 6380
    },
    {
      "epoch": 0.77145961608113,
      "grad_norm": 6.784679889678955,
      "learning_rate": 3.7294436138833896e-05,
      "loss": 0.9839,
      "step": 6390
    },
    {
      "epoch": 0.7726669081250754,
      "grad_norm": 21.060497283935547,
      "learning_rate": 3.7274233302355654e-05,
      "loss": 1.7711,
      "step": 6400
    },
    {
      "epoch": 0.7738742001690209,
      "grad_norm": 12.580795288085938,
      "learning_rate": 3.725403046587741e-05,
      "loss": 0.9956,
      "step": 6410
    },
    {
      "epoch": 0.7750814922129663,
      "grad_norm": 3.2638285160064697,
      "learning_rate": 3.723382762939917e-05,
      "loss": 1.299,
      "step": 6420
    },
    {
      "epoch": 0.7762887842569117,
      "grad_norm": 6.428546905517578,
      "learning_rate": 3.721362479292092e-05,
      "loss": 1.3667,
      "step": 6430
    },
    {
      "epoch": 0.7774960763008572,
      "grad_norm": 5.612183094024658,
      "learning_rate": 3.719342195644269e-05,
      "loss": 1.6076,
      "step": 6440
    },
    {
      "epoch": 0.7787033683448026,
      "grad_norm": 23.472747802734375,
      "learning_rate": 3.717321911996445e-05,
      "loss": 1.1346,
      "step": 6450
    },
    {
      "epoch": 0.779910660388748,
      "grad_norm": 10.587603569030762,
      "learning_rate": 3.7153016283486205e-05,
      "loss": 1.4035,
      "step": 6460
    },
    {
      "epoch": 0.7811179524326934,
      "grad_norm": 10.357770919799805,
      "learning_rate": 3.7132813447007964e-05,
      "loss": 0.9828,
      "step": 6470
    },
    {
      "epoch": 0.7823252444766389,
      "grad_norm": 25.28893280029297,
      "learning_rate": 3.711261061052972e-05,
      "loss": 1.3453,
      "step": 6480
    },
    {
      "epoch": 0.7835325365205843,
      "grad_norm": 26.423730850219727,
      "learning_rate": 3.709240777405148e-05,
      "loss": 0.9322,
      "step": 6490
    },
    {
      "epoch": 0.7847398285645297,
      "grad_norm": 21.293434143066406,
      "learning_rate": 3.707220493757323e-05,
      "loss": 1.398,
      "step": 6500
    },
    {
      "epoch": 0.7859471206084752,
      "grad_norm": 6.445140361785889,
      "learning_rate": 3.7052002101095e-05,
      "loss": 1.0628,
      "step": 6510
    },
    {
      "epoch": 0.7871544126524206,
      "grad_norm": 8.619717597961426,
      "learning_rate": 3.7031799264616756e-05,
      "loss": 1.0456,
      "step": 6520
    },
    {
      "epoch": 0.788361704696366,
      "grad_norm": 6.591585159301758,
      "learning_rate": 3.7011596428138514e-05,
      "loss": 0.8229,
      "step": 6530
    },
    {
      "epoch": 0.7895689967403114,
      "grad_norm": 11.97983455657959,
      "learning_rate": 3.699139359166027e-05,
      "loss": 1.3401,
      "step": 6540
    },
    {
      "epoch": 0.7907762887842569,
      "grad_norm": 2.6246821880340576,
      "learning_rate": 3.6971190755182025e-05,
      "loss": 1.2847,
      "step": 6550
    },
    {
      "epoch": 0.7919835808282023,
      "grad_norm": 15.185606002807617,
      "learning_rate": 3.695098791870379e-05,
      "loss": 1.4465,
      "step": 6560
    },
    {
      "epoch": 0.7931908728721477,
      "grad_norm": 23.015501022338867,
      "learning_rate": 3.693078508222554e-05,
      "loss": 1.0372,
      "step": 6570
    },
    {
      "epoch": 0.7943981649160932,
      "grad_norm": 3.9782052040100098,
      "learning_rate": 3.691058224574731e-05,
      "loss": 1.1665,
      "step": 6580
    },
    {
      "epoch": 0.7956054569600386,
      "grad_norm": 7.766810894012451,
      "learning_rate": 3.6890379409269065e-05,
      "loss": 1.333,
      "step": 6590
    },
    {
      "epoch": 0.796812749003984,
      "grad_norm": 8.568608283996582,
      "learning_rate": 3.6870176572790824e-05,
      "loss": 1.0275,
      "step": 6600
    },
    {
      "epoch": 0.7980200410479295,
      "grad_norm": 11.476324081420898,
      "learning_rate": 3.684997373631258e-05,
      "loss": 1.0721,
      "step": 6610
    },
    {
      "epoch": 0.7992273330918749,
      "grad_norm": 11.087934494018555,
      "learning_rate": 3.6829770899834334e-05,
      "loss": 1.0646,
      "step": 6620
    },
    {
      "epoch": 0.8004346251358203,
      "grad_norm": 8.237784385681152,
      "learning_rate": 3.68095680633561e-05,
      "loss": 1.156,
      "step": 6630
    },
    {
      "epoch": 0.8016419171797657,
      "grad_norm": 12.454843521118164,
      "learning_rate": 3.678936522687785e-05,
      "loss": 1.5213,
      "step": 6640
    },
    {
      "epoch": 0.8028492092237112,
      "grad_norm": 16.190034866333008,
      "learning_rate": 3.6769162390399616e-05,
      "loss": 1.086,
      "step": 6650
    },
    {
      "epoch": 0.8040565012676566,
      "grad_norm": 8.783921241760254,
      "learning_rate": 3.6748959553921375e-05,
      "loss": 1.4831,
      "step": 6660
    },
    {
      "epoch": 0.805263793311602,
      "grad_norm": 15.900962829589844,
      "learning_rate": 3.672875671744313e-05,
      "loss": 1.4304,
      "step": 6670
    },
    {
      "epoch": 0.8064710853555475,
      "grad_norm": 8.945877075195312,
      "learning_rate": 3.670855388096489e-05,
      "loss": 0.7573,
      "step": 6680
    },
    {
      "epoch": 0.8076783773994929,
      "grad_norm": 6.700409412384033,
      "learning_rate": 3.668835104448664e-05,
      "loss": 1.4799,
      "step": 6690
    },
    {
      "epoch": 0.8088856694434383,
      "grad_norm": 10.808086395263672,
      "learning_rate": 3.666814820800841e-05,
      "loss": 1.3668,
      "step": 6700
    },
    {
      "epoch": 0.8100929614873837,
      "grad_norm": 19.012516021728516,
      "learning_rate": 3.664794537153016e-05,
      "loss": 0.9632,
      "step": 6710
    },
    {
      "epoch": 0.8113002535313292,
      "grad_norm": 8.940380096435547,
      "learning_rate": 3.6627742535051925e-05,
      "loss": 0.7908,
      "step": 6720
    },
    {
      "epoch": 0.8125075455752747,
      "grad_norm": 11.524357795715332,
      "learning_rate": 3.6607539698573684e-05,
      "loss": 1.4009,
      "step": 6730
    },
    {
      "epoch": 0.8137148376192201,
      "grad_norm": 34.07101058959961,
      "learning_rate": 3.658733686209544e-05,
      "loss": 1.2298,
      "step": 6740
    },
    {
      "epoch": 0.8149221296631656,
      "grad_norm": 20.03511619567871,
      "learning_rate": 3.65671340256172e-05,
      "loss": 1.2053,
      "step": 6750
    },
    {
      "epoch": 0.816129421707111,
      "grad_norm": 10.519394874572754,
      "learning_rate": 3.654693118913895e-05,
      "loss": 0.9883,
      "step": 6760
    },
    {
      "epoch": 0.8173367137510564,
      "grad_norm": 23.005714416503906,
      "learning_rate": 3.652672835266072e-05,
      "loss": 1.1121,
      "step": 6770
    },
    {
      "epoch": 0.8185440057950019,
      "grad_norm": 10.819910049438477,
      "learning_rate": 3.650652551618247e-05,
      "loss": 1.6912,
      "step": 6780
    },
    {
      "epoch": 0.8197512978389473,
      "grad_norm": 4.602050304412842,
      "learning_rate": 3.6486322679704235e-05,
      "loss": 1.2946,
      "step": 6790
    },
    {
      "epoch": 0.8209585898828927,
      "grad_norm": 12.130800247192383,
      "learning_rate": 3.646611984322599e-05,
      "loss": 0.8795,
      "step": 6800
    },
    {
      "epoch": 0.8221658819268381,
      "grad_norm": 7.219020366668701,
      "learning_rate": 3.644591700674775e-05,
      "loss": 1.1741,
      "step": 6810
    },
    {
      "epoch": 0.8233731739707836,
      "grad_norm": 6.882835865020752,
      "learning_rate": 3.642571417026951e-05,
      "loss": 0.7891,
      "step": 6820
    },
    {
      "epoch": 0.824580466014729,
      "grad_norm": 16.500715255737305,
      "learning_rate": 3.640551133379126e-05,
      "loss": 1.1796,
      "step": 6830
    },
    {
      "epoch": 0.8257877580586744,
      "grad_norm": 8.792635917663574,
      "learning_rate": 3.638530849731303e-05,
      "loss": 1.1132,
      "step": 6840
    },
    {
      "epoch": 0.8269950501026199,
      "grad_norm": 16.490474700927734,
      "learning_rate": 3.636510566083478e-05,
      "loss": 1.0773,
      "step": 6850
    },
    {
      "epoch": 0.8282023421465653,
      "grad_norm": 3.6070940494537354,
      "learning_rate": 3.6344902824356544e-05,
      "loss": 1.1621,
      "step": 6860
    },
    {
      "epoch": 0.8294096341905107,
      "grad_norm": 16.744394302368164,
      "learning_rate": 3.63246999878783e-05,
      "loss": 0.7763,
      "step": 6870
    },
    {
      "epoch": 0.8306169262344562,
      "grad_norm": 10.162827491760254,
      "learning_rate": 3.6304497151400054e-05,
      "loss": 1.1176,
      "step": 6880
    },
    {
      "epoch": 0.8318242182784016,
      "grad_norm": 14.822830200195312,
      "learning_rate": 3.628429431492182e-05,
      "loss": 1.0326,
      "step": 6890
    },
    {
      "epoch": 0.833031510322347,
      "grad_norm": 7.598111629486084,
      "learning_rate": 3.626409147844357e-05,
      "loss": 1.1403,
      "step": 6900
    },
    {
      "epoch": 0.8342388023662924,
      "grad_norm": 8.557995796203613,
      "learning_rate": 3.6243888641965337e-05,
      "loss": 1.1759,
      "step": 6910
    },
    {
      "epoch": 0.8354460944102379,
      "grad_norm": 16.53129005432129,
      "learning_rate": 3.622368580548709e-05,
      "loss": 1.0934,
      "step": 6920
    },
    {
      "epoch": 0.8366533864541833,
      "grad_norm": 37.44465255737305,
      "learning_rate": 3.6203482969008853e-05,
      "loss": 0.9243,
      "step": 6930
    },
    {
      "epoch": 0.8378606784981287,
      "grad_norm": 18.452354431152344,
      "learning_rate": 3.618328013253061e-05,
      "loss": 1.2455,
      "step": 6940
    },
    {
      "epoch": 0.8390679705420742,
      "grad_norm": 16.42274284362793,
      "learning_rate": 3.6163077296052364e-05,
      "loss": 1.4989,
      "step": 6950
    },
    {
      "epoch": 0.8402752625860196,
      "grad_norm": 21.33744239807129,
      "learning_rate": 3.614287445957413e-05,
      "loss": 1.1859,
      "step": 6960
    },
    {
      "epoch": 0.841482554629965,
      "grad_norm": 13.678173065185547,
      "learning_rate": 3.612267162309588e-05,
      "loss": 1.0704,
      "step": 6970
    },
    {
      "epoch": 0.8426898466739104,
      "grad_norm": 9.459888458251953,
      "learning_rate": 3.6102468786617646e-05,
      "loss": 1.1808,
      "step": 6980
    },
    {
      "epoch": 0.8438971387178559,
      "grad_norm": 6.979079246520996,
      "learning_rate": 3.60822659501394e-05,
      "loss": 1.5001,
      "step": 6990
    },
    {
      "epoch": 0.8451044307618013,
      "grad_norm": 27.3658390045166,
      "learning_rate": 3.606206311366116e-05,
      "loss": 1.2884,
      "step": 7000
    },
    {
      "epoch": 0.8463117228057467,
      "grad_norm": 15.316617012023926,
      "learning_rate": 3.604186027718292e-05,
      "loss": 1.3865,
      "step": 7010
    },
    {
      "epoch": 0.8475190148496922,
      "grad_norm": 5.345479965209961,
      "learning_rate": 3.602165744070467e-05,
      "loss": 1.3919,
      "step": 7020
    },
    {
      "epoch": 0.8487263068936376,
      "grad_norm": 18.665302276611328,
      "learning_rate": 3.600145460422644e-05,
      "loss": 1.2199,
      "step": 7030
    },
    {
      "epoch": 0.849933598937583,
      "grad_norm": 3.7863454818725586,
      "learning_rate": 3.598125176774819e-05,
      "loss": 0.6516,
      "step": 7040
    },
    {
      "epoch": 0.8511408909815285,
      "grad_norm": 10.453804016113281,
      "learning_rate": 3.5961048931269955e-05,
      "loss": 0.8875,
      "step": 7050
    },
    {
      "epoch": 0.8523481830254739,
      "grad_norm": 16.547687530517578,
      "learning_rate": 3.594084609479171e-05,
      "loss": 0.9794,
      "step": 7060
    },
    {
      "epoch": 0.8535554750694193,
      "grad_norm": 10.916485786437988,
      "learning_rate": 3.592064325831347e-05,
      "loss": 1.02,
      "step": 7070
    },
    {
      "epoch": 0.8547627671133647,
      "grad_norm": 18.049457550048828,
      "learning_rate": 3.590044042183523e-05,
      "loss": 1.2219,
      "step": 7080
    },
    {
      "epoch": 0.8559700591573102,
      "grad_norm": 7.398848056793213,
      "learning_rate": 3.588023758535698e-05,
      "loss": 1.1602,
      "step": 7090
    },
    {
      "epoch": 0.8571773512012556,
      "grad_norm": 9.703118324279785,
      "learning_rate": 3.586003474887875e-05,
      "loss": 1.2709,
      "step": 7100
    },
    {
      "epoch": 0.858384643245201,
      "grad_norm": 7.91157865524292,
      "learning_rate": 3.58398319124005e-05,
      "loss": 1.1199,
      "step": 7110
    },
    {
      "epoch": 0.8595919352891465,
      "grad_norm": 2.9030418395996094,
      "learning_rate": 3.5819629075922264e-05,
      "loss": 1.1562,
      "step": 7120
    },
    {
      "epoch": 0.8607992273330919,
      "grad_norm": 12.654620170593262,
      "learning_rate": 3.5799426239444016e-05,
      "loss": 1.3412,
      "step": 7130
    },
    {
      "epoch": 0.8620065193770373,
      "grad_norm": 20.14617919921875,
      "learning_rate": 3.5779223402965775e-05,
      "loss": 1.136,
      "step": 7140
    },
    {
      "epoch": 0.8632138114209827,
      "grad_norm": 15.538508415222168,
      "learning_rate": 3.575902056648754e-05,
      "loss": 0.9302,
      "step": 7150
    },
    {
      "epoch": 0.8644211034649282,
      "grad_norm": 1.9748427867889404,
      "learning_rate": 3.573881773000929e-05,
      "loss": 1.325,
      "step": 7160
    },
    {
      "epoch": 0.8656283955088736,
      "grad_norm": 8.648490905761719,
      "learning_rate": 3.571861489353106e-05,
      "loss": 1.1461,
      "step": 7170
    },
    {
      "epoch": 0.866835687552819,
      "grad_norm": 4.039603233337402,
      "learning_rate": 3.569841205705281e-05,
      "loss": 1.0726,
      "step": 7180
    },
    {
      "epoch": 0.8680429795967645,
      "grad_norm": 8.22611141204834,
      "learning_rate": 3.5678209220574574e-05,
      "loss": 1.3468,
      "step": 7190
    },
    {
      "epoch": 0.8692502716407099,
      "grad_norm": 17.665164947509766,
      "learning_rate": 3.5658006384096326e-05,
      "loss": 1.0602,
      "step": 7200
    },
    {
      "epoch": 0.8704575636846553,
      "grad_norm": 13.192081451416016,
      "learning_rate": 3.5637803547618084e-05,
      "loss": 0.7894,
      "step": 7210
    },
    {
      "epoch": 0.8716648557286008,
      "grad_norm": 20.664249420166016,
      "learning_rate": 3.561760071113985e-05,
      "loss": 1.1514,
      "step": 7220
    },
    {
      "epoch": 0.8728721477725462,
      "grad_norm": 15.55339527130127,
      "learning_rate": 3.55973978746616e-05,
      "loss": 1.5981,
      "step": 7230
    },
    {
      "epoch": 0.8740794398164916,
      "grad_norm": 24.421602249145508,
      "learning_rate": 3.5577195038183366e-05,
      "loss": 0.8232,
      "step": 7240
    },
    {
      "epoch": 0.875286731860437,
      "grad_norm": 9.89637279510498,
      "learning_rate": 3.555699220170512e-05,
      "loss": 0.8157,
      "step": 7250
    },
    {
      "epoch": 0.8764940239043825,
      "grad_norm": 14.998745918273926,
      "learning_rate": 3.553678936522688e-05,
      "loss": 1.4917,
      "step": 7260
    },
    {
      "epoch": 0.8777013159483279,
      "grad_norm": 20.63814926147461,
      "learning_rate": 3.5516586528748635e-05,
      "loss": 1.3111,
      "step": 7270
    },
    {
      "epoch": 0.8789086079922733,
      "grad_norm": 19.067289352416992,
      "learning_rate": 3.549638369227039e-05,
      "loss": 1.103,
      "step": 7280
    },
    {
      "epoch": 0.8801159000362188,
      "grad_norm": 14.51915168762207,
      "learning_rate": 3.547618085579216e-05,
      "loss": 0.9856,
      "step": 7290
    },
    {
      "epoch": 0.8813231920801642,
      "grad_norm": 7.631899833679199,
      "learning_rate": 3.545597801931391e-05,
      "loss": 0.9646,
      "step": 7300
    },
    {
      "epoch": 0.8825304841241096,
      "grad_norm": 13.418293952941895,
      "learning_rate": 3.5435775182835676e-05,
      "loss": 1.5088,
      "step": 7310
    },
    {
      "epoch": 0.883737776168055,
      "grad_norm": 11.499563217163086,
      "learning_rate": 3.541557234635743e-05,
      "loss": 1.4403,
      "step": 7320
    },
    {
      "epoch": 0.8849450682120005,
      "grad_norm": 15.919937133789062,
      "learning_rate": 3.539536950987919e-05,
      "loss": 1.2229,
      "step": 7330
    },
    {
      "epoch": 0.8861523602559459,
      "grad_norm": 24.40229034423828,
      "learning_rate": 3.5375166673400944e-05,
      "loss": 1.0436,
      "step": 7340
    },
    {
      "epoch": 0.8873596522998913,
      "grad_norm": 9.962295532226562,
      "learning_rate": 3.53549638369227e-05,
      "loss": 0.8884,
      "step": 7350
    },
    {
      "epoch": 0.8885669443438368,
      "grad_norm": 9.973992347717285,
      "learning_rate": 3.533476100044447e-05,
      "loss": 0.898,
      "step": 7360
    },
    {
      "epoch": 0.8897742363877822,
      "grad_norm": 35.84889602661133,
      "learning_rate": 3.531455816396622e-05,
      "loss": 1.2909,
      "step": 7370
    },
    {
      "epoch": 0.8909815284317276,
      "grad_norm": 19.416950225830078,
      "learning_rate": 3.5294355327487985e-05,
      "loss": 1.2304,
      "step": 7380
    },
    {
      "epoch": 0.892188820475673,
      "grad_norm": 6.675816535949707,
      "learning_rate": 3.5274152491009737e-05,
      "loss": 1.0671,
      "step": 7390
    },
    {
      "epoch": 0.8933961125196185,
      "grad_norm": 9.728459358215332,
      "learning_rate": 3.52539496545315e-05,
      "loss": 1.2201,
      "step": 7400
    },
    {
      "epoch": 0.8946034045635639,
      "grad_norm": 9.942370414733887,
      "learning_rate": 3.5233746818053253e-05,
      "loss": 0.7414,
      "step": 7410
    },
    {
      "epoch": 0.8958106966075093,
      "grad_norm": 6.215778350830078,
      "learning_rate": 3.521354398157501e-05,
      "loss": 1.1055,
      "step": 7420
    },
    {
      "epoch": 0.8970179886514548,
      "grad_norm": 31.132892608642578,
      "learning_rate": 3.519334114509678e-05,
      "loss": 1.1825,
      "step": 7430
    },
    {
      "epoch": 0.8982252806954002,
      "grad_norm": 11.309321403503418,
      "learning_rate": 3.517313830861853e-05,
      "loss": 1.2905,
      "step": 7440
    },
    {
      "epoch": 0.8994325727393456,
      "grad_norm": 15.365930557250977,
      "learning_rate": 3.5152935472140294e-05,
      "loss": 1.2928,
      "step": 7450
    },
    {
      "epoch": 0.900639864783291,
      "grad_norm": 8.302011489868164,
      "learning_rate": 3.5132732635662046e-05,
      "loss": 0.6266,
      "step": 7460
    },
    {
      "epoch": 0.9018471568272365,
      "grad_norm": 17.765174865722656,
      "learning_rate": 3.5112529799183804e-05,
      "loss": 1.1719,
      "step": 7470
    },
    {
      "epoch": 0.9030544488711819,
      "grad_norm": 10.360702514648438,
      "learning_rate": 3.509232696270556e-05,
      "loss": 1.0348,
      "step": 7480
    },
    {
      "epoch": 0.9042617409151273,
      "grad_norm": 15.0255126953125,
      "learning_rate": 3.507212412622732e-05,
      "loss": 0.8026,
      "step": 7490
    },
    {
      "epoch": 0.9054690329590728,
      "grad_norm": 32.968650817871094,
      "learning_rate": 3.5051921289749087e-05,
      "loss": 1.0765,
      "step": 7500
    },
    {
      "epoch": 0.9066763250030182,
      "grad_norm": 9.72695541381836,
      "learning_rate": 3.503171845327084e-05,
      "loss": 1.242,
      "step": 7510
    },
    {
      "epoch": 0.9078836170469636,
      "grad_norm": 3.4234838485717773,
      "learning_rate": 3.5011515616792603e-05,
      "loss": 0.9607,
      "step": 7520
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 15.997836112976074,
      "learning_rate": 3.4991312780314355e-05,
      "loss": 1.3708,
      "step": 7530
    },
    {
      "epoch": 0.9102982011348545,
      "grad_norm": 5.904785633087158,
      "learning_rate": 3.4971109943836114e-05,
      "loss": 1.1813,
      "step": 7540
    },
    {
      "epoch": 0.9115054931787999,
      "grad_norm": 9.463635444641113,
      "learning_rate": 3.495090710735787e-05,
      "loss": 1.6051,
      "step": 7550
    },
    {
      "epoch": 0.9127127852227453,
      "grad_norm": 16.61902618408203,
      "learning_rate": 3.493070427087963e-05,
      "loss": 1.2508,
      "step": 7560
    },
    {
      "epoch": 0.9139200772666908,
      "grad_norm": 15.488524436950684,
      "learning_rate": 3.4910501434401396e-05,
      "loss": 1.2154,
      "step": 7570
    },
    {
      "epoch": 0.9151273693106362,
      "grad_norm": 12.61644172668457,
      "learning_rate": 3.489029859792315e-05,
      "loss": 0.9664,
      "step": 7580
    },
    {
      "epoch": 0.9163346613545816,
      "grad_norm": 20.289548873901367,
      "learning_rate": 3.487009576144491e-05,
      "loss": 0.7491,
      "step": 7590
    },
    {
      "epoch": 0.9175419533985271,
      "grad_norm": 11.547300338745117,
      "learning_rate": 3.4849892924966665e-05,
      "loss": 1.0742,
      "step": 7600
    },
    {
      "epoch": 0.9187492454424725,
      "grad_norm": 6.234065532684326,
      "learning_rate": 3.482969008848842e-05,
      "loss": 0.9263,
      "step": 7610
    },
    {
      "epoch": 0.9199565374864179,
      "grad_norm": 24.507972717285156,
      "learning_rate": 3.480948725201018e-05,
      "loss": 1.1268,
      "step": 7620
    },
    {
      "epoch": 0.9211638295303634,
      "grad_norm": 14.860640525817871,
      "learning_rate": 3.478928441553194e-05,
      "loss": 1.1757,
      "step": 7630
    },
    {
      "epoch": 0.9223711215743088,
      "grad_norm": 24.424785614013672,
      "learning_rate": 3.4769081579053705e-05,
      "loss": 0.9603,
      "step": 7640
    },
    {
      "epoch": 0.9235784136182542,
      "grad_norm": 3.6783227920532227,
      "learning_rate": 3.474887874257546e-05,
      "loss": 1.0023,
      "step": 7650
    },
    {
      "epoch": 0.9247857056621996,
      "grad_norm": 10.580509185791016,
      "learning_rate": 3.472867590609722e-05,
      "loss": 1.2017,
      "step": 7660
    },
    {
      "epoch": 0.9259929977061451,
      "grad_norm": 8.9448823928833,
      "learning_rate": 3.4708473069618974e-05,
      "loss": 0.7989,
      "step": 7670
    },
    {
      "epoch": 0.9272002897500905,
      "grad_norm": 24.44384765625,
      "learning_rate": 3.468827023314073e-05,
      "loss": 1.3081,
      "step": 7680
    },
    {
      "epoch": 0.9284075817940359,
      "grad_norm": 10.523001670837402,
      "learning_rate": 3.46680673966625e-05,
      "loss": 1.0089,
      "step": 7690
    },
    {
      "epoch": 0.9296148738379814,
      "grad_norm": 3.547475814819336,
      "learning_rate": 3.464786456018425e-05,
      "loss": 0.9911,
      "step": 7700
    },
    {
      "epoch": 0.9308221658819268,
      "grad_norm": 2.236950397491455,
      "learning_rate": 3.4627661723706015e-05,
      "loss": 1.1123,
      "step": 7710
    },
    {
      "epoch": 0.9320294579258722,
      "grad_norm": 12.699675559997559,
      "learning_rate": 3.4607458887227766e-05,
      "loss": 0.8852,
      "step": 7720
    },
    {
      "epoch": 0.9332367499698176,
      "grad_norm": 16.795616149902344,
      "learning_rate": 3.4587256050749525e-05,
      "loss": 1.4138,
      "step": 7730
    },
    {
      "epoch": 0.9344440420137631,
      "grad_norm": 6.64914083480835,
      "learning_rate": 3.456705321427128e-05,
      "loss": 0.7982,
      "step": 7740
    },
    {
      "epoch": 0.9356513340577085,
      "grad_norm": 7.5566911697387695,
      "learning_rate": 3.454685037779304e-05,
      "loss": 1.3593,
      "step": 7750
    },
    {
      "epoch": 0.9368586261016539,
      "grad_norm": 9.405370712280273,
      "learning_rate": 3.452664754131481e-05,
      "loss": 1.1769,
      "step": 7760
    },
    {
      "epoch": 0.9380659181455995,
      "grad_norm": 6.690150737762451,
      "learning_rate": 3.450644470483656e-05,
      "loss": 1.2611,
      "step": 7770
    },
    {
      "epoch": 0.9392732101895449,
      "grad_norm": 5.840714454650879,
      "learning_rate": 3.4486241868358324e-05,
      "loss": 0.9542,
      "step": 7780
    },
    {
      "epoch": 0.9404805022334903,
      "grad_norm": 16.895051956176758,
      "learning_rate": 3.4466039031880076e-05,
      "loss": 1.1474,
      "step": 7790
    },
    {
      "epoch": 0.9416877942774358,
      "grad_norm": 9.6728515625,
      "learning_rate": 3.4445836195401834e-05,
      "loss": 1.0034,
      "step": 7800
    },
    {
      "epoch": 0.9428950863213812,
      "grad_norm": 15.010249137878418,
      "learning_rate": 3.442563335892359e-05,
      "loss": 1.4592,
      "step": 7810
    },
    {
      "epoch": 0.9441023783653266,
      "grad_norm": 19.72796058654785,
      "learning_rate": 3.440543052244535e-05,
      "loss": 1.2509,
      "step": 7820
    },
    {
      "epoch": 0.945309670409272,
      "grad_norm": 3.812546730041504,
      "learning_rate": 3.4385227685967116e-05,
      "loss": 1.0666,
      "step": 7830
    },
    {
      "epoch": 0.9465169624532175,
      "grad_norm": 11.254084587097168,
      "learning_rate": 3.436502484948887e-05,
      "loss": 1.0219,
      "step": 7840
    },
    {
      "epoch": 0.9477242544971629,
      "grad_norm": 20.908905029296875,
      "learning_rate": 3.434482201301063e-05,
      "loss": 1.1935,
      "step": 7850
    },
    {
      "epoch": 0.9489315465411083,
      "grad_norm": 4.841529369354248,
      "learning_rate": 3.4324619176532385e-05,
      "loss": 1.1918,
      "step": 7860
    },
    {
      "epoch": 0.9501388385850538,
      "grad_norm": 11.415703773498535,
      "learning_rate": 3.430441634005414e-05,
      "loss": 1.0197,
      "step": 7870
    },
    {
      "epoch": 0.9513461306289992,
      "grad_norm": 10.902266502380371,
      "learning_rate": 3.42842135035759e-05,
      "loss": 1.3457,
      "step": 7880
    },
    {
      "epoch": 0.9525534226729446,
      "grad_norm": 3.260730266571045,
      "learning_rate": 3.426401066709766e-05,
      "loss": 0.9663,
      "step": 7890
    },
    {
      "epoch": 0.9537607147168901,
      "grad_norm": 10.595190048217773,
      "learning_rate": 3.4243807830619426e-05,
      "loss": 0.9816,
      "step": 7900
    },
    {
      "epoch": 0.9549680067608355,
      "grad_norm": 10.105086326599121,
      "learning_rate": 3.422360499414118e-05,
      "loss": 1.0815,
      "step": 7910
    },
    {
      "epoch": 0.9561752988047809,
      "grad_norm": 24.862998962402344,
      "learning_rate": 3.420340215766294e-05,
      "loss": 1.1322,
      "step": 7920
    },
    {
      "epoch": 0.9573825908487263,
      "grad_norm": 7.542875289916992,
      "learning_rate": 3.4183199321184694e-05,
      "loss": 1.0678,
      "step": 7930
    },
    {
      "epoch": 0.9585898828926718,
      "grad_norm": 6.022435665130615,
      "learning_rate": 3.416299648470645e-05,
      "loss": 1.0853,
      "step": 7940
    },
    {
      "epoch": 0.9597971749366172,
      "grad_norm": 13.221710205078125,
      "learning_rate": 3.414279364822821e-05,
      "loss": 1.2282,
      "step": 7950
    },
    {
      "epoch": 0.9610044669805626,
      "grad_norm": 6.738224983215332,
      "learning_rate": 3.412259081174997e-05,
      "loss": 0.8464,
      "step": 7960
    },
    {
      "epoch": 0.9622117590245081,
      "grad_norm": 12.501680374145508,
      "learning_rate": 3.4102387975271735e-05,
      "loss": 1.4085,
      "step": 7970
    },
    {
      "epoch": 0.9634190510684535,
      "grad_norm": 23.3294620513916,
      "learning_rate": 3.4082185138793487e-05,
      "loss": 0.7528,
      "step": 7980
    },
    {
      "epoch": 0.9646263431123989,
      "grad_norm": 9.737671852111816,
      "learning_rate": 3.406198230231525e-05,
      "loss": 0.8751,
      "step": 7990
    },
    {
      "epoch": 0.9658336351563444,
      "grad_norm": 5.778087139129639,
      "learning_rate": 3.4041779465837004e-05,
      "loss": 0.8996,
      "step": 8000
    },
    {
      "epoch": 0.9670409272002898,
      "grad_norm": 15.529086112976074,
      "learning_rate": 3.402157662935876e-05,
      "loss": 1.265,
      "step": 8010
    },
    {
      "epoch": 0.9682482192442352,
      "grad_norm": 5.8196001052856445,
      "learning_rate": 3.400137379288052e-05,
      "loss": 1.2047,
      "step": 8020
    },
    {
      "epoch": 0.9694555112881806,
      "grad_norm": 5.194161415100098,
      "learning_rate": 3.398117095640228e-05,
      "loss": 1.213,
      "step": 8030
    },
    {
      "epoch": 0.9706628033321261,
      "grad_norm": 24.955963134765625,
      "learning_rate": 3.3960968119924044e-05,
      "loss": 1.235,
      "step": 8040
    },
    {
      "epoch": 0.9718700953760715,
      "grad_norm": 10.759511947631836,
      "learning_rate": 3.3940765283445796e-05,
      "loss": 1.2642,
      "step": 8050
    },
    {
      "epoch": 0.9730773874200169,
      "grad_norm": 4.011023998260498,
      "learning_rate": 3.3920562446967554e-05,
      "loss": 0.6851,
      "step": 8060
    },
    {
      "epoch": 0.9742846794639624,
      "grad_norm": 15.086312294006348,
      "learning_rate": 3.390035961048931e-05,
      "loss": 0.8035,
      "step": 8070
    },
    {
      "epoch": 0.9754919715079078,
      "grad_norm": 30.346830368041992,
      "learning_rate": 3.388015677401107e-05,
      "loss": 1.3552,
      "step": 8080
    },
    {
      "epoch": 0.9766992635518532,
      "grad_norm": 11.880096435546875,
      "learning_rate": 3.385995393753283e-05,
      "loss": 0.7942,
      "step": 8090
    },
    {
      "epoch": 0.9779065555957986,
      "grad_norm": 12.465813636779785,
      "learning_rate": 3.383975110105459e-05,
      "loss": 1.0592,
      "step": 8100
    },
    {
      "epoch": 0.9791138476397441,
      "grad_norm": 12.467931747436523,
      "learning_rate": 3.3819548264576354e-05,
      "loss": 0.838,
      "step": 8110
    },
    {
      "epoch": 0.9803211396836895,
      "grad_norm": 3.3872504234313965,
      "learning_rate": 3.3799345428098105e-05,
      "loss": 0.8799,
      "step": 8120
    },
    {
      "epoch": 0.9815284317276349,
      "grad_norm": 14.841547012329102,
      "learning_rate": 3.3779142591619864e-05,
      "loss": 1.3313,
      "step": 8130
    },
    {
      "epoch": 0.9827357237715804,
      "grad_norm": 8.42226505279541,
      "learning_rate": 3.375893975514162e-05,
      "loss": 0.9511,
      "step": 8140
    },
    {
      "epoch": 0.9839430158155258,
      "grad_norm": 14.52470588684082,
      "learning_rate": 3.373873691866338e-05,
      "loss": 1.0462,
      "step": 8150
    },
    {
      "epoch": 0.9851503078594712,
      "grad_norm": 48.48129653930664,
      "learning_rate": 3.371853408218514e-05,
      "loss": 1.2178,
      "step": 8160
    },
    {
      "epoch": 0.9863575999034166,
      "grad_norm": 7.801813125610352,
      "learning_rate": 3.36983312457069e-05,
      "loss": 0.9525,
      "step": 8170
    },
    {
      "epoch": 0.9875648919473621,
      "grad_norm": 9.257281303405762,
      "learning_rate": 3.367812840922866e-05,
      "loss": 0.8886,
      "step": 8180
    },
    {
      "epoch": 0.9887721839913075,
      "grad_norm": 19.416545867919922,
      "learning_rate": 3.3657925572750415e-05,
      "loss": 1.0107,
      "step": 8190
    },
    {
      "epoch": 0.9899794760352529,
      "grad_norm": 7.69077205657959,
      "learning_rate": 3.363772273627217e-05,
      "loss": 1.5358,
      "step": 8200
    },
    {
      "epoch": 0.9911867680791984,
      "grad_norm": 8.921051979064941,
      "learning_rate": 3.361751989979393e-05,
      "loss": 1.2141,
      "step": 8210
    },
    {
      "epoch": 0.9923940601231438,
      "grad_norm": 7.142158508300781,
      "learning_rate": 3.359731706331569e-05,
      "loss": 1.4836,
      "step": 8220
    },
    {
      "epoch": 0.9936013521670892,
      "grad_norm": 3.9266159534454346,
      "learning_rate": 3.357711422683745e-05,
      "loss": 1.1926,
      "step": 8230
    },
    {
      "epoch": 0.9948086442110347,
      "grad_norm": 17.71778106689453,
      "learning_rate": 3.355691139035921e-05,
      "loss": 0.9932,
      "step": 8240
    },
    {
      "epoch": 0.9960159362549801,
      "grad_norm": 5.80845832824707,
      "learning_rate": 3.353670855388097e-05,
      "loss": 0.7823,
      "step": 8250
    },
    {
      "epoch": 0.9972232282989255,
      "grad_norm": 15.150306701660156,
      "learning_rate": 3.3516505717402724e-05,
      "loss": 0.7466,
      "step": 8260
    },
    {
      "epoch": 0.9984305203428709,
      "grad_norm": 7.93013858795166,
      "learning_rate": 3.349630288092448e-05,
      "loss": 1.2383,
      "step": 8270
    },
    {
      "epoch": 0.9996378123868164,
      "grad_norm": 12.874200820922852,
      "learning_rate": 3.347610004444624e-05,
      "loss": 0.9072,
      "step": 8280
    },
    {
      "epoch": 1.0008451044307618,
      "grad_norm": 17.499820709228516,
      "learning_rate": 3.3455897207968e-05,
      "loss": 0.9442,
      "step": 8290
    },
    {
      "epoch": 1.0020523964747072,
      "grad_norm": 3.4708168506622314,
      "learning_rate": 3.343569437148976e-05,
      "loss": 0.8122,
      "step": 8300
    },
    {
      "epoch": 1.0032596885186527,
      "grad_norm": 11.277446746826172,
      "learning_rate": 3.3415491535011516e-05,
      "loss": 0.5659,
      "step": 8310
    },
    {
      "epoch": 1.004466980562598,
      "grad_norm": 4.8075408935546875,
      "learning_rate": 3.3395288698533275e-05,
      "loss": 1.0592,
      "step": 8320
    },
    {
      "epoch": 1.0056742726065435,
      "grad_norm": 23.001083374023438,
      "learning_rate": 3.337508586205503e-05,
      "loss": 0.6969,
      "step": 8330
    },
    {
      "epoch": 1.006881564650489,
      "grad_norm": 8.951621055603027,
      "learning_rate": 3.335488302557679e-05,
      "loss": 0.8489,
      "step": 8340
    },
    {
      "epoch": 1.0080888566944344,
      "grad_norm": 12.376330375671387,
      "learning_rate": 3.333468018909855e-05,
      "loss": 0.864,
      "step": 8350
    },
    {
      "epoch": 1.0092961487383798,
      "grad_norm": 7.239992618560791,
      "learning_rate": 3.331447735262031e-05,
      "loss": 0.8759,
      "step": 8360
    },
    {
      "epoch": 1.0105034407823252,
      "grad_norm": 26.56719207763672,
      "learning_rate": 3.329427451614207e-05,
      "loss": 1.0977,
      "step": 8370
    },
    {
      "epoch": 1.0117107328262707,
      "grad_norm": 17.076066970825195,
      "learning_rate": 3.3274071679663826e-05,
      "loss": 0.9572,
      "step": 8380
    },
    {
      "epoch": 1.012918024870216,
      "grad_norm": 7.010860919952393,
      "learning_rate": 3.3253868843185584e-05,
      "loss": 0.9164,
      "step": 8390
    },
    {
      "epoch": 1.0141253169141615,
      "grad_norm": 1.875085711479187,
      "learning_rate": 3.323366600670734e-05,
      "loss": 1.373,
      "step": 8400
    },
    {
      "epoch": 1.015332608958107,
      "grad_norm": 7.668385982513428,
      "learning_rate": 3.32134631702291e-05,
      "loss": 0.993,
      "step": 8410
    },
    {
      "epoch": 1.0165399010020524,
      "grad_norm": 10.797591209411621,
      "learning_rate": 3.319326033375086e-05,
      "loss": 1.0178,
      "step": 8420
    },
    {
      "epoch": 1.0177471930459978,
      "grad_norm": 5.546599864959717,
      "learning_rate": 3.317305749727262e-05,
      "loss": 1.2845,
      "step": 8430
    },
    {
      "epoch": 1.0189544850899432,
      "grad_norm": 23.0700740814209,
      "learning_rate": 3.3152854660794376e-05,
      "loss": 1.1538,
      "step": 8440
    },
    {
      "epoch": 1.0201617771338887,
      "grad_norm": 3.4416863918304443,
      "learning_rate": 3.3132651824316135e-05,
      "loss": 0.8053,
      "step": 8450
    },
    {
      "epoch": 1.021369069177834,
      "grad_norm": 11.88143539428711,
      "learning_rate": 3.311244898783789e-05,
      "loss": 1.2329,
      "step": 8460
    },
    {
      "epoch": 1.0225763612217795,
      "grad_norm": 15.954208374023438,
      "learning_rate": 3.309224615135965e-05,
      "loss": 1.7085,
      "step": 8470
    },
    {
      "epoch": 1.023783653265725,
      "grad_norm": 12.155426979064941,
      "learning_rate": 3.307204331488141e-05,
      "loss": 1.2021,
      "step": 8480
    },
    {
      "epoch": 1.0249909453096704,
      "grad_norm": 6.788132190704346,
      "learning_rate": 3.305184047840317e-05,
      "loss": 1.4198,
      "step": 8490
    },
    {
      "epoch": 1.0261982373536158,
      "grad_norm": 13.629963874816895,
      "learning_rate": 3.303163764192493e-05,
      "loss": 0.838,
      "step": 8500
    },
    {
      "epoch": 1.0274055293975612,
      "grad_norm": 9.267754554748535,
      "learning_rate": 3.3011434805446686e-05,
      "loss": 0.8093,
      "step": 8510
    },
    {
      "epoch": 1.0286128214415067,
      "grad_norm": 9.556741714477539,
      "learning_rate": 3.2991231968968444e-05,
      "loss": 0.5135,
      "step": 8520
    },
    {
      "epoch": 1.029820113485452,
      "grad_norm": 9.26260757446289,
      "learning_rate": 3.29710291324902e-05,
      "loss": 0.6223,
      "step": 8530
    },
    {
      "epoch": 1.0310274055293975,
      "grad_norm": 9.960397720336914,
      "learning_rate": 3.295082629601196e-05,
      "loss": 0.9756,
      "step": 8540
    },
    {
      "epoch": 1.032234697573343,
      "grad_norm": 27.06209373474121,
      "learning_rate": 3.293062345953372e-05,
      "loss": 1.3257,
      "step": 8550
    },
    {
      "epoch": 1.0334419896172884,
      "grad_norm": 18.708824157714844,
      "learning_rate": 3.291042062305548e-05,
      "loss": 1.2838,
      "step": 8560
    },
    {
      "epoch": 1.0346492816612338,
      "grad_norm": 27.101490020751953,
      "learning_rate": 3.2890217786577237e-05,
      "loss": 1.1043,
      "step": 8570
    },
    {
      "epoch": 1.0358565737051793,
      "grad_norm": 2.325141191482544,
      "learning_rate": 3.2870014950098995e-05,
      "loss": 0.7942,
      "step": 8580
    },
    {
      "epoch": 1.0370638657491247,
      "grad_norm": 2.8271889686584473,
      "learning_rate": 3.2849812113620754e-05,
      "loss": 1.0874,
      "step": 8590
    },
    {
      "epoch": 1.03827115779307,
      "grad_norm": 6.959092617034912,
      "learning_rate": 3.282960927714251e-05,
      "loss": 0.9203,
      "step": 8600
    },
    {
      "epoch": 1.0394784498370155,
      "grad_norm": 12.12652587890625,
      "learning_rate": 3.280940644066427e-05,
      "loss": 1.0594,
      "step": 8610
    },
    {
      "epoch": 1.040685741880961,
      "grad_norm": 57.616973876953125,
      "learning_rate": 3.278920360418603e-05,
      "loss": 0.9153,
      "step": 8620
    },
    {
      "epoch": 1.0418930339249064,
      "grad_norm": 8.785550117492676,
      "learning_rate": 3.276900076770779e-05,
      "loss": 0.5792,
      "step": 8630
    },
    {
      "epoch": 1.0431003259688518,
      "grad_norm": 18.17445945739746,
      "learning_rate": 3.2748797931229546e-05,
      "loss": 1.0482,
      "step": 8640
    },
    {
      "epoch": 1.0443076180127973,
      "grad_norm": 17.887046813964844,
      "learning_rate": 3.2728595094751304e-05,
      "loss": 1.2461,
      "step": 8650
    },
    {
      "epoch": 1.0455149100567427,
      "grad_norm": 34.10347366333008,
      "learning_rate": 3.270839225827306e-05,
      "loss": 1.0609,
      "step": 8660
    },
    {
      "epoch": 1.0467222021006881,
      "grad_norm": 0.8636152744293213,
      "learning_rate": 3.268818942179482e-05,
      "loss": 0.7573,
      "step": 8670
    },
    {
      "epoch": 1.0479294941446335,
      "grad_norm": 17.538061141967773,
      "learning_rate": 3.266798658531658e-05,
      "loss": 0.7863,
      "step": 8680
    },
    {
      "epoch": 1.049136786188579,
      "grad_norm": 8.66651725769043,
      "learning_rate": 3.264778374883834e-05,
      "loss": 1.4049,
      "step": 8690
    },
    {
      "epoch": 1.0503440782325244,
      "grad_norm": 37.72056198120117,
      "learning_rate": 3.26275809123601e-05,
      "loss": 1.1409,
      "step": 8700
    },
    {
      "epoch": 1.0515513702764698,
      "grad_norm": 17.057472229003906,
      "learning_rate": 3.2607378075881855e-05,
      "loss": 1.6954,
      "step": 8710
    },
    {
      "epoch": 1.0527586623204153,
      "grad_norm": 6.885904788970947,
      "learning_rate": 3.2587175239403614e-05,
      "loss": 1.0687,
      "step": 8720
    },
    {
      "epoch": 1.0539659543643607,
      "grad_norm": 14.720796585083008,
      "learning_rate": 3.256697240292537e-05,
      "loss": 0.9811,
      "step": 8730
    },
    {
      "epoch": 1.0551732464083061,
      "grad_norm": 7.962707996368408,
      "learning_rate": 3.254676956644713e-05,
      "loss": 0.8777,
      "step": 8740
    },
    {
      "epoch": 1.0563805384522515,
      "grad_norm": 16.575162887573242,
      "learning_rate": 3.252656672996889e-05,
      "loss": 1.1448,
      "step": 8750
    },
    {
      "epoch": 1.057587830496197,
      "grad_norm": 6.950376987457275,
      "learning_rate": 3.250636389349065e-05,
      "loss": 1.2605,
      "step": 8760
    },
    {
      "epoch": 1.0587951225401424,
      "grad_norm": 12.373239517211914,
      "learning_rate": 3.2486161057012406e-05,
      "loss": 1.1612,
      "step": 8770
    },
    {
      "epoch": 1.0600024145840878,
      "grad_norm": 21.290374755859375,
      "learning_rate": 3.2465958220534165e-05,
      "loss": 0.8522,
      "step": 8780
    },
    {
      "epoch": 1.0612097066280333,
      "grad_norm": 9.621726036071777,
      "learning_rate": 3.244575538405592e-05,
      "loss": 0.6158,
      "step": 8790
    },
    {
      "epoch": 1.0624169986719787,
      "grad_norm": 14.010881423950195,
      "learning_rate": 3.242555254757768e-05,
      "loss": 1.1506,
      "step": 8800
    },
    {
      "epoch": 1.0636242907159241,
      "grad_norm": 25.666791915893555,
      "learning_rate": 3.240534971109944e-05,
      "loss": 1.1566,
      "step": 8810
    },
    {
      "epoch": 1.0648315827598696,
      "grad_norm": 11.46548080444336,
      "learning_rate": 3.23851468746212e-05,
      "loss": 1.1025,
      "step": 8820
    },
    {
      "epoch": 1.066038874803815,
      "grad_norm": 17.23679542541504,
      "learning_rate": 3.236494403814296e-05,
      "loss": 0.7132,
      "step": 8830
    },
    {
      "epoch": 1.0672461668477604,
      "grad_norm": 17.705677032470703,
      "learning_rate": 3.2344741201664715e-05,
      "loss": 1.0616,
      "step": 8840
    },
    {
      "epoch": 1.0684534588917058,
      "grad_norm": 21.743337631225586,
      "learning_rate": 3.2324538365186474e-05,
      "loss": 0.9253,
      "step": 8850
    },
    {
      "epoch": 1.0696607509356513,
      "grad_norm": 9.25418758392334,
      "learning_rate": 3.230433552870823e-05,
      "loss": 0.9592,
      "step": 8860
    },
    {
      "epoch": 1.0708680429795967,
      "grad_norm": 10.688076972961426,
      "learning_rate": 3.228413269222999e-05,
      "loss": 1.5658,
      "step": 8870
    },
    {
      "epoch": 1.0720753350235421,
      "grad_norm": 4.158071041107178,
      "learning_rate": 3.226392985575175e-05,
      "loss": 0.8525,
      "step": 8880
    },
    {
      "epoch": 1.0732826270674876,
      "grad_norm": 10.590166091918945,
      "learning_rate": 3.224372701927351e-05,
      "loss": 1.2327,
      "step": 8890
    },
    {
      "epoch": 1.074489919111433,
      "grad_norm": 13.779923439025879,
      "learning_rate": 3.2223524182795266e-05,
      "loss": 1.1564,
      "step": 8900
    },
    {
      "epoch": 1.0756972111553784,
      "grad_norm": 18.28632164001465,
      "learning_rate": 3.2203321346317025e-05,
      "loss": 0.9528,
      "step": 8910
    },
    {
      "epoch": 1.0769045031993238,
      "grad_norm": 15.73055648803711,
      "learning_rate": 3.218311850983878e-05,
      "loss": 1.4725,
      "step": 8920
    },
    {
      "epoch": 1.0781117952432693,
      "grad_norm": 11.596927642822266,
      "learning_rate": 3.216291567336054e-05,
      "loss": 0.8011,
      "step": 8930
    },
    {
      "epoch": 1.0793190872872147,
      "grad_norm": 9.389619827270508,
      "learning_rate": 3.21427128368823e-05,
      "loss": 1.1378,
      "step": 8940
    },
    {
      "epoch": 1.0805263793311601,
      "grad_norm": 4.224643707275391,
      "learning_rate": 3.212251000040406e-05,
      "loss": 0.8325,
      "step": 8950
    },
    {
      "epoch": 1.0817336713751056,
      "grad_norm": 7.635295391082764,
      "learning_rate": 3.210230716392582e-05,
      "loss": 1.2528,
      "step": 8960
    },
    {
      "epoch": 1.082940963419051,
      "grad_norm": 19.735958099365234,
      "learning_rate": 3.2082104327447576e-05,
      "loss": 1.1895,
      "step": 8970
    },
    {
      "epoch": 1.0841482554629964,
      "grad_norm": 6.686542987823486,
      "learning_rate": 3.2061901490969334e-05,
      "loss": 0.8889,
      "step": 8980
    },
    {
      "epoch": 1.0853555475069419,
      "grad_norm": 31.56827735900879,
      "learning_rate": 3.204169865449109e-05,
      "loss": 0.9487,
      "step": 8990
    },
    {
      "epoch": 1.0865628395508873,
      "grad_norm": 18.930339813232422,
      "learning_rate": 3.202149581801285e-05,
      "loss": 1.3087,
      "step": 9000
    },
    {
      "epoch": 1.0877701315948327,
      "grad_norm": 3.0784895420074463,
      "learning_rate": 3.200129298153461e-05,
      "loss": 0.6919,
      "step": 9010
    },
    {
      "epoch": 1.0889774236387781,
      "grad_norm": 3.755145788192749,
      "learning_rate": 3.198109014505637e-05,
      "loss": 0.9477,
      "step": 9020
    },
    {
      "epoch": 1.0901847156827236,
      "grad_norm": 15.612567901611328,
      "learning_rate": 3.1960887308578126e-05,
      "loss": 1.0356,
      "step": 9030
    },
    {
      "epoch": 1.091392007726669,
      "grad_norm": 3.925875425338745,
      "learning_rate": 3.1940684472099885e-05,
      "loss": 0.8205,
      "step": 9040
    },
    {
      "epoch": 1.0925992997706144,
      "grad_norm": 11.74939250946045,
      "learning_rate": 3.192048163562164e-05,
      "loss": 0.9014,
      "step": 9050
    },
    {
      "epoch": 1.0938065918145599,
      "grad_norm": 10.823087692260742,
      "learning_rate": 3.19002787991434e-05,
      "loss": 0.894,
      "step": 9060
    },
    {
      "epoch": 1.0950138838585053,
      "grad_norm": 12.108199119567871,
      "learning_rate": 3.188007596266516e-05,
      "loss": 0.8587,
      "step": 9070
    },
    {
      "epoch": 1.0962211759024507,
      "grad_norm": 32.45283889770508,
      "learning_rate": 3.185987312618692e-05,
      "loss": 1.1818,
      "step": 9080
    },
    {
      "epoch": 1.0974284679463961,
      "grad_norm": 24.492156982421875,
      "learning_rate": 3.183967028970868e-05,
      "loss": 0.8781,
      "step": 9090
    },
    {
      "epoch": 1.0986357599903416,
      "grad_norm": 17.485851287841797,
      "learning_rate": 3.1819467453230436e-05,
      "loss": 1.0862,
      "step": 9100
    },
    {
      "epoch": 1.099843052034287,
      "grad_norm": 18.51313591003418,
      "learning_rate": 3.1799264616752194e-05,
      "loss": 0.8755,
      "step": 9110
    },
    {
      "epoch": 1.1010503440782324,
      "grad_norm": 33.59788131713867,
      "learning_rate": 3.177906178027395e-05,
      "loss": 1.1558,
      "step": 9120
    },
    {
      "epoch": 1.1022576361221779,
      "grad_norm": 25.604366302490234,
      "learning_rate": 3.175885894379571e-05,
      "loss": 0.8319,
      "step": 9130
    },
    {
      "epoch": 1.1034649281661233,
      "grad_norm": 15.357072830200195,
      "learning_rate": 3.173865610731747e-05,
      "loss": 1.0917,
      "step": 9140
    },
    {
      "epoch": 1.1046722202100687,
      "grad_norm": 16.471113204956055,
      "learning_rate": 3.171845327083923e-05,
      "loss": 0.7525,
      "step": 9150
    },
    {
      "epoch": 1.1058795122540142,
      "grad_norm": 11.064594268798828,
      "learning_rate": 3.1698250434360987e-05,
      "loss": 0.889,
      "step": 9160
    },
    {
      "epoch": 1.1070868042979596,
      "grad_norm": 8.191085815429688,
      "learning_rate": 3.1678047597882745e-05,
      "loss": 1.0637,
      "step": 9170
    },
    {
      "epoch": 1.108294096341905,
      "grad_norm": 22.639585494995117,
      "learning_rate": 3.1657844761404504e-05,
      "loss": 1.1181,
      "step": 9180
    },
    {
      "epoch": 1.1095013883858504,
      "grad_norm": 11.537515640258789,
      "learning_rate": 3.163764192492626e-05,
      "loss": 0.6787,
      "step": 9190
    },
    {
      "epoch": 1.1107086804297959,
      "grad_norm": 16.244659423828125,
      "learning_rate": 3.161743908844802e-05,
      "loss": 1.467,
      "step": 9200
    },
    {
      "epoch": 1.1119159724737413,
      "grad_norm": 7.640618801116943,
      "learning_rate": 3.159723625196978e-05,
      "loss": 0.9924,
      "step": 9210
    },
    {
      "epoch": 1.1131232645176867,
      "grad_norm": 10.02643871307373,
      "learning_rate": 3.157703341549154e-05,
      "loss": 1.0916,
      "step": 9220
    },
    {
      "epoch": 1.1143305565616322,
      "grad_norm": 21.139366149902344,
      "learning_rate": 3.1556830579013296e-05,
      "loss": 1.4281,
      "step": 9230
    },
    {
      "epoch": 1.1155378486055776,
      "grad_norm": 8.054068565368652,
      "learning_rate": 3.1536627742535054e-05,
      "loss": 1.1035,
      "step": 9240
    },
    {
      "epoch": 1.116745140649523,
      "grad_norm": 40.69295120239258,
      "learning_rate": 3.151642490605681e-05,
      "loss": 1.2807,
      "step": 9250
    },
    {
      "epoch": 1.1179524326934684,
      "grad_norm": 12.031417846679688,
      "learning_rate": 3.149622206957857e-05,
      "loss": 0.9086,
      "step": 9260
    },
    {
      "epoch": 1.1191597247374139,
      "grad_norm": 18.498943328857422,
      "learning_rate": 3.147601923310033e-05,
      "loss": 0.675,
      "step": 9270
    },
    {
      "epoch": 1.1203670167813593,
      "grad_norm": 6.884809494018555,
      "learning_rate": 3.145581639662209e-05,
      "loss": 0.891,
      "step": 9280
    },
    {
      "epoch": 1.1215743088253047,
      "grad_norm": 11.512457847595215,
      "learning_rate": 3.143561356014385e-05,
      "loss": 1.2131,
      "step": 9290
    },
    {
      "epoch": 1.1227816008692502,
      "grad_norm": 21.098676681518555,
      "learning_rate": 3.1415410723665605e-05,
      "loss": 1.3034,
      "step": 9300
    },
    {
      "epoch": 1.1239888929131956,
      "grad_norm": 7.311788082122803,
      "learning_rate": 3.1395207887187364e-05,
      "loss": 0.8638,
      "step": 9310
    },
    {
      "epoch": 1.125196184957141,
      "grad_norm": 16.128341674804688,
      "learning_rate": 3.137500505070912e-05,
      "loss": 1.1427,
      "step": 9320
    },
    {
      "epoch": 1.1264034770010865,
      "grad_norm": 16.070186614990234,
      "learning_rate": 3.1354802214230874e-05,
      "loss": 0.6432,
      "step": 9330
    },
    {
      "epoch": 1.1276107690450319,
      "grad_norm": 12.750826835632324,
      "learning_rate": 3.133459937775264e-05,
      "loss": 0.6728,
      "step": 9340
    },
    {
      "epoch": 1.1288180610889773,
      "grad_norm": 8.498040199279785,
      "learning_rate": 3.13143965412744e-05,
      "loss": 0.6827,
      "step": 9350
    },
    {
      "epoch": 1.1300253531329227,
      "grad_norm": 17.661264419555664,
      "learning_rate": 3.1294193704796156e-05,
      "loss": 1.1123,
      "step": 9360
    },
    {
      "epoch": 1.1312326451768682,
      "grad_norm": 13.131952285766602,
      "learning_rate": 3.1273990868317915e-05,
      "loss": 1.2673,
      "step": 9370
    },
    {
      "epoch": 1.1324399372208136,
      "grad_norm": 11.955333709716797,
      "learning_rate": 3.125378803183967e-05,
      "loss": 0.7202,
      "step": 9380
    },
    {
      "epoch": 1.133647229264759,
      "grad_norm": 12.095769882202148,
      "learning_rate": 3.123358519536143e-05,
      "loss": 0.8862,
      "step": 9390
    },
    {
      "epoch": 1.1348545213087045,
      "grad_norm": 20.072246551513672,
      "learning_rate": 3.121338235888318e-05,
      "loss": 1.1785,
      "step": 9400
    },
    {
      "epoch": 1.1360618133526499,
      "grad_norm": 11.522872924804688,
      "learning_rate": 3.119317952240495e-05,
      "loss": 1.3878,
      "step": 9410
    },
    {
      "epoch": 1.1372691053965953,
      "grad_norm": 7.952047348022461,
      "learning_rate": 3.117297668592671e-05,
      "loss": 0.8223,
      "step": 9420
    },
    {
      "epoch": 1.1384763974405407,
      "grad_norm": 15.856884002685547,
      "learning_rate": 3.1152773849448465e-05,
      "loss": 1.2263,
      "step": 9430
    },
    {
      "epoch": 1.1396836894844862,
      "grad_norm": 18.0610294342041,
      "learning_rate": 3.1132571012970224e-05,
      "loss": 1.1473,
      "step": 9440
    },
    {
      "epoch": 1.1408909815284318,
      "grad_norm": 1.0427806377410889,
      "learning_rate": 3.111236817649198e-05,
      "loss": 0.3509,
      "step": 9450
    },
    {
      "epoch": 1.1420982735723773,
      "grad_norm": 5.240071773529053,
      "learning_rate": 3.109216534001374e-05,
      "loss": 0.9483,
      "step": 9460
    },
    {
      "epoch": 1.1433055656163227,
      "grad_norm": 13.137255668640137,
      "learning_rate": 3.107196250353549e-05,
      "loss": 1.1238,
      "step": 9470
    },
    {
      "epoch": 1.1445128576602681,
      "grad_norm": 14.501502990722656,
      "learning_rate": 3.105175966705726e-05,
      "loss": 0.9154,
      "step": 9480
    },
    {
      "epoch": 1.1457201497042135,
      "grad_norm": 13.993549346923828,
      "learning_rate": 3.1031556830579016e-05,
      "loss": 0.7101,
      "step": 9490
    },
    {
      "epoch": 1.146927441748159,
      "grad_norm": 11.564323425292969,
      "learning_rate": 3.1011353994100775e-05,
      "loss": 1.0481,
      "step": 9500
    },
    {
      "epoch": 1.1481347337921044,
      "grad_norm": 9.62912368774414,
      "learning_rate": 3.099115115762253e-05,
      "loss": 0.9866,
      "step": 9510
    },
    {
      "epoch": 1.1493420258360498,
      "grad_norm": 6.026259899139404,
      "learning_rate": 3.097094832114429e-05,
      "loss": 0.8073,
      "step": 9520
    },
    {
      "epoch": 1.1505493178799953,
      "grad_norm": 8.678258895874023,
      "learning_rate": 3.095074548466605e-05,
      "loss": 1.2897,
      "step": 9530
    },
    {
      "epoch": 1.1517566099239407,
      "grad_norm": 9.835529327392578,
      "learning_rate": 3.09305426481878e-05,
      "loss": 0.5258,
      "step": 9540
    },
    {
      "epoch": 1.1529639019678861,
      "grad_norm": 8.955848693847656,
      "learning_rate": 3.091033981170957e-05,
      "loss": 1.056,
      "step": 9550
    },
    {
      "epoch": 1.1541711940118315,
      "grad_norm": 38.22920227050781,
      "learning_rate": 3.0890136975231326e-05,
      "loss": 1.0793,
      "step": 9560
    },
    {
      "epoch": 1.155378486055777,
      "grad_norm": 8.031079292297363,
      "learning_rate": 3.0869934138753084e-05,
      "loss": 0.7368,
      "step": 9570
    },
    {
      "epoch": 1.1565857780997224,
      "grad_norm": 14.365914344787598,
      "learning_rate": 3.084973130227484e-05,
      "loss": 1.2086,
      "step": 9580
    },
    {
      "epoch": 1.1577930701436678,
      "grad_norm": 25.524805068969727,
      "learning_rate": 3.08295284657966e-05,
      "loss": 1.031,
      "step": 9590
    },
    {
      "epoch": 1.1590003621876133,
      "grad_norm": 7.761993885040283,
      "learning_rate": 3.080932562931836e-05,
      "loss": 0.9285,
      "step": 9600
    },
    {
      "epoch": 1.1602076542315587,
      "grad_norm": 3.058943271636963,
      "learning_rate": 3.078912279284011e-05,
      "loss": 0.8417,
      "step": 9610
    },
    {
      "epoch": 1.1614149462755041,
      "grad_norm": 10.032872200012207,
      "learning_rate": 3.0768919956361876e-05,
      "loss": 0.754,
      "step": 9620
    },
    {
      "epoch": 1.1626222383194496,
      "grad_norm": 19.871986389160156,
      "learning_rate": 3.0748717119883635e-05,
      "loss": 1.138,
      "step": 9630
    },
    {
      "epoch": 1.163829530363395,
      "grad_norm": 13.141674995422363,
      "learning_rate": 3.0728514283405393e-05,
      "loss": 0.7937,
      "step": 9640
    },
    {
      "epoch": 1.1650368224073404,
      "grad_norm": 5.974830150604248,
      "learning_rate": 3.070831144692715e-05,
      "loss": 1.1305,
      "step": 9650
    },
    {
      "epoch": 1.1662441144512858,
      "grad_norm": 4.2159423828125,
      "learning_rate": 3.0688108610448904e-05,
      "loss": 1.0835,
      "step": 9660
    },
    {
      "epoch": 1.1674514064952313,
      "grad_norm": 4.949521541595459,
      "learning_rate": 3.066790577397067e-05,
      "loss": 0.6693,
      "step": 9670
    },
    {
      "epoch": 1.1686586985391767,
      "grad_norm": 12.315423965454102,
      "learning_rate": 3.064770293749242e-05,
      "loss": 0.9722,
      "step": 9680
    },
    {
      "epoch": 1.1698659905831221,
      "grad_norm": 31.383773803710938,
      "learning_rate": 3.0627500101014186e-05,
      "loss": 1.1408,
      "step": 9690
    },
    {
      "epoch": 1.1710732826270676,
      "grad_norm": 15.61435604095459,
      "learning_rate": 3.0607297264535944e-05,
      "loss": 1.3245,
      "step": 9700
    },
    {
      "epoch": 1.172280574671013,
      "grad_norm": 15.831605911254883,
      "learning_rate": 3.05870944280577e-05,
      "loss": 1.6717,
      "step": 9710
    },
    {
      "epoch": 1.1734878667149584,
      "grad_norm": 14.382612228393555,
      "learning_rate": 3.056689159157946e-05,
      "loss": 1.1627,
      "step": 9720
    },
    {
      "epoch": 1.1746951587589038,
      "grad_norm": 13.541892051696777,
      "learning_rate": 3.054668875510121e-05,
      "loss": 0.9011,
      "step": 9730
    },
    {
      "epoch": 1.1759024508028493,
      "grad_norm": 10.550148010253906,
      "learning_rate": 3.052648591862298e-05,
      "loss": 0.6439,
      "step": 9740
    },
    {
      "epoch": 1.1771097428467947,
      "grad_norm": 10.471038818359375,
      "learning_rate": 3.0506283082144733e-05,
      "loss": 0.9614,
      "step": 9750
    },
    {
      "epoch": 1.1783170348907401,
      "grad_norm": 5.0482497215271,
      "learning_rate": 3.0486080245666492e-05,
      "loss": 1.0288,
      "step": 9760
    },
    {
      "epoch": 1.1795243269346856,
      "grad_norm": 21.180553436279297,
      "learning_rate": 3.0465877409188254e-05,
      "loss": 1.276,
      "step": 9770
    },
    {
      "epoch": 1.180731618978631,
      "grad_norm": 27.964380264282227,
      "learning_rate": 3.044567457271001e-05,
      "loss": 1.0247,
      "step": 9780
    },
    {
      "epoch": 1.1819389110225764,
      "grad_norm": 5.817234992980957,
      "learning_rate": 3.042547173623177e-05,
      "loss": 1.0799,
      "step": 9790
    },
    {
      "epoch": 1.1831462030665219,
      "grad_norm": 17.114952087402344,
      "learning_rate": 3.0405268899753526e-05,
      "loss": 0.9836,
      "step": 9800
    },
    {
      "epoch": 1.1843534951104673,
      "grad_norm": 7.679940223693848,
      "learning_rate": 3.0385066063275287e-05,
      "loss": 0.9844,
      "step": 9810
    },
    {
      "epoch": 1.1855607871544127,
      "grad_norm": 40.663089752197266,
      "learning_rate": 3.0364863226797043e-05,
      "loss": 0.877,
      "step": 9820
    },
    {
      "epoch": 1.1867680791983581,
      "grad_norm": 10.978057861328125,
      "learning_rate": 3.03446603903188e-05,
      "loss": 0.9888,
      "step": 9830
    },
    {
      "epoch": 1.1879753712423036,
      "grad_norm": 9.658713340759277,
      "learning_rate": 3.0324457553840563e-05,
      "loss": 0.8981,
      "step": 9840
    },
    {
      "epoch": 1.189182663286249,
      "grad_norm": 11.62814998626709,
      "learning_rate": 3.0304254717362318e-05,
      "loss": 0.8642,
      "step": 9850
    },
    {
      "epoch": 1.1903899553301944,
      "grad_norm": 7.151744365692139,
      "learning_rate": 3.028405188088408e-05,
      "loss": 0.841,
      "step": 9860
    },
    {
      "epoch": 1.1915972473741399,
      "grad_norm": 15.86739444732666,
      "learning_rate": 3.0263849044405835e-05,
      "loss": 0.8125,
      "step": 9870
    },
    {
      "epoch": 1.1928045394180853,
      "grad_norm": 16.08427619934082,
      "learning_rate": 3.0243646207927593e-05,
      "loss": 0.4835,
      "step": 9880
    },
    {
      "epoch": 1.1940118314620307,
      "grad_norm": 1.8394172191619873,
      "learning_rate": 3.0223443371449352e-05,
      "loss": 1.248,
      "step": 9890
    },
    {
      "epoch": 1.1952191235059761,
      "grad_norm": 35.39223861694336,
      "learning_rate": 3.020324053497111e-05,
      "loss": 1.0878,
      "step": 9900
    },
    {
      "epoch": 1.1964264155499216,
      "grad_norm": 6.731860160827637,
      "learning_rate": 3.0183037698492872e-05,
      "loss": 1.2265,
      "step": 9910
    },
    {
      "epoch": 1.197633707593867,
      "grad_norm": 33.284603118896484,
      "learning_rate": 3.0162834862014627e-05,
      "loss": 0.8111,
      "step": 9920
    },
    {
      "epoch": 1.1988409996378124,
      "grad_norm": 26.193042755126953,
      "learning_rate": 3.014263202553639e-05,
      "loss": 1.111,
      "step": 9930
    },
    {
      "epoch": 1.2000482916817579,
      "grad_norm": 29.843185424804688,
      "learning_rate": 3.0122429189058144e-05,
      "loss": 1.0254,
      "step": 9940
    },
    {
      "epoch": 1.2012555837257033,
      "grad_norm": 4.945756912231445,
      "learning_rate": 3.0102226352579903e-05,
      "loss": 0.8038,
      "step": 9950
    },
    {
      "epoch": 1.2024628757696487,
      "grad_norm": 39.03877258300781,
      "learning_rate": 3.0082023516101658e-05,
      "loss": 0.9674,
      "step": 9960
    },
    {
      "epoch": 1.2036701678135941,
      "grad_norm": 3.7611145973205566,
      "learning_rate": 3.006182067962342e-05,
      "loss": 1.1281,
      "step": 9970
    },
    {
      "epoch": 1.2048774598575396,
      "grad_norm": 20.437875747680664,
      "learning_rate": 3.004161784314518e-05,
      "loss": 1.1985,
      "step": 9980
    },
    {
      "epoch": 1.206084751901485,
      "grad_norm": 3.584418773651123,
      "learning_rate": 3.0021415006666937e-05,
      "loss": 1.3711,
      "step": 9990
    },
    {
      "epoch": 1.2072920439454304,
      "grad_norm": 27.66411781311035,
      "learning_rate": 3.00012121701887e-05,
      "loss": 1.1042,
      "step": 10000
    },
    {
      "epoch": 1.2084993359893759,
      "grad_norm": 25.46059226989746,
      "learning_rate": 2.9981009333710454e-05,
      "loss": 1.1491,
      "step": 10010
    },
    {
      "epoch": 1.2097066280333213,
      "grad_norm": 14.975367546081543,
      "learning_rate": 2.9960806497232212e-05,
      "loss": 1.1743,
      "step": 10020
    },
    {
      "epoch": 1.2109139200772667,
      "grad_norm": 16.737258911132812,
      "learning_rate": 2.9940603660753967e-05,
      "loss": 1.018,
      "step": 10030
    },
    {
      "epoch": 1.2121212121212122,
      "grad_norm": 22.547584533691406,
      "learning_rate": 2.992040082427573e-05,
      "loss": 1.0336,
      "step": 10040
    },
    {
      "epoch": 1.2133285041651576,
      "grad_norm": 1.2200275659561157,
      "learning_rate": 2.990019798779749e-05,
      "loss": 0.7518,
      "step": 10050
    },
    {
      "epoch": 1.214535796209103,
      "grad_norm": 9.9761381149292,
      "learning_rate": 2.9879995151319246e-05,
      "loss": 0.7286,
      "step": 10060
    },
    {
      "epoch": 1.2157430882530484,
      "grad_norm": 20.326250076293945,
      "learning_rate": 2.9859792314841008e-05,
      "loss": 0.7857,
      "step": 10070
    },
    {
      "epoch": 1.2169503802969939,
      "grad_norm": 25.362218856811523,
      "learning_rate": 2.9839589478362763e-05,
      "loss": 1.1071,
      "step": 10080
    },
    {
      "epoch": 1.2181576723409393,
      "grad_norm": 26.50567626953125,
      "learning_rate": 2.981938664188452e-05,
      "loss": 1.3012,
      "step": 10090
    },
    {
      "epoch": 1.2193649643848847,
      "grad_norm": 10.431090354919434,
      "learning_rate": 2.9799183805406276e-05,
      "loss": 1.2481,
      "step": 10100
    },
    {
      "epoch": 1.2205722564288302,
      "grad_norm": 14.325708389282227,
      "learning_rate": 2.977898096892804e-05,
      "loss": 0.9282,
      "step": 10110
    },
    {
      "epoch": 1.2217795484727756,
      "grad_norm": 14.173785209655762,
      "learning_rate": 2.97587781324498e-05,
      "loss": 1.3626,
      "step": 10120
    },
    {
      "epoch": 1.222986840516721,
      "grad_norm": 15.503037452697754,
      "learning_rate": 2.9738575295971555e-05,
      "loss": 0.6952,
      "step": 10130
    },
    {
      "epoch": 1.2241941325606664,
      "grad_norm": 22.807451248168945,
      "learning_rate": 2.9718372459493317e-05,
      "loss": 0.7351,
      "step": 10140
    },
    {
      "epoch": 1.2254014246046119,
      "grad_norm": 18.455652236938477,
      "learning_rate": 2.9698169623015072e-05,
      "loss": 0.6723,
      "step": 10150
    },
    {
      "epoch": 1.2266087166485573,
      "grad_norm": 9.5655517578125,
      "learning_rate": 2.967796678653683e-05,
      "loss": 1.0973,
      "step": 10160
    },
    {
      "epoch": 1.2278160086925027,
      "grad_norm": 12.969175338745117,
      "learning_rate": 2.9657763950058586e-05,
      "loss": 1.0255,
      "step": 10170
    },
    {
      "epoch": 1.2290233007364482,
      "grad_norm": 16.98894691467285,
      "learning_rate": 2.9637561113580348e-05,
      "loss": 0.874,
      "step": 10180
    },
    {
      "epoch": 1.2302305927803936,
      "grad_norm": 41.59648132324219,
      "learning_rate": 2.961735827710211e-05,
      "loss": 0.8871,
      "step": 10190
    },
    {
      "epoch": 1.231437884824339,
      "grad_norm": 8.7612886428833,
      "learning_rate": 2.9597155440623865e-05,
      "loss": 0.9527,
      "step": 10200
    },
    {
      "epoch": 1.2326451768682845,
      "grad_norm": 22.448444366455078,
      "learning_rate": 2.9576952604145623e-05,
      "loss": 1.1906,
      "step": 10210
    },
    {
      "epoch": 1.2338524689122299,
      "grad_norm": 1.846740961074829,
      "learning_rate": 2.955674976766738e-05,
      "loss": 1.1659,
      "step": 10220
    },
    {
      "epoch": 1.2350597609561753,
      "grad_norm": 27.685565948486328,
      "learning_rate": 2.953654693118914e-05,
      "loss": 1.3035,
      "step": 10230
    },
    {
      "epoch": 1.2362670530001207,
      "grad_norm": 23.27533531188965,
      "learning_rate": 2.9516344094710902e-05,
      "loss": 1.0055,
      "step": 10240
    },
    {
      "epoch": 1.2374743450440662,
      "grad_norm": 23.520645141601562,
      "learning_rate": 2.9496141258232657e-05,
      "loss": 1.2655,
      "step": 10250
    },
    {
      "epoch": 1.2386816370880116,
      "grad_norm": 16.89566421508789,
      "learning_rate": 2.947593842175442e-05,
      "loss": 1.1201,
      "step": 10260
    },
    {
      "epoch": 1.239888929131957,
      "grad_norm": 13.912094116210938,
      "learning_rate": 2.9455735585276174e-05,
      "loss": 0.6636,
      "step": 10270
    },
    {
      "epoch": 1.2410962211759025,
      "grad_norm": 10.915513038635254,
      "learning_rate": 2.9435532748797932e-05,
      "loss": 0.8483,
      "step": 10280
    },
    {
      "epoch": 1.2423035132198479,
      "grad_norm": 17.651744842529297,
      "learning_rate": 2.9415329912319688e-05,
      "loss": 0.4717,
      "step": 10290
    },
    {
      "epoch": 1.2435108052637933,
      "grad_norm": 24.30554962158203,
      "learning_rate": 2.939512707584145e-05,
      "loss": 1.1423,
      "step": 10300
    },
    {
      "epoch": 1.2447180973077387,
      "grad_norm": 13.621794700622559,
      "learning_rate": 2.937492423936321e-05,
      "loss": 1.2908,
      "step": 10310
    },
    {
      "epoch": 1.2459253893516842,
      "grad_norm": 13.355534553527832,
      "learning_rate": 2.9354721402884966e-05,
      "loss": 1.1439,
      "step": 10320
    },
    {
      "epoch": 1.2471326813956296,
      "grad_norm": 12.777816772460938,
      "learning_rate": 2.9334518566406728e-05,
      "loss": 0.7045,
      "step": 10330
    },
    {
      "epoch": 1.248339973439575,
      "grad_norm": 13.350201606750488,
      "learning_rate": 2.9314315729928483e-05,
      "loss": 1.2229,
      "step": 10340
    },
    {
      "epoch": 1.2495472654835205,
      "grad_norm": 17.5137882232666,
      "learning_rate": 2.9294112893450242e-05,
      "loss": 0.959,
      "step": 10350
    },
    {
      "epoch": 1.250754557527466,
      "grad_norm": 13.160019874572754,
      "learning_rate": 2.9273910056971997e-05,
      "loss": 1.6109,
      "step": 10360
    },
    {
      "epoch": 1.2519618495714113,
      "grad_norm": 12.261622428894043,
      "learning_rate": 2.925370722049376e-05,
      "loss": 1.0913,
      "step": 10370
    },
    {
      "epoch": 1.2531691416153568,
      "grad_norm": 13.654335975646973,
      "learning_rate": 2.923350438401552e-05,
      "loss": 1.015,
      "step": 10380
    },
    {
      "epoch": 1.2543764336593022,
      "grad_norm": 7.92751932144165,
      "learning_rate": 2.9213301547537276e-05,
      "loss": 1.1017,
      "step": 10390
    },
    {
      "epoch": 1.2555837257032476,
      "grad_norm": 30.521326065063477,
      "learning_rate": 2.9193098711059038e-05,
      "loss": 1.2624,
      "step": 10400
    },
    {
      "epoch": 1.256791017747193,
      "grad_norm": 4.326565742492676,
      "learning_rate": 2.9172895874580793e-05,
      "loss": 0.5268,
      "step": 10410
    },
    {
      "epoch": 1.2579983097911385,
      "grad_norm": 2.9163568019866943,
      "learning_rate": 2.915269303810255e-05,
      "loss": 1.0636,
      "step": 10420
    },
    {
      "epoch": 1.259205601835084,
      "grad_norm": 18.74502182006836,
      "learning_rate": 2.9132490201624306e-05,
      "loss": 1.1553,
      "step": 10430
    },
    {
      "epoch": 1.2604128938790293,
      "grad_norm": 13.326951026916504,
      "learning_rate": 2.9112287365146068e-05,
      "loss": 0.9187,
      "step": 10440
    },
    {
      "epoch": 1.2616201859229748,
      "grad_norm": 11.392986297607422,
      "learning_rate": 2.909208452866783e-05,
      "loss": 1.0373,
      "step": 10450
    },
    {
      "epoch": 1.2628274779669202,
      "grad_norm": 7.1728010177612305,
      "learning_rate": 2.9071881692189585e-05,
      "loss": 1.0786,
      "step": 10460
    },
    {
      "epoch": 1.2640347700108656,
      "grad_norm": 13.722111701965332,
      "learning_rate": 2.9051678855711343e-05,
      "loss": 0.8474,
      "step": 10470
    },
    {
      "epoch": 1.265242062054811,
      "grad_norm": 14.521607398986816,
      "learning_rate": 2.9031476019233102e-05,
      "loss": 0.9703,
      "step": 10480
    },
    {
      "epoch": 1.2664493540987565,
      "grad_norm": 3.436641216278076,
      "learning_rate": 2.901127318275486e-05,
      "loss": 0.8144,
      "step": 10490
    },
    {
      "epoch": 1.267656646142702,
      "grad_norm": 15.052314758300781,
      "learning_rate": 2.8991070346276615e-05,
      "loss": 0.5739,
      "step": 10500
    },
    {
      "epoch": 1.2688639381866473,
      "grad_norm": 12.402728080749512,
      "learning_rate": 2.8970867509798377e-05,
      "loss": 1.1793,
      "step": 10510
    },
    {
      "epoch": 1.2700712302305928,
      "grad_norm": 11.205432891845703,
      "learning_rate": 2.895066467332014e-05,
      "loss": 0.9383,
      "step": 10520
    },
    {
      "epoch": 1.2712785222745382,
      "grad_norm": 23.724849700927734,
      "learning_rate": 2.8930461836841894e-05,
      "loss": 1.0052,
      "step": 10530
    },
    {
      "epoch": 1.2724858143184836,
      "grad_norm": 12.953574180603027,
      "learning_rate": 2.8910259000363653e-05,
      "loss": 1.018,
      "step": 10540
    },
    {
      "epoch": 1.273693106362429,
      "grad_norm": 18.084707260131836,
      "learning_rate": 2.8890056163885408e-05,
      "loss": 0.843,
      "step": 10550
    },
    {
      "epoch": 1.2749003984063745,
      "grad_norm": 11.003893852233887,
      "learning_rate": 2.886985332740717e-05,
      "loss": 1.0501,
      "step": 10560
    },
    {
      "epoch": 1.27610769045032,
      "grad_norm": 21.789888381958008,
      "learning_rate": 2.8849650490928925e-05,
      "loss": 1.4615,
      "step": 10570
    },
    {
      "epoch": 1.2773149824942653,
      "grad_norm": 5.081242084503174,
      "learning_rate": 2.8829447654450687e-05,
      "loss": 1.0769,
      "step": 10580
    },
    {
      "epoch": 1.2785222745382108,
      "grad_norm": 8.569925308227539,
      "learning_rate": 2.880924481797245e-05,
      "loss": 0.7638,
      "step": 10590
    },
    {
      "epoch": 1.2797295665821562,
      "grad_norm": 5.054373741149902,
      "learning_rate": 2.8789041981494204e-05,
      "loss": 0.8337,
      "step": 10600
    },
    {
      "epoch": 1.2809368586261016,
      "grad_norm": 11.793610572814941,
      "learning_rate": 2.8768839145015962e-05,
      "loss": 1.1101,
      "step": 10610
    },
    {
      "epoch": 1.282144150670047,
      "grad_norm": 11.978378295898438,
      "learning_rate": 2.8748636308537717e-05,
      "loss": 1.3301,
      "step": 10620
    },
    {
      "epoch": 1.2833514427139925,
      "grad_norm": 11.044843673706055,
      "learning_rate": 2.872843347205948e-05,
      "loss": 1.3122,
      "step": 10630
    },
    {
      "epoch": 1.284558734757938,
      "grad_norm": 15.262765884399414,
      "learning_rate": 2.8708230635581234e-05,
      "loss": 0.9442,
      "step": 10640
    },
    {
      "epoch": 1.2857660268018833,
      "grad_norm": 2.147256374359131,
      "learning_rate": 2.8688027799102996e-05,
      "loss": 0.9631,
      "step": 10650
    },
    {
      "epoch": 1.2869733188458288,
      "grad_norm": 22.726469039916992,
      "learning_rate": 2.8667824962624758e-05,
      "loss": 1.1153,
      "step": 10660
    },
    {
      "epoch": 1.2881806108897742,
      "grad_norm": 16.828105926513672,
      "learning_rate": 2.8647622126146513e-05,
      "loss": 1.0666,
      "step": 10670
    },
    {
      "epoch": 1.2893879029337196,
      "grad_norm": 9.28690242767334,
      "learning_rate": 2.862741928966827e-05,
      "loss": 0.9749,
      "step": 10680
    },
    {
      "epoch": 1.290595194977665,
      "grad_norm": 11.045960426330566,
      "learning_rate": 2.8607216453190027e-05,
      "loss": 0.9411,
      "step": 10690
    },
    {
      "epoch": 1.2918024870216105,
      "grad_norm": 7.554493427276611,
      "learning_rate": 2.858701361671179e-05,
      "loss": 0.6866,
      "step": 10700
    },
    {
      "epoch": 1.293009779065556,
      "grad_norm": 11.409550666809082,
      "learning_rate": 2.8566810780233543e-05,
      "loss": 0.8563,
      "step": 10710
    },
    {
      "epoch": 1.2942170711095013,
      "grad_norm": 7.495802879333496,
      "learning_rate": 2.8546607943755305e-05,
      "loss": 0.6783,
      "step": 10720
    },
    {
      "epoch": 1.2954243631534468,
      "grad_norm": 11.021071434020996,
      "learning_rate": 2.8526405107277067e-05,
      "loss": 1.4293,
      "step": 10730
    },
    {
      "epoch": 1.2966316551973922,
      "grad_norm": 12.000422477722168,
      "learning_rate": 2.8506202270798822e-05,
      "loss": 1.0092,
      "step": 10740
    },
    {
      "epoch": 1.2978389472413376,
      "grad_norm": 7.531917572021484,
      "learning_rate": 2.848599943432058e-05,
      "loss": 1.0733,
      "step": 10750
    },
    {
      "epoch": 1.299046239285283,
      "grad_norm": 65.93876647949219,
      "learning_rate": 2.8465796597842336e-05,
      "loss": 0.9415,
      "step": 10760
    },
    {
      "epoch": 1.3002535313292285,
      "grad_norm": 15.656793594360352,
      "learning_rate": 2.8445593761364098e-05,
      "loss": 0.8072,
      "step": 10770
    },
    {
      "epoch": 1.301460823373174,
      "grad_norm": 2.901967763900757,
      "learning_rate": 2.8425390924885853e-05,
      "loss": 0.8974,
      "step": 10780
    },
    {
      "epoch": 1.3026681154171194,
      "grad_norm": 15.051213264465332,
      "learning_rate": 2.8405188088407615e-05,
      "loss": 1.0984,
      "step": 10790
    },
    {
      "epoch": 1.3038754074610648,
      "grad_norm": 22.035776138305664,
      "learning_rate": 2.8384985251929373e-05,
      "loss": 1.0462,
      "step": 10800
    },
    {
      "epoch": 1.3050826995050102,
      "grad_norm": 7.240533828735352,
      "learning_rate": 2.836478241545113e-05,
      "loss": 0.8722,
      "step": 10810
    },
    {
      "epoch": 1.3062899915489556,
      "grad_norm": 3.4082961082458496,
      "learning_rate": 2.834457957897289e-05,
      "loss": 0.5202,
      "step": 10820
    },
    {
      "epoch": 1.307497283592901,
      "grad_norm": 11.766298294067383,
      "learning_rate": 2.8324376742494645e-05,
      "loss": 1.1169,
      "step": 10830
    },
    {
      "epoch": 1.3087045756368465,
      "grad_norm": 29.66962432861328,
      "learning_rate": 2.8304173906016407e-05,
      "loss": 1.2543,
      "step": 10840
    },
    {
      "epoch": 1.309911867680792,
      "grad_norm": 13.616974830627441,
      "learning_rate": 2.8283971069538162e-05,
      "loss": 1.0641,
      "step": 10850
    },
    {
      "epoch": 1.3111191597247374,
      "grad_norm": 22.127925872802734,
      "learning_rate": 2.8263768233059924e-05,
      "loss": 1.129,
      "step": 10860
    },
    {
      "epoch": 1.3123264517686828,
      "grad_norm": 29.38013458251953,
      "learning_rate": 2.8243565396581682e-05,
      "loss": 0.9767,
      "step": 10870
    },
    {
      "epoch": 1.3135337438126282,
      "grad_norm": 18.552650451660156,
      "learning_rate": 2.8223362560103438e-05,
      "loss": 1.38,
      "step": 10880
    },
    {
      "epoch": 1.3147410358565736,
      "grad_norm": 21.162704467773438,
      "learning_rate": 2.82031597236252e-05,
      "loss": 0.8251,
      "step": 10890
    },
    {
      "epoch": 1.315948327900519,
      "grad_norm": 6.115911960601807,
      "learning_rate": 2.8182956887146954e-05,
      "loss": 0.7795,
      "step": 10900
    },
    {
      "epoch": 1.3171556199444645,
      "grad_norm": 15.618997573852539,
      "learning_rate": 2.8162754050668716e-05,
      "loss": 0.8341,
      "step": 10910
    },
    {
      "epoch": 1.31836291198841,
      "grad_norm": 20.966411590576172,
      "learning_rate": 2.814255121419047e-05,
      "loss": 1.3459,
      "step": 10920
    },
    {
      "epoch": 1.3195702040323554,
      "grad_norm": 33.6614875793457,
      "learning_rate": 2.8122348377712233e-05,
      "loss": 0.9261,
      "step": 10930
    },
    {
      "epoch": 1.3207774960763008,
      "grad_norm": 30.76746368408203,
      "learning_rate": 2.8102145541233992e-05,
      "loss": 0.8539,
      "step": 10940
    },
    {
      "epoch": 1.3219847881202462,
      "grad_norm": 4.396899700164795,
      "learning_rate": 2.8081942704755747e-05,
      "loss": 0.7214,
      "step": 10950
    },
    {
      "epoch": 1.3231920801641917,
      "grad_norm": 7.705544948577881,
      "learning_rate": 2.806173986827751e-05,
      "loss": 1.4987,
      "step": 10960
    },
    {
      "epoch": 1.324399372208137,
      "grad_norm": 24.6973819732666,
      "learning_rate": 2.8041537031799264e-05,
      "loss": 1.1377,
      "step": 10970
    },
    {
      "epoch": 1.3256066642520825,
      "grad_norm": 13.796104431152344,
      "learning_rate": 2.8021334195321026e-05,
      "loss": 1.0125,
      "step": 10980
    },
    {
      "epoch": 1.326813956296028,
      "grad_norm": 8.157767295837402,
      "learning_rate": 2.800113135884278e-05,
      "loss": 1.3229,
      "step": 10990
    },
    {
      "epoch": 1.3280212483399734,
      "grad_norm": 23.772903442382812,
      "learning_rate": 2.7980928522364543e-05,
      "loss": 0.8035,
      "step": 11000
    },
    {
      "epoch": 1.3292285403839188,
      "grad_norm": 14.393952369689941,
      "learning_rate": 2.79607256858863e-05,
      "loss": 0.9438,
      "step": 11010
    },
    {
      "epoch": 1.3304358324278642,
      "grad_norm": 6.1330389976501465,
      "learning_rate": 2.7940522849408056e-05,
      "loss": 1.2076,
      "step": 11020
    },
    {
      "epoch": 1.3316431244718097,
      "grad_norm": 17.571073532104492,
      "learning_rate": 2.7920320012929818e-05,
      "loss": 1.0862,
      "step": 11030
    },
    {
      "epoch": 1.332850416515755,
      "grad_norm": 2.56974458694458,
      "learning_rate": 2.7900117176451573e-05,
      "loss": 1.0703,
      "step": 11040
    },
    {
      "epoch": 1.3340577085597005,
      "grad_norm": 20.080556869506836,
      "learning_rate": 2.7879914339973335e-05,
      "loss": 1.2441,
      "step": 11050
    },
    {
      "epoch": 1.335265000603646,
      "grad_norm": 10.9102783203125,
      "learning_rate": 2.785971150349509e-05,
      "loss": 0.9696,
      "step": 11060
    },
    {
      "epoch": 1.3364722926475914,
      "grad_norm": 10.571517944335938,
      "learning_rate": 2.7839508667016852e-05,
      "loss": 0.9065,
      "step": 11070
    },
    {
      "epoch": 1.3376795846915368,
      "grad_norm": 17.771595001220703,
      "learning_rate": 2.781930583053861e-05,
      "loss": 0.9896,
      "step": 11080
    },
    {
      "epoch": 1.3388868767354822,
      "grad_norm": 10.662449836730957,
      "learning_rate": 2.7799102994060366e-05,
      "loss": 0.8151,
      "step": 11090
    },
    {
      "epoch": 1.3400941687794277,
      "grad_norm": 6.23698091506958,
      "learning_rate": 2.7778900157582127e-05,
      "loss": 0.7415,
      "step": 11100
    },
    {
      "epoch": 1.341301460823373,
      "grad_norm": 9.069952964782715,
      "learning_rate": 2.7758697321103882e-05,
      "loss": 0.5986,
      "step": 11110
    },
    {
      "epoch": 1.3425087528673185,
      "grad_norm": 18.8253173828125,
      "learning_rate": 2.7738494484625644e-05,
      "loss": 0.9475,
      "step": 11120
    },
    {
      "epoch": 1.343716044911264,
      "grad_norm": 6.353711128234863,
      "learning_rate": 2.77182916481474e-05,
      "loss": 1.0523,
      "step": 11130
    },
    {
      "epoch": 1.3449233369552096,
      "grad_norm": 11.33154010772705,
      "learning_rate": 2.7698088811669158e-05,
      "loss": 0.8535,
      "step": 11140
    },
    {
      "epoch": 1.346130628999155,
      "grad_norm": 4.385570526123047,
      "learning_rate": 2.767788597519092e-05,
      "loss": 1.2007,
      "step": 11150
    },
    {
      "epoch": 1.3473379210431005,
      "grad_norm": 0.9558895826339722,
      "learning_rate": 2.7657683138712675e-05,
      "loss": 1.2063,
      "step": 11160
    },
    {
      "epoch": 1.348545213087046,
      "grad_norm": 4.337576866149902,
      "learning_rate": 2.7637480302234437e-05,
      "loss": 0.6449,
      "step": 11170
    },
    {
      "epoch": 1.3497525051309913,
      "grad_norm": 9.384725570678711,
      "learning_rate": 2.7617277465756192e-05,
      "loss": 1.4582,
      "step": 11180
    },
    {
      "epoch": 1.3509597971749367,
      "grad_norm": 22.009775161743164,
      "learning_rate": 2.7597074629277954e-05,
      "loss": 1.0388,
      "step": 11190
    },
    {
      "epoch": 1.3521670892188822,
      "grad_norm": 33.61045837402344,
      "learning_rate": 2.757687179279971e-05,
      "loss": 1.1949,
      "step": 11200
    },
    {
      "epoch": 1.3533743812628276,
      "grad_norm": 14.072471618652344,
      "learning_rate": 2.7556668956321467e-05,
      "loss": 0.9791,
      "step": 11210
    },
    {
      "epoch": 1.354581673306773,
      "grad_norm": 9.001229286193848,
      "learning_rate": 2.753646611984323e-05,
      "loss": 0.9743,
      "step": 11220
    },
    {
      "epoch": 1.3557889653507185,
      "grad_norm": 25.860034942626953,
      "learning_rate": 2.7516263283364984e-05,
      "loss": 0.8807,
      "step": 11230
    },
    {
      "epoch": 1.356996257394664,
      "grad_norm": 20.0762882232666,
      "learning_rate": 2.7496060446886746e-05,
      "loss": 1.1512,
      "step": 11240
    },
    {
      "epoch": 1.3582035494386093,
      "grad_norm": 9.305818557739258,
      "learning_rate": 2.74758576104085e-05,
      "loss": 1.1957,
      "step": 11250
    },
    {
      "epoch": 1.3594108414825548,
      "grad_norm": 5.922593116760254,
      "learning_rate": 2.7455654773930263e-05,
      "loss": 0.9248,
      "step": 11260
    },
    {
      "epoch": 1.3606181335265002,
      "grad_norm": 23.67429542541504,
      "learning_rate": 2.7435451937452018e-05,
      "loss": 1.077,
      "step": 11270
    },
    {
      "epoch": 1.3618254255704456,
      "grad_norm": 6.9376325607299805,
      "learning_rate": 2.7415249100973777e-05,
      "loss": 1.1672,
      "step": 11280
    },
    {
      "epoch": 1.363032717614391,
      "grad_norm": 11.399659156799316,
      "learning_rate": 2.739504626449554e-05,
      "loss": 1.1652,
      "step": 11290
    },
    {
      "epoch": 1.3642400096583365,
      "grad_norm": 18.323150634765625,
      "learning_rate": 2.7374843428017293e-05,
      "loss": 0.5597,
      "step": 11300
    },
    {
      "epoch": 1.365447301702282,
      "grad_norm": 10.625784873962402,
      "learning_rate": 2.7354640591539055e-05,
      "loss": 1.0147,
      "step": 11310
    },
    {
      "epoch": 1.3666545937462273,
      "grad_norm": 1.4498348236083984,
      "learning_rate": 2.733443775506081e-05,
      "loss": 0.6642,
      "step": 11320
    },
    {
      "epoch": 1.3678618857901728,
      "grad_norm": 11.763860702514648,
      "learning_rate": 2.7314234918582572e-05,
      "loss": 0.8547,
      "step": 11330
    },
    {
      "epoch": 1.3690691778341182,
      "grad_norm": 1.085267186164856,
      "learning_rate": 2.7294032082104327e-05,
      "loss": 1.3915,
      "step": 11340
    },
    {
      "epoch": 1.3702764698780636,
      "grad_norm": 10.557280540466309,
      "learning_rate": 2.7273829245626086e-05,
      "loss": 1.3002,
      "step": 11350
    },
    {
      "epoch": 1.371483761922009,
      "grad_norm": 1.6762815713882446,
      "learning_rate": 2.7253626409147848e-05,
      "loss": 1.085,
      "step": 11360
    },
    {
      "epoch": 1.3726910539659545,
      "grad_norm": 3.2384989261627197,
      "learning_rate": 2.7233423572669603e-05,
      "loss": 0.9218,
      "step": 11370
    },
    {
      "epoch": 1.3738983460099,
      "grad_norm": 19.0007381439209,
      "learning_rate": 2.7213220736191365e-05,
      "loss": 0.9884,
      "step": 11380
    },
    {
      "epoch": 1.3751056380538453,
      "grad_norm": 8.018393516540527,
      "learning_rate": 2.719301789971312e-05,
      "loss": 0.5292,
      "step": 11390
    },
    {
      "epoch": 1.3763129300977908,
      "grad_norm": 6.342172145843506,
      "learning_rate": 2.7172815063234878e-05,
      "loss": 1.4095,
      "step": 11400
    },
    {
      "epoch": 1.3775202221417362,
      "grad_norm": 36.95235824584961,
      "learning_rate": 2.7152612226756637e-05,
      "loss": 1.3856,
      "step": 11410
    },
    {
      "epoch": 1.3787275141856816,
      "grad_norm": 17.66057014465332,
      "learning_rate": 2.7132409390278395e-05,
      "loss": 1.2038,
      "step": 11420
    },
    {
      "epoch": 1.379934806229627,
      "grad_norm": 7.10762357711792,
      "learning_rate": 2.7112206553800157e-05,
      "loss": 0.6612,
      "step": 11430
    },
    {
      "epoch": 1.3811420982735725,
      "grad_norm": 11.674662590026855,
      "learning_rate": 2.7092003717321912e-05,
      "loss": 0.7442,
      "step": 11440
    },
    {
      "epoch": 1.382349390317518,
      "grad_norm": 5.312351703643799,
      "learning_rate": 2.7071800880843674e-05,
      "loss": 0.7786,
      "step": 11450
    },
    {
      "epoch": 1.3835566823614633,
      "grad_norm": 15.94638729095459,
      "learning_rate": 2.705159804436543e-05,
      "loss": 1.0611,
      "step": 11460
    },
    {
      "epoch": 1.3847639744054088,
      "grad_norm": 9.794593811035156,
      "learning_rate": 2.7031395207887188e-05,
      "loss": 1.4015,
      "step": 11470
    },
    {
      "epoch": 1.3859712664493542,
      "grad_norm": 14.49980354309082,
      "learning_rate": 2.701119237140895e-05,
      "loss": 1.0429,
      "step": 11480
    },
    {
      "epoch": 1.3871785584932996,
      "grad_norm": 12.538017272949219,
      "learning_rate": 2.6990989534930705e-05,
      "loss": 1.2268,
      "step": 11490
    },
    {
      "epoch": 1.388385850537245,
      "grad_norm": 13.047897338867188,
      "learning_rate": 2.6970786698452466e-05,
      "loss": 1.2528,
      "step": 11500
    },
    {
      "epoch": 1.3895931425811905,
      "grad_norm": 19.14700698852539,
      "learning_rate": 2.695058386197422e-05,
      "loss": 1.3829,
      "step": 11510
    },
    {
      "epoch": 1.390800434625136,
      "grad_norm": 30.38543128967285,
      "learning_rate": 2.6930381025495983e-05,
      "loss": 1.0051,
      "step": 11520
    },
    {
      "epoch": 1.3920077266690813,
      "grad_norm": 9.702961921691895,
      "learning_rate": 2.691017818901774e-05,
      "loss": 1.0976,
      "step": 11530
    },
    {
      "epoch": 1.3932150187130268,
      "grad_norm": 3.4884934425354004,
      "learning_rate": 2.6889975352539497e-05,
      "loss": 0.9361,
      "step": 11540
    },
    {
      "epoch": 1.3944223107569722,
      "grad_norm": 6.438786506652832,
      "learning_rate": 2.686977251606126e-05,
      "loss": 0.9566,
      "step": 11550
    },
    {
      "epoch": 1.3956296028009176,
      "grad_norm": 1.088739275932312,
      "learning_rate": 2.6849569679583014e-05,
      "loss": 0.7983,
      "step": 11560
    },
    {
      "epoch": 1.396836894844863,
      "grad_norm": 22.716075897216797,
      "learning_rate": 2.6829366843104776e-05,
      "loss": 0.8871,
      "step": 11570
    },
    {
      "epoch": 1.3980441868888085,
      "grad_norm": 16.187776565551758,
      "learning_rate": 2.680916400662653e-05,
      "loss": 0.8316,
      "step": 11580
    },
    {
      "epoch": 1.399251478932754,
      "grad_norm": 26.6857852935791,
      "learning_rate": 2.6788961170148293e-05,
      "loss": 1.2505,
      "step": 11590
    },
    {
      "epoch": 1.4004587709766994,
      "grad_norm": 9.63068675994873,
      "learning_rate": 2.6768758333670048e-05,
      "loss": 1.3464,
      "step": 11600
    },
    {
      "epoch": 1.4016660630206448,
      "grad_norm": 8.388972282409668,
      "learning_rate": 2.6748555497191806e-05,
      "loss": 0.9363,
      "step": 11610
    },
    {
      "epoch": 1.4028733550645902,
      "grad_norm": 16.617469787597656,
      "learning_rate": 2.6728352660713568e-05,
      "loss": 0.8929,
      "step": 11620
    },
    {
      "epoch": 1.4040806471085356,
      "grad_norm": 11.084049224853516,
      "learning_rate": 2.6708149824235323e-05,
      "loss": 1.1311,
      "step": 11630
    },
    {
      "epoch": 1.405287939152481,
      "grad_norm": 9.033219337463379,
      "learning_rate": 2.6687946987757085e-05,
      "loss": 0.7096,
      "step": 11640
    },
    {
      "epoch": 1.4064952311964265,
      "grad_norm": 9.509979248046875,
      "learning_rate": 2.666774415127884e-05,
      "loss": 0.841,
      "step": 11650
    },
    {
      "epoch": 1.407702523240372,
      "grad_norm": 10.659226417541504,
      "learning_rate": 2.6647541314800602e-05,
      "loss": 0.718,
      "step": 11660
    },
    {
      "epoch": 1.4089098152843174,
      "grad_norm": 3.05588698387146,
      "learning_rate": 2.6627338478322357e-05,
      "loss": 0.7772,
      "step": 11670
    },
    {
      "epoch": 1.4101171073282628,
      "grad_norm": 19.4785099029541,
      "learning_rate": 2.6607135641844116e-05,
      "loss": 0.9677,
      "step": 11680
    },
    {
      "epoch": 1.4113243993722082,
      "grad_norm": 8.409974098205566,
      "learning_rate": 2.6586932805365877e-05,
      "loss": 0.9443,
      "step": 11690
    },
    {
      "epoch": 1.4125316914161536,
      "grad_norm": 29.00954246520996,
      "learning_rate": 2.6566729968887632e-05,
      "loss": 1.0661,
      "step": 11700
    },
    {
      "epoch": 1.413738983460099,
      "grad_norm": 8.52160358428955,
      "learning_rate": 2.6546527132409394e-05,
      "loss": 1.0616,
      "step": 11710
    },
    {
      "epoch": 1.4149462755040445,
      "grad_norm": 18.08308982849121,
      "learning_rate": 2.652632429593115e-05,
      "loss": 1.6026,
      "step": 11720
    },
    {
      "epoch": 1.41615356754799,
      "grad_norm": 19.79472541809082,
      "learning_rate": 2.6506121459452908e-05,
      "loss": 0.7729,
      "step": 11730
    },
    {
      "epoch": 1.4173608595919354,
      "grad_norm": 19.07781219482422,
      "learning_rate": 2.6485918622974666e-05,
      "loss": 0.719,
      "step": 11740
    },
    {
      "epoch": 1.4185681516358808,
      "grad_norm": 7.588253021240234,
      "learning_rate": 2.6465715786496425e-05,
      "loss": 0.7052,
      "step": 11750
    },
    {
      "epoch": 1.4197754436798262,
      "grad_norm": 7.873288631439209,
      "learning_rate": 2.6445512950018187e-05,
      "loss": 0.9911,
      "step": 11760
    },
    {
      "epoch": 1.4209827357237717,
      "grad_norm": 18.796165466308594,
      "learning_rate": 2.6425310113539942e-05,
      "loss": 1.0524,
      "step": 11770
    },
    {
      "epoch": 1.422190027767717,
      "grad_norm": 6.348322868347168,
      "learning_rate": 2.6405107277061704e-05,
      "loss": 1.0182,
      "step": 11780
    },
    {
      "epoch": 1.4233973198116625,
      "grad_norm": 12.049840927124023,
      "learning_rate": 2.638490444058346e-05,
      "loss": 0.7923,
      "step": 11790
    },
    {
      "epoch": 1.424604611855608,
      "grad_norm": 13.726216316223145,
      "learning_rate": 2.6364701604105217e-05,
      "loss": 0.8581,
      "step": 11800
    },
    {
      "epoch": 1.4258119038995534,
      "grad_norm": 8.874062538146973,
      "learning_rate": 2.6344498767626972e-05,
      "loss": 0.8827,
      "step": 11810
    },
    {
      "epoch": 1.4270191959434988,
      "grad_norm": 6.278963565826416,
      "learning_rate": 2.6324295931148734e-05,
      "loss": 1.0115,
      "step": 11820
    },
    {
      "epoch": 1.4282264879874442,
      "grad_norm": 21.460277557373047,
      "learning_rate": 2.6304093094670496e-05,
      "loss": 1.0526,
      "step": 11830
    },
    {
      "epoch": 1.4294337800313897,
      "grad_norm": 5.560268878936768,
      "learning_rate": 2.628389025819225e-05,
      "loss": 0.917,
      "step": 11840
    },
    {
      "epoch": 1.430641072075335,
      "grad_norm": 12.685395240783691,
      "learning_rate": 2.6263687421714013e-05,
      "loss": 1.0008,
      "step": 11850
    },
    {
      "epoch": 1.4318483641192805,
      "grad_norm": 5.698336601257324,
      "learning_rate": 2.6243484585235768e-05,
      "loss": 1.2412,
      "step": 11860
    },
    {
      "epoch": 1.433055656163226,
      "grad_norm": 8.255685806274414,
      "learning_rate": 2.6223281748757527e-05,
      "loss": 1.1789,
      "step": 11870
    },
    {
      "epoch": 1.4342629482071714,
      "grad_norm": 19.377670288085938,
      "learning_rate": 2.620307891227928e-05,
      "loss": 1.355,
      "step": 11880
    },
    {
      "epoch": 1.4354702402511168,
      "grad_norm": 13.589982986450195,
      "learning_rate": 2.6182876075801044e-05,
      "loss": 1.3545,
      "step": 11890
    },
    {
      "epoch": 1.4366775322950622,
      "grad_norm": 6.3388190269470215,
      "learning_rate": 2.6162673239322805e-05,
      "loss": 1.1789,
      "step": 11900
    },
    {
      "epoch": 1.4378848243390077,
      "grad_norm": 17.411270141601562,
      "learning_rate": 2.614247040284456e-05,
      "loss": 0.9374,
      "step": 11910
    },
    {
      "epoch": 1.439092116382953,
      "grad_norm": 13.883177757263184,
      "learning_rate": 2.6122267566366322e-05,
      "loss": 0.8119,
      "step": 11920
    },
    {
      "epoch": 1.4402994084268985,
      "grad_norm": 16.896669387817383,
      "learning_rate": 2.6102064729888077e-05,
      "loss": 0.9603,
      "step": 11930
    },
    {
      "epoch": 1.441506700470844,
      "grad_norm": 34.53330612182617,
      "learning_rate": 2.6081861893409836e-05,
      "loss": 0.8406,
      "step": 11940
    },
    {
      "epoch": 1.4427139925147894,
      "grad_norm": 29.477140426635742,
      "learning_rate": 2.606165905693159e-05,
      "loss": 1.1111,
      "step": 11950
    },
    {
      "epoch": 1.4439212845587348,
      "grad_norm": 19.603923797607422,
      "learning_rate": 2.6041456220453353e-05,
      "loss": 1.1024,
      "step": 11960
    },
    {
      "epoch": 1.4451285766026802,
      "grad_norm": 4.542745590209961,
      "learning_rate": 2.6021253383975115e-05,
      "loss": 0.8212,
      "step": 11970
    },
    {
      "epoch": 1.4463358686466257,
      "grad_norm": 8.496732711791992,
      "learning_rate": 2.600105054749687e-05,
      "loss": 1.0396,
      "step": 11980
    },
    {
      "epoch": 1.447543160690571,
      "grad_norm": 10.248193740844727,
      "learning_rate": 2.5980847711018628e-05,
      "loss": 0.7361,
      "step": 11990
    },
    {
      "epoch": 1.4487504527345165,
      "grad_norm": 22.104103088378906,
      "learning_rate": 2.5960644874540387e-05,
      "loss": 1.2986,
      "step": 12000
    },
    {
      "epoch": 1.449957744778462,
      "grad_norm": 8.440836906433105,
      "learning_rate": 2.5940442038062145e-05,
      "loss": 0.9378,
      "step": 12010
    },
    {
      "epoch": 1.4511650368224074,
      "grad_norm": 15.12961196899414,
      "learning_rate": 2.59202392015839e-05,
      "loss": 1.181,
      "step": 12020
    },
    {
      "epoch": 1.4523723288663528,
      "grad_norm": 11.034793853759766,
      "learning_rate": 2.5900036365105662e-05,
      "loss": 0.925,
      "step": 12030
    },
    {
      "epoch": 1.4535796209102982,
      "grad_norm": 12.97219181060791,
      "learning_rate": 2.5879833528627424e-05,
      "loss": 1.18,
      "step": 12040
    },
    {
      "epoch": 1.4547869129542437,
      "grad_norm": 15.662128448486328,
      "learning_rate": 2.585963069214918e-05,
      "loss": 0.7231,
      "step": 12050
    },
    {
      "epoch": 1.455994204998189,
      "grad_norm": 16.287620544433594,
      "learning_rate": 2.5839427855670938e-05,
      "loss": 1.0507,
      "step": 12060
    },
    {
      "epoch": 1.4572014970421345,
      "grad_norm": 14.300922393798828,
      "learning_rate": 2.5819225019192693e-05,
      "loss": 0.998,
      "step": 12070
    },
    {
      "epoch": 1.45840878908608,
      "grad_norm": 12.365530967712402,
      "learning_rate": 2.5799022182714455e-05,
      "loss": 0.8321,
      "step": 12080
    },
    {
      "epoch": 1.4596160811300254,
      "grad_norm": 7.471349716186523,
      "learning_rate": 2.577881934623621e-05,
      "loss": 0.7816,
      "step": 12090
    },
    {
      "epoch": 1.4608233731739708,
      "grad_norm": 2.2803614139556885,
      "learning_rate": 2.575861650975797e-05,
      "loss": 0.9364,
      "step": 12100
    },
    {
      "epoch": 1.4620306652179162,
      "grad_norm": 24.29170799255371,
      "learning_rate": 2.5738413673279733e-05,
      "loss": 1.268,
      "step": 12110
    },
    {
      "epoch": 1.4632379572618617,
      "grad_norm": 15.395285606384277,
      "learning_rate": 2.571821083680149e-05,
      "loss": 1.0983,
      "step": 12120
    },
    {
      "epoch": 1.464445249305807,
      "grad_norm": 8.634528160095215,
      "learning_rate": 2.5698008000323247e-05,
      "loss": 0.9411,
      "step": 12130
    },
    {
      "epoch": 1.4656525413497525,
      "grad_norm": 7.118587017059326,
      "learning_rate": 2.5677805163845002e-05,
      "loss": 0.4723,
      "step": 12140
    },
    {
      "epoch": 1.466859833393698,
      "grad_norm": 20.84715461730957,
      "learning_rate": 2.5657602327366764e-05,
      "loss": 0.5365,
      "step": 12150
    },
    {
      "epoch": 1.4680671254376434,
      "grad_norm": 10.073576927185059,
      "learning_rate": 2.563739949088852e-05,
      "loss": 0.7965,
      "step": 12160
    },
    {
      "epoch": 1.4692744174815888,
      "grad_norm": 17.300209045410156,
      "learning_rate": 2.561719665441028e-05,
      "loss": 0.9254,
      "step": 12170
    },
    {
      "epoch": 1.4704817095255343,
      "grad_norm": 10.880805015563965,
      "learning_rate": 2.5596993817932043e-05,
      "loss": 0.9536,
      "step": 12180
    },
    {
      "epoch": 1.4716890015694797,
      "grad_norm": 13.620851516723633,
      "learning_rate": 2.5576790981453798e-05,
      "loss": 0.8189,
      "step": 12190
    },
    {
      "epoch": 1.4728962936134251,
      "grad_norm": 53.503517150878906,
      "learning_rate": 2.5556588144975556e-05,
      "loss": 1.0456,
      "step": 12200
    },
    {
      "epoch": 1.4741035856573705,
      "grad_norm": 4.518499374389648,
      "learning_rate": 2.553638530849731e-05,
      "loss": 0.6303,
      "step": 12210
    },
    {
      "epoch": 1.475310877701316,
      "grad_norm": 6.4362874031066895,
      "learning_rate": 2.5516182472019073e-05,
      "loss": 0.8813,
      "step": 12220
    },
    {
      "epoch": 1.4765181697452614,
      "grad_norm": 13.194111824035645,
      "learning_rate": 2.5495979635540828e-05,
      "loss": 1.0771,
      "step": 12230
    },
    {
      "epoch": 1.4777254617892068,
      "grad_norm": 9.814130783081055,
      "learning_rate": 2.547577679906259e-05,
      "loss": 0.5456,
      "step": 12240
    },
    {
      "epoch": 1.4789327538331523,
      "grad_norm": 11.216462135314941,
      "learning_rate": 2.5455573962584352e-05,
      "loss": 1.1463,
      "step": 12250
    },
    {
      "epoch": 1.4801400458770977,
      "grad_norm": 5.819253921508789,
      "learning_rate": 2.5435371126106107e-05,
      "loss": 0.9981,
      "step": 12260
    },
    {
      "epoch": 1.4813473379210431,
      "grad_norm": 23.311189651489258,
      "learning_rate": 2.5415168289627866e-05,
      "loss": 1.1123,
      "step": 12270
    },
    {
      "epoch": 1.4825546299649885,
      "grad_norm": 20.400733947753906,
      "learning_rate": 2.539496545314962e-05,
      "loss": 1.3111,
      "step": 12280
    },
    {
      "epoch": 1.483761922008934,
      "grad_norm": 5.336943626403809,
      "learning_rate": 2.5374762616671383e-05,
      "loss": 0.6864,
      "step": 12290
    },
    {
      "epoch": 1.4849692140528794,
      "grad_norm": 13.440387725830078,
      "learning_rate": 2.5354559780193138e-05,
      "loss": 1.1789,
      "step": 12300
    },
    {
      "epoch": 1.4861765060968248,
      "grad_norm": 7.650569915771484,
      "learning_rate": 2.53343569437149e-05,
      "loss": 0.7493,
      "step": 12310
    },
    {
      "epoch": 1.4873837981407703,
      "grad_norm": 8.196991920471191,
      "learning_rate": 2.5314154107236658e-05,
      "loss": 0.9492,
      "step": 12320
    },
    {
      "epoch": 1.4885910901847157,
      "grad_norm": 16.495811462402344,
      "learning_rate": 2.5293951270758416e-05,
      "loss": 0.9306,
      "step": 12330
    },
    {
      "epoch": 1.4897983822286611,
      "grad_norm": 14.659077644348145,
      "learning_rate": 2.5273748434280175e-05,
      "loss": 1.2241,
      "step": 12340
    },
    {
      "epoch": 1.4910056742726066,
      "grad_norm": 19.636096954345703,
      "learning_rate": 2.525354559780193e-05,
      "loss": 1.2835,
      "step": 12350
    },
    {
      "epoch": 1.492212966316552,
      "grad_norm": 26.941059112548828,
      "learning_rate": 2.5233342761323692e-05,
      "loss": 0.8447,
      "step": 12360
    },
    {
      "epoch": 1.4934202583604974,
      "grad_norm": 18.631271362304688,
      "learning_rate": 2.5213139924845447e-05,
      "loss": 0.7809,
      "step": 12370
    },
    {
      "epoch": 1.4946275504044428,
      "grad_norm": 16.809574127197266,
      "learning_rate": 2.519293708836721e-05,
      "loss": 0.9118,
      "step": 12380
    },
    {
      "epoch": 1.4958348424483883,
      "grad_norm": 3.290919303894043,
      "learning_rate": 2.5172734251888967e-05,
      "loss": 0.891,
      "step": 12390
    },
    {
      "epoch": 1.4970421344923337,
      "grad_norm": 5.528446674346924,
      "learning_rate": 2.5152531415410722e-05,
      "loss": 0.9562,
      "step": 12400
    },
    {
      "epoch": 1.4982494265362791,
      "grad_norm": 29.602312088012695,
      "learning_rate": 2.5132328578932484e-05,
      "loss": 0.8223,
      "step": 12410
    },
    {
      "epoch": 1.4994567185802246,
      "grad_norm": 11.013571739196777,
      "learning_rate": 2.511212574245424e-05,
      "loss": 0.7631,
      "step": 12420
    },
    {
      "epoch": 1.50066401062417,
      "grad_norm": 2.4692986011505127,
      "learning_rate": 2.5091922905976e-05,
      "loss": 1.5223,
      "step": 12430
    },
    {
      "epoch": 1.5018713026681154,
      "grad_norm": 25.87965202331543,
      "learning_rate": 2.5071720069497756e-05,
      "loss": 0.5788,
      "step": 12440
    },
    {
      "epoch": 1.5030785947120608,
      "grad_norm": 27.30670928955078,
      "learning_rate": 2.5051517233019518e-05,
      "loss": 0.5621,
      "step": 12450
    },
    {
      "epoch": 1.5042858867560063,
      "grad_norm": 8.33663272857666,
      "learning_rate": 2.5031314396541277e-05,
      "loss": 0.8716,
      "step": 12460
    },
    {
      "epoch": 1.5054931787999517,
      "grad_norm": 7.630097389221191,
      "learning_rate": 2.501111156006303e-05,
      "loss": 0.7911,
      "step": 12470
    },
    {
      "epoch": 1.5067004708438971,
      "grad_norm": 22.59745979309082,
      "learning_rate": 2.4990908723584794e-05,
      "loss": 1.0986,
      "step": 12480
    },
    {
      "epoch": 1.5079077628878426,
      "grad_norm": 14.445764541625977,
      "learning_rate": 2.4970705887106552e-05,
      "loss": 0.818,
      "step": 12490
    },
    {
      "epoch": 1.509115054931788,
      "grad_norm": 1.9972405433654785,
      "learning_rate": 2.495050305062831e-05,
      "loss": 0.827,
      "step": 12500
    },
    {
      "epoch": 1.5103223469757334,
      "grad_norm": 23.005596160888672,
      "learning_rate": 2.493030021415007e-05,
      "loss": 1.2548,
      "step": 12510
    },
    {
      "epoch": 1.5115296390196789,
      "grad_norm": 9.784256935119629,
      "learning_rate": 2.4910097377671827e-05,
      "loss": 1.25,
      "step": 12520
    },
    {
      "epoch": 1.5127369310636243,
      "grad_norm": 8.710111618041992,
      "learning_rate": 2.4889894541193583e-05,
      "loss": 0.7174,
      "step": 12530
    },
    {
      "epoch": 1.5139442231075697,
      "grad_norm": 5.778669357299805,
      "learning_rate": 2.486969170471534e-05,
      "loss": 1.1002,
      "step": 12540
    },
    {
      "epoch": 1.5151515151515151,
      "grad_norm": 7.964632511138916,
      "learning_rate": 2.4849488868237103e-05,
      "loss": 0.9869,
      "step": 12550
    },
    {
      "epoch": 1.5163588071954606,
      "grad_norm": 7.29714298248291,
      "learning_rate": 2.482928603175886e-05,
      "loss": 0.7915,
      "step": 12560
    },
    {
      "epoch": 1.517566099239406,
      "grad_norm": 30.658390045166016,
      "learning_rate": 2.480908319528062e-05,
      "loss": 0.6494,
      "step": 12570
    },
    {
      "epoch": 1.5187733912833514,
      "grad_norm": 21.58065414428711,
      "learning_rate": 2.4788880358802378e-05,
      "loss": 1.0558,
      "step": 12580
    },
    {
      "epoch": 1.5199806833272969,
      "grad_norm": 39.18333053588867,
      "learning_rate": 2.4768677522324137e-05,
      "loss": 0.7365,
      "step": 12590
    },
    {
      "epoch": 1.5211879753712423,
      "grad_norm": 15.89927864074707,
      "learning_rate": 2.4748474685845892e-05,
      "loss": 0.79,
      "step": 12600
    },
    {
      "epoch": 1.5223952674151877,
      "grad_norm": 11.1023530960083,
      "learning_rate": 2.472827184936765e-05,
      "loss": 1.5867,
      "step": 12610
    },
    {
      "epoch": 1.5236025594591331,
      "grad_norm": 18.426467895507812,
      "learning_rate": 2.4708069012889412e-05,
      "loss": 0.8361,
      "step": 12620
    },
    {
      "epoch": 1.5248098515030786,
      "grad_norm": 32.87694549560547,
      "learning_rate": 2.468786617641117e-05,
      "loss": 0.7752,
      "step": 12630
    },
    {
      "epoch": 1.526017143547024,
      "grad_norm": 2.7786951065063477,
      "learning_rate": 2.466766333993293e-05,
      "loss": 0.7742,
      "step": 12640
    },
    {
      "epoch": 1.5272244355909694,
      "grad_norm": 27.54779052734375,
      "learning_rate": 2.4647460503454688e-05,
      "loss": 1.1691,
      "step": 12650
    },
    {
      "epoch": 1.5284317276349149,
      "grad_norm": 14.189085960388184,
      "learning_rate": 2.4627257666976443e-05,
      "loss": 0.8379,
      "step": 12660
    },
    {
      "epoch": 1.5296390196788603,
      "grad_norm": 22.006126403808594,
      "learning_rate": 2.46070548304982e-05,
      "loss": 0.9701,
      "step": 12670
    },
    {
      "epoch": 1.5308463117228057,
      "grad_norm": 35.659324645996094,
      "learning_rate": 2.458685199401996e-05,
      "loss": 1.8063,
      "step": 12680
    },
    {
      "epoch": 1.5320536037667511,
      "grad_norm": 9.468942642211914,
      "learning_rate": 2.456664915754172e-05,
      "loss": 0.836,
      "step": 12690
    },
    {
      "epoch": 1.5332608958106966,
      "grad_norm": 33.64291763305664,
      "learning_rate": 2.454644632106348e-05,
      "loss": 1.0226,
      "step": 12700
    },
    {
      "epoch": 1.534468187854642,
      "grad_norm": 22.404062271118164,
      "learning_rate": 2.452624348458524e-05,
      "loss": 1.1213,
      "step": 12710
    },
    {
      "epoch": 1.5356754798985874,
      "grad_norm": 11.57475471496582,
      "learning_rate": 2.4506040648106997e-05,
      "loss": 1.1028,
      "step": 12720
    },
    {
      "epoch": 1.5368827719425329,
      "grad_norm": 4.611263275146484,
      "learning_rate": 2.4485837811628752e-05,
      "loss": 0.9281,
      "step": 12730
    },
    {
      "epoch": 1.5380900639864783,
      "grad_norm": 8.566828727722168,
      "learning_rate": 2.446563497515051e-05,
      "loss": 0.9674,
      "step": 12740
    },
    {
      "epoch": 1.5392973560304237,
      "grad_norm": 17.912899017333984,
      "learning_rate": 2.444543213867227e-05,
      "loss": 1.4976,
      "step": 12750
    },
    {
      "epoch": 1.5405046480743692,
      "grad_norm": 8.223530769348145,
      "learning_rate": 2.442522930219403e-05,
      "loss": 0.8456,
      "step": 12760
    },
    {
      "epoch": 1.5417119401183146,
      "grad_norm": 10.493898391723633,
      "learning_rate": 2.440502646571579e-05,
      "loss": 0.9587,
      "step": 12770
    },
    {
      "epoch": 1.54291923216226,
      "grad_norm": 105.8651123046875,
      "learning_rate": 2.4384823629237548e-05,
      "loss": 0.7102,
      "step": 12780
    },
    {
      "epoch": 1.5441265242062054,
      "grad_norm": 18.585983276367188,
      "learning_rate": 2.4364620792759303e-05,
      "loss": 1.2232,
      "step": 12790
    },
    {
      "epoch": 1.5453338162501509,
      "grad_norm": 9.311059951782227,
      "learning_rate": 2.434441795628106e-05,
      "loss": 1.0705,
      "step": 12800
    },
    {
      "epoch": 1.5465411082940963,
      "grad_norm": 28.74772071838379,
      "learning_rate": 2.432421511980282e-05,
      "loss": 0.6692,
      "step": 12810
    },
    {
      "epoch": 1.5477484003380417,
      "grad_norm": 10.699061393737793,
      "learning_rate": 2.4304012283324578e-05,
      "loss": 1.1041,
      "step": 12820
    },
    {
      "epoch": 1.5489556923819872,
      "grad_norm": 6.6941022872924805,
      "learning_rate": 2.428380944684634e-05,
      "loss": 0.6717,
      "step": 12830
    },
    {
      "epoch": 1.5501629844259326,
      "grad_norm": 9.871674537658691,
      "learning_rate": 2.42636066103681e-05,
      "loss": 1.025,
      "step": 12840
    },
    {
      "epoch": 1.551370276469878,
      "grad_norm": 13.384499549865723,
      "learning_rate": 2.4243403773889857e-05,
      "loss": 0.9535,
      "step": 12850
    },
    {
      "epoch": 1.5525775685138234,
      "grad_norm": 2.0843405723571777,
      "learning_rate": 2.4223200937411612e-05,
      "loss": 0.8285,
      "step": 12860
    },
    {
      "epoch": 1.5537848605577689,
      "grad_norm": 9.985257148742676,
      "learning_rate": 2.420299810093337e-05,
      "loss": 0.6873,
      "step": 12870
    },
    {
      "epoch": 1.5549921526017143,
      "grad_norm": 24.681337356567383,
      "learning_rate": 2.418279526445513e-05,
      "loss": 1.0036,
      "step": 12880
    },
    {
      "epoch": 1.5561994446456597,
      "grad_norm": 11.02968978881836,
      "learning_rate": 2.4162592427976888e-05,
      "loss": 0.7391,
      "step": 12890
    },
    {
      "epoch": 1.5574067366896052,
      "grad_norm": 20.62665367126465,
      "learning_rate": 2.414238959149865e-05,
      "loss": 1.0457,
      "step": 12900
    },
    {
      "epoch": 1.5586140287335506,
      "grad_norm": 15.079858779907227,
      "learning_rate": 2.4122186755020408e-05,
      "loss": 0.7982,
      "step": 12910
    },
    {
      "epoch": 1.559821320777496,
      "grad_norm": 16.773672103881836,
      "learning_rate": 2.4101983918542166e-05,
      "loss": 0.8342,
      "step": 12920
    },
    {
      "epoch": 1.5610286128214415,
      "grad_norm": 12.188947677612305,
      "learning_rate": 2.408178108206392e-05,
      "loss": 0.9826,
      "step": 12930
    },
    {
      "epoch": 1.5622359048653869,
      "grad_norm": 24.253185272216797,
      "learning_rate": 2.406157824558568e-05,
      "loss": 0.8256,
      "step": 12940
    },
    {
      "epoch": 1.5634431969093323,
      "grad_norm": 23.74017333984375,
      "learning_rate": 2.404137540910744e-05,
      "loss": 1.4451,
      "step": 12950
    },
    {
      "epoch": 1.5646504889532777,
      "grad_norm": 21.438034057617188,
      "learning_rate": 2.4021172572629197e-05,
      "loss": 0.9016,
      "step": 12960
    },
    {
      "epoch": 1.5658577809972232,
      "grad_norm": 13.676427841186523,
      "learning_rate": 2.400096973615096e-05,
      "loss": 0.5833,
      "step": 12970
    },
    {
      "epoch": 1.5670650730411686,
      "grad_norm": 14.987502098083496,
      "learning_rate": 2.3980766899672717e-05,
      "loss": 1.1144,
      "step": 12980
    },
    {
      "epoch": 1.568272365085114,
      "grad_norm": 8.937113761901855,
      "learning_rate": 2.3960564063194472e-05,
      "loss": 1.055,
      "step": 12990
    },
    {
      "epoch": 1.5694796571290595,
      "grad_norm": 45.10525131225586,
      "learning_rate": 2.394036122671623e-05,
      "loss": 0.9751,
      "step": 13000
    },
    {
      "epoch": 1.5706869491730049,
      "grad_norm": 15.386812210083008,
      "learning_rate": 2.392015839023799e-05,
      "loss": 1.1391,
      "step": 13010
    },
    {
      "epoch": 1.5718942412169503,
      "grad_norm": 6.2233710289001465,
      "learning_rate": 2.3899955553759748e-05,
      "loss": 1.2053,
      "step": 13020
    },
    {
      "epoch": 1.5731015332608957,
      "grad_norm": 1.9963107109069824,
      "learning_rate": 2.3879752717281506e-05,
      "loss": 1.1812,
      "step": 13030
    },
    {
      "epoch": 1.5743088253048412,
      "grad_norm": 14.074752807617188,
      "learning_rate": 2.3859549880803268e-05,
      "loss": 1.048,
      "step": 13040
    },
    {
      "epoch": 1.5755161173487866,
      "grad_norm": 13.240731239318848,
      "learning_rate": 2.3839347044325027e-05,
      "loss": 1.0318,
      "step": 13050
    },
    {
      "epoch": 1.576723409392732,
      "grad_norm": 59.671791076660156,
      "learning_rate": 2.381914420784678e-05,
      "loss": 1.0318,
      "step": 13060
    },
    {
      "epoch": 1.5779307014366775,
      "grad_norm": 2.4815292358398438,
      "learning_rate": 2.379894137136854e-05,
      "loss": 1.0248,
      "step": 13070
    },
    {
      "epoch": 1.579137993480623,
      "grad_norm": 22.804128646850586,
      "learning_rate": 2.37787385348903e-05,
      "loss": 1.2101,
      "step": 13080
    },
    {
      "epoch": 1.5803452855245683,
      "grad_norm": 9.052618980407715,
      "learning_rate": 2.3758535698412057e-05,
      "loss": 1.0799,
      "step": 13090
    },
    {
      "epoch": 1.5815525775685138,
      "grad_norm": 16.22858428955078,
      "learning_rate": 2.373833286193382e-05,
      "loss": 1.1683,
      "step": 13100
    },
    {
      "epoch": 1.5827598696124592,
      "grad_norm": 7.947248935699463,
      "learning_rate": 2.3718130025455577e-05,
      "loss": 0.5809,
      "step": 13110
    },
    {
      "epoch": 1.5839671616564046,
      "grad_norm": 17.61154556274414,
      "learning_rate": 2.3697927188977333e-05,
      "loss": 0.9381,
      "step": 13120
    },
    {
      "epoch": 1.58517445370035,
      "grad_norm": 12.542841911315918,
      "learning_rate": 2.367772435249909e-05,
      "loss": 1.0764,
      "step": 13130
    },
    {
      "epoch": 1.5863817457442955,
      "grad_norm": 20.040864944458008,
      "learning_rate": 2.365752151602085e-05,
      "loss": 1.3705,
      "step": 13140
    },
    {
      "epoch": 1.587589037788241,
      "grad_norm": 15.363109588623047,
      "learning_rate": 2.3637318679542608e-05,
      "loss": 1.1804,
      "step": 13150
    },
    {
      "epoch": 1.5887963298321863,
      "grad_norm": 14.266369819641113,
      "learning_rate": 2.3617115843064366e-05,
      "loss": 1.1983,
      "step": 13160
    },
    {
      "epoch": 1.5900036218761318,
      "grad_norm": 4.230973243713379,
      "learning_rate": 2.3596913006586128e-05,
      "loss": 0.7959,
      "step": 13170
    },
    {
      "epoch": 1.5912109139200772,
      "grad_norm": 17.247922897338867,
      "learning_rate": 2.3576710170107887e-05,
      "loss": 0.9202,
      "step": 13180
    },
    {
      "epoch": 1.5924182059640226,
      "grad_norm": 22.223697662353516,
      "learning_rate": 2.3556507333629642e-05,
      "loss": 0.9344,
      "step": 13190
    },
    {
      "epoch": 1.593625498007968,
      "grad_norm": 2.0698440074920654,
      "learning_rate": 2.35363044971514e-05,
      "loss": 1.2625,
      "step": 13200
    },
    {
      "epoch": 1.5948327900519135,
      "grad_norm": 7.315591335296631,
      "learning_rate": 2.351610166067316e-05,
      "loss": 1.5454,
      "step": 13210
    },
    {
      "epoch": 1.596040082095859,
      "grad_norm": 19.357032775878906,
      "learning_rate": 2.3495898824194917e-05,
      "loss": 1.0245,
      "step": 13220
    },
    {
      "epoch": 1.5972473741398043,
      "grad_norm": 14.681772232055664,
      "learning_rate": 2.3475695987716676e-05,
      "loss": 0.7305,
      "step": 13230
    },
    {
      "epoch": 1.5984546661837498,
      "grad_norm": 8.600985527038574,
      "learning_rate": 2.3455493151238438e-05,
      "loss": 1.3338,
      "step": 13240
    },
    {
      "epoch": 1.5996619582276952,
      "grad_norm": 13.664219856262207,
      "learning_rate": 2.3435290314760193e-05,
      "loss": 0.9293,
      "step": 13250
    },
    {
      "epoch": 1.6008692502716406,
      "grad_norm": 34.83662796020508,
      "learning_rate": 2.341508747828195e-05,
      "loss": 0.9226,
      "step": 13260
    },
    {
      "epoch": 1.602076542315586,
      "grad_norm": 14.523893356323242,
      "learning_rate": 2.339488464180371e-05,
      "loss": 1.0645,
      "step": 13270
    },
    {
      "epoch": 1.6032838343595315,
      "grad_norm": 11.626002311706543,
      "learning_rate": 2.3374681805325468e-05,
      "loss": 1.1844,
      "step": 13280
    },
    {
      "epoch": 1.604491126403477,
      "grad_norm": 26.694042205810547,
      "learning_rate": 2.3354478968847227e-05,
      "loss": 0.8888,
      "step": 13290
    },
    {
      "epoch": 1.6056984184474223,
      "grad_norm": 12.316693305969238,
      "learning_rate": 2.3334276132368985e-05,
      "loss": 0.982,
      "step": 13300
    },
    {
      "epoch": 1.6069057104913678,
      "grad_norm": 9.726806640625,
      "learning_rate": 2.3314073295890747e-05,
      "loss": 0.8858,
      "step": 13310
    },
    {
      "epoch": 1.6081130025353132,
      "grad_norm": 16.23672103881836,
      "learning_rate": 2.3293870459412502e-05,
      "loss": 0.9964,
      "step": 13320
    },
    {
      "epoch": 1.6093202945792586,
      "grad_norm": 11.876401901245117,
      "learning_rate": 2.327366762293426e-05,
      "loss": 0.928,
      "step": 13330
    },
    {
      "epoch": 1.610527586623204,
      "grad_norm": 9.617765426635742,
      "learning_rate": 2.325346478645602e-05,
      "loss": 0.8333,
      "step": 13340
    },
    {
      "epoch": 1.6117348786671495,
      "grad_norm": 8.953312873840332,
      "learning_rate": 2.3233261949977777e-05,
      "loss": 1.1917,
      "step": 13350
    },
    {
      "epoch": 1.612942170711095,
      "grad_norm": 16.47088050842285,
      "learning_rate": 2.3213059113499536e-05,
      "loss": 1.1031,
      "step": 13360
    },
    {
      "epoch": 1.6141494627550403,
      "grad_norm": 7.868622779846191,
      "learning_rate": 2.3192856277021294e-05,
      "loss": 0.7582,
      "step": 13370
    },
    {
      "epoch": 1.6153567547989858,
      "grad_norm": 6.6080498695373535,
      "learning_rate": 2.3172653440543053e-05,
      "loss": 1.0426,
      "step": 13380
    },
    {
      "epoch": 1.6165640468429312,
      "grad_norm": 8.790926933288574,
      "learning_rate": 2.315245060406481e-05,
      "loss": 0.9964,
      "step": 13390
    },
    {
      "epoch": 1.6177713388868766,
      "grad_norm": 12.570124626159668,
      "learning_rate": 2.313224776758657e-05,
      "loss": 1.096,
      "step": 13400
    },
    {
      "epoch": 1.618978630930822,
      "grad_norm": 23.011608123779297,
      "learning_rate": 2.311204493110833e-05,
      "loss": 0.941,
      "step": 13410
    },
    {
      "epoch": 1.6201859229747675,
      "grad_norm": 12.124536514282227,
      "learning_rate": 2.3091842094630087e-05,
      "loss": 1.2821,
      "step": 13420
    },
    {
      "epoch": 1.621393215018713,
      "grad_norm": 20.796037673950195,
      "learning_rate": 2.3071639258151845e-05,
      "loss": 0.5878,
      "step": 13430
    },
    {
      "epoch": 1.6226005070626583,
      "grad_norm": 3.544663429260254,
      "learning_rate": 2.3051436421673604e-05,
      "loss": 0.8342,
      "step": 13440
    },
    {
      "epoch": 1.6238077991066038,
      "grad_norm": 33.12778091430664,
      "learning_rate": 2.3031233585195362e-05,
      "loss": 1.2739,
      "step": 13450
    },
    {
      "epoch": 1.6250150911505492,
      "grad_norm": 12.098657608032227,
      "learning_rate": 2.301103074871712e-05,
      "loss": 0.8531,
      "step": 13460
    },
    {
      "epoch": 1.6262223831944946,
      "grad_norm": 9.138688087463379,
      "learning_rate": 2.299082791223888e-05,
      "loss": 0.7349,
      "step": 13470
    },
    {
      "epoch": 1.62742967523844,
      "grad_norm": 16.85980796813965,
      "learning_rate": 2.2970625075760638e-05,
      "loss": 1.2105,
      "step": 13480
    },
    {
      "epoch": 1.6286369672823855,
      "grad_norm": 8.028451919555664,
      "learning_rate": 2.2950422239282396e-05,
      "loss": 0.485,
      "step": 13490
    },
    {
      "epoch": 1.629844259326331,
      "grad_norm": 15.99238395690918,
      "learning_rate": 2.2930219402804155e-05,
      "loss": 0.7483,
      "step": 13500
    },
    {
      "epoch": 1.6310515513702764,
      "grad_norm": 22.833744049072266,
      "learning_rate": 2.2910016566325913e-05,
      "loss": 1.3733,
      "step": 13510
    },
    {
      "epoch": 1.6322588434142218,
      "grad_norm": 13.137679100036621,
      "learning_rate": 2.288981372984767e-05,
      "loss": 1.1018,
      "step": 13520
    },
    {
      "epoch": 1.6334661354581672,
      "grad_norm": 29.51358985900879,
      "learning_rate": 2.286961089336943e-05,
      "loss": 0.8458,
      "step": 13530
    },
    {
      "epoch": 1.6346734275021126,
      "grad_norm": 19.93012809753418,
      "learning_rate": 2.284940805689119e-05,
      "loss": 0.947,
      "step": 13540
    },
    {
      "epoch": 1.635880719546058,
      "grad_norm": 6.311330318450928,
      "learning_rate": 2.2829205220412947e-05,
      "loss": 1.153,
      "step": 13550
    },
    {
      "epoch": 1.6370880115900035,
      "grad_norm": 8.019070625305176,
      "learning_rate": 2.2809002383934705e-05,
      "loss": 1.1262,
      "step": 13560
    },
    {
      "epoch": 1.638295303633949,
      "grad_norm": 7.740245819091797,
      "learning_rate": 2.2788799547456464e-05,
      "loss": 0.8058,
      "step": 13570
    },
    {
      "epoch": 1.6395025956778944,
      "grad_norm": 10.956445693969727,
      "learning_rate": 2.2768596710978222e-05,
      "loss": 1.005,
      "step": 13580
    },
    {
      "epoch": 1.6407098877218398,
      "grad_norm": 26.359437942504883,
      "learning_rate": 2.274839387449998e-05,
      "loss": 1.5201,
      "step": 13590
    },
    {
      "epoch": 1.6419171797657852,
      "grad_norm": 17.426504135131836,
      "learning_rate": 2.272819103802174e-05,
      "loss": 0.8788,
      "step": 13600
    },
    {
      "epoch": 1.6431244718097306,
      "grad_norm": 42.23351287841797,
      "learning_rate": 2.2707988201543498e-05,
      "loss": 0.8726,
      "step": 13610
    },
    {
      "epoch": 1.644331763853676,
      "grad_norm": 22.411479949951172,
      "learning_rate": 2.2687785365065256e-05,
      "loss": 0.906,
      "step": 13620
    },
    {
      "epoch": 1.6455390558976215,
      "grad_norm": 23.604290008544922,
      "learning_rate": 2.2667582528587015e-05,
      "loss": 0.8381,
      "step": 13630
    },
    {
      "epoch": 1.646746347941567,
      "grad_norm": 15.437644004821777,
      "learning_rate": 2.2647379692108773e-05,
      "loss": 1.0922,
      "step": 13640
    },
    {
      "epoch": 1.6479536399855124,
      "grad_norm": 6.444321155548096,
      "learning_rate": 2.2627176855630532e-05,
      "loss": 0.584,
      "step": 13650
    },
    {
      "epoch": 1.6491609320294578,
      "grad_norm": 39.6758918762207,
      "learning_rate": 2.260697401915229e-05,
      "loss": 0.9173,
      "step": 13660
    },
    {
      "epoch": 1.6503682240734032,
      "grad_norm": 4.188887596130371,
      "learning_rate": 2.258677118267405e-05,
      "loss": 0.834,
      "step": 13670
    },
    {
      "epoch": 1.6515755161173487,
      "grad_norm": 8.16096305847168,
      "learning_rate": 2.2566568346195807e-05,
      "loss": 0.6925,
      "step": 13680
    },
    {
      "epoch": 1.652782808161294,
      "grad_norm": 13.550382614135742,
      "learning_rate": 2.2546365509717566e-05,
      "loss": 0.6301,
      "step": 13690
    },
    {
      "epoch": 1.6539901002052395,
      "grad_norm": 11.686091423034668,
      "learning_rate": 2.2526162673239324e-05,
      "loss": 0.9607,
      "step": 13700
    },
    {
      "epoch": 1.655197392249185,
      "grad_norm": 14.355452537536621,
      "learning_rate": 2.2505959836761083e-05,
      "loss": 1.0099,
      "step": 13710
    },
    {
      "epoch": 1.6564046842931304,
      "grad_norm": 13.40802001953125,
      "learning_rate": 2.248575700028284e-05,
      "loss": 0.7194,
      "step": 13720
    },
    {
      "epoch": 1.6576119763370758,
      "grad_norm": 12.374186515808105,
      "learning_rate": 2.24655541638046e-05,
      "loss": 0.6476,
      "step": 13730
    },
    {
      "epoch": 1.6588192683810212,
      "grad_norm": 30.377370834350586,
      "learning_rate": 2.2445351327326358e-05,
      "loss": 0.9428,
      "step": 13740
    },
    {
      "epoch": 1.6600265604249667,
      "grad_norm": 6.464419841766357,
      "learning_rate": 2.2425148490848116e-05,
      "loss": 0.6187,
      "step": 13750
    },
    {
      "epoch": 1.661233852468912,
      "grad_norm": 13.547152519226074,
      "learning_rate": 2.2404945654369875e-05,
      "loss": 0.8339,
      "step": 13760
    },
    {
      "epoch": 1.6624411445128575,
      "grad_norm": 14.134217262268066,
      "learning_rate": 2.2384742817891633e-05,
      "loss": 0.7133,
      "step": 13770
    },
    {
      "epoch": 1.663648436556803,
      "grad_norm": 9.939017295837402,
      "learning_rate": 2.2364539981413392e-05,
      "loss": 0.8954,
      "step": 13780
    },
    {
      "epoch": 1.6648557286007484,
      "grad_norm": 1.3239110708236694,
      "learning_rate": 2.234433714493515e-05,
      "loss": 0.9871,
      "step": 13790
    },
    {
      "epoch": 1.6660630206446938,
      "grad_norm": 19.799806594848633,
      "learning_rate": 2.232413430845691e-05,
      "loss": 1.4528,
      "step": 13800
    },
    {
      "epoch": 1.6672703126886392,
      "grad_norm": 12.171647071838379,
      "learning_rate": 2.2303931471978667e-05,
      "loss": 0.6524,
      "step": 13810
    },
    {
      "epoch": 1.6684776047325847,
      "grad_norm": 30.40402603149414,
      "learning_rate": 2.2283728635500426e-05,
      "loss": 0.9992,
      "step": 13820
    },
    {
      "epoch": 1.66968489677653,
      "grad_norm": 25.173843383789062,
      "learning_rate": 2.2263525799022184e-05,
      "loss": 0.9533,
      "step": 13830
    },
    {
      "epoch": 1.6708921888204755,
      "grad_norm": 4.97238826751709,
      "learning_rate": 2.2243322962543943e-05,
      "loss": 0.7436,
      "step": 13840
    },
    {
      "epoch": 1.672099480864421,
      "grad_norm": 0.9660268425941467,
      "learning_rate": 2.22231201260657e-05,
      "loss": 0.784,
      "step": 13850
    },
    {
      "epoch": 1.6733067729083664,
      "grad_norm": 3.124803304672241,
      "learning_rate": 2.220291728958746e-05,
      "loss": 0.6366,
      "step": 13860
    },
    {
      "epoch": 1.6745140649523118,
      "grad_norm": 6.332460403442383,
      "learning_rate": 2.2182714453109218e-05,
      "loss": 1.1004,
      "step": 13870
    },
    {
      "epoch": 1.6757213569962572,
      "grad_norm": 20.935338973999023,
      "learning_rate": 2.2162511616630977e-05,
      "loss": 1.3572,
      "step": 13880
    },
    {
      "epoch": 1.6769286490402027,
      "grad_norm": 13.622316360473633,
      "learning_rate": 2.2142308780152735e-05,
      "loss": 0.5562,
      "step": 13890
    },
    {
      "epoch": 1.678135941084148,
      "grad_norm": 33.324462890625,
      "learning_rate": 2.2122105943674494e-05,
      "loss": 0.9871,
      "step": 13900
    },
    {
      "epoch": 1.6793432331280935,
      "grad_norm": 13.532309532165527,
      "learning_rate": 2.2101903107196252e-05,
      "loss": 0.8996,
      "step": 13910
    },
    {
      "epoch": 1.680550525172039,
      "grad_norm": 11.970497131347656,
      "learning_rate": 2.2081700270718007e-05,
      "loss": 1.1948,
      "step": 13920
    },
    {
      "epoch": 1.6817578172159844,
      "grad_norm": 9.08559513092041,
      "learning_rate": 2.206149743423977e-05,
      "loss": 1.7273,
      "step": 13930
    },
    {
      "epoch": 1.6829651092599298,
      "grad_norm": 25.80976104736328,
      "learning_rate": 2.2041294597761527e-05,
      "loss": 1.0978,
      "step": 13940
    },
    {
      "epoch": 1.6841724013038752,
      "grad_norm": 17.801647186279297,
      "learning_rate": 2.2021091761283286e-05,
      "loss": 1.154,
      "step": 13950
    },
    {
      "epoch": 1.6853796933478207,
      "grad_norm": 13.908089637756348,
      "learning_rate": 2.2000888924805044e-05,
      "loss": 1.0837,
      "step": 13960
    },
    {
      "epoch": 1.686586985391766,
      "grad_norm": 12.824328422546387,
      "learning_rate": 2.1980686088326803e-05,
      "loss": 1.4097,
      "step": 13970
    },
    {
      "epoch": 1.6877942774357118,
      "grad_norm": 46.52275085449219,
      "learning_rate": 2.196048325184856e-05,
      "loss": 1.0626,
      "step": 13980
    },
    {
      "epoch": 1.6890015694796572,
      "grad_norm": 11.653787612915039,
      "learning_rate": 2.1940280415370316e-05,
      "loss": 1.1489,
      "step": 13990
    },
    {
      "epoch": 1.6902088615236026,
      "grad_norm": 19.111507415771484,
      "learning_rate": 2.192007757889208e-05,
      "loss": 1.0691,
      "step": 14000
    },
    {
      "epoch": 1.691416153567548,
      "grad_norm": 5.187194347381592,
      "learning_rate": 2.1899874742413837e-05,
      "loss": 0.592,
      "step": 14010
    },
    {
      "epoch": 1.6926234456114935,
      "grad_norm": 19.45875358581543,
      "learning_rate": 2.1879671905935595e-05,
      "loss": 0.7122,
      "step": 14020
    },
    {
      "epoch": 1.693830737655439,
      "grad_norm": 3.7464101314544678,
      "learning_rate": 2.1859469069457354e-05,
      "loss": 1.0563,
      "step": 14030
    },
    {
      "epoch": 1.6950380296993843,
      "grad_norm": 1.3404264450073242,
      "learning_rate": 2.1839266232979112e-05,
      "loss": 0.8386,
      "step": 14040
    },
    {
      "epoch": 1.6962453217433298,
      "grad_norm": 14.064988136291504,
      "learning_rate": 2.1819063396500867e-05,
      "loss": 0.9905,
      "step": 14050
    },
    {
      "epoch": 1.6974526137872752,
      "grad_norm": 12.935839653015137,
      "learning_rate": 2.1798860560022626e-05,
      "loss": 0.8328,
      "step": 14060
    },
    {
      "epoch": 1.6986599058312206,
      "grad_norm": 1.593340516090393,
      "learning_rate": 2.1778657723544388e-05,
      "loss": 0.6393,
      "step": 14070
    },
    {
      "epoch": 1.699867197875166,
      "grad_norm": 10.461077690124512,
      "learning_rate": 2.1758454887066146e-05,
      "loss": 0.6393,
      "step": 14080
    },
    {
      "epoch": 1.7010744899191115,
      "grad_norm": 4.059604644775391,
      "learning_rate": 2.1738252050587905e-05,
      "loss": 0.5748,
      "step": 14090
    },
    {
      "epoch": 1.702281781963057,
      "grad_norm": 13.8495454788208,
      "learning_rate": 2.1718049214109663e-05,
      "loss": 1.1036,
      "step": 14100
    },
    {
      "epoch": 1.7034890740070023,
      "grad_norm": 11.054835319519043,
      "learning_rate": 2.169784637763142e-05,
      "loss": 0.8121,
      "step": 14110
    },
    {
      "epoch": 1.7046963660509478,
      "grad_norm": 10.42788314819336,
      "learning_rate": 2.1677643541153177e-05,
      "loss": 1.391,
      "step": 14120
    },
    {
      "epoch": 1.7059036580948932,
      "grad_norm": 12.132906913757324,
      "learning_rate": 2.1657440704674935e-05,
      "loss": 1.2234,
      "step": 14130
    },
    {
      "epoch": 1.7071109501388386,
      "grad_norm": 1.6585572957992554,
      "learning_rate": 2.1637237868196697e-05,
      "loss": 0.8281,
      "step": 14140
    },
    {
      "epoch": 1.708318242182784,
      "grad_norm": 19.82203483581543,
      "learning_rate": 2.1617035031718455e-05,
      "loss": 0.9178,
      "step": 14150
    },
    {
      "epoch": 1.7095255342267295,
      "grad_norm": 8.212645530700684,
      "learning_rate": 2.1596832195240214e-05,
      "loss": 1.0443,
      "step": 14160
    },
    {
      "epoch": 1.710732826270675,
      "grad_norm": 7.626570224761963,
      "learning_rate": 2.1576629358761972e-05,
      "loss": 0.807,
      "step": 14170
    },
    {
      "epoch": 1.7119401183146203,
      "grad_norm": 16.804288864135742,
      "learning_rate": 2.1556426522283728e-05,
      "loss": 1.4167,
      "step": 14180
    },
    {
      "epoch": 1.7131474103585658,
      "grad_norm": 12.984159469604492,
      "learning_rate": 2.1536223685805486e-05,
      "loss": 1.3756,
      "step": 14190
    },
    {
      "epoch": 1.7143547024025112,
      "grad_norm": 24.64720344543457,
      "learning_rate": 2.1516020849327244e-05,
      "loss": 0.675,
      "step": 14200
    },
    {
      "epoch": 1.7155619944464566,
      "grad_norm": 8.061629295349121,
      "learning_rate": 2.1495818012849006e-05,
      "loss": 0.5665,
      "step": 14210
    },
    {
      "epoch": 1.716769286490402,
      "grad_norm": 31.559816360473633,
      "learning_rate": 2.1475615176370765e-05,
      "loss": 0.9324,
      "step": 14220
    },
    {
      "epoch": 1.7179765785343475,
      "grad_norm": 8.00545597076416,
      "learning_rate": 2.1455412339892523e-05,
      "loss": 1.2156,
      "step": 14230
    },
    {
      "epoch": 1.719183870578293,
      "grad_norm": 27.650924682617188,
      "learning_rate": 2.1435209503414282e-05,
      "loss": 0.7775,
      "step": 14240
    },
    {
      "epoch": 1.7203911626222383,
      "grad_norm": 13.825690269470215,
      "learning_rate": 2.1415006666936037e-05,
      "loss": 0.6909,
      "step": 14250
    },
    {
      "epoch": 1.7215984546661838,
      "grad_norm": 24.670408248901367,
      "learning_rate": 2.1394803830457795e-05,
      "loss": 1.3538,
      "step": 14260
    },
    {
      "epoch": 1.7228057467101292,
      "grad_norm": 11.615914344787598,
      "learning_rate": 2.1374600993979554e-05,
      "loss": 1.0397,
      "step": 14270
    },
    {
      "epoch": 1.7240130387540746,
      "grad_norm": 14.769317626953125,
      "learning_rate": 2.1354398157501316e-05,
      "loss": 0.948,
      "step": 14280
    },
    {
      "epoch": 1.72522033079802,
      "grad_norm": 9.16063404083252,
      "learning_rate": 2.1334195321023074e-05,
      "loss": 0.6404,
      "step": 14290
    },
    {
      "epoch": 1.7264276228419655,
      "grad_norm": 1.8388195037841797,
      "learning_rate": 2.1313992484544833e-05,
      "loss": 0.859,
      "step": 14300
    },
    {
      "epoch": 1.727634914885911,
      "grad_norm": 17.761383056640625,
      "learning_rate": 2.129378964806659e-05,
      "loss": 1.2858,
      "step": 14310
    },
    {
      "epoch": 1.7288422069298564,
      "grad_norm": 21.89221954345703,
      "learning_rate": 2.1273586811588346e-05,
      "loss": 0.8782,
      "step": 14320
    },
    {
      "epoch": 1.7300494989738018,
      "grad_norm": 9.560425758361816,
      "learning_rate": 2.1253383975110105e-05,
      "loss": 1.2233,
      "step": 14330
    },
    {
      "epoch": 1.7312567910177472,
      "grad_norm": 70.22447204589844,
      "learning_rate": 2.1233181138631866e-05,
      "loss": 0.9179,
      "step": 14340
    },
    {
      "epoch": 1.7324640830616926,
      "grad_norm": 9.450370788574219,
      "learning_rate": 2.1212978302153625e-05,
      "loss": 1.2089,
      "step": 14350
    },
    {
      "epoch": 1.733671375105638,
      "grad_norm": 7.807147026062012,
      "learning_rate": 2.1192775465675383e-05,
      "loss": 0.8255,
      "step": 14360
    },
    {
      "epoch": 1.7348786671495835,
      "grad_norm": 11.584732055664062,
      "learning_rate": 2.1172572629197142e-05,
      "loss": 0.9348,
      "step": 14370
    },
    {
      "epoch": 1.736085959193529,
      "grad_norm": 7.963894844055176,
      "learning_rate": 2.1152369792718897e-05,
      "loss": 0.7975,
      "step": 14380
    },
    {
      "epoch": 1.7372932512374744,
      "grad_norm": 15.232930183410645,
      "learning_rate": 2.1132166956240655e-05,
      "loss": 0.8737,
      "step": 14390
    },
    {
      "epoch": 1.7385005432814198,
      "grad_norm": 22.795238494873047,
      "learning_rate": 2.1111964119762414e-05,
      "loss": 1.1474,
      "step": 14400
    },
    {
      "epoch": 1.7397078353253652,
      "grad_norm": 6.377875328063965,
      "learning_rate": 2.1091761283284176e-05,
      "loss": 0.7165,
      "step": 14410
    },
    {
      "epoch": 1.7409151273693106,
      "grad_norm": 14.302197456359863,
      "learning_rate": 2.1071558446805934e-05,
      "loss": 1.0654,
      "step": 14420
    },
    {
      "epoch": 1.742122419413256,
      "grad_norm": 15.838897705078125,
      "learning_rate": 2.1051355610327693e-05,
      "loss": 1.2992,
      "step": 14430
    },
    {
      "epoch": 1.7433297114572015,
      "grad_norm": 16.805164337158203,
      "learning_rate": 2.103115277384945e-05,
      "loss": 1.2657,
      "step": 14440
    },
    {
      "epoch": 1.744537003501147,
      "grad_norm": 1.1849497556686401,
      "learning_rate": 2.1010949937371206e-05,
      "loss": 0.6824,
      "step": 14450
    },
    {
      "epoch": 1.7457442955450924,
      "grad_norm": 7.740742206573486,
      "learning_rate": 2.0990747100892965e-05,
      "loss": 1.04,
      "step": 14460
    },
    {
      "epoch": 1.7469515875890378,
      "grad_norm": 16.90540313720703,
      "learning_rate": 2.0970544264414723e-05,
      "loss": 1.0391,
      "step": 14470
    },
    {
      "epoch": 1.7481588796329832,
      "grad_norm": 25.19227409362793,
      "learning_rate": 2.0950341427936485e-05,
      "loss": 1.141,
      "step": 14480
    },
    {
      "epoch": 1.7493661716769286,
      "grad_norm": 6.837967872619629,
      "learning_rate": 2.0930138591458244e-05,
      "loss": 0.7415,
      "step": 14490
    },
    {
      "epoch": 1.750573463720874,
      "grad_norm": 41.75093460083008,
      "learning_rate": 2.0909935754980002e-05,
      "loss": 1.2331,
      "step": 14500
    },
    {
      "epoch": 1.7517807557648195,
      "grad_norm": 6.410999774932861,
      "learning_rate": 2.0889732918501757e-05,
      "loss": 1.0452,
      "step": 14510
    },
    {
      "epoch": 1.752988047808765,
      "grad_norm": 9.0108642578125,
      "learning_rate": 2.0869530082023516e-05,
      "loss": 0.567,
      "step": 14520
    },
    {
      "epoch": 1.7541953398527104,
      "grad_norm": 13.057019233703613,
      "learning_rate": 2.0849327245545274e-05,
      "loss": 0.6651,
      "step": 14530
    },
    {
      "epoch": 1.7554026318966558,
      "grad_norm": 7.868969440460205,
      "learning_rate": 2.0829124409067033e-05,
      "loss": 0.6085,
      "step": 14540
    },
    {
      "epoch": 1.7566099239406012,
      "grad_norm": 29.81682777404785,
      "learning_rate": 2.0808921572588794e-05,
      "loss": 1.1424,
      "step": 14550
    },
    {
      "epoch": 1.7578172159845467,
      "grad_norm": 13.56236743927002,
      "learning_rate": 2.0788718736110553e-05,
      "loss": 1.2331,
      "step": 14560
    },
    {
      "epoch": 1.759024508028492,
      "grad_norm": 15.611265182495117,
      "learning_rate": 2.076851589963231e-05,
      "loss": 0.9108,
      "step": 14570
    },
    {
      "epoch": 1.7602318000724375,
      "grad_norm": 6.4561238288879395,
      "learning_rate": 2.0748313063154067e-05,
      "loss": 1.1368,
      "step": 14580
    },
    {
      "epoch": 1.761439092116383,
      "grad_norm": 10.27342700958252,
      "learning_rate": 2.0728110226675825e-05,
      "loss": 0.7822,
      "step": 14590
    },
    {
      "epoch": 1.7626463841603284,
      "grad_norm": 7.306971073150635,
      "learning_rate": 2.0707907390197583e-05,
      "loss": 0.7661,
      "step": 14600
    },
    {
      "epoch": 1.7638536762042738,
      "grad_norm": 4.410816192626953,
      "learning_rate": 2.0687704553719342e-05,
      "loss": 0.8302,
      "step": 14610
    },
    {
      "epoch": 1.7650609682482192,
      "grad_norm": 8.250892639160156,
      "learning_rate": 2.0667501717241104e-05,
      "loss": 0.5664,
      "step": 14620
    },
    {
      "epoch": 1.7662682602921647,
      "grad_norm": 17.472814559936523,
      "learning_rate": 2.0647298880762862e-05,
      "loss": 0.6636,
      "step": 14630
    },
    {
      "epoch": 1.76747555233611,
      "grad_norm": 13.501619338989258,
      "learning_rate": 2.0627096044284617e-05,
      "loss": 0.6675,
      "step": 14640
    },
    {
      "epoch": 1.7686828443800555,
      "grad_norm": 22.93730354309082,
      "learning_rate": 2.0606893207806376e-05,
      "loss": 1.0259,
      "step": 14650
    },
    {
      "epoch": 1.769890136424001,
      "grad_norm": 25.756715774536133,
      "learning_rate": 2.0586690371328134e-05,
      "loss": 0.7943,
      "step": 14660
    },
    {
      "epoch": 1.7710974284679464,
      "grad_norm": 19.91506576538086,
      "learning_rate": 2.0566487534849893e-05,
      "loss": 0.8004,
      "step": 14670
    },
    {
      "epoch": 1.7723047205118918,
      "grad_norm": 12.088974952697754,
      "learning_rate": 2.054628469837165e-05,
      "loss": 1.2866,
      "step": 14680
    },
    {
      "epoch": 1.7735120125558372,
      "grad_norm": 28.47678565979004,
      "learning_rate": 2.0526081861893413e-05,
      "loss": 0.981,
      "step": 14690
    },
    {
      "epoch": 1.7747193045997827,
      "grad_norm": 0.9213241338729858,
      "learning_rate": 2.050587902541517e-05,
      "loss": 0.6116,
      "step": 14700
    },
    {
      "epoch": 1.775926596643728,
      "grad_norm": 4.64879846572876,
      "learning_rate": 2.0485676188936927e-05,
      "loss": 0.5764,
      "step": 14710
    },
    {
      "epoch": 1.7771338886876735,
      "grad_norm": 3.298039436340332,
      "learning_rate": 2.0465473352458685e-05,
      "loss": 0.7645,
      "step": 14720
    },
    {
      "epoch": 1.778341180731619,
      "grad_norm": 7.440815448760986,
      "learning_rate": 2.0445270515980444e-05,
      "loss": 1.0085,
      "step": 14730
    },
    {
      "epoch": 1.7795484727755644,
      "grad_norm": 15.477529525756836,
      "learning_rate": 2.0425067679502202e-05,
      "loss": 1.2665,
      "step": 14740
    },
    {
      "epoch": 1.7807557648195098,
      "grad_norm": 14.54726505279541,
      "learning_rate": 2.040486484302396e-05,
      "loss": 0.7757,
      "step": 14750
    },
    {
      "epoch": 1.7819630568634552,
      "grad_norm": 0.8004035353660583,
      "learning_rate": 2.0384662006545722e-05,
      "loss": 1.1295,
      "step": 14760
    },
    {
      "epoch": 1.7831703489074007,
      "grad_norm": 11.371805191040039,
      "learning_rate": 2.0364459170067478e-05,
      "loss": 0.9839,
      "step": 14770
    },
    {
      "epoch": 1.784377640951346,
      "grad_norm": 4.73362922668457,
      "learning_rate": 2.0344256333589236e-05,
      "loss": 1.1406,
      "step": 14780
    },
    {
      "epoch": 1.7855849329952915,
      "grad_norm": 23.37311553955078,
      "learning_rate": 2.0324053497110994e-05,
      "loss": 1.2272,
      "step": 14790
    },
    {
      "epoch": 1.786792225039237,
      "grad_norm": 4.561320781707764,
      "learning_rate": 2.0303850660632753e-05,
      "loss": 0.8049,
      "step": 14800
    },
    {
      "epoch": 1.7879995170831824,
      "grad_norm": 17.0560359954834,
      "learning_rate": 2.028364782415451e-05,
      "loss": 0.5872,
      "step": 14810
    },
    {
      "epoch": 1.7892068091271278,
      "grad_norm": 1.6257303953170776,
      "learning_rate": 2.026344498767627e-05,
      "loss": 0.9291,
      "step": 14820
    },
    {
      "epoch": 1.7904141011710732,
      "grad_norm": 2.3298773765563965,
      "learning_rate": 2.0243242151198032e-05,
      "loss": 0.7908,
      "step": 14830
    },
    {
      "epoch": 1.7916213932150187,
      "grad_norm": 29.995590209960938,
      "learning_rate": 2.0223039314719787e-05,
      "loss": 0.9667,
      "step": 14840
    },
    {
      "epoch": 1.792828685258964,
      "grad_norm": 21.729393005371094,
      "learning_rate": 2.0202836478241545e-05,
      "loss": 0.9927,
      "step": 14850
    },
    {
      "epoch": 1.7940359773029095,
      "grad_norm": 9.959272384643555,
      "learning_rate": 2.0182633641763304e-05,
      "loss": 0.9422,
      "step": 14860
    },
    {
      "epoch": 1.795243269346855,
      "grad_norm": 8.134075164794922,
      "learning_rate": 2.0162430805285062e-05,
      "loss": 0.7671,
      "step": 14870
    },
    {
      "epoch": 1.7964505613908004,
      "grad_norm": 8.74778938293457,
      "learning_rate": 2.014222796880682e-05,
      "loss": 0.745,
      "step": 14880
    },
    {
      "epoch": 1.7976578534347458,
      "grad_norm": 9.036081314086914,
      "learning_rate": 2.012202513232858e-05,
      "loss": 0.9538,
      "step": 14890
    },
    {
      "epoch": 1.7988651454786913,
      "grad_norm": 10.179512977600098,
      "learning_rate": 2.0101822295850338e-05,
      "loss": 0.7222,
      "step": 14900
    },
    {
      "epoch": 1.8000724375226367,
      "grad_norm": 24.807527542114258,
      "learning_rate": 2.0081619459372096e-05,
      "loss": 1.1817,
      "step": 14910
    },
    {
      "epoch": 1.801279729566582,
      "grad_norm": 27.231996536254883,
      "learning_rate": 2.0061416622893855e-05,
      "loss": 0.9812,
      "step": 14920
    },
    {
      "epoch": 1.8024870216105275,
      "grad_norm": 23.048234939575195,
      "learning_rate": 2.0041213786415613e-05,
      "loss": 0.867,
      "step": 14930
    },
    {
      "epoch": 1.803694313654473,
      "grad_norm": 17.49505615234375,
      "learning_rate": 2.002101094993737e-05,
      "loss": 0.8964,
      "step": 14940
    },
    {
      "epoch": 1.8049016056984184,
      "grad_norm": 17.804149627685547,
      "learning_rate": 2.000080811345913e-05,
      "loss": 1.3181,
      "step": 14950
    },
    {
      "epoch": 1.8061088977423638,
      "grad_norm": 22.720077514648438,
      "learning_rate": 1.998060527698089e-05,
      "loss": 1.1046,
      "step": 14960
    },
    {
      "epoch": 1.8073161897863093,
      "grad_norm": 43.2933235168457,
      "learning_rate": 1.9960402440502647e-05,
      "loss": 1.398,
      "step": 14970
    },
    {
      "epoch": 1.8085234818302547,
      "grad_norm": 14.12105655670166,
      "learning_rate": 1.9940199604024406e-05,
      "loss": 0.9199,
      "step": 14980
    },
    {
      "epoch": 1.8097307738742001,
      "grad_norm": 9.08963680267334,
      "learning_rate": 1.9919996767546164e-05,
      "loss": 0.9529,
      "step": 14990
    },
    {
      "epoch": 1.8109380659181455,
      "grad_norm": 15.730798721313477,
      "learning_rate": 1.9899793931067922e-05,
      "loss": 0.9344,
      "step": 15000
    },
    {
      "epoch": 1.812145357962091,
      "grad_norm": 10.643024444580078,
      "learning_rate": 1.987959109458968e-05,
      "loss": 0.929,
      "step": 15010
    },
    {
      "epoch": 1.8133526500060366,
      "grad_norm": 5.970413684844971,
      "learning_rate": 1.985938825811144e-05,
      "loss": 1.3376,
      "step": 15020
    },
    {
      "epoch": 1.814559942049982,
      "grad_norm": 15.051450729370117,
      "learning_rate": 1.98391854216332e-05,
      "loss": 0.7359,
      "step": 15030
    },
    {
      "epoch": 1.8157672340939275,
      "grad_norm": 9.00806999206543,
      "learning_rate": 1.9818982585154956e-05,
      "loss": 1.0901,
      "step": 15040
    },
    {
      "epoch": 1.816974526137873,
      "grad_norm": 9.157861709594727,
      "learning_rate": 1.9798779748676715e-05,
      "loss": 1.217,
      "step": 15050
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 8.586831092834473,
      "learning_rate": 1.9778576912198473e-05,
      "loss": 0.8643,
      "step": 15060
    },
    {
      "epoch": 1.8193891102257638,
      "grad_norm": 24.749156951904297,
      "learning_rate": 1.9758374075720232e-05,
      "loss": 1.5968,
      "step": 15070
    },
    {
      "epoch": 1.8205964022697092,
      "grad_norm": 13.251930236816406,
      "learning_rate": 1.973817123924199e-05,
      "loss": 0.7538,
      "step": 15080
    },
    {
      "epoch": 1.8218036943136546,
      "grad_norm": 3.628690004348755,
      "learning_rate": 1.971796840276375e-05,
      "loss": 0.7905,
      "step": 15090
    },
    {
      "epoch": 1.8230109863576,
      "grad_norm": 15.424034118652344,
      "learning_rate": 1.9697765566285507e-05,
      "loss": 1.1237,
      "step": 15100
    },
    {
      "epoch": 1.8242182784015455,
      "grad_norm": 41.127567291259766,
      "learning_rate": 1.9677562729807266e-05,
      "loss": 0.8517,
      "step": 15110
    },
    {
      "epoch": 1.825425570445491,
      "grad_norm": 8.259191513061523,
      "learning_rate": 1.9657359893329024e-05,
      "loss": 0.7457,
      "step": 15120
    },
    {
      "epoch": 1.8266328624894363,
      "grad_norm": 34.22703552246094,
      "learning_rate": 1.9637157056850783e-05,
      "loss": 0.8927,
      "step": 15130
    },
    {
      "epoch": 1.8278401545333818,
      "grad_norm": 14.537532806396484,
      "learning_rate": 1.961695422037254e-05,
      "loss": 1.1811,
      "step": 15140
    },
    {
      "epoch": 1.8290474465773272,
      "grad_norm": 18.039409637451172,
      "learning_rate": 1.95967513838943e-05,
      "loss": 0.6149,
      "step": 15150
    },
    {
      "epoch": 1.8302547386212726,
      "grad_norm": 14.426883697509766,
      "learning_rate": 1.9576548547416058e-05,
      "loss": 1.1558,
      "step": 15160
    },
    {
      "epoch": 1.831462030665218,
      "grad_norm": 22.146896362304688,
      "learning_rate": 1.9556345710937817e-05,
      "loss": 1.0684,
      "step": 15170
    },
    {
      "epoch": 1.8326693227091635,
      "grad_norm": 1.7266159057617188,
      "learning_rate": 1.9536142874459575e-05,
      "loss": 1.085,
      "step": 15180
    },
    {
      "epoch": 1.833876614753109,
      "grad_norm": 18.965375900268555,
      "learning_rate": 1.9515940037981333e-05,
      "loss": 1.1608,
      "step": 15190
    },
    {
      "epoch": 1.8350839067970544,
      "grad_norm": 1.117147445678711,
      "learning_rate": 1.9495737201503092e-05,
      "loss": 0.9596,
      "step": 15200
    },
    {
      "epoch": 1.8362911988409998,
      "grad_norm": 4.571306228637695,
      "learning_rate": 1.947553436502485e-05,
      "loss": 1.0703,
      "step": 15210
    },
    {
      "epoch": 1.8374984908849452,
      "grad_norm": 8.815680503845215,
      "learning_rate": 1.945533152854661e-05,
      "loss": 0.8597,
      "step": 15220
    },
    {
      "epoch": 1.8387057829288906,
      "grad_norm": 31.564912796020508,
      "learning_rate": 1.9435128692068367e-05,
      "loss": 1.0207,
      "step": 15230
    },
    {
      "epoch": 1.839913074972836,
      "grad_norm": 3.387814521789551,
      "learning_rate": 1.9414925855590126e-05,
      "loss": 1.2889,
      "step": 15240
    },
    {
      "epoch": 1.8411203670167815,
      "grad_norm": 12.443814277648926,
      "learning_rate": 1.9394723019111884e-05,
      "loss": 1.3323,
      "step": 15250
    },
    {
      "epoch": 1.842327659060727,
      "grad_norm": 24.316762924194336,
      "learning_rate": 1.9374520182633643e-05,
      "loss": 1.108,
      "step": 15260
    },
    {
      "epoch": 1.8435349511046724,
      "grad_norm": 9.995509147644043,
      "learning_rate": 1.93543173461554e-05,
      "loss": 1.0677,
      "step": 15270
    },
    {
      "epoch": 1.8447422431486178,
      "grad_norm": 16.66081428527832,
      "learning_rate": 1.933411450967716e-05,
      "loss": 1.1572,
      "step": 15280
    },
    {
      "epoch": 1.8459495351925632,
      "grad_norm": 8.504645347595215,
      "learning_rate": 1.9313911673198918e-05,
      "loss": 1.1419,
      "step": 15290
    },
    {
      "epoch": 1.8471568272365086,
      "grad_norm": 19.37603759765625,
      "learning_rate": 1.9293708836720677e-05,
      "loss": 0.7155,
      "step": 15300
    },
    {
      "epoch": 1.848364119280454,
      "grad_norm": 9.62389087677002,
      "learning_rate": 1.9273506000242435e-05,
      "loss": 1.0175,
      "step": 15310
    },
    {
      "epoch": 1.8495714113243995,
      "grad_norm": 7.583249568939209,
      "learning_rate": 1.9253303163764194e-05,
      "loss": 0.9666,
      "step": 15320
    },
    {
      "epoch": 1.850778703368345,
      "grad_norm": 15.249643325805664,
      "learning_rate": 1.9233100327285952e-05,
      "loss": 0.9015,
      "step": 15330
    },
    {
      "epoch": 1.8519859954122904,
      "grad_norm": 9.825966835021973,
      "learning_rate": 1.921289749080771e-05,
      "loss": 0.8634,
      "step": 15340
    },
    {
      "epoch": 1.8531932874562358,
      "grad_norm": 9.752689361572266,
      "learning_rate": 1.919269465432947e-05,
      "loss": 1.0168,
      "step": 15350
    },
    {
      "epoch": 1.8544005795001812,
      "grad_norm": 12.141910552978516,
      "learning_rate": 1.9172491817851228e-05,
      "loss": 1.0659,
      "step": 15360
    },
    {
      "epoch": 1.8556078715441267,
      "grad_norm": 9.747432708740234,
      "learning_rate": 1.9152288981372986e-05,
      "loss": 0.7823,
      "step": 15370
    },
    {
      "epoch": 1.856815163588072,
      "grad_norm": 3.2435388565063477,
      "learning_rate": 1.9132086144894745e-05,
      "loss": 0.7164,
      "step": 15380
    },
    {
      "epoch": 1.8580224556320175,
      "grad_norm": 13.574966430664062,
      "learning_rate": 1.9111883308416503e-05,
      "loss": 0.8597,
      "step": 15390
    },
    {
      "epoch": 1.859229747675963,
      "grad_norm": 9.802249908447266,
      "learning_rate": 1.909168047193826e-05,
      "loss": 0.8982,
      "step": 15400
    },
    {
      "epoch": 1.8604370397199084,
      "grad_norm": 11.184064865112305,
      "learning_rate": 1.907147763546002e-05,
      "loss": 0.6676,
      "step": 15410
    },
    {
      "epoch": 1.8616443317638538,
      "grad_norm": 7.31282377243042,
      "learning_rate": 1.905127479898178e-05,
      "loss": 1.0297,
      "step": 15420
    },
    {
      "epoch": 1.8628516238077992,
      "grad_norm": 7.036930084228516,
      "learning_rate": 1.9031071962503537e-05,
      "loss": 1.0556,
      "step": 15430
    },
    {
      "epoch": 1.8640589158517447,
      "grad_norm": 8.212656021118164,
      "learning_rate": 1.9010869126025292e-05,
      "loss": 1.2469,
      "step": 15440
    },
    {
      "epoch": 1.86526620789569,
      "grad_norm": 5.839860439300537,
      "learning_rate": 1.8990666289547054e-05,
      "loss": 0.8483,
      "step": 15450
    },
    {
      "epoch": 1.8664734999396355,
      "grad_norm": 23.591678619384766,
      "learning_rate": 1.8970463453068812e-05,
      "loss": 0.8395,
      "step": 15460
    },
    {
      "epoch": 1.867680791983581,
      "grad_norm": 16.95004653930664,
      "learning_rate": 1.895026061659057e-05,
      "loss": 1.0615,
      "step": 15470
    },
    {
      "epoch": 1.8688880840275264,
      "grad_norm": 7.747603416442871,
      "learning_rate": 1.893005778011233e-05,
      "loss": 0.7664,
      "step": 15480
    },
    {
      "epoch": 1.8700953760714718,
      "grad_norm": 38.45075988769531,
      "learning_rate": 1.8909854943634088e-05,
      "loss": 0.8129,
      "step": 15490
    },
    {
      "epoch": 1.8713026681154172,
      "grad_norm": 30.572542190551758,
      "learning_rate": 1.8889652107155846e-05,
      "loss": 0.92,
      "step": 15500
    },
    {
      "epoch": 1.8725099601593627,
      "grad_norm": 8.172870635986328,
      "learning_rate": 1.88694492706776e-05,
      "loss": 0.8685,
      "step": 15510
    },
    {
      "epoch": 1.873717252203308,
      "grad_norm": 14.328533172607422,
      "learning_rate": 1.8849246434199363e-05,
      "loss": 1.1772,
      "step": 15520
    },
    {
      "epoch": 1.8749245442472535,
      "grad_norm": 49.916194915771484,
      "learning_rate": 1.882904359772112e-05,
      "loss": 1.0234,
      "step": 15530
    },
    {
      "epoch": 1.876131836291199,
      "grad_norm": 38.677650451660156,
      "learning_rate": 1.880884076124288e-05,
      "loss": 1.1969,
      "step": 15540
    },
    {
      "epoch": 1.8773391283351444,
      "grad_norm": 12.645341873168945,
      "learning_rate": 1.878863792476464e-05,
      "loss": 1.1452,
      "step": 15550
    },
    {
      "epoch": 1.8785464203790898,
      "grad_norm": 8.381092071533203,
      "learning_rate": 1.8768435088286397e-05,
      "loss": 0.909,
      "step": 15560
    },
    {
      "epoch": 1.8797537124230352,
      "grad_norm": 5.332622051239014,
      "learning_rate": 1.8748232251808152e-05,
      "loss": 0.9303,
      "step": 15570
    },
    {
      "epoch": 1.8809610044669807,
      "grad_norm": 7.114144802093506,
      "learning_rate": 1.872802941532991e-05,
      "loss": 1.2515,
      "step": 15580
    },
    {
      "epoch": 1.882168296510926,
      "grad_norm": 13.727352142333984,
      "learning_rate": 1.8707826578851672e-05,
      "loss": 0.7442,
      "step": 15590
    },
    {
      "epoch": 1.8833755885548715,
      "grad_norm": 2.0718250274658203,
      "learning_rate": 1.868762374237343e-05,
      "loss": 1.0298,
      "step": 15600
    },
    {
      "epoch": 1.884582880598817,
      "grad_norm": 13.235681533813477,
      "learning_rate": 1.866742090589519e-05,
      "loss": 1.0452,
      "step": 15610
    },
    {
      "epoch": 1.8857901726427624,
      "grad_norm": 8.784310340881348,
      "learning_rate": 1.8647218069416948e-05,
      "loss": 1.0794,
      "step": 15620
    },
    {
      "epoch": 1.8869974646867078,
      "grad_norm": 6.664924144744873,
      "learning_rate": 1.8627015232938706e-05,
      "loss": 0.5552,
      "step": 15630
    },
    {
      "epoch": 1.8882047567306532,
      "grad_norm": 17.160789489746094,
      "learning_rate": 1.860681239646046e-05,
      "loss": 1.1018,
      "step": 15640
    },
    {
      "epoch": 1.8894120487745987,
      "grad_norm": 6.670936107635498,
      "learning_rate": 1.8586609559982223e-05,
      "loss": 0.8528,
      "step": 15650
    },
    {
      "epoch": 1.890619340818544,
      "grad_norm": 1.7814059257507324,
      "learning_rate": 1.8566406723503982e-05,
      "loss": 0.8007,
      "step": 15660
    },
    {
      "epoch": 1.8918266328624895,
      "grad_norm": 10.578251838684082,
      "learning_rate": 1.854620388702574e-05,
      "loss": 0.858,
      "step": 15670
    },
    {
      "epoch": 1.893033924906435,
      "grad_norm": 8.487356185913086,
      "learning_rate": 1.85260010505475e-05,
      "loss": 0.8423,
      "step": 15680
    },
    {
      "epoch": 1.8942412169503804,
      "grad_norm": 31.545780181884766,
      "learning_rate": 1.8505798214069257e-05,
      "loss": 1.2669,
      "step": 15690
    },
    {
      "epoch": 1.8954485089943258,
      "grad_norm": 64.44950103759766,
      "learning_rate": 1.8485595377591012e-05,
      "loss": 1.0467,
      "step": 15700
    },
    {
      "epoch": 1.8966558010382712,
      "grad_norm": 17.450794219970703,
      "learning_rate": 1.846539254111277e-05,
      "loss": 1.1902,
      "step": 15710
    },
    {
      "epoch": 1.8978630930822167,
      "grad_norm": 15.477787017822266,
      "learning_rate": 1.8445189704634533e-05,
      "loss": 0.7153,
      "step": 15720
    },
    {
      "epoch": 1.899070385126162,
      "grad_norm": 15.375127792358398,
      "learning_rate": 1.842498686815629e-05,
      "loss": 0.9619,
      "step": 15730
    },
    {
      "epoch": 1.9002776771701075,
      "grad_norm": 1.8868755102157593,
      "learning_rate": 1.840478403167805e-05,
      "loss": 1.1968,
      "step": 15740
    },
    {
      "epoch": 1.901484969214053,
      "grad_norm": 5.271520137786865,
      "learning_rate": 1.8384581195199808e-05,
      "loss": 1.0301,
      "step": 15750
    },
    {
      "epoch": 1.9026922612579984,
      "grad_norm": 8.181119918823242,
      "learning_rate": 1.8364378358721567e-05,
      "loss": 0.9733,
      "step": 15760
    },
    {
      "epoch": 1.9038995533019438,
      "grad_norm": 35.83622741699219,
      "learning_rate": 1.834417552224332e-05,
      "loss": 0.7553,
      "step": 15770
    },
    {
      "epoch": 1.9051068453458893,
      "grad_norm": 2.8831772804260254,
      "learning_rate": 1.832397268576508e-05,
      "loss": 1.0797,
      "step": 15780
    },
    {
      "epoch": 1.9063141373898347,
      "grad_norm": 10.602365493774414,
      "learning_rate": 1.8303769849286842e-05,
      "loss": 1.0065,
      "step": 15790
    },
    {
      "epoch": 1.9075214294337801,
      "grad_norm": 17.860260009765625,
      "learning_rate": 1.82835670128086e-05,
      "loss": 0.9259,
      "step": 15800
    },
    {
      "epoch": 1.9087287214777255,
      "grad_norm": 14.498353958129883,
      "learning_rate": 1.826336417633036e-05,
      "loss": 0.9713,
      "step": 15810
    },
    {
      "epoch": 1.909936013521671,
      "grad_norm": 9.800497055053711,
      "learning_rate": 1.8243161339852117e-05,
      "loss": 0.809,
      "step": 15820
    },
    {
      "epoch": 1.9111433055656164,
      "grad_norm": 2.961550712585449,
      "learning_rate": 1.8222958503373876e-05,
      "loss": 0.6828,
      "step": 15830
    },
    {
      "epoch": 1.9123505976095618,
      "grad_norm": 17.58814239501953,
      "learning_rate": 1.820275566689563e-05,
      "loss": 0.8749,
      "step": 15840
    },
    {
      "epoch": 1.9135578896535073,
      "grad_norm": 21.863779067993164,
      "learning_rate": 1.818255283041739e-05,
      "loss": 0.7549,
      "step": 15850
    },
    {
      "epoch": 1.9147651816974527,
      "grad_norm": 1.5730432271957397,
      "learning_rate": 1.816234999393915e-05,
      "loss": 1.2327,
      "step": 15860
    },
    {
      "epoch": 1.9159724737413981,
      "grad_norm": 8.717631340026855,
      "learning_rate": 1.814214715746091e-05,
      "loss": 1.1099,
      "step": 15870
    },
    {
      "epoch": 1.9171797657853435,
      "grad_norm": 15.070160865783691,
      "learning_rate": 1.8121944320982668e-05,
      "loss": 1.0964,
      "step": 15880
    },
    {
      "epoch": 1.918387057829289,
      "grad_norm": 16.245752334594727,
      "learning_rate": 1.8101741484504427e-05,
      "loss": 1.2221,
      "step": 15890
    },
    {
      "epoch": 1.9195943498732344,
      "grad_norm": 11.823589324951172,
      "learning_rate": 1.8081538648026182e-05,
      "loss": 1.2433,
      "step": 15900
    },
    {
      "epoch": 1.9208016419171798,
      "grad_norm": 2.965940475463867,
      "learning_rate": 1.806133581154794e-05,
      "loss": 1.1141,
      "step": 15910
    },
    {
      "epoch": 1.9220089339611253,
      "grad_norm": 2.771385431289673,
      "learning_rate": 1.80411329750697e-05,
      "loss": 1.0569,
      "step": 15920
    },
    {
      "epoch": 1.9232162260050707,
      "grad_norm": 14.863994598388672,
      "learning_rate": 1.802093013859146e-05,
      "loss": 1.1436,
      "step": 15930
    },
    {
      "epoch": 1.9244235180490161,
      "grad_norm": 18.9285831451416,
      "learning_rate": 1.800072730211322e-05,
      "loss": 0.5884,
      "step": 15940
    },
    {
      "epoch": 1.9256308100929616,
      "grad_norm": 16.86488151550293,
      "learning_rate": 1.7980524465634978e-05,
      "loss": 1.543,
      "step": 15950
    },
    {
      "epoch": 1.926838102136907,
      "grad_norm": 9.212276458740234,
      "learning_rate": 1.7960321629156736e-05,
      "loss": 0.8608,
      "step": 15960
    },
    {
      "epoch": 1.9280453941808524,
      "grad_norm": 7.0313873291015625,
      "learning_rate": 1.794011879267849e-05,
      "loss": 0.9468,
      "step": 15970
    },
    {
      "epoch": 1.9292526862247978,
      "grad_norm": 10.263205528259277,
      "learning_rate": 1.791991595620025e-05,
      "loss": 1.0929,
      "step": 15980
    },
    {
      "epoch": 1.9304599782687433,
      "grad_norm": 10.761801719665527,
      "learning_rate": 1.7899713119722008e-05,
      "loss": 0.7818,
      "step": 15990
    },
    {
      "epoch": 1.9316672703126887,
      "grad_norm": 28.14299964904785,
      "learning_rate": 1.787951028324377e-05,
      "loss": 0.89,
      "step": 16000
    },
    {
      "epoch": 1.9328745623566341,
      "grad_norm": 29.4427547454834,
      "learning_rate": 1.785930744676553e-05,
      "loss": 0.8728,
      "step": 16010
    },
    {
      "epoch": 1.9340818544005796,
      "grad_norm": 22.582801818847656,
      "learning_rate": 1.7839104610287287e-05,
      "loss": 1.1779,
      "step": 16020
    },
    {
      "epoch": 1.935289146444525,
      "grad_norm": 11.004524230957031,
      "learning_rate": 1.7818901773809042e-05,
      "loss": 0.4973,
      "step": 16030
    },
    {
      "epoch": 1.9364964384884704,
      "grad_norm": 14.67165756225586,
      "learning_rate": 1.77986989373308e-05,
      "loss": 1.2385,
      "step": 16040
    },
    {
      "epoch": 1.9377037305324158,
      "grad_norm": 11.969593048095703,
      "learning_rate": 1.777849610085256e-05,
      "loss": 1.4362,
      "step": 16050
    },
    {
      "epoch": 1.9389110225763613,
      "grad_norm": 4.806218147277832,
      "learning_rate": 1.7758293264374317e-05,
      "loss": 0.5793,
      "step": 16060
    },
    {
      "epoch": 1.9401183146203067,
      "grad_norm": 1.802122950553894,
      "learning_rate": 1.773809042789608e-05,
      "loss": 0.8938,
      "step": 16070
    },
    {
      "epoch": 1.9413256066642521,
      "grad_norm": 12.46435832977295,
      "learning_rate": 1.7717887591417838e-05,
      "loss": 0.934,
      "step": 16080
    },
    {
      "epoch": 1.9425328987081976,
      "grad_norm": 14.25204086303711,
      "learning_rate": 1.7697684754939596e-05,
      "loss": 1.2363,
      "step": 16090
    },
    {
      "epoch": 1.943740190752143,
      "grad_norm": 2.8684699535369873,
      "learning_rate": 1.767748191846135e-05,
      "loss": 0.8657,
      "step": 16100
    },
    {
      "epoch": 1.9449474827960884,
      "grad_norm": 44.31208419799805,
      "learning_rate": 1.765727908198311e-05,
      "loss": 1.2392,
      "step": 16110
    },
    {
      "epoch": 1.9461547748400339,
      "grad_norm": 18.238449096679688,
      "learning_rate": 1.7637076245504868e-05,
      "loss": 0.6768,
      "step": 16120
    },
    {
      "epoch": 1.9473620668839793,
      "grad_norm": 8.362272262573242,
      "learning_rate": 1.7616873409026627e-05,
      "loss": 1.1655,
      "step": 16130
    },
    {
      "epoch": 1.9485693589279247,
      "grad_norm": 18.08146095275879,
      "learning_rate": 1.759667057254839e-05,
      "loss": 1.3344,
      "step": 16140
    },
    {
      "epoch": 1.9497766509718701,
      "grad_norm": 16.039348602294922,
      "learning_rate": 1.7576467736070147e-05,
      "loss": 0.7806,
      "step": 16150
    },
    {
      "epoch": 1.9509839430158156,
      "grad_norm": 14.318255424499512,
      "learning_rate": 1.7556264899591902e-05,
      "loss": 0.8725,
      "step": 16160
    },
    {
      "epoch": 1.952191235059761,
      "grad_norm": 58.32987976074219,
      "learning_rate": 1.753606206311366e-05,
      "loss": 0.807,
      "step": 16170
    },
    {
      "epoch": 1.9533985271037064,
      "grad_norm": 29.41402244567871,
      "learning_rate": 1.751585922663542e-05,
      "loss": 1.1398,
      "step": 16180
    },
    {
      "epoch": 1.9546058191476519,
      "grad_norm": 16.453357696533203,
      "learning_rate": 1.7495656390157178e-05,
      "loss": 0.8823,
      "step": 16190
    },
    {
      "epoch": 1.9558131111915973,
      "grad_norm": 14.172119140625,
      "learning_rate": 1.7475453553678936e-05,
      "loss": 1.1412,
      "step": 16200
    },
    {
      "epoch": 1.9570204032355427,
      "grad_norm": 9.576529502868652,
      "learning_rate": 1.7455250717200698e-05,
      "loss": 0.6894,
      "step": 16210
    },
    {
      "epoch": 1.9582276952794881,
      "grad_norm": 17.851757049560547,
      "learning_rate": 1.7435047880722456e-05,
      "loss": 0.8232,
      "step": 16220
    },
    {
      "epoch": 1.9594349873234336,
      "grad_norm": 27.916576385498047,
      "learning_rate": 1.741484504424421e-05,
      "loss": 0.981,
      "step": 16230
    },
    {
      "epoch": 1.960642279367379,
      "grad_norm": 29.04560661315918,
      "learning_rate": 1.739464220776597e-05,
      "loss": 1.1359,
      "step": 16240
    },
    {
      "epoch": 1.9618495714113244,
      "grad_norm": 41.001258850097656,
      "learning_rate": 1.737443937128773e-05,
      "loss": 0.6377,
      "step": 16250
    },
    {
      "epoch": 1.9630568634552699,
      "grad_norm": 15.211069107055664,
      "learning_rate": 1.7354236534809487e-05,
      "loss": 0.8669,
      "step": 16260
    },
    {
      "epoch": 1.9642641554992153,
      "grad_norm": 15.959657669067383,
      "learning_rate": 1.733403369833125e-05,
      "loss": 0.3921,
      "step": 16270
    },
    {
      "epoch": 1.9654714475431607,
      "grad_norm": 34.17222595214844,
      "learning_rate": 1.7313830861853007e-05,
      "loss": 0.9525,
      "step": 16280
    },
    {
      "epoch": 1.9666787395871062,
      "grad_norm": 24.3832950592041,
      "learning_rate": 1.7293628025374762e-05,
      "loss": 1.5535,
      "step": 16290
    },
    {
      "epoch": 1.9678860316310516,
      "grad_norm": 18.90343475341797,
      "learning_rate": 1.727342518889652e-05,
      "loss": 1.0163,
      "step": 16300
    },
    {
      "epoch": 1.969093323674997,
      "grad_norm": 20.694841384887695,
      "learning_rate": 1.725322235241828e-05,
      "loss": 1.2427,
      "step": 16310
    },
    {
      "epoch": 1.9703006157189424,
      "grad_norm": 4.955702781677246,
      "learning_rate": 1.7233019515940038e-05,
      "loss": 0.7236,
      "step": 16320
    },
    {
      "epoch": 1.9715079077628879,
      "grad_norm": 11.15919017791748,
      "learning_rate": 1.7212816679461796e-05,
      "loss": 1.2031,
      "step": 16330
    },
    {
      "epoch": 1.9727151998068333,
      "grad_norm": 7.393800258636475,
      "learning_rate": 1.7192613842983558e-05,
      "loss": 0.9235,
      "step": 16340
    },
    {
      "epoch": 1.9739224918507787,
      "grad_norm": 34.80161666870117,
      "learning_rate": 1.7172411006505317e-05,
      "loss": 0.9562,
      "step": 16350
    },
    {
      "epoch": 1.9751297838947242,
      "grad_norm": 5.257025241851807,
      "learning_rate": 1.715220817002707e-05,
      "loss": 0.8025,
      "step": 16360
    },
    {
      "epoch": 1.9763370759386696,
      "grad_norm": 13.087923049926758,
      "learning_rate": 1.713200533354883e-05,
      "loss": 0.9605,
      "step": 16370
    },
    {
      "epoch": 1.977544367982615,
      "grad_norm": 4.16162109375,
      "learning_rate": 1.711180249707059e-05,
      "loss": 0.6636,
      "step": 16380
    },
    {
      "epoch": 1.9787516600265604,
      "grad_norm": 13.018779754638672,
      "learning_rate": 1.7091599660592347e-05,
      "loss": 0.8653,
      "step": 16390
    },
    {
      "epoch": 1.9799589520705059,
      "grad_norm": 22.189651489257812,
      "learning_rate": 1.7071396824114106e-05,
      "loss": 0.9553,
      "step": 16400
    },
    {
      "epoch": 1.9811662441144513,
      "grad_norm": 26.792625427246094,
      "learning_rate": 1.7051193987635867e-05,
      "loss": 0.8027,
      "step": 16410
    },
    {
      "epoch": 1.9823735361583967,
      "grad_norm": 19.72887420654297,
      "learning_rate": 1.7030991151157626e-05,
      "loss": 1.0377,
      "step": 16420
    },
    {
      "epoch": 1.9835808282023422,
      "grad_norm": 10.871551513671875,
      "learning_rate": 1.701078831467938e-05,
      "loss": 1.1592,
      "step": 16430
    },
    {
      "epoch": 1.9847881202462876,
      "grad_norm": 11.354755401611328,
      "learning_rate": 1.699058547820114e-05,
      "loss": 1.057,
      "step": 16440
    },
    {
      "epoch": 1.985995412290233,
      "grad_norm": 11.592129707336426,
      "learning_rate": 1.6970382641722898e-05,
      "loss": 0.9517,
      "step": 16450
    },
    {
      "epoch": 1.9872027043341784,
      "grad_norm": 9.2820463180542,
      "learning_rate": 1.6950179805244656e-05,
      "loss": 1.0823,
      "step": 16460
    },
    {
      "epoch": 1.9884099963781239,
      "grad_norm": 12.326805114746094,
      "learning_rate": 1.6929976968766415e-05,
      "loss": 1.053,
      "step": 16470
    },
    {
      "epoch": 1.9896172884220693,
      "grad_norm": 17.970630645751953,
      "learning_rate": 1.6909774132288177e-05,
      "loss": 0.816,
      "step": 16480
    },
    {
      "epoch": 1.9908245804660147,
      "grad_norm": 13.872296333312988,
      "learning_rate": 1.6889571295809932e-05,
      "loss": 1.1822,
      "step": 16490
    },
    {
      "epoch": 1.9920318725099602,
      "grad_norm": 18.83074378967285,
      "learning_rate": 1.686936845933169e-05,
      "loss": 1.0723,
      "step": 16500
    },
    {
      "epoch": 1.9932391645539056,
      "grad_norm": 17.088010787963867,
      "learning_rate": 1.684916562285345e-05,
      "loss": 1.2013,
      "step": 16510
    },
    {
      "epoch": 1.994446456597851,
      "grad_norm": 12.333873748779297,
      "learning_rate": 1.6828962786375207e-05,
      "loss": 0.8458,
      "step": 16520
    },
    {
      "epoch": 1.9956537486417965,
      "grad_norm": 4.257887363433838,
      "learning_rate": 1.6808759949896966e-05,
      "loss": 0.9528,
      "step": 16530
    },
    {
      "epoch": 1.9968610406857419,
      "grad_norm": 28.225923538208008,
      "learning_rate": 1.6788557113418724e-05,
      "loss": 0.9796,
      "step": 16540
    },
    {
      "epoch": 1.9980683327296873,
      "grad_norm": 19.239513397216797,
      "learning_rate": 1.6768354276940486e-05,
      "loss": 0.6476,
      "step": 16550
    },
    {
      "epoch": 1.9992756247736327,
      "grad_norm": 11.48581600189209,
      "learning_rate": 1.674815144046224e-05,
      "loss": 1.121,
      "step": 16560
    },
    {
      "epoch": 2.000482916817578,
      "grad_norm": 11.204534530639648,
      "learning_rate": 1.6727948603984e-05,
      "loss": 1.0597,
      "step": 16570
    },
    {
      "epoch": 2.0016902088615236,
      "grad_norm": 1.9322247505187988,
      "learning_rate": 1.6707745767505758e-05,
      "loss": 0.615,
      "step": 16580
    },
    {
      "epoch": 2.002897500905469,
      "grad_norm": 2.848154067993164,
      "learning_rate": 1.6687542931027517e-05,
      "loss": 0.7143,
      "step": 16590
    },
    {
      "epoch": 2.0041047929494145,
      "grad_norm": 1.4145792722702026,
      "learning_rate": 1.6667340094549275e-05,
      "loss": 0.7441,
      "step": 16600
    },
    {
      "epoch": 2.00531208499336,
      "grad_norm": 17.27072525024414,
      "learning_rate": 1.6647137258071034e-05,
      "loss": 1.0817,
      "step": 16610
    },
    {
      "epoch": 2.0065193770373053,
      "grad_norm": 13.437647819519043,
      "learning_rate": 1.6626934421592792e-05,
      "loss": 0.7756,
      "step": 16620
    },
    {
      "epoch": 2.0077266690812507,
      "grad_norm": 11.10927677154541,
      "learning_rate": 1.660673158511455e-05,
      "loss": 0.9717,
      "step": 16630
    },
    {
      "epoch": 2.008933961125196,
      "grad_norm": 2.261584997177124,
      "learning_rate": 1.658652874863631e-05,
      "loss": 0.5258,
      "step": 16640
    },
    {
      "epoch": 2.0101412531691416,
      "grad_norm": 4.199213981628418,
      "learning_rate": 1.6566325912158067e-05,
      "loss": 0.9239,
      "step": 16650
    },
    {
      "epoch": 2.011348545213087,
      "grad_norm": 5.18364953994751,
      "learning_rate": 1.6546123075679826e-05,
      "loss": 0.5057,
      "step": 16660
    },
    {
      "epoch": 2.0125558372570325,
      "grad_norm": 14.735456466674805,
      "learning_rate": 1.6525920239201584e-05,
      "loss": 0.7957,
      "step": 16670
    },
    {
      "epoch": 2.013763129300978,
      "grad_norm": 15.752409934997559,
      "learning_rate": 1.6505717402723343e-05,
      "loss": 0.8373,
      "step": 16680
    },
    {
      "epoch": 2.0149704213449233,
      "grad_norm": 12.57786750793457,
      "learning_rate": 1.64855145662451e-05,
      "loss": 0.9032,
      "step": 16690
    },
    {
      "epoch": 2.0161777133888688,
      "grad_norm": 12.0497465133667,
      "learning_rate": 1.646531172976686e-05,
      "loss": 0.966,
      "step": 16700
    },
    {
      "epoch": 2.017385005432814,
      "grad_norm": 20.355249404907227,
      "learning_rate": 1.6445108893288618e-05,
      "loss": 0.5467,
      "step": 16710
    },
    {
      "epoch": 2.0185922974767596,
      "grad_norm": 10.236241340637207,
      "learning_rate": 1.6424906056810377e-05,
      "loss": 0.8909,
      "step": 16720
    },
    {
      "epoch": 2.019799589520705,
      "grad_norm": 6.772604465484619,
      "learning_rate": 1.6404703220332135e-05,
      "loss": 0.6907,
      "step": 16730
    },
    {
      "epoch": 2.0210068815646505,
      "grad_norm": 22.87328338623047,
      "learning_rate": 1.6384500383853894e-05,
      "loss": 0.7488,
      "step": 16740
    },
    {
      "epoch": 2.022214173608596,
      "grad_norm": 30.42987823486328,
      "learning_rate": 1.6364297547375652e-05,
      "loss": 0.4588,
      "step": 16750
    },
    {
      "epoch": 2.0234214656525413,
      "grad_norm": 19.17439079284668,
      "learning_rate": 1.634409471089741e-05,
      "loss": 0.8404,
      "step": 16760
    },
    {
      "epoch": 2.0246287576964868,
      "grad_norm": 17.143516540527344,
      "learning_rate": 1.632389187441917e-05,
      "loss": 0.9573,
      "step": 16770
    },
    {
      "epoch": 2.025836049740432,
      "grad_norm": 14.465167999267578,
      "learning_rate": 1.6303689037940928e-05,
      "loss": 0.894,
      "step": 16780
    },
    {
      "epoch": 2.0270433417843776,
      "grad_norm": 18.959985733032227,
      "learning_rate": 1.6283486201462686e-05,
      "loss": 0.76,
      "step": 16790
    },
    {
      "epoch": 2.028250633828323,
      "grad_norm": 23.79229736328125,
      "learning_rate": 1.6263283364984445e-05,
      "loss": 1.0106,
      "step": 16800
    },
    {
      "epoch": 2.0294579258722685,
      "grad_norm": 25.360883712768555,
      "learning_rate": 1.6243080528506203e-05,
      "loss": 1.0084,
      "step": 16810
    },
    {
      "epoch": 2.030665217916214,
      "grad_norm": 2.3727517127990723,
      "learning_rate": 1.622287769202796e-05,
      "loss": 1.0392,
      "step": 16820
    },
    {
      "epoch": 2.0318725099601593,
      "grad_norm": 12.92117977142334,
      "learning_rate": 1.620267485554972e-05,
      "loss": 0.6899,
      "step": 16830
    },
    {
      "epoch": 2.0330798020041048,
      "grad_norm": 10.590582847595215,
      "learning_rate": 1.618247201907148e-05,
      "loss": 0.8649,
      "step": 16840
    },
    {
      "epoch": 2.03428709404805,
      "grad_norm": 12.96107006072998,
      "learning_rate": 1.6162269182593237e-05,
      "loss": 0.5845,
      "step": 16850
    },
    {
      "epoch": 2.0354943860919956,
      "grad_norm": 17.964584350585938,
      "learning_rate": 1.6142066346114995e-05,
      "loss": 1.1703,
      "step": 16860
    },
    {
      "epoch": 2.036701678135941,
      "grad_norm": 19.94473648071289,
      "learning_rate": 1.6121863509636754e-05,
      "loss": 0.9079,
      "step": 16870
    },
    {
      "epoch": 2.0379089701798865,
      "grad_norm": 14.336570739746094,
      "learning_rate": 1.6101660673158512e-05,
      "loss": 1.0174,
      "step": 16880
    },
    {
      "epoch": 2.039116262223832,
      "grad_norm": 21.329273223876953,
      "learning_rate": 1.608145783668027e-05,
      "loss": 0.5631,
      "step": 16890
    },
    {
      "epoch": 2.0403235542677773,
      "grad_norm": 4.851711273193359,
      "learning_rate": 1.606125500020203e-05,
      "loss": 1.05,
      "step": 16900
    },
    {
      "epoch": 2.0415308463117228,
      "grad_norm": 15.166211128234863,
      "learning_rate": 1.6041052163723788e-05,
      "loss": 0.9026,
      "step": 16910
    },
    {
      "epoch": 2.042738138355668,
      "grad_norm": 15.620976448059082,
      "learning_rate": 1.6020849327245546e-05,
      "loss": 0.8476,
      "step": 16920
    },
    {
      "epoch": 2.0439454303996136,
      "grad_norm": 10.178659439086914,
      "learning_rate": 1.6000646490767305e-05,
      "loss": 0.7239,
      "step": 16930
    },
    {
      "epoch": 2.045152722443559,
      "grad_norm": 40.63846969604492,
      "learning_rate": 1.5980443654289063e-05,
      "loss": 0.8526,
      "step": 16940
    },
    {
      "epoch": 2.0463600144875045,
      "grad_norm": 27.071401596069336,
      "learning_rate": 1.596024081781082e-05,
      "loss": 1.0756,
      "step": 16950
    },
    {
      "epoch": 2.04756730653145,
      "grad_norm": 21.692148208618164,
      "learning_rate": 1.594003798133258e-05,
      "loss": 0.7585,
      "step": 16960
    },
    {
      "epoch": 2.0487745985753953,
      "grad_norm": 3.2209410667419434,
      "learning_rate": 1.591983514485434e-05,
      "loss": 0.4755,
      "step": 16970
    },
    {
      "epoch": 2.0499818906193408,
      "grad_norm": 4.03912353515625,
      "learning_rate": 1.5899632308376097e-05,
      "loss": 0.7504,
      "step": 16980
    },
    {
      "epoch": 2.051189182663286,
      "grad_norm": 10.731917381286621,
      "learning_rate": 1.5879429471897856e-05,
      "loss": 0.653,
      "step": 16990
    },
    {
      "epoch": 2.0523964747072316,
      "grad_norm": 1.0930027961730957,
      "learning_rate": 1.5859226635419614e-05,
      "loss": 0.509,
      "step": 17000
    },
    {
      "epoch": 2.053603766751177,
      "grad_norm": 21.33726692199707,
      "learning_rate": 1.5839023798941373e-05,
      "loss": 1.0521,
      "step": 17010
    },
    {
      "epoch": 2.0548110587951225,
      "grad_norm": 33.06449508666992,
      "learning_rate": 1.581882096246313e-05,
      "loss": 1.034,
      "step": 17020
    },
    {
      "epoch": 2.056018350839068,
      "grad_norm": 5.3030171394348145,
      "learning_rate": 1.579861812598489e-05,
      "loss": 0.754,
      "step": 17030
    },
    {
      "epoch": 2.0572256428830134,
      "grad_norm": 17.76031494140625,
      "learning_rate": 1.5778415289506648e-05,
      "loss": 0.739,
      "step": 17040
    },
    {
      "epoch": 2.058432934926959,
      "grad_norm": 8.221938133239746,
      "learning_rate": 1.5758212453028406e-05,
      "loss": 0.7645,
      "step": 17050
    },
    {
      "epoch": 2.059640226970904,
      "grad_norm": 25.747190475463867,
      "learning_rate": 1.5738009616550165e-05,
      "loss": 1.1173,
      "step": 17060
    },
    {
      "epoch": 2.0608475190148496,
      "grad_norm": 7.1144256591796875,
      "learning_rate": 1.5717806780071923e-05,
      "loss": 0.8543,
      "step": 17070
    },
    {
      "epoch": 2.062054811058795,
      "grad_norm": 27.781005859375,
      "learning_rate": 1.5697603943593682e-05,
      "loss": 0.6702,
      "step": 17080
    },
    {
      "epoch": 2.0632621031027405,
      "grad_norm": 9.176952362060547,
      "learning_rate": 1.5677401107115437e-05,
      "loss": 0.8014,
      "step": 17090
    },
    {
      "epoch": 2.064469395146686,
      "grad_norm": 45.8448486328125,
      "learning_rate": 1.56571982706372e-05,
      "loss": 0.7133,
      "step": 17100
    },
    {
      "epoch": 2.0656766871906314,
      "grad_norm": 7.640994548797607,
      "learning_rate": 1.5636995434158957e-05,
      "loss": 0.6951,
      "step": 17110
    },
    {
      "epoch": 2.066883979234577,
      "grad_norm": 7.188880443572998,
      "learning_rate": 1.5616792597680716e-05,
      "loss": 0.6618,
      "step": 17120
    },
    {
      "epoch": 2.068091271278522,
      "grad_norm": 10.14450740814209,
      "learning_rate": 1.5596589761202474e-05,
      "loss": 0.8981,
      "step": 17130
    },
    {
      "epoch": 2.0692985633224676,
      "grad_norm": 21.559301376342773,
      "learning_rate": 1.5576386924724233e-05,
      "loss": 0.5758,
      "step": 17140
    },
    {
      "epoch": 2.070505855366413,
      "grad_norm": 1.229810118675232,
      "learning_rate": 1.555618408824599e-05,
      "loss": 0.5176,
      "step": 17150
    },
    {
      "epoch": 2.0717131474103585,
      "grad_norm": 4.552802562713623,
      "learning_rate": 1.5535981251767746e-05,
      "loss": 0.9215,
      "step": 17160
    },
    {
      "epoch": 2.072920439454304,
      "grad_norm": 54.189823150634766,
      "learning_rate": 1.5515778415289508e-05,
      "loss": 0.8437,
      "step": 17170
    },
    {
      "epoch": 2.0741277314982494,
      "grad_norm": 5.617227077484131,
      "learning_rate": 1.5495575578811267e-05,
      "loss": 1.0099,
      "step": 17180
    },
    {
      "epoch": 2.075335023542195,
      "grad_norm": 18.5819034576416,
      "learning_rate": 1.5475372742333025e-05,
      "loss": 0.7981,
      "step": 17190
    },
    {
      "epoch": 2.07654231558614,
      "grad_norm": 6.5086870193481445,
      "learning_rate": 1.5455169905854784e-05,
      "loss": 0.7354,
      "step": 17200
    },
    {
      "epoch": 2.0777496076300856,
      "grad_norm": 14.70653247833252,
      "learning_rate": 1.5434967069376542e-05,
      "loss": 0.8112,
      "step": 17210
    },
    {
      "epoch": 2.078956899674031,
      "grad_norm": 34.763126373291016,
      "learning_rate": 1.54147642328983e-05,
      "loss": 0.6528,
      "step": 17220
    },
    {
      "epoch": 2.0801641917179765,
      "grad_norm": 23.54616928100586,
      "learning_rate": 1.5394561396420056e-05,
      "loss": 0.9892,
      "step": 17230
    },
    {
      "epoch": 2.081371483761922,
      "grad_norm": 28.49018669128418,
      "learning_rate": 1.5374358559941817e-05,
      "loss": 0.79,
      "step": 17240
    },
    {
      "epoch": 2.0825787758058674,
      "grad_norm": 10.6107759475708,
      "learning_rate": 1.5354155723463576e-05,
      "loss": 0.9962,
      "step": 17250
    },
    {
      "epoch": 2.083786067849813,
      "grad_norm": 9.321063995361328,
      "learning_rate": 1.5333952886985334e-05,
      "loss": 0.6507,
      "step": 17260
    },
    {
      "epoch": 2.0849933598937582,
      "grad_norm": 7.027838706970215,
      "learning_rate": 1.5313750050507093e-05,
      "loss": 0.7281,
      "step": 17270
    },
    {
      "epoch": 2.0862006519377037,
      "grad_norm": 16.1539363861084,
      "learning_rate": 1.529354721402885e-05,
      "loss": 1.0747,
      "step": 17280
    },
    {
      "epoch": 2.087407943981649,
      "grad_norm": 13.686274528503418,
      "learning_rate": 1.5273344377550606e-05,
      "loss": 0.6267,
      "step": 17290
    },
    {
      "epoch": 2.0886152360255945,
      "grad_norm": 18.75580406188965,
      "learning_rate": 1.5253141541072367e-05,
      "loss": 0.7459,
      "step": 17300
    },
    {
      "epoch": 2.08982252806954,
      "grad_norm": 21.400331497192383,
      "learning_rate": 1.5232938704594127e-05,
      "loss": 1.0422,
      "step": 17310
    },
    {
      "epoch": 2.0910298201134854,
      "grad_norm": 23.63475227355957,
      "learning_rate": 1.5212735868115885e-05,
      "loss": 0.7057,
      "step": 17320
    },
    {
      "epoch": 2.092237112157431,
      "grad_norm": 21.91584014892578,
      "learning_rate": 1.5192533031637644e-05,
      "loss": 0.985,
      "step": 17330
    },
    {
      "epoch": 2.0934444042013762,
      "grad_norm": 15.997300148010254,
      "learning_rate": 1.51723301951594e-05,
      "loss": 1.0303,
      "step": 17340
    },
    {
      "epoch": 2.0946516962453217,
      "grad_norm": 8.867905616760254,
      "learning_rate": 1.5152127358681159e-05,
      "loss": 0.9498,
      "step": 17350
    },
    {
      "epoch": 2.095858988289267,
      "grad_norm": 10.700113296508789,
      "learning_rate": 1.5131924522202917e-05,
      "loss": 0.882,
      "step": 17360
    },
    {
      "epoch": 2.0970662803332125,
      "grad_norm": 18.750574111938477,
      "learning_rate": 1.5111721685724676e-05,
      "loss": 0.8452,
      "step": 17370
    },
    {
      "epoch": 2.098273572377158,
      "grad_norm": 7.178251266479492,
      "learning_rate": 1.5091518849246436e-05,
      "loss": 0.6039,
      "step": 17380
    },
    {
      "epoch": 2.0994808644211034,
      "grad_norm": 18.899789810180664,
      "learning_rate": 1.5071316012768195e-05,
      "loss": 0.7664,
      "step": 17390
    },
    {
      "epoch": 2.100688156465049,
      "grad_norm": 10.44929027557373,
      "learning_rate": 1.5051113176289951e-05,
      "loss": 0.7074,
      "step": 17400
    },
    {
      "epoch": 2.1018954485089942,
      "grad_norm": 8.029425621032715,
      "learning_rate": 1.503091033981171e-05,
      "loss": 0.5291,
      "step": 17410
    },
    {
      "epoch": 2.1031027405529397,
      "grad_norm": 20.69525909423828,
      "learning_rate": 1.5010707503333468e-05,
      "loss": 0.5599,
      "step": 17420
    },
    {
      "epoch": 2.104310032596885,
      "grad_norm": 18.676055908203125,
      "learning_rate": 1.4990504666855227e-05,
      "loss": 1.1684,
      "step": 17430
    },
    {
      "epoch": 2.1055173246408305,
      "grad_norm": 11.941705703735352,
      "learning_rate": 1.4970301830376984e-05,
      "loss": 1.2824,
      "step": 17440
    },
    {
      "epoch": 2.106724616684776,
      "grad_norm": 24.770526885986328,
      "learning_rate": 1.4950098993898745e-05,
      "loss": 0.7554,
      "step": 17450
    },
    {
      "epoch": 2.1079319087287214,
      "grad_norm": 7.657273292541504,
      "learning_rate": 1.4929896157420504e-05,
      "loss": 0.6569,
      "step": 17460
    },
    {
      "epoch": 2.109139200772667,
      "grad_norm": 4.229912281036377,
      "learning_rate": 1.490969332094226e-05,
      "loss": 0.6836,
      "step": 17470
    },
    {
      "epoch": 2.1103464928166122,
      "grad_norm": 10.534623146057129,
      "learning_rate": 1.488949048446402e-05,
      "loss": 0.5738,
      "step": 17480
    },
    {
      "epoch": 2.1115537848605577,
      "grad_norm": 49.16779708862305,
      "learning_rate": 1.4869287647985778e-05,
      "loss": 1.2383,
      "step": 17490
    },
    {
      "epoch": 2.112761076904503,
      "grad_norm": 13.703643798828125,
      "learning_rate": 1.4849084811507536e-05,
      "loss": 0.7886,
      "step": 17500
    },
    {
      "epoch": 2.1139683689484485,
      "grad_norm": 9.85037899017334,
      "learning_rate": 1.4828881975029293e-05,
      "loss": 0.794,
      "step": 17510
    },
    {
      "epoch": 2.115175660992394,
      "grad_norm": 7.779178619384766,
      "learning_rate": 1.4808679138551055e-05,
      "loss": 0.714,
      "step": 17520
    },
    {
      "epoch": 2.1163829530363394,
      "grad_norm": 15.147582054138184,
      "learning_rate": 1.4788476302072812e-05,
      "loss": 0.6536,
      "step": 17530
    },
    {
      "epoch": 2.117590245080285,
      "grad_norm": 34.97456741333008,
      "learning_rate": 1.476827346559457e-05,
      "loss": 0.7287,
      "step": 17540
    },
    {
      "epoch": 2.1187975371242302,
      "grad_norm": 11.190325736999512,
      "learning_rate": 1.4748070629116328e-05,
      "loss": 0.9307,
      "step": 17550
    },
    {
      "epoch": 2.1200048291681757,
      "grad_norm": 21.983604431152344,
      "learning_rate": 1.4727867792638087e-05,
      "loss": 0.5012,
      "step": 17560
    },
    {
      "epoch": 2.121212121212121,
      "grad_norm": 25.410675048828125,
      "learning_rate": 1.4707664956159844e-05,
      "loss": 0.7627,
      "step": 17570
    },
    {
      "epoch": 2.1224194132560665,
      "grad_norm": 17.690792083740234,
      "learning_rate": 1.4687462119681606e-05,
      "loss": 1.0516,
      "step": 17580
    },
    {
      "epoch": 2.123626705300012,
      "grad_norm": 3.994797706604004,
      "learning_rate": 1.4667259283203364e-05,
      "loss": 0.4136,
      "step": 17590
    },
    {
      "epoch": 2.1248339973439574,
      "grad_norm": 20.35289764404297,
      "learning_rate": 1.4647056446725121e-05,
      "loss": 0.8976,
      "step": 17600
    },
    {
      "epoch": 2.126041289387903,
      "grad_norm": 23.8492488861084,
      "learning_rate": 1.462685361024688e-05,
      "loss": 1.1226,
      "step": 17610
    },
    {
      "epoch": 2.1272485814318483,
      "grad_norm": 7.302371025085449,
      "learning_rate": 1.4606650773768638e-05,
      "loss": 0.6431,
      "step": 17620
    },
    {
      "epoch": 2.1284558734757937,
      "grad_norm": 15.020218849182129,
      "learning_rate": 1.4586447937290396e-05,
      "loss": 0.7472,
      "step": 17630
    },
    {
      "epoch": 2.129663165519739,
      "grad_norm": 8.64437484741211,
      "learning_rate": 1.4566245100812153e-05,
      "loss": 0.9156,
      "step": 17640
    },
    {
      "epoch": 2.1308704575636845,
      "grad_norm": 13.911958694458008,
      "learning_rate": 1.4546042264333915e-05,
      "loss": 1.0499,
      "step": 17650
    },
    {
      "epoch": 2.13207774960763,
      "grad_norm": 50.731258392333984,
      "learning_rate": 1.4525839427855672e-05,
      "loss": 0.8434,
      "step": 17660
    },
    {
      "epoch": 2.1332850416515754,
      "grad_norm": 1.5797107219696045,
      "learning_rate": 1.450563659137743e-05,
      "loss": 0.71,
      "step": 17670
    },
    {
      "epoch": 2.134492333695521,
      "grad_norm": 15.437755584716797,
      "learning_rate": 1.4485433754899189e-05,
      "loss": 1.162,
      "step": 17680
    },
    {
      "epoch": 2.1356996257394663,
      "grad_norm": 11.393067359924316,
      "learning_rate": 1.4465230918420947e-05,
      "loss": 0.6552,
      "step": 17690
    },
    {
      "epoch": 2.1369069177834117,
      "grad_norm": 4.2356767654418945,
      "learning_rate": 1.4445028081942704e-05,
      "loss": 0.8757,
      "step": 17700
    },
    {
      "epoch": 2.138114209827357,
      "grad_norm": 14.935529708862305,
      "learning_rate": 1.4424825245464462e-05,
      "loss": 1.1765,
      "step": 17710
    },
    {
      "epoch": 2.1393215018713025,
      "grad_norm": 23.062068939208984,
      "learning_rate": 1.4404622408986224e-05,
      "loss": 0.616,
      "step": 17720
    },
    {
      "epoch": 2.140528793915248,
      "grad_norm": 11.1925630569458,
      "learning_rate": 1.4384419572507981e-05,
      "loss": 0.58,
      "step": 17730
    },
    {
      "epoch": 2.1417360859591934,
      "grad_norm": 2.4594790935516357,
      "learning_rate": 1.436421673602974e-05,
      "loss": 0.7681,
      "step": 17740
    },
    {
      "epoch": 2.142943378003139,
      "grad_norm": 11.473057746887207,
      "learning_rate": 1.4344013899551498e-05,
      "loss": 0.4883,
      "step": 17750
    },
    {
      "epoch": 2.1441506700470843,
      "grad_norm": 13.688669204711914,
      "learning_rate": 1.4323811063073256e-05,
      "loss": 1.3637,
      "step": 17760
    },
    {
      "epoch": 2.1453579620910297,
      "grad_norm": 1.1208726167678833,
      "learning_rate": 1.4303608226595013e-05,
      "loss": 0.8282,
      "step": 17770
    },
    {
      "epoch": 2.146565254134975,
      "grad_norm": 37.75774383544922,
      "learning_rate": 1.4283405390116772e-05,
      "loss": 0.6075,
      "step": 17780
    },
    {
      "epoch": 2.1477725461789205,
      "grad_norm": 6.581281661987305,
      "learning_rate": 1.4263202553638534e-05,
      "loss": 1.1991,
      "step": 17790
    },
    {
      "epoch": 2.148979838222866,
      "grad_norm": 20.48369026184082,
      "learning_rate": 1.424299971716029e-05,
      "loss": 0.6466,
      "step": 17800
    },
    {
      "epoch": 2.1501871302668114,
      "grad_norm": 5.7696733474731445,
      "learning_rate": 1.4222796880682049e-05,
      "loss": 0.6233,
      "step": 17810
    },
    {
      "epoch": 2.151394422310757,
      "grad_norm": 5.172064781188965,
      "learning_rate": 1.4202594044203807e-05,
      "loss": 0.6058,
      "step": 17820
    },
    {
      "epoch": 2.1526017143547023,
      "grad_norm": 8.911687850952148,
      "learning_rate": 1.4182391207725566e-05,
      "loss": 1.0711,
      "step": 17830
    },
    {
      "epoch": 2.1538090063986477,
      "grad_norm": 6.636078834533691,
      "learning_rate": 1.4162188371247323e-05,
      "loss": 0.8102,
      "step": 17840
    },
    {
      "epoch": 2.155016298442593,
      "grad_norm": 39.97214889526367,
      "learning_rate": 1.4141985534769081e-05,
      "loss": 1.2365,
      "step": 17850
    },
    {
      "epoch": 2.1562235904865386,
      "grad_norm": 7.455310344696045,
      "learning_rate": 1.4121782698290841e-05,
      "loss": 0.7082,
      "step": 17860
    },
    {
      "epoch": 2.157430882530484,
      "grad_norm": 5.124485969543457,
      "learning_rate": 1.41015798618126e-05,
      "loss": 0.8274,
      "step": 17870
    },
    {
      "epoch": 2.1586381745744294,
      "grad_norm": 40.457088470458984,
      "learning_rate": 1.4081377025334358e-05,
      "loss": 0.8898,
      "step": 17880
    },
    {
      "epoch": 2.159845466618375,
      "grad_norm": 16.452823638916016,
      "learning_rate": 1.4061174188856117e-05,
      "loss": 0.9698,
      "step": 17890
    },
    {
      "epoch": 2.1610527586623203,
      "grad_norm": 17.67535400390625,
      "learning_rate": 1.4040971352377873e-05,
      "loss": 1.1113,
      "step": 17900
    },
    {
      "epoch": 2.1622600507062657,
      "grad_norm": 6.103643894195557,
      "learning_rate": 1.4020768515899632e-05,
      "loss": 0.5403,
      "step": 17910
    },
    {
      "epoch": 2.163467342750211,
      "grad_norm": 8.44028377532959,
      "learning_rate": 1.400056567942139e-05,
      "loss": 0.6406,
      "step": 17920
    },
    {
      "epoch": 2.1646746347941566,
      "grad_norm": 24.591442108154297,
      "learning_rate": 1.398036284294315e-05,
      "loss": 0.8049,
      "step": 17930
    },
    {
      "epoch": 2.165881926838102,
      "grad_norm": 45.004974365234375,
      "learning_rate": 1.3960160006464909e-05,
      "loss": 0.9196,
      "step": 17940
    },
    {
      "epoch": 2.1670892188820474,
      "grad_norm": 12.552878379821777,
      "learning_rate": 1.3939957169986667e-05,
      "loss": 0.5905,
      "step": 17950
    },
    {
      "epoch": 2.168296510925993,
      "grad_norm": 9.43107795715332,
      "learning_rate": 1.3919754333508426e-05,
      "loss": 0.7214,
      "step": 17960
    },
    {
      "epoch": 2.1695038029699383,
      "grad_norm": 25.628856658935547,
      "learning_rate": 1.3899551497030183e-05,
      "loss": 0.5164,
      "step": 17970
    },
    {
      "epoch": 2.1707110950138837,
      "grad_norm": 10.631377220153809,
      "learning_rate": 1.3879348660551941e-05,
      "loss": 0.6324,
      "step": 17980
    },
    {
      "epoch": 2.171918387057829,
      "grad_norm": 23.983469009399414,
      "learning_rate": 1.38591458240737e-05,
      "loss": 0.9438,
      "step": 17990
    },
    {
      "epoch": 2.1731256791017746,
      "grad_norm": 5.040485382080078,
      "learning_rate": 1.383894298759546e-05,
      "loss": 0.833,
      "step": 18000
    },
    {
      "epoch": 2.17433297114572,
      "grad_norm": 9.345149993896484,
      "learning_rate": 1.3818740151117218e-05,
      "loss": 0.8098,
      "step": 18010
    },
    {
      "epoch": 2.1755402631896654,
      "grad_norm": 9.675541877746582,
      "learning_rate": 1.3798537314638977e-05,
      "loss": 0.6758,
      "step": 18020
    },
    {
      "epoch": 2.176747555233611,
      "grad_norm": 1.7567671537399292,
      "learning_rate": 1.3778334478160734e-05,
      "loss": 0.615,
      "step": 18030
    },
    {
      "epoch": 2.1779548472775563,
      "grad_norm": 41.21391296386719,
      "learning_rate": 1.3758131641682492e-05,
      "loss": 0.7999,
      "step": 18040
    },
    {
      "epoch": 2.1791621393215017,
      "grad_norm": 9.60523509979248,
      "learning_rate": 1.373792880520425e-05,
      "loss": 0.897,
      "step": 18050
    },
    {
      "epoch": 2.180369431365447,
      "grad_norm": 12.622248649597168,
      "learning_rate": 1.3717725968726009e-05,
      "loss": 0.8239,
      "step": 18060
    },
    {
      "epoch": 2.1815767234093926,
      "grad_norm": 23.6464900970459,
      "learning_rate": 1.369752313224777e-05,
      "loss": 0.5567,
      "step": 18070
    },
    {
      "epoch": 2.182784015453338,
      "grad_norm": 4.276128768920898,
      "learning_rate": 1.3677320295769528e-05,
      "loss": 0.613,
      "step": 18080
    },
    {
      "epoch": 2.1839913074972834,
      "grad_norm": 14.974501609802246,
      "learning_rate": 1.3657117459291286e-05,
      "loss": 0.7228,
      "step": 18090
    },
    {
      "epoch": 2.185198599541229,
      "grad_norm": 40.32563400268555,
      "learning_rate": 1.3636914622813043e-05,
      "loss": 0.5618,
      "step": 18100
    },
    {
      "epoch": 2.1864058915851743,
      "grad_norm": 14.944623947143555,
      "learning_rate": 1.3616711786334801e-05,
      "loss": 0.9254,
      "step": 18110
    },
    {
      "epoch": 2.1876131836291197,
      "grad_norm": 16.224828720092773,
      "learning_rate": 1.359650894985656e-05,
      "loss": 0.7214,
      "step": 18120
    },
    {
      "epoch": 2.188820475673065,
      "grad_norm": 13.531900405883789,
      "learning_rate": 1.3576306113378318e-05,
      "loss": 0.7372,
      "step": 18130
    },
    {
      "epoch": 2.1900277677170106,
      "grad_norm": 9.118074417114258,
      "learning_rate": 1.3556103276900079e-05,
      "loss": 0.7365,
      "step": 18140
    },
    {
      "epoch": 2.191235059760956,
      "grad_norm": 22.491064071655273,
      "learning_rate": 1.3535900440421837e-05,
      "loss": 0.7786,
      "step": 18150
    },
    {
      "epoch": 2.1924423518049014,
      "grad_norm": 24.325027465820312,
      "learning_rate": 1.3515697603943594e-05,
      "loss": 0.8341,
      "step": 18160
    },
    {
      "epoch": 2.193649643848847,
      "grad_norm": 10.971212387084961,
      "learning_rate": 1.3495494767465352e-05,
      "loss": 0.5644,
      "step": 18170
    },
    {
      "epoch": 2.1948569358927923,
      "grad_norm": 16.779006958007812,
      "learning_rate": 1.347529193098711e-05,
      "loss": 0.7668,
      "step": 18180
    },
    {
      "epoch": 2.1960642279367377,
      "grad_norm": 3.3489224910736084,
      "learning_rate": 1.345508909450887e-05,
      "loss": 0.6352,
      "step": 18190
    },
    {
      "epoch": 2.197271519980683,
      "grad_norm": 32.05180740356445,
      "learning_rate": 1.343488625803063e-05,
      "loss": 1.1906,
      "step": 18200
    },
    {
      "epoch": 2.1984788120246286,
      "grad_norm": 50.479000091552734,
      "learning_rate": 1.3414683421552388e-05,
      "loss": 0.9825,
      "step": 18210
    },
    {
      "epoch": 2.199686104068574,
      "grad_norm": 10.453009605407715,
      "learning_rate": 1.3394480585074146e-05,
      "loss": 0.9182,
      "step": 18220
    },
    {
      "epoch": 2.2008933961125194,
      "grad_norm": 16.16997718811035,
      "learning_rate": 1.3374277748595903e-05,
      "loss": 0.4835,
      "step": 18230
    },
    {
      "epoch": 2.202100688156465,
      "grad_norm": 13.59000301361084,
      "learning_rate": 1.3354074912117662e-05,
      "loss": 0.6077,
      "step": 18240
    },
    {
      "epoch": 2.2033079802004103,
      "grad_norm": 15.654610633850098,
      "learning_rate": 1.333387207563942e-05,
      "loss": 0.706,
      "step": 18250
    },
    {
      "epoch": 2.2045152722443557,
      "grad_norm": 38.206504821777344,
      "learning_rate": 1.3313669239161179e-05,
      "loss": 0.6427,
      "step": 18260
    },
    {
      "epoch": 2.205722564288301,
      "grad_norm": 11.482831954956055,
      "learning_rate": 1.3293466402682939e-05,
      "loss": 0.8935,
      "step": 18270
    },
    {
      "epoch": 2.2069298563322466,
      "grad_norm": 40.508602142333984,
      "learning_rate": 1.3273263566204697e-05,
      "loss": 0.8957,
      "step": 18280
    },
    {
      "epoch": 2.208137148376192,
      "grad_norm": 31.09208869934082,
      "learning_rate": 1.3253060729726454e-05,
      "loss": 0.8038,
      "step": 18290
    },
    {
      "epoch": 2.2093444404201374,
      "grad_norm": 38.510955810546875,
      "learning_rate": 1.3232857893248212e-05,
      "loss": 0.8882,
      "step": 18300
    },
    {
      "epoch": 2.210551732464083,
      "grad_norm": 26.985017776489258,
      "learning_rate": 1.3212655056769971e-05,
      "loss": 1.0025,
      "step": 18310
    },
    {
      "epoch": 2.2117590245080283,
      "grad_norm": 20.137245178222656,
      "learning_rate": 1.319245222029173e-05,
      "loss": 0.6956,
      "step": 18320
    },
    {
      "epoch": 2.2129663165519737,
      "grad_norm": 24.74285316467285,
      "learning_rate": 1.3172249383813486e-05,
      "loss": 1.017,
      "step": 18330
    },
    {
      "epoch": 2.214173608595919,
      "grad_norm": 0.4861941337585449,
      "learning_rate": 1.3152046547335248e-05,
      "loss": 0.9028,
      "step": 18340
    },
    {
      "epoch": 2.2153809006398646,
      "grad_norm": 11.138595581054688,
      "learning_rate": 1.3131843710857006e-05,
      "loss": 0.9421,
      "step": 18350
    },
    {
      "epoch": 2.21658819268381,
      "grad_norm": 22.870742797851562,
      "learning_rate": 1.3111640874378763e-05,
      "loss": 0.818,
      "step": 18360
    },
    {
      "epoch": 2.2177954847277555,
      "grad_norm": 3.2891600131988525,
      "learning_rate": 1.3091438037900522e-05,
      "loss": 0.8283,
      "step": 18370
    },
    {
      "epoch": 2.219002776771701,
      "grad_norm": 1.915550947189331,
      "learning_rate": 1.307123520142228e-05,
      "loss": 1.4022,
      "step": 18380
    },
    {
      "epoch": 2.2202100688156463,
      "grad_norm": 10.983954429626465,
      "learning_rate": 1.3051032364944039e-05,
      "loss": 0.8048,
      "step": 18390
    },
    {
      "epoch": 2.2214173608595917,
      "grad_norm": 20.669042587280273,
      "learning_rate": 1.3030829528465795e-05,
      "loss": 0.7519,
      "step": 18400
    },
    {
      "epoch": 2.222624652903537,
      "grad_norm": 19.313892364501953,
      "learning_rate": 1.3010626691987557e-05,
      "loss": 0.8387,
      "step": 18410
    },
    {
      "epoch": 2.2238319449474826,
      "grad_norm": 21.020755767822266,
      "learning_rate": 1.2990423855509314e-05,
      "loss": 0.6043,
      "step": 18420
    },
    {
      "epoch": 2.225039236991428,
      "grad_norm": 18.626312255859375,
      "learning_rate": 1.2970221019031073e-05,
      "loss": 0.8976,
      "step": 18430
    },
    {
      "epoch": 2.2262465290353735,
      "grad_norm": 7.25297212600708,
      "learning_rate": 1.2950018182552831e-05,
      "loss": 1.1381,
      "step": 18440
    },
    {
      "epoch": 2.227453821079319,
      "grad_norm": 1.2249683141708374,
      "learning_rate": 1.292981534607459e-05,
      "loss": 0.7744,
      "step": 18450
    },
    {
      "epoch": 2.2286611131232643,
      "grad_norm": 27.548126220703125,
      "learning_rate": 1.2909612509596346e-05,
      "loss": 0.7216,
      "step": 18460
    },
    {
      "epoch": 2.2298684051672097,
      "grad_norm": 23.402971267700195,
      "learning_rate": 1.2889409673118105e-05,
      "loss": 0.5718,
      "step": 18470
    },
    {
      "epoch": 2.231075697211155,
      "grad_norm": 12.51275634765625,
      "learning_rate": 1.2869206836639867e-05,
      "loss": 0.6616,
      "step": 18480
    },
    {
      "epoch": 2.2322829892551006,
      "grad_norm": 5.784684181213379,
      "learning_rate": 1.2849004000161623e-05,
      "loss": 0.9213,
      "step": 18490
    },
    {
      "epoch": 2.233490281299046,
      "grad_norm": 19.775638580322266,
      "learning_rate": 1.2828801163683382e-05,
      "loss": 0.9926,
      "step": 18500
    },
    {
      "epoch": 2.2346975733429915,
      "grad_norm": 19.000457763671875,
      "learning_rate": 1.280859832720514e-05,
      "loss": 0.9622,
      "step": 18510
    },
    {
      "epoch": 2.235904865386937,
      "grad_norm": 18.9771785736084,
      "learning_rate": 1.2788395490726899e-05,
      "loss": 0.9247,
      "step": 18520
    },
    {
      "epoch": 2.2371121574308823,
      "grad_norm": 1.6305564641952515,
      "learning_rate": 1.2768192654248656e-05,
      "loss": 0.8766,
      "step": 18530
    },
    {
      "epoch": 2.2383194494748277,
      "grad_norm": 0.7858160734176636,
      "learning_rate": 1.2747989817770414e-05,
      "loss": 0.7249,
      "step": 18540
    },
    {
      "epoch": 2.239526741518773,
      "grad_norm": 37.52659225463867,
      "learning_rate": 1.2727786981292176e-05,
      "loss": 1.0444,
      "step": 18550
    },
    {
      "epoch": 2.2407340335627186,
      "grad_norm": 8.91016674041748,
      "learning_rate": 1.2707584144813933e-05,
      "loss": 0.8514,
      "step": 18560
    },
    {
      "epoch": 2.241941325606664,
      "grad_norm": 6.6682939529418945,
      "learning_rate": 1.2687381308335691e-05,
      "loss": 0.5793,
      "step": 18570
    },
    {
      "epoch": 2.2431486176506095,
      "grad_norm": 11.841216087341309,
      "learning_rate": 1.266717847185745e-05,
      "loss": 0.8949,
      "step": 18580
    },
    {
      "epoch": 2.244355909694555,
      "grad_norm": 20.26390838623047,
      "learning_rate": 1.2646975635379208e-05,
      "loss": 0.8225,
      "step": 18590
    },
    {
      "epoch": 2.2455632017385003,
      "grad_norm": 18.266027450561523,
      "learning_rate": 1.2626772798900965e-05,
      "loss": 0.8566,
      "step": 18600
    },
    {
      "epoch": 2.2467704937824458,
      "grad_norm": 6.756629467010498,
      "learning_rate": 1.2606569962422723e-05,
      "loss": 0.5301,
      "step": 18610
    },
    {
      "epoch": 2.247977785826391,
      "grad_norm": 9.2037935256958,
      "learning_rate": 1.2586367125944484e-05,
      "loss": 1.3485,
      "step": 18620
    },
    {
      "epoch": 2.2491850778703366,
      "grad_norm": 2.1858253479003906,
      "learning_rate": 1.2566164289466242e-05,
      "loss": 0.934,
      "step": 18630
    },
    {
      "epoch": 2.250392369914282,
      "grad_norm": 18.260591506958008,
      "learning_rate": 1.2545961452988e-05,
      "loss": 1.2218,
      "step": 18640
    },
    {
      "epoch": 2.2515996619582275,
      "grad_norm": 9.902510643005371,
      "learning_rate": 1.2525758616509759e-05,
      "loss": 0.7932,
      "step": 18650
    },
    {
      "epoch": 2.252806954002173,
      "grad_norm": 28.81163215637207,
      "learning_rate": 1.2505555780031516e-05,
      "loss": 0.722,
      "step": 18660
    },
    {
      "epoch": 2.2540142460461183,
      "grad_norm": 17.742998123168945,
      "learning_rate": 1.2485352943553276e-05,
      "loss": 0.7038,
      "step": 18670
    },
    {
      "epoch": 2.2552215380900638,
      "grad_norm": 8.940312385559082,
      "learning_rate": 1.2465150107075034e-05,
      "loss": 0.781,
      "step": 18680
    },
    {
      "epoch": 2.256428830134009,
      "grad_norm": 2.319079637527466,
      "learning_rate": 1.2444947270596791e-05,
      "loss": 0.7066,
      "step": 18690
    },
    {
      "epoch": 2.2576361221779546,
      "grad_norm": 6.2851762771606445,
      "learning_rate": 1.2424744434118551e-05,
      "loss": 0.6438,
      "step": 18700
    },
    {
      "epoch": 2.2588434142219,
      "grad_norm": 11.304967880249023,
      "learning_rate": 1.240454159764031e-05,
      "loss": 0.8262,
      "step": 18710
    },
    {
      "epoch": 2.2600507062658455,
      "grad_norm": 12.079258918762207,
      "learning_rate": 1.2384338761162068e-05,
      "loss": 0.7466,
      "step": 18720
    },
    {
      "epoch": 2.261257998309791,
      "grad_norm": 3.05283784866333,
      "learning_rate": 1.2364135924683825e-05,
      "loss": 0.6864,
      "step": 18730
    },
    {
      "epoch": 2.2624652903537363,
      "grad_norm": 18.14229965209961,
      "learning_rate": 1.2343933088205585e-05,
      "loss": 1.0062,
      "step": 18740
    },
    {
      "epoch": 2.2636725823976818,
      "grad_norm": 20.997716903686523,
      "learning_rate": 1.2323730251727344e-05,
      "loss": 0.8813,
      "step": 18750
    },
    {
      "epoch": 2.264879874441627,
      "grad_norm": 28.78122329711914,
      "learning_rate": 1.23035274152491e-05,
      "loss": 0.7335,
      "step": 18760
    },
    {
      "epoch": 2.2660871664855726,
      "grad_norm": 6.455710411071777,
      "learning_rate": 1.228332457877086e-05,
      "loss": 0.6938,
      "step": 18770
    },
    {
      "epoch": 2.267294458529518,
      "grad_norm": 24.837207794189453,
      "learning_rate": 1.226312174229262e-05,
      "loss": 1.295,
      "step": 18780
    },
    {
      "epoch": 2.2685017505734635,
      "grad_norm": 11.676143646240234,
      "learning_rate": 1.2242918905814376e-05,
      "loss": 1.0784,
      "step": 18790
    },
    {
      "epoch": 2.269709042617409,
      "grad_norm": 19.48927116394043,
      "learning_rate": 1.2222716069336134e-05,
      "loss": 1.0406,
      "step": 18800
    },
    {
      "epoch": 2.2709163346613543,
      "grad_norm": 3.2064597606658936,
      "learning_rate": 1.2202513232857895e-05,
      "loss": 0.3359,
      "step": 18810
    },
    {
      "epoch": 2.2721236267052998,
      "grad_norm": 4.340291500091553,
      "learning_rate": 1.2182310396379651e-05,
      "loss": 0.799,
      "step": 18820
    },
    {
      "epoch": 2.273330918749245,
      "grad_norm": 3.6109747886657715,
      "learning_rate": 1.216210755990141e-05,
      "loss": 0.5742,
      "step": 18830
    },
    {
      "epoch": 2.2745382107931906,
      "grad_norm": 26.274089813232422,
      "learning_rate": 1.214190472342317e-05,
      "loss": 1.0743,
      "step": 18840
    },
    {
      "epoch": 2.275745502837136,
      "grad_norm": 20.50476837158203,
      "learning_rate": 1.2121701886944929e-05,
      "loss": 0.8803,
      "step": 18850
    },
    {
      "epoch": 2.2769527948810815,
      "grad_norm": 10.62234878540039,
      "learning_rate": 1.2101499050466685e-05,
      "loss": 0.7826,
      "step": 18860
    },
    {
      "epoch": 2.278160086925027,
      "grad_norm": 16.091548919677734,
      "learning_rate": 1.2081296213988444e-05,
      "loss": 0.6375,
      "step": 18870
    },
    {
      "epoch": 2.2793673789689723,
      "grad_norm": 31.9377384185791,
      "learning_rate": 1.2061093377510204e-05,
      "loss": 0.7298,
      "step": 18880
    },
    {
      "epoch": 2.2805746710129178,
      "grad_norm": 26.399539947509766,
      "learning_rate": 1.204089054103196e-05,
      "loss": 0.722,
      "step": 18890
    },
    {
      "epoch": 2.2817819630568636,
      "grad_norm": 14.06995964050293,
      "learning_rate": 1.202068770455372e-05,
      "loss": 0.9531,
      "step": 18900
    },
    {
      "epoch": 2.282989255100809,
      "grad_norm": 37.5555419921875,
      "learning_rate": 1.200048486807548e-05,
      "loss": 0.8836,
      "step": 18910
    },
    {
      "epoch": 2.2841965471447545,
      "grad_norm": 0.3462620973587036,
      "learning_rate": 1.1980282031597236e-05,
      "loss": 0.7808,
      "step": 18920
    },
    {
      "epoch": 2.2854038391887,
      "grad_norm": 24.584606170654297,
      "learning_rate": 1.1960079195118995e-05,
      "loss": 0.9385,
      "step": 18930
    },
    {
      "epoch": 2.2866111312326454,
      "grad_norm": 38.975921630859375,
      "learning_rate": 1.1939876358640753e-05,
      "loss": 0.9726,
      "step": 18940
    },
    {
      "epoch": 2.287818423276591,
      "grad_norm": 10.054938316345215,
      "learning_rate": 1.1919673522162513e-05,
      "loss": 0.9473,
      "step": 18950
    },
    {
      "epoch": 2.2890257153205362,
      "grad_norm": 37.718414306640625,
      "learning_rate": 1.189947068568427e-05,
      "loss": 0.8843,
      "step": 18960
    },
    {
      "epoch": 2.2902330073644817,
      "grad_norm": 12.43630599975586,
      "learning_rate": 1.1879267849206029e-05,
      "loss": 0.7552,
      "step": 18970
    },
    {
      "epoch": 2.291440299408427,
      "grad_norm": 12.605921745300293,
      "learning_rate": 1.1859065012727789e-05,
      "loss": 0.8717,
      "step": 18980
    },
    {
      "epoch": 2.2926475914523725,
      "grad_norm": 28.118450164794922,
      "learning_rate": 1.1838862176249546e-05,
      "loss": 0.8579,
      "step": 18990
    },
    {
      "epoch": 2.293854883496318,
      "grad_norm": 58.03962707519531,
      "learning_rate": 1.1818659339771304e-05,
      "loss": 1.4107,
      "step": 19000
    },
    {
      "epoch": 2.2950621755402634,
      "grad_norm": 7.621524810791016,
      "learning_rate": 1.1798456503293064e-05,
      "loss": 0.5853,
      "step": 19010
    },
    {
      "epoch": 2.296269467584209,
      "grad_norm": 44.90055847167969,
      "learning_rate": 1.1778253666814821e-05,
      "loss": 0.7919,
      "step": 19020
    },
    {
      "epoch": 2.2974767596281542,
      "grad_norm": 4.630683422088623,
      "learning_rate": 1.175805083033658e-05,
      "loss": 0.5644,
      "step": 19030
    },
    {
      "epoch": 2.2986840516720997,
      "grad_norm": 0.7553032040596008,
      "learning_rate": 1.1737847993858338e-05,
      "loss": 1.0701,
      "step": 19040
    },
    {
      "epoch": 2.299891343716045,
      "grad_norm": 16.378938674926758,
      "learning_rate": 1.1717645157380096e-05,
      "loss": 0.6344,
      "step": 19050
    },
    {
      "epoch": 2.3010986357599905,
      "grad_norm": 10.321065902709961,
      "learning_rate": 1.1697442320901855e-05,
      "loss": 0.538,
      "step": 19060
    },
    {
      "epoch": 2.302305927803936,
      "grad_norm": 24.53988265991211,
      "learning_rate": 1.1677239484423613e-05,
      "loss": 0.7577,
      "step": 19070
    },
    {
      "epoch": 2.3035132198478814,
      "grad_norm": 21.341472625732422,
      "learning_rate": 1.1657036647945373e-05,
      "loss": 0.7553,
      "step": 19080
    },
    {
      "epoch": 2.304720511891827,
      "grad_norm": 3.76430082321167,
      "learning_rate": 1.163683381146713e-05,
      "loss": 0.8301,
      "step": 19090
    },
    {
      "epoch": 2.3059278039357722,
      "grad_norm": 14.712066650390625,
      "learning_rate": 1.1616630974988889e-05,
      "loss": 0.4607,
      "step": 19100
    },
    {
      "epoch": 2.3071350959797177,
      "grad_norm": 31.313573837280273,
      "learning_rate": 1.1596428138510647e-05,
      "loss": 0.6752,
      "step": 19110
    },
    {
      "epoch": 2.308342388023663,
      "grad_norm": 19.639629364013672,
      "learning_rate": 1.1576225302032406e-05,
      "loss": 1.1913,
      "step": 19120
    },
    {
      "epoch": 2.3095496800676085,
      "grad_norm": 3.3071770668029785,
      "learning_rate": 1.1556022465554164e-05,
      "loss": 0.7071,
      "step": 19130
    },
    {
      "epoch": 2.310756972111554,
      "grad_norm": 25.736379623413086,
      "learning_rate": 1.1535819629075923e-05,
      "loss": 0.9594,
      "step": 19140
    },
    {
      "epoch": 2.3119642641554994,
      "grad_norm": 1.097156286239624,
      "learning_rate": 1.1515616792597681e-05,
      "loss": 0.7225,
      "step": 19150
    },
    {
      "epoch": 2.313171556199445,
      "grad_norm": 43.66969299316406,
      "learning_rate": 1.149541395611944e-05,
      "loss": 0.8721,
      "step": 19160
    },
    {
      "epoch": 2.3143788482433902,
      "grad_norm": 9.46072769165039,
      "learning_rate": 1.1475211119641198e-05,
      "loss": 0.7067,
      "step": 19170
    },
    {
      "epoch": 2.3155861402873357,
      "grad_norm": 53.012672424316406,
      "learning_rate": 1.1455008283162957e-05,
      "loss": 0.6899,
      "step": 19180
    },
    {
      "epoch": 2.316793432331281,
      "grad_norm": 9.685965538024902,
      "learning_rate": 1.1434805446684715e-05,
      "loss": 1.4198,
      "step": 19190
    },
    {
      "epoch": 2.3180007243752265,
      "grad_norm": 13.407829284667969,
      "learning_rate": 1.1414602610206473e-05,
      "loss": 1.2682,
      "step": 19200
    },
    {
      "epoch": 2.319208016419172,
      "grad_norm": 22.249544143676758,
      "learning_rate": 1.1394399773728232e-05,
      "loss": 0.8979,
      "step": 19210
    },
    {
      "epoch": 2.3204153084631174,
      "grad_norm": 29.708438873291016,
      "learning_rate": 1.137419693724999e-05,
      "loss": 0.8566,
      "step": 19220
    },
    {
      "epoch": 2.321622600507063,
      "grad_norm": 28.829708099365234,
      "learning_rate": 1.1353994100771749e-05,
      "loss": 0.9817,
      "step": 19230
    },
    {
      "epoch": 2.3228298925510082,
      "grad_norm": 10.351776123046875,
      "learning_rate": 1.1333791264293507e-05,
      "loss": 0.5743,
      "step": 19240
    },
    {
      "epoch": 2.3240371845949537,
      "grad_norm": 9.677617073059082,
      "learning_rate": 1.1313588427815266e-05,
      "loss": 0.8898,
      "step": 19250
    },
    {
      "epoch": 2.325244476638899,
      "grad_norm": 5.356738567352295,
      "learning_rate": 1.1293385591337024e-05,
      "loss": 0.563,
      "step": 19260
    },
    {
      "epoch": 2.3264517686828445,
      "grad_norm": 15.992120742797852,
      "learning_rate": 1.1273182754858783e-05,
      "loss": 0.778,
      "step": 19270
    },
    {
      "epoch": 2.32765906072679,
      "grad_norm": 10.398543357849121,
      "learning_rate": 1.1252979918380541e-05,
      "loss": 0.5576,
      "step": 19280
    },
    {
      "epoch": 2.3288663527707354,
      "grad_norm": 12.654329299926758,
      "learning_rate": 1.12327770819023e-05,
      "loss": 0.705,
      "step": 19290
    },
    {
      "epoch": 2.330073644814681,
      "grad_norm": 16.786977767944336,
      "learning_rate": 1.1212574245424058e-05,
      "loss": 0.7733,
      "step": 19300
    },
    {
      "epoch": 2.3312809368586263,
      "grad_norm": 31.786752700805664,
      "learning_rate": 1.1192371408945817e-05,
      "loss": 0.6727,
      "step": 19310
    },
    {
      "epoch": 2.3324882289025717,
      "grad_norm": 9.324110984802246,
      "learning_rate": 1.1172168572467575e-05,
      "loss": 0.3643,
      "step": 19320
    },
    {
      "epoch": 2.333695520946517,
      "grad_norm": 33.577423095703125,
      "learning_rate": 1.1151965735989334e-05,
      "loss": 0.9572,
      "step": 19330
    },
    {
      "epoch": 2.3349028129904625,
      "grad_norm": 18.046842575073242,
      "learning_rate": 1.1131762899511092e-05,
      "loss": 0.9643,
      "step": 19340
    },
    {
      "epoch": 2.336110105034408,
      "grad_norm": 18.908357620239258,
      "learning_rate": 1.111156006303285e-05,
      "loss": 1.223,
      "step": 19350
    },
    {
      "epoch": 2.3373173970783534,
      "grad_norm": 7.137038230895996,
      "learning_rate": 1.1091357226554609e-05,
      "loss": 0.8422,
      "step": 19360
    },
    {
      "epoch": 2.338524689122299,
      "grad_norm": 25.516979217529297,
      "learning_rate": 1.1071154390076368e-05,
      "loss": 0.9418,
      "step": 19370
    },
    {
      "epoch": 2.3397319811662443,
      "grad_norm": 46.190757751464844,
      "learning_rate": 1.1050951553598126e-05,
      "loss": 0.7569,
      "step": 19380
    },
    {
      "epoch": 2.3409392732101897,
      "grad_norm": 8.329981803894043,
      "learning_rate": 1.1030748717119885e-05,
      "loss": 1.0702,
      "step": 19390
    },
    {
      "epoch": 2.342146565254135,
      "grad_norm": 36.87815856933594,
      "learning_rate": 1.1010545880641643e-05,
      "loss": 0.7994,
      "step": 19400
    },
    {
      "epoch": 2.3433538572980805,
      "grad_norm": 23.208057403564453,
      "learning_rate": 1.0990343044163401e-05,
      "loss": 0.6364,
      "step": 19410
    },
    {
      "epoch": 2.344561149342026,
      "grad_norm": 5.430502891540527,
      "learning_rate": 1.0970140207685158e-05,
      "loss": 0.8618,
      "step": 19420
    },
    {
      "epoch": 2.3457684413859714,
      "grad_norm": 9.956197738647461,
      "learning_rate": 1.0949937371206918e-05,
      "loss": 0.9068,
      "step": 19430
    },
    {
      "epoch": 2.346975733429917,
      "grad_norm": 24.08917808532715,
      "learning_rate": 1.0929734534728677e-05,
      "loss": 0.6619,
      "step": 19440
    },
    {
      "epoch": 2.3481830254738623,
      "grad_norm": 4.660079002380371,
      "learning_rate": 1.0909531698250434e-05,
      "loss": 0.9128,
      "step": 19450
    },
    {
      "epoch": 2.3493903175178077,
      "grad_norm": 23.698122024536133,
      "learning_rate": 1.0889328861772194e-05,
      "loss": 1.3948,
      "step": 19460
    },
    {
      "epoch": 2.350597609561753,
      "grad_norm": 1.159500002861023,
      "learning_rate": 1.0869126025293952e-05,
      "loss": 0.8058,
      "step": 19470
    },
    {
      "epoch": 2.3518049016056986,
      "grad_norm": 1.0941994190216064,
      "learning_rate": 1.084892318881571e-05,
      "loss": 1.0117,
      "step": 19480
    },
    {
      "epoch": 2.353012193649644,
      "grad_norm": 23.348562240600586,
      "learning_rate": 1.0828720352337468e-05,
      "loss": 0.761,
      "step": 19490
    },
    {
      "epoch": 2.3542194856935894,
      "grad_norm": 12.794855117797852,
      "learning_rate": 1.0808517515859228e-05,
      "loss": 1.0511,
      "step": 19500
    },
    {
      "epoch": 2.355426777737535,
      "grad_norm": 2.273796558380127,
      "learning_rate": 1.0788314679380986e-05,
      "loss": 1.2408,
      "step": 19510
    },
    {
      "epoch": 2.3566340697814803,
      "grad_norm": 10.670280456542969,
      "learning_rate": 1.0768111842902743e-05,
      "loss": 0.439,
      "step": 19520
    },
    {
      "epoch": 2.3578413618254257,
      "grad_norm": 0.8197128176689148,
      "learning_rate": 1.0747909006424503e-05,
      "loss": 0.5122,
      "step": 19530
    },
    {
      "epoch": 2.359048653869371,
      "grad_norm": 15.964715957641602,
      "learning_rate": 1.0727706169946262e-05,
      "loss": 0.6971,
      "step": 19540
    },
    {
      "epoch": 2.3602559459133166,
      "grad_norm": 15.159314155578613,
      "learning_rate": 1.0707503333468018e-05,
      "loss": 0.9624,
      "step": 19550
    },
    {
      "epoch": 2.361463237957262,
      "grad_norm": 0.7506232857704163,
      "learning_rate": 1.0687300496989777e-05,
      "loss": 0.8073,
      "step": 19560
    },
    {
      "epoch": 2.3626705300012074,
      "grad_norm": 28.554126739501953,
      "learning_rate": 1.0667097660511537e-05,
      "loss": 0.8807,
      "step": 19570
    },
    {
      "epoch": 2.363877822045153,
      "grad_norm": 13.9207181930542,
      "learning_rate": 1.0646894824033296e-05,
      "loss": 1.2928,
      "step": 19580
    },
    {
      "epoch": 2.3650851140890983,
      "grad_norm": 12.59956169128418,
      "learning_rate": 1.0626691987555052e-05,
      "loss": 0.8336,
      "step": 19590
    },
    {
      "epoch": 2.3662924061330437,
      "grad_norm": 15.669353485107422,
      "learning_rate": 1.0606489151076812e-05,
      "loss": 0.9332,
      "step": 19600
    },
    {
      "epoch": 2.367499698176989,
      "grad_norm": 20.12005615234375,
      "learning_rate": 1.0586286314598571e-05,
      "loss": 0.6871,
      "step": 19610
    },
    {
      "epoch": 2.3687069902209346,
      "grad_norm": 12.989645957946777,
      "learning_rate": 1.0566083478120328e-05,
      "loss": 1.0185,
      "step": 19620
    },
    {
      "epoch": 2.36991428226488,
      "grad_norm": 24.703237533569336,
      "learning_rate": 1.0545880641642088e-05,
      "loss": 0.7951,
      "step": 19630
    },
    {
      "epoch": 2.3711215743088254,
      "grad_norm": 32.057716369628906,
      "learning_rate": 1.0525677805163846e-05,
      "loss": 0.6378,
      "step": 19640
    },
    {
      "epoch": 2.372328866352771,
      "grad_norm": 52.5326042175293,
      "learning_rate": 1.0505474968685603e-05,
      "loss": 0.8554,
      "step": 19650
    },
    {
      "epoch": 2.3735361583967163,
      "grad_norm": 7.080691337585449,
      "learning_rate": 1.0485272132207362e-05,
      "loss": 1.008,
      "step": 19660
    },
    {
      "epoch": 2.3747434504406617,
      "grad_norm": 30.631620407104492,
      "learning_rate": 1.0465069295729122e-05,
      "loss": 1.0435,
      "step": 19670
    },
    {
      "epoch": 2.375950742484607,
      "grad_norm": 21.679853439331055,
      "learning_rate": 1.0444866459250879e-05,
      "loss": 1.3389,
      "step": 19680
    },
    {
      "epoch": 2.3771580345285526,
      "grad_norm": 43.917076110839844,
      "learning_rate": 1.0424663622772637e-05,
      "loss": 1.0528,
      "step": 19690
    },
    {
      "epoch": 2.378365326572498,
      "grad_norm": 23.34821891784668,
      "learning_rate": 1.0404460786294397e-05,
      "loss": 0.8039,
      "step": 19700
    },
    {
      "epoch": 2.3795726186164434,
      "grad_norm": 20.065799713134766,
      "learning_rate": 1.0384257949816156e-05,
      "loss": 1.2887,
      "step": 19710
    },
    {
      "epoch": 2.380779910660389,
      "grad_norm": 13.888937950134277,
      "learning_rate": 1.0364055113337912e-05,
      "loss": 0.9731,
      "step": 19720
    },
    {
      "epoch": 2.3819872027043343,
      "grad_norm": 26.059534072875977,
      "learning_rate": 1.0343852276859671e-05,
      "loss": 0.9712,
      "step": 19730
    },
    {
      "epoch": 2.3831944947482797,
      "grad_norm": 32.54450225830078,
      "learning_rate": 1.0323649440381431e-05,
      "loss": 1.141,
      "step": 19740
    },
    {
      "epoch": 2.384401786792225,
      "grad_norm": 2.8565661907196045,
      "learning_rate": 1.0303446603903188e-05,
      "loss": 0.6837,
      "step": 19750
    },
    {
      "epoch": 2.3856090788361706,
      "grad_norm": 24.80270004272461,
      "learning_rate": 1.0283243767424946e-05,
      "loss": 0.9333,
      "step": 19760
    },
    {
      "epoch": 2.386816370880116,
      "grad_norm": 31.633743286132812,
      "learning_rate": 1.0263040930946707e-05,
      "loss": 0.9901,
      "step": 19770
    },
    {
      "epoch": 2.3880236629240614,
      "grad_norm": 23.37162971496582,
      "learning_rate": 1.0242838094468463e-05,
      "loss": 0.476,
      "step": 19780
    },
    {
      "epoch": 2.389230954968007,
      "grad_norm": 29.865346908569336,
      "learning_rate": 1.0222635257990222e-05,
      "loss": 1.1467,
      "step": 19790
    },
    {
      "epoch": 2.3904382470119523,
      "grad_norm": 53.47759246826172,
      "learning_rate": 1.020243242151198e-05,
      "loss": 1.4453,
      "step": 19800
    },
    {
      "epoch": 2.3916455390558977,
      "grad_norm": 22.044464111328125,
      "learning_rate": 1.0182229585033739e-05,
      "loss": 1.0378,
      "step": 19810
    },
    {
      "epoch": 2.392852831099843,
      "grad_norm": 4.8116865158081055,
      "learning_rate": 1.0162026748555497e-05,
      "loss": 0.8704,
      "step": 19820
    },
    {
      "epoch": 2.3940601231437886,
      "grad_norm": 19.70046043395996,
      "learning_rate": 1.0141823912077256e-05,
      "loss": 0.6401,
      "step": 19830
    },
    {
      "epoch": 2.395267415187734,
      "grad_norm": 27.7103328704834,
      "learning_rate": 1.0121621075599016e-05,
      "loss": 0.4885,
      "step": 19840
    },
    {
      "epoch": 2.3964747072316794,
      "grad_norm": 2.7382404804229736,
      "learning_rate": 1.0101418239120773e-05,
      "loss": 0.5755,
      "step": 19850
    },
    {
      "epoch": 2.397681999275625,
      "grad_norm": 12.869295120239258,
      "learning_rate": 1.0081215402642531e-05,
      "loss": 0.4765,
      "step": 19860
    },
    {
      "epoch": 2.3988892913195703,
      "grad_norm": 2.3791065216064453,
      "learning_rate": 1.006101256616429e-05,
      "loss": 0.3932,
      "step": 19870
    },
    {
      "epoch": 2.4000965833635157,
      "grad_norm": 20.59455680847168,
      "learning_rate": 1.0040809729686048e-05,
      "loss": 0.796,
      "step": 19880
    },
    {
      "epoch": 2.401303875407461,
      "grad_norm": 19.591373443603516,
      "learning_rate": 1.0020606893207807e-05,
      "loss": 0.7256,
      "step": 19890
    },
    {
      "epoch": 2.4025111674514066,
      "grad_norm": 41.158653259277344,
      "learning_rate": 1.0000404056729565e-05,
      "loss": 0.8641,
      "step": 19900
    },
    {
      "epoch": 2.403718459495352,
      "grad_norm": 7.4664998054504395,
      "learning_rate": 9.980201220251324e-06,
      "loss": 0.8919,
      "step": 19910
    },
    {
      "epoch": 2.4049257515392974,
      "grad_norm": 13.103621482849121,
      "learning_rate": 9.959998383773082e-06,
      "loss": 0.7578,
      "step": 19920
    },
    {
      "epoch": 2.406133043583243,
      "grad_norm": 9.77096176147461,
      "learning_rate": 9.93979554729484e-06,
      "loss": 1.0401,
      "step": 19930
    },
    {
      "epoch": 2.4073403356271883,
      "grad_norm": 33.54362869262695,
      "learning_rate": 9.9195927108166e-06,
      "loss": 0.732,
      "step": 19940
    },
    {
      "epoch": 2.4085476276711337,
      "grad_norm": 39.2683219909668,
      "learning_rate": 9.899389874338357e-06,
      "loss": 0.5398,
      "step": 19950
    },
    {
      "epoch": 2.409754919715079,
      "grad_norm": 19.005624771118164,
      "learning_rate": 9.879187037860116e-06,
      "loss": 0.7048,
      "step": 19960
    },
    {
      "epoch": 2.4109622117590246,
      "grad_norm": 23.95442008972168,
      "learning_rate": 9.858984201381874e-06,
      "loss": 1.1171,
      "step": 19970
    },
    {
      "epoch": 2.41216950380297,
      "grad_norm": 32.34153747558594,
      "learning_rate": 9.838781364903633e-06,
      "loss": 1.007,
      "step": 19980
    },
    {
      "epoch": 2.4133767958469154,
      "grad_norm": 10.321290969848633,
      "learning_rate": 9.818578528425391e-06,
      "loss": 1.2474,
      "step": 19990
    },
    {
      "epoch": 2.414584087890861,
      "grad_norm": 16.50524139404297,
      "learning_rate": 9.79837569194715e-06,
      "loss": 0.5866,
      "step": 20000
    },
    {
      "epoch": 2.4157913799348063,
      "grad_norm": 6.225854873657227,
      "learning_rate": 9.778172855468908e-06,
      "loss": 1.078,
      "step": 20010
    },
    {
      "epoch": 2.4169986719787517,
      "grad_norm": 63.936668395996094,
      "learning_rate": 9.757970018990667e-06,
      "loss": 0.9224,
      "step": 20020
    },
    {
      "epoch": 2.418205964022697,
      "grad_norm": 14.949941635131836,
      "learning_rate": 9.737767182512425e-06,
      "loss": 0.7759,
      "step": 20030
    },
    {
      "epoch": 2.4194132560666426,
      "grad_norm": 20.331859588623047,
      "learning_rate": 9.717564346034184e-06,
      "loss": 0.7499,
      "step": 20040
    },
    {
      "epoch": 2.420620548110588,
      "grad_norm": 27.208948135375977,
      "learning_rate": 9.697361509555942e-06,
      "loss": 0.9922,
      "step": 20050
    },
    {
      "epoch": 2.4218278401545335,
      "grad_norm": 23.60843849182129,
      "learning_rate": 9.6771586730777e-06,
      "loss": 0.5465,
      "step": 20060
    },
    {
      "epoch": 2.423035132198479,
      "grad_norm": 10.383758544921875,
      "learning_rate": 9.656955836599459e-06,
      "loss": 0.8448,
      "step": 20070
    },
    {
      "epoch": 2.4242424242424243,
      "grad_norm": 10.194206237792969,
      "learning_rate": 9.636753000121218e-06,
      "loss": 0.5055,
      "step": 20080
    },
    {
      "epoch": 2.4254497162863697,
      "grad_norm": 21.079566955566406,
      "learning_rate": 9.616550163642976e-06,
      "loss": 1.1223,
      "step": 20090
    },
    {
      "epoch": 2.426657008330315,
      "grad_norm": 23.990219116210938,
      "learning_rate": 9.596347327164735e-06,
      "loss": 0.7529,
      "step": 20100
    },
    {
      "epoch": 2.4278643003742606,
      "grad_norm": 6.923336029052734,
      "learning_rate": 9.576144490686493e-06,
      "loss": 0.4874,
      "step": 20110
    },
    {
      "epoch": 2.429071592418206,
      "grad_norm": 19.58660316467285,
      "learning_rate": 9.555941654208251e-06,
      "loss": 1.1378,
      "step": 20120
    },
    {
      "epoch": 2.4302788844621515,
      "grad_norm": 1.627100944519043,
      "learning_rate": 9.53573881773001e-06,
      "loss": 0.957,
      "step": 20130
    },
    {
      "epoch": 2.431486176506097,
      "grad_norm": 3.620029926300049,
      "learning_rate": 9.515535981251768e-06,
      "loss": 0.7488,
      "step": 20140
    },
    {
      "epoch": 2.4326934685500423,
      "grad_norm": 34.80943298339844,
      "learning_rate": 9.495333144773527e-06,
      "loss": 1.1685,
      "step": 20150
    },
    {
      "epoch": 2.4339007605939877,
      "grad_norm": 9.464836120605469,
      "learning_rate": 9.475130308295285e-06,
      "loss": 0.72,
      "step": 20160
    },
    {
      "epoch": 2.435108052637933,
      "grad_norm": 1.4106800556182861,
      "learning_rate": 9.454927471817044e-06,
      "loss": 0.9515,
      "step": 20170
    },
    {
      "epoch": 2.4363153446818786,
      "grad_norm": 27.196388244628906,
      "learning_rate": 9.4347246353388e-06,
      "loss": 0.8486,
      "step": 20180
    },
    {
      "epoch": 2.437522636725824,
      "grad_norm": 6.809576034545898,
      "learning_rate": 9.41452179886056e-06,
      "loss": 0.4637,
      "step": 20190
    },
    {
      "epoch": 2.4387299287697695,
      "grad_norm": 22.182113647460938,
      "learning_rate": 9.39431896238232e-06,
      "loss": 0.6827,
      "step": 20200
    },
    {
      "epoch": 2.439937220813715,
      "grad_norm": 58.51742172241211,
      "learning_rate": 9.374116125904076e-06,
      "loss": 0.755,
      "step": 20210
    },
    {
      "epoch": 2.4411445128576603,
      "grad_norm": 23.005605697631836,
      "learning_rate": 9.353913289425836e-06,
      "loss": 1.0513,
      "step": 20220
    },
    {
      "epoch": 2.4423518049016057,
      "grad_norm": 16.11469078063965,
      "learning_rate": 9.333710452947595e-06,
      "loss": 0.6412,
      "step": 20230
    },
    {
      "epoch": 2.443559096945551,
      "grad_norm": 8.049489974975586,
      "learning_rate": 9.313507616469353e-06,
      "loss": 0.5592,
      "step": 20240
    },
    {
      "epoch": 2.4447663889894966,
      "grad_norm": 32.64936065673828,
      "learning_rate": 9.293304779991112e-06,
      "loss": 0.5538,
      "step": 20250
    },
    {
      "epoch": 2.445973681033442,
      "grad_norm": 9.601767539978027,
      "learning_rate": 9.27310194351287e-06,
      "loss": 1.0563,
      "step": 20260
    },
    {
      "epoch": 2.4471809730773875,
      "grad_norm": 8.580645561218262,
      "learning_rate": 9.252899107034629e-06,
      "loss": 0.8711,
      "step": 20270
    },
    {
      "epoch": 2.448388265121333,
      "grad_norm": 19.37752914428711,
      "learning_rate": 9.232696270556385e-06,
      "loss": 1.2107,
      "step": 20280
    },
    {
      "epoch": 2.4495955571652783,
      "grad_norm": 3.6990082263946533,
      "learning_rate": 9.212493434078146e-06,
      "loss": 0.6568,
      "step": 20290
    },
    {
      "epoch": 2.4508028492092238,
      "grad_norm": 35.99006271362305,
      "learning_rate": 9.192290597599904e-06,
      "loss": 0.7963,
      "step": 20300
    },
    {
      "epoch": 2.452010141253169,
      "grad_norm": 40.47325134277344,
      "learning_rate": 9.17208776112166e-06,
      "loss": 0.6464,
      "step": 20310
    },
    {
      "epoch": 2.4532174332971146,
      "grad_norm": 5.889370441436768,
      "learning_rate": 9.151884924643421e-06,
      "loss": 0.6594,
      "step": 20320
    },
    {
      "epoch": 2.45442472534106,
      "grad_norm": 10.270655632019043,
      "learning_rate": 9.13168208816518e-06,
      "loss": 0.4758,
      "step": 20330
    },
    {
      "epoch": 2.4556320173850055,
      "grad_norm": 27.79999542236328,
      "learning_rate": 9.111479251686938e-06,
      "loss": 0.9493,
      "step": 20340
    },
    {
      "epoch": 2.456839309428951,
      "grad_norm": 49.16884994506836,
      "learning_rate": 9.091276415208695e-06,
      "loss": 0.784,
      "step": 20350
    },
    {
      "epoch": 2.4580466014728963,
      "grad_norm": 0.3499263823032379,
      "learning_rate": 9.071073578730455e-06,
      "loss": 1.0121,
      "step": 20360
    },
    {
      "epoch": 2.4592538935168418,
      "grad_norm": 3.594970226287842,
      "learning_rate": 9.050870742252213e-06,
      "loss": 0.8102,
      "step": 20370
    },
    {
      "epoch": 2.460461185560787,
      "grad_norm": 12.422776222229004,
      "learning_rate": 9.03066790577397e-06,
      "loss": 1.0878,
      "step": 20380
    },
    {
      "epoch": 2.4616684776047326,
      "grad_norm": 23.849550247192383,
      "learning_rate": 9.01046506929573e-06,
      "loss": 0.6714,
      "step": 20390
    },
    {
      "epoch": 2.462875769648678,
      "grad_norm": 21.655475616455078,
      "learning_rate": 8.990262232817489e-06,
      "loss": 0.7876,
      "step": 20400
    },
    {
      "epoch": 2.4640830616926235,
      "grad_norm": 10.033309936523438,
      "learning_rate": 8.970059396339246e-06,
      "loss": 0.5344,
      "step": 20410
    },
    {
      "epoch": 2.465290353736569,
      "grad_norm": 0.7608277201652527,
      "learning_rate": 8.949856559861004e-06,
      "loss": 0.7193,
      "step": 20420
    },
    {
      "epoch": 2.4664976457805143,
      "grad_norm": 22.855037689208984,
      "learning_rate": 8.929653723382764e-06,
      "loss": 0.4676,
      "step": 20430
    },
    {
      "epoch": 2.4677049378244598,
      "grad_norm": 16.473852157592773,
      "learning_rate": 8.909450886904521e-06,
      "loss": 0.7232,
      "step": 20440
    },
    {
      "epoch": 2.468912229868405,
      "grad_norm": 2.172999858856201,
      "learning_rate": 8.88924805042628e-06,
      "loss": 1.0748,
      "step": 20450
    },
    {
      "epoch": 2.4701195219123506,
      "grad_norm": 12.725136756896973,
      "learning_rate": 8.86904521394804e-06,
      "loss": 0.468,
      "step": 20460
    },
    {
      "epoch": 2.471326813956296,
      "grad_norm": 29.27521324157715,
      "learning_rate": 8.848842377469798e-06,
      "loss": 0.808,
      "step": 20470
    },
    {
      "epoch": 2.4725341060002415,
      "grad_norm": 0.3766450881958008,
      "learning_rate": 8.828639540991555e-06,
      "loss": 0.7565,
      "step": 20480
    },
    {
      "epoch": 2.473741398044187,
      "grad_norm": 16.200538635253906,
      "learning_rate": 8.808436704513313e-06,
      "loss": 1.0033,
      "step": 20490
    },
    {
      "epoch": 2.4749486900881323,
      "grad_norm": 58.431331634521484,
      "learning_rate": 8.788233868035074e-06,
      "loss": 0.7418,
      "step": 20500
    },
    {
      "epoch": 2.4761559821320778,
      "grad_norm": 1.2682474851608276,
      "learning_rate": 8.76803103155683e-06,
      "loss": 1.1027,
      "step": 20510
    },
    {
      "epoch": 2.477363274176023,
      "grad_norm": 13.579154014587402,
      "learning_rate": 8.747828195078589e-06,
      "loss": 0.942,
      "step": 20520
    },
    {
      "epoch": 2.4785705662199686,
      "grad_norm": 21.28840446472168,
      "learning_rate": 8.727625358600349e-06,
      "loss": 0.6957,
      "step": 20530
    },
    {
      "epoch": 2.479777858263914,
      "grad_norm": 14.879831314086914,
      "learning_rate": 8.707422522122106e-06,
      "loss": 0.4651,
      "step": 20540
    },
    {
      "epoch": 2.4809851503078595,
      "grad_norm": 10.568427085876465,
      "learning_rate": 8.687219685643864e-06,
      "loss": 0.5387,
      "step": 20550
    },
    {
      "epoch": 2.482192442351805,
      "grad_norm": 17.919723510742188,
      "learning_rate": 8.667016849165624e-06,
      "loss": 1.0221,
      "step": 20560
    },
    {
      "epoch": 2.4833997343957503,
      "grad_norm": 18.87501335144043,
      "learning_rate": 8.646814012687381e-06,
      "loss": 1.1126,
      "step": 20570
    },
    {
      "epoch": 2.4846070264396958,
      "grad_norm": 18.634273529052734,
      "learning_rate": 8.62661117620914e-06,
      "loss": 0.7648,
      "step": 20580
    },
    {
      "epoch": 2.485814318483641,
      "grad_norm": 12.87051010131836,
      "learning_rate": 8.606408339730898e-06,
      "loss": 0.7679,
      "step": 20590
    },
    {
      "epoch": 2.4870216105275866,
      "grad_norm": 11.77473258972168,
      "learning_rate": 8.586205503252658e-06,
      "loss": 0.6772,
      "step": 20600
    },
    {
      "epoch": 2.488228902571532,
      "grad_norm": 32.671321868896484,
      "learning_rate": 8.566002666774415e-06,
      "loss": 0.79,
      "step": 20610
    },
    {
      "epoch": 2.4894361946154775,
      "grad_norm": 1.6939870119094849,
      "learning_rate": 8.545799830296174e-06,
      "loss": 0.7376,
      "step": 20620
    },
    {
      "epoch": 2.490643486659423,
      "grad_norm": 1.4648795127868652,
      "learning_rate": 8.525596993817934e-06,
      "loss": 0.5227,
      "step": 20630
    },
    {
      "epoch": 2.4918507787033684,
      "grad_norm": 12.31555461883545,
      "learning_rate": 8.50539415733969e-06,
      "loss": 1.3715,
      "step": 20640
    },
    {
      "epoch": 2.493058070747314,
      "grad_norm": 11.914531707763672,
      "learning_rate": 8.485191320861449e-06,
      "loss": 0.5625,
      "step": 20650
    },
    {
      "epoch": 2.494265362791259,
      "grad_norm": 14.410798072814941,
      "learning_rate": 8.464988484383207e-06,
      "loss": 0.792,
      "step": 20660
    },
    {
      "epoch": 2.4954726548352046,
      "grad_norm": 14.349925994873047,
      "learning_rate": 8.444785647904966e-06,
      "loss": 0.91,
      "step": 20670
    },
    {
      "epoch": 2.49667994687915,
      "grad_norm": 13.984639167785645,
      "learning_rate": 8.424582811426724e-06,
      "loss": 1.1628,
      "step": 20680
    },
    {
      "epoch": 2.4978872389230955,
      "grad_norm": 36.73181915283203,
      "learning_rate": 8.404379974948483e-06,
      "loss": 0.7023,
      "step": 20690
    },
    {
      "epoch": 2.499094530967041,
      "grad_norm": 18.422996520996094,
      "learning_rate": 8.384177138470243e-06,
      "loss": 1.0438,
      "step": 20700
    },
    {
      "epoch": 2.5003018230109864,
      "grad_norm": 12.4799165725708,
      "learning_rate": 8.363974301992e-06,
      "loss": 0.7355,
      "step": 20710
    },
    {
      "epoch": 2.501509115054932,
      "grad_norm": 5.224386692047119,
      "learning_rate": 8.343771465513758e-06,
      "loss": 0.3713,
      "step": 20720
    },
    {
      "epoch": 2.502716407098877,
      "grad_norm": 25.518512725830078,
      "learning_rate": 8.323568629035517e-06,
      "loss": 0.6013,
      "step": 20730
    },
    {
      "epoch": 2.5039236991428226,
      "grad_norm": 15.83566951751709,
      "learning_rate": 8.303365792557275e-06,
      "loss": 1.1279,
      "step": 20740
    },
    {
      "epoch": 2.505130991186768,
      "grad_norm": 22.7626895904541,
      "learning_rate": 8.283162956079034e-06,
      "loss": 0.8112,
      "step": 20750
    },
    {
      "epoch": 2.5063382832307135,
      "grad_norm": 19.109132766723633,
      "learning_rate": 8.262960119600792e-06,
      "loss": 0.8416,
      "step": 20760
    },
    {
      "epoch": 2.507545575274659,
      "grad_norm": 9.886226654052734,
      "learning_rate": 8.24275728312255e-06,
      "loss": 0.7639,
      "step": 20770
    },
    {
      "epoch": 2.5087528673186044,
      "grad_norm": 14.067451477050781,
      "learning_rate": 8.222554446644309e-06,
      "loss": 0.6493,
      "step": 20780
    },
    {
      "epoch": 2.50996015936255,
      "grad_norm": 16.03148651123047,
      "learning_rate": 8.202351610166068e-06,
      "loss": 0.9007,
      "step": 20790
    },
    {
      "epoch": 2.511167451406495,
      "grad_norm": 15.664069175720215,
      "learning_rate": 8.182148773687826e-06,
      "loss": 1.0639,
      "step": 20800
    },
    {
      "epoch": 2.5123747434504407,
      "grad_norm": 14.832355499267578,
      "learning_rate": 8.161945937209585e-06,
      "loss": 0.947,
      "step": 20810
    },
    {
      "epoch": 2.513582035494386,
      "grad_norm": 15.165544509887695,
      "learning_rate": 8.141743100731343e-06,
      "loss": 0.6569,
      "step": 20820
    },
    {
      "epoch": 2.5147893275383315,
      "grad_norm": 17.382768630981445,
      "learning_rate": 8.121540264253102e-06,
      "loss": 0.4884,
      "step": 20830
    },
    {
      "epoch": 2.515996619582277,
      "grad_norm": 14.272574424743652,
      "learning_rate": 8.10133742777486e-06,
      "loss": 0.9502,
      "step": 20840
    },
    {
      "epoch": 2.5172039116262224,
      "grad_norm": 29.963232040405273,
      "learning_rate": 8.081134591296618e-06,
      "loss": 0.782,
      "step": 20850
    },
    {
      "epoch": 2.518411203670168,
      "grad_norm": 18.211957931518555,
      "learning_rate": 8.060931754818377e-06,
      "loss": 0.7017,
      "step": 20860
    },
    {
      "epoch": 2.5196184957141132,
      "grad_norm": 15.096739768981934,
      "learning_rate": 8.040728918340135e-06,
      "loss": 0.6045,
      "step": 20870
    },
    {
      "epoch": 2.5208257877580587,
      "grad_norm": 15.699153900146484,
      "learning_rate": 8.020526081861894e-06,
      "loss": 0.9796,
      "step": 20880
    },
    {
      "epoch": 2.522033079802004,
      "grad_norm": 0.2418670356273651,
      "learning_rate": 8.000323245383652e-06,
      "loss": 0.6352,
      "step": 20890
    },
    {
      "epoch": 2.5232403718459495,
      "grad_norm": 27.86214256286621,
      "learning_rate": 7.98012040890541e-06,
      "loss": 0.6732,
      "step": 20900
    },
    {
      "epoch": 2.524447663889895,
      "grad_norm": 1.0731587409973145,
      "learning_rate": 7.95991757242717e-06,
      "loss": 0.8222,
      "step": 20910
    },
    {
      "epoch": 2.5256549559338404,
      "grad_norm": 8.364757537841797,
      "learning_rate": 7.939714735948928e-06,
      "loss": 0.7526,
      "step": 20920
    },
    {
      "epoch": 2.526862247977786,
      "grad_norm": 12.873478889465332,
      "learning_rate": 7.919511899470686e-06,
      "loss": 0.512,
      "step": 20930
    },
    {
      "epoch": 2.5280695400217312,
      "grad_norm": 12.288305282592773,
      "learning_rate": 7.899309062992445e-06,
      "loss": 0.7,
      "step": 20940
    },
    {
      "epoch": 2.5292768320656767,
      "grad_norm": 10.543802261352539,
      "learning_rate": 7.879106226514203e-06,
      "loss": 0.7666,
      "step": 20950
    },
    {
      "epoch": 2.530484124109622,
      "grad_norm": 10.119218826293945,
      "learning_rate": 7.858903390035962e-06,
      "loss": 0.6627,
      "step": 20960
    },
    {
      "epoch": 2.5316914161535675,
      "grad_norm": 17.649805068969727,
      "learning_rate": 7.838700553557718e-06,
      "loss": 0.4714,
      "step": 20970
    },
    {
      "epoch": 2.532898708197513,
      "grad_norm": 19.682077407836914,
      "learning_rate": 7.818497717079479e-06,
      "loss": 0.641,
      "step": 20980
    },
    {
      "epoch": 2.5341060002414584,
      "grad_norm": 13.203289985656738,
      "learning_rate": 7.798294880601237e-06,
      "loss": 1.042,
      "step": 20990
    },
    {
      "epoch": 2.535313292285404,
      "grad_norm": 5.681002140045166,
      "learning_rate": 7.778092044122996e-06,
      "loss": 0.5737,
      "step": 21000
    },
    {
      "epoch": 2.5365205843293492,
      "grad_norm": 17.433528900146484,
      "learning_rate": 7.757889207644754e-06,
      "loss": 0.8796,
      "step": 21010
    },
    {
      "epoch": 2.5377278763732947,
      "grad_norm": 9.02531623840332,
      "learning_rate": 7.737686371166513e-06,
      "loss": 1.1943,
      "step": 21020
    },
    {
      "epoch": 2.53893516841724,
      "grad_norm": 22.013656616210938,
      "learning_rate": 7.717483534688271e-06,
      "loss": 1.0357,
      "step": 21030
    },
    {
      "epoch": 2.5401424604611855,
      "grad_norm": 42.19314193725586,
      "learning_rate": 7.697280698210028e-06,
      "loss": 0.9184,
      "step": 21040
    },
    {
      "epoch": 2.541349752505131,
      "grad_norm": 19.969501495361328,
      "learning_rate": 7.677077861731788e-06,
      "loss": 0.8445,
      "step": 21050
    },
    {
      "epoch": 2.5425570445490764,
      "grad_norm": 7.5420241355896,
      "learning_rate": 7.656875025253546e-06,
      "loss": 0.9934,
      "step": 21060
    },
    {
      "epoch": 2.543764336593022,
      "grad_norm": 34.81424331665039,
      "learning_rate": 7.636672188775303e-06,
      "loss": 0.8694,
      "step": 21070
    },
    {
      "epoch": 2.5449716286369672,
      "grad_norm": 9.339410781860352,
      "learning_rate": 7.616469352297063e-06,
      "loss": 0.9002,
      "step": 21080
    },
    {
      "epoch": 2.5461789206809127,
      "grad_norm": 18.08611488342285,
      "learning_rate": 7.596266515818822e-06,
      "loss": 0.6595,
      "step": 21090
    },
    {
      "epoch": 2.547386212724858,
      "grad_norm": 33.095947265625,
      "learning_rate": 7.5760636793405795e-06,
      "loss": 1.287,
      "step": 21100
    },
    {
      "epoch": 2.5485935047688035,
      "grad_norm": 14.032745361328125,
      "learning_rate": 7.555860842862338e-06,
      "loss": 0.8973,
      "step": 21110
    },
    {
      "epoch": 2.549800796812749,
      "grad_norm": 15.679180145263672,
      "learning_rate": 7.535658006384097e-06,
      "loss": 0.8488,
      "step": 21120
    },
    {
      "epoch": 2.5510080888566944,
      "grad_norm": 10.421177864074707,
      "learning_rate": 7.515455169905855e-06,
      "loss": 0.9424,
      "step": 21130
    },
    {
      "epoch": 2.55221538090064,
      "grad_norm": 19.8222599029541,
      "learning_rate": 7.495252333427613e-06,
      "loss": 0.8558,
      "step": 21140
    },
    {
      "epoch": 2.5534226729445852,
      "grad_norm": 34.61710739135742,
      "learning_rate": 7.475049496949373e-06,
      "loss": 0.7978,
      "step": 21150
    },
    {
      "epoch": 2.5546299649885307,
      "grad_norm": 23.281923294067383,
      "learning_rate": 7.45484666047113e-06,
      "loss": 1.0593,
      "step": 21160
    },
    {
      "epoch": 2.555837257032476,
      "grad_norm": 9.526586532592773,
      "learning_rate": 7.434643823992889e-06,
      "loss": 0.7167,
      "step": 21170
    },
    {
      "epoch": 2.5570445490764215,
      "grad_norm": 1.7221665382385254,
      "learning_rate": 7.4144409875146465e-06,
      "loss": 0.6353,
      "step": 21180
    },
    {
      "epoch": 2.558251841120367,
      "grad_norm": 7.90825080871582,
      "learning_rate": 7.394238151036406e-06,
      "loss": 0.8109,
      "step": 21190
    },
    {
      "epoch": 2.5594591331643124,
      "grad_norm": 1.6548681259155273,
      "learning_rate": 7.374035314558164e-06,
      "loss": 0.5614,
      "step": 21200
    },
    {
      "epoch": 2.560666425208258,
      "grad_norm": 7.425678730010986,
      "learning_rate": 7.353832478079922e-06,
      "loss": 0.941,
      "step": 21210
    },
    {
      "epoch": 2.5618737172522033,
      "grad_norm": 18.70587921142578,
      "learning_rate": 7.333629641601682e-06,
      "loss": 0.7559,
      "step": 21220
    },
    {
      "epoch": 2.5630810092961487,
      "grad_norm": 42.80250549316406,
      "learning_rate": 7.31342680512344e-06,
      "loss": 0.4001,
      "step": 21230
    },
    {
      "epoch": 2.564288301340094,
      "grad_norm": 1.8414965867996216,
      "learning_rate": 7.293223968645198e-06,
      "loss": 0.7491,
      "step": 21240
    },
    {
      "epoch": 2.5654955933840395,
      "grad_norm": 22.916446685791016,
      "learning_rate": 7.2730211321669575e-06,
      "loss": 0.3434,
      "step": 21250
    },
    {
      "epoch": 2.566702885427985,
      "grad_norm": 11.789255142211914,
      "learning_rate": 7.252818295688715e-06,
      "loss": 0.8725,
      "step": 21260
    },
    {
      "epoch": 2.5679101774719304,
      "grad_norm": 10.416468620300293,
      "learning_rate": 7.2326154592104736e-06,
      "loss": 0.788,
      "step": 21270
    },
    {
      "epoch": 2.569117469515876,
      "grad_norm": 25.362934112548828,
      "learning_rate": 7.212412622732231e-06,
      "loss": 1.1259,
      "step": 21280
    },
    {
      "epoch": 2.5703247615598213,
      "grad_norm": 25.47882652282715,
      "learning_rate": 7.1922097862539905e-06,
      "loss": 1.2122,
      "step": 21290
    },
    {
      "epoch": 2.5715320536037667,
      "grad_norm": 11.477636337280273,
      "learning_rate": 7.172006949775749e-06,
      "loss": 0.4335,
      "step": 21300
    },
    {
      "epoch": 2.572739345647712,
      "grad_norm": 24.8778133392334,
      "learning_rate": 7.151804113297507e-06,
      "loss": 0.6335,
      "step": 21310
    },
    {
      "epoch": 2.5739466376916575,
      "grad_norm": 4.9896931648254395,
      "learning_rate": 7.131601276819267e-06,
      "loss": 0.5271,
      "step": 21320
    },
    {
      "epoch": 2.575153929735603,
      "grad_norm": 19.90097427368164,
      "learning_rate": 7.111398440341024e-06,
      "loss": 0.6697,
      "step": 21330
    },
    {
      "epoch": 2.5763612217795484,
      "grad_norm": 6.671165466308594,
      "learning_rate": 7.091195603862783e-06,
      "loss": 0.5466,
      "step": 21340
    },
    {
      "epoch": 2.577568513823494,
      "grad_norm": 15.014126777648926,
      "learning_rate": 7.0709927673845405e-06,
      "loss": 0.4604,
      "step": 21350
    },
    {
      "epoch": 2.5787758058674393,
      "grad_norm": 16.973134994506836,
      "learning_rate": 7.0507899309063e-06,
      "loss": 1.183,
      "step": 21360
    },
    {
      "epoch": 2.5799830979113847,
      "grad_norm": 24.751440048217773,
      "learning_rate": 7.030587094428058e-06,
      "loss": 1.0679,
      "step": 21370
    },
    {
      "epoch": 2.58119038995533,
      "grad_norm": 10.663359642028809,
      "learning_rate": 7.010384257949816e-06,
      "loss": 0.5889,
      "step": 21380
    },
    {
      "epoch": 2.5823976819992756,
      "grad_norm": 24.676036834716797,
      "learning_rate": 6.990181421471575e-06,
      "loss": 0.9283,
      "step": 21390
    },
    {
      "epoch": 2.583604974043221,
      "grad_norm": 37.493709564208984,
      "learning_rate": 6.969978584993334e-06,
      "loss": 0.6721,
      "step": 21400
    },
    {
      "epoch": 2.5848122660871664,
      "grad_norm": 34.83772659301758,
      "learning_rate": 6.949775748515091e-06,
      "loss": 0.4764,
      "step": 21410
    },
    {
      "epoch": 2.586019558131112,
      "grad_norm": 50.03681182861328,
      "learning_rate": 6.92957291203685e-06,
      "loss": 1.1852,
      "step": 21420
    },
    {
      "epoch": 2.5872268501750573,
      "grad_norm": 20.607744216918945,
      "learning_rate": 6.909370075558609e-06,
      "loss": 0.7311,
      "step": 21430
    },
    {
      "epoch": 2.5884341422190027,
      "grad_norm": 35.95832061767578,
      "learning_rate": 6.889167239080367e-06,
      "loss": 0.5612,
      "step": 21440
    },
    {
      "epoch": 2.589641434262948,
      "grad_norm": 0.6677225828170776,
      "learning_rate": 6.868964402602125e-06,
      "loss": 0.4974,
      "step": 21450
    },
    {
      "epoch": 2.5908487263068936,
      "grad_norm": 2.140782117843628,
      "learning_rate": 6.848761566123885e-06,
      "loss": 0.6996,
      "step": 21460
    },
    {
      "epoch": 2.592056018350839,
      "grad_norm": 18.28948974609375,
      "learning_rate": 6.828558729645643e-06,
      "loss": 0.8823,
      "step": 21470
    },
    {
      "epoch": 2.5932633103947844,
      "grad_norm": 21.93641471862793,
      "learning_rate": 6.808355893167401e-06,
      "loss": 0.5619,
      "step": 21480
    },
    {
      "epoch": 2.59447060243873,
      "grad_norm": 11.315455436706543,
      "learning_rate": 6.788153056689159e-06,
      "loss": 0.4752,
      "step": 21490
    },
    {
      "epoch": 2.5956778944826753,
      "grad_norm": 25.855695724487305,
      "learning_rate": 6.7679502202109185e-06,
      "loss": 1.2514,
      "step": 21500
    },
    {
      "epoch": 2.5968851865266207,
      "grad_norm": 40.19979476928711,
      "learning_rate": 6.747747383732676e-06,
      "loss": 0.9986,
      "step": 21510
    },
    {
      "epoch": 2.598092478570566,
      "grad_norm": 7.479718208312988,
      "learning_rate": 6.727544547254435e-06,
      "loss": 0.8575,
      "step": 21520
    },
    {
      "epoch": 2.5992997706145116,
      "grad_norm": 5.172685146331787,
      "learning_rate": 6.707341710776194e-06,
      "loss": 0.6925,
      "step": 21530
    },
    {
      "epoch": 2.600507062658457,
      "grad_norm": 37.2103271484375,
      "learning_rate": 6.6871388742979516e-06,
      "loss": 0.9787,
      "step": 21540
    },
    {
      "epoch": 2.6017143547024024,
      "grad_norm": 4.390345096588135,
      "learning_rate": 6.66693603781971e-06,
      "loss": 1.3472,
      "step": 21550
    },
    {
      "epoch": 2.602921646746348,
      "grad_norm": 28.289165496826172,
      "learning_rate": 6.646733201341469e-06,
      "loss": 0.6121,
      "step": 21560
    },
    {
      "epoch": 2.6041289387902933,
      "grad_norm": 15.711234092712402,
      "learning_rate": 6.626530364863227e-06,
      "loss": 0.8643,
      "step": 21570
    },
    {
      "epoch": 2.6053362308342387,
      "grad_norm": 3.849323034286499,
      "learning_rate": 6.6063275283849855e-06,
      "loss": 1.0102,
      "step": 21580
    },
    {
      "epoch": 2.606543522878184,
      "grad_norm": 1.8752388954162598,
      "learning_rate": 6.586124691906743e-06,
      "loss": 0.3514,
      "step": 21590
    },
    {
      "epoch": 2.6077508149221296,
      "grad_norm": 23.641170501708984,
      "learning_rate": 6.565921855428503e-06,
      "loss": 0.8645,
      "step": 21600
    },
    {
      "epoch": 2.608958106966075,
      "grad_norm": 3.818894624710083,
      "learning_rate": 6.545719018950261e-06,
      "loss": 0.5107,
      "step": 21610
    },
    {
      "epoch": 2.6101653990100204,
      "grad_norm": 3.587418794631958,
      "learning_rate": 6.525516182472019e-06,
      "loss": 1.2436,
      "step": 21620
    },
    {
      "epoch": 2.611372691053966,
      "grad_norm": 20.501976013183594,
      "learning_rate": 6.505313345993779e-06,
      "loss": 0.6203,
      "step": 21630
    },
    {
      "epoch": 2.6125799830979113,
      "grad_norm": 1.8819905519485474,
      "learning_rate": 6.485110509515536e-06,
      "loss": 0.5239,
      "step": 21640
    },
    {
      "epoch": 2.6137872751418567,
      "grad_norm": 10.06116771697998,
      "learning_rate": 6.464907673037295e-06,
      "loss": 0.5731,
      "step": 21650
    },
    {
      "epoch": 2.614994567185802,
      "grad_norm": 10.655029296875,
      "learning_rate": 6.444704836559052e-06,
      "loss": 0.8072,
      "step": 21660
    },
    {
      "epoch": 2.6162018592297476,
      "grad_norm": 16.763492584228516,
      "learning_rate": 6.424502000080812e-06,
      "loss": 0.5444,
      "step": 21670
    },
    {
      "epoch": 2.617409151273693,
      "grad_norm": 5.307007789611816,
      "learning_rate": 6.40429916360257e-06,
      "loss": 0.7661,
      "step": 21680
    },
    {
      "epoch": 2.6186164433176384,
      "grad_norm": 10.584131240844727,
      "learning_rate": 6.384096327124328e-06,
      "loss": 1.0618,
      "step": 21690
    },
    {
      "epoch": 2.619823735361584,
      "grad_norm": 37.31357192993164,
      "learning_rate": 6.363893490646088e-06,
      "loss": 0.6135,
      "step": 21700
    },
    {
      "epoch": 2.6210310274055293,
      "grad_norm": 39.747859954833984,
      "learning_rate": 6.343690654167846e-06,
      "loss": 1.1229,
      "step": 21710
    },
    {
      "epoch": 2.6222383194494747,
      "grad_norm": 42.61591339111328,
      "learning_rate": 6.323487817689604e-06,
      "loss": 0.6014,
      "step": 21720
    },
    {
      "epoch": 2.62344561149342,
      "grad_norm": 13.81668758392334,
      "learning_rate": 6.303284981211362e-06,
      "loss": 0.6797,
      "step": 21730
    },
    {
      "epoch": 2.6246529035373656,
      "grad_norm": 22.121850967407227,
      "learning_rate": 6.283082144733121e-06,
      "loss": 0.6744,
      "step": 21740
    },
    {
      "epoch": 2.625860195581311,
      "grad_norm": 23.151451110839844,
      "learning_rate": 6.2628793082548795e-06,
      "loss": 0.8662,
      "step": 21750
    },
    {
      "epoch": 2.6270674876252564,
      "grad_norm": 15.004400253295898,
      "learning_rate": 6.242676471776638e-06,
      "loss": 0.7629,
      "step": 21760
    },
    {
      "epoch": 2.628274779669202,
      "grad_norm": 16.18265724182129,
      "learning_rate": 6.222473635298396e-06,
      "loss": 0.5219,
      "step": 21770
    },
    {
      "epoch": 2.6294820717131473,
      "grad_norm": 37.128536224365234,
      "learning_rate": 6.202270798820155e-06,
      "loss": 0.9139,
      "step": 21780
    },
    {
      "epoch": 2.6306893637570927,
      "grad_norm": 9.686922073364258,
      "learning_rate": 6.182067962341913e-06,
      "loss": 0.5173,
      "step": 21790
    },
    {
      "epoch": 2.631896655801038,
      "grad_norm": 39.615509033203125,
      "learning_rate": 6.161865125863672e-06,
      "loss": 0.683,
      "step": 21800
    },
    {
      "epoch": 2.6331039478449836,
      "grad_norm": 2.2661585807800293,
      "learning_rate": 6.14166228938543e-06,
      "loss": 0.7564,
      "step": 21810
    },
    {
      "epoch": 2.634311239888929,
      "grad_norm": 37.90367126464844,
      "learning_rate": 6.121459452907188e-06,
      "loss": 0.7923,
      "step": 21820
    },
    {
      "epoch": 2.6355185319328744,
      "grad_norm": 22.052204132080078,
      "learning_rate": 6.101256616428947e-06,
      "loss": 1.1014,
      "step": 21830
    },
    {
      "epoch": 2.63672582397682,
      "grad_norm": 52.07131576538086,
      "learning_rate": 6.081053779950705e-06,
      "loss": 0.9888,
      "step": 21840
    },
    {
      "epoch": 2.6379331160207653,
      "grad_norm": 8.747702598571777,
      "learning_rate": 6.060850943472464e-06,
      "loss": 0.6013,
      "step": 21850
    },
    {
      "epoch": 2.6391404080647107,
      "grad_norm": 11.929267883300781,
      "learning_rate": 6.040648106994222e-06,
      "loss": 1.0268,
      "step": 21860
    },
    {
      "epoch": 2.640347700108656,
      "grad_norm": 17.035816192626953,
      "learning_rate": 6.02044527051598e-06,
      "loss": 0.5285,
      "step": 21870
    },
    {
      "epoch": 2.6415549921526016,
      "grad_norm": 43.598087310791016,
      "learning_rate": 6.00024243403774e-06,
      "loss": 0.596,
      "step": 21880
    },
    {
      "epoch": 2.642762284196547,
      "grad_norm": 16.66098403930664,
      "learning_rate": 5.980039597559497e-06,
      "loss": 0.7788,
      "step": 21890
    },
    {
      "epoch": 2.6439695762404924,
      "grad_norm": 90.47927856445312,
      "learning_rate": 5.959836761081257e-06,
      "loss": 0.6598,
      "step": 21900
    },
    {
      "epoch": 2.645176868284438,
      "grad_norm": 9.756113052368164,
      "learning_rate": 5.939633924603014e-06,
      "loss": 1.0394,
      "step": 21910
    },
    {
      "epoch": 2.6463841603283833,
      "grad_norm": 46.75081253051758,
      "learning_rate": 5.919431088124773e-06,
      "loss": 1.0741,
      "step": 21920
    },
    {
      "epoch": 2.6475914523723287,
      "grad_norm": 10.826495170593262,
      "learning_rate": 5.899228251646532e-06,
      "loss": 0.8872,
      "step": 21930
    },
    {
      "epoch": 2.648798744416274,
      "grad_norm": 34.67647171020508,
      "learning_rate": 5.87902541516829e-06,
      "loss": 0.84,
      "step": 21940
    },
    {
      "epoch": 2.6500060364602196,
      "grad_norm": 13.072074890136719,
      "learning_rate": 5.858822578690048e-06,
      "loss": 0.7792,
      "step": 21950
    },
    {
      "epoch": 2.651213328504165,
      "grad_norm": 10.584209442138672,
      "learning_rate": 5.838619742211807e-06,
      "loss": 0.5378,
      "step": 21960
    },
    {
      "epoch": 2.6524206205481105,
      "grad_norm": 7.0539093017578125,
      "learning_rate": 5.818416905733565e-06,
      "loss": 0.9544,
      "step": 21970
    },
    {
      "epoch": 2.653627912592056,
      "grad_norm": 26.449941635131836,
      "learning_rate": 5.798214069255324e-06,
      "loss": 1.0013,
      "step": 21980
    },
    {
      "epoch": 2.6548352046360013,
      "grad_norm": 11.0072603225708,
      "learning_rate": 5.778011232777082e-06,
      "loss": 0.6792,
      "step": 21990
    },
    {
      "epoch": 2.6560424966799467,
      "grad_norm": 3.823835611343384,
      "learning_rate": 5.7578083962988406e-06,
      "loss": 1.1295,
      "step": 22000
    },
    {
      "epoch": 2.657249788723892,
      "grad_norm": 31.331003189086914,
      "learning_rate": 5.737605559820599e-06,
      "loss": 1.163,
      "step": 22010
    },
    {
      "epoch": 2.6584570807678376,
      "grad_norm": 6.871838092803955,
      "learning_rate": 5.7174027233423575e-06,
      "loss": 0.6119,
      "step": 22020
    },
    {
      "epoch": 2.659664372811783,
      "grad_norm": 7.261454105377197,
      "learning_rate": 5.697199886864116e-06,
      "loss": 0.4298,
      "step": 22030
    },
    {
      "epoch": 2.6608716648557285,
      "grad_norm": 9.041485786437988,
      "learning_rate": 5.6769970503858745e-06,
      "loss": 0.2934,
      "step": 22040
    },
    {
      "epoch": 2.662078956899674,
      "grad_norm": 3.6454291343688965,
      "learning_rate": 5.656794213907633e-06,
      "loss": 0.7424,
      "step": 22050
    },
    {
      "epoch": 2.6632862489436193,
      "grad_norm": 8.570298194885254,
      "learning_rate": 5.636591377429391e-06,
      "loss": 0.9196,
      "step": 22060
    },
    {
      "epoch": 2.6644935409875647,
      "grad_norm": 10.062475204467773,
      "learning_rate": 5.61638854095115e-06,
      "loss": 0.723,
      "step": 22070
    },
    {
      "epoch": 2.66570083303151,
      "grad_norm": 11.89646053314209,
      "learning_rate": 5.596185704472908e-06,
      "loss": 0.7289,
      "step": 22080
    },
    {
      "epoch": 2.6669081250754556,
      "grad_norm": 4.510172367095947,
      "learning_rate": 5.575982867994667e-06,
      "loss": 0.5003,
      "step": 22090
    },
    {
      "epoch": 2.668115417119401,
      "grad_norm": 28.0794677734375,
      "learning_rate": 5.555780031516425e-06,
      "loss": 0.688,
      "step": 22100
    },
    {
      "epoch": 2.6693227091633465,
      "grad_norm": 18.899871826171875,
      "learning_rate": 5.535577195038184e-06,
      "loss": 0.5538,
      "step": 22110
    },
    {
      "epoch": 2.670530001207292,
      "grad_norm": 16.590370178222656,
      "learning_rate": 5.515374358559942e-06,
      "loss": 0.9647,
      "step": 22120
    },
    {
      "epoch": 2.6717372932512373,
      "grad_norm": 4.004922866821289,
      "learning_rate": 5.495171522081701e-06,
      "loss": 0.6413,
      "step": 22130
    },
    {
      "epoch": 2.6729445852951828,
      "grad_norm": 9.163948059082031,
      "learning_rate": 5.474968685603459e-06,
      "loss": 0.8385,
      "step": 22140
    },
    {
      "epoch": 2.674151877339128,
      "grad_norm": 10.122897148132324,
      "learning_rate": 5.454765849125217e-06,
      "loss": 0.9421,
      "step": 22150
    },
    {
      "epoch": 2.6753591693830736,
      "grad_norm": 8.353696823120117,
      "learning_rate": 5.434563012646976e-06,
      "loss": 0.9162,
      "step": 22160
    },
    {
      "epoch": 2.676566461427019,
      "grad_norm": 13.771352767944336,
      "learning_rate": 5.414360176168734e-06,
      "loss": 0.5213,
      "step": 22170
    },
    {
      "epoch": 2.6777737534709645,
      "grad_norm": 13.587998390197754,
      "learning_rate": 5.394157339690493e-06,
      "loss": 0.9663,
      "step": 22180
    },
    {
      "epoch": 2.67898104551491,
      "grad_norm": 16.773483276367188,
      "learning_rate": 5.373954503212252e-06,
      "loss": 0.7415,
      "step": 22190
    },
    {
      "epoch": 2.6801883375588553,
      "grad_norm": 5.1120123863220215,
      "learning_rate": 5.353751666734009e-06,
      "loss": 0.9006,
      "step": 22200
    },
    {
      "epoch": 2.6813956296028008,
      "grad_norm": 43.48530960083008,
      "learning_rate": 5.3335488302557685e-06,
      "loss": 1.0392,
      "step": 22210
    },
    {
      "epoch": 2.682602921646746,
      "grad_norm": 21.95634651184082,
      "learning_rate": 5.313345993777526e-06,
      "loss": 0.5027,
      "step": 22220
    },
    {
      "epoch": 2.6838102136906916,
      "grad_norm": 12.064799308776855,
      "learning_rate": 5.2931431572992855e-06,
      "loss": 0.4748,
      "step": 22230
    },
    {
      "epoch": 2.685017505734637,
      "grad_norm": 30.969308853149414,
      "learning_rate": 5.272940320821044e-06,
      "loss": 0.9852,
      "step": 22240
    },
    {
      "epoch": 2.6862247977785825,
      "grad_norm": 45.49223709106445,
      "learning_rate": 5.252737484342802e-06,
      "loss": 0.6054,
      "step": 22250
    },
    {
      "epoch": 2.687432089822528,
      "grad_norm": 3.3966240882873535,
      "learning_rate": 5.232534647864561e-06,
      "loss": 0.5686,
      "step": 22260
    },
    {
      "epoch": 2.6886393818664738,
      "grad_norm": 33.65557861328125,
      "learning_rate": 5.2123318113863185e-06,
      "loss": 1.2363,
      "step": 22270
    },
    {
      "epoch": 2.689846673910419,
      "grad_norm": 18.9597225189209,
      "learning_rate": 5.192128974908078e-06,
      "loss": 0.7587,
      "step": 22280
    },
    {
      "epoch": 2.6910539659543646,
      "grad_norm": 17.125408172607422,
      "learning_rate": 5.1719261384298355e-06,
      "loss": 0.8244,
      "step": 22290
    },
    {
      "epoch": 2.69226125799831,
      "grad_norm": 15.988890647888184,
      "learning_rate": 5.151723301951594e-06,
      "loss": 0.7396,
      "step": 22300
    },
    {
      "epoch": 2.6934685500422555,
      "grad_norm": 4.345247745513916,
      "learning_rate": 5.131520465473353e-06,
      "loss": 0.7626,
      "step": 22310
    },
    {
      "epoch": 2.694675842086201,
      "grad_norm": 32.8530387878418,
      "learning_rate": 5.111317628995111e-06,
      "loss": 0.5415,
      "step": 22320
    },
    {
      "epoch": 2.6958831341301464,
      "grad_norm": 4.431756973266602,
      "learning_rate": 5.091114792516869e-06,
      "loss": 0.496,
      "step": 22330
    },
    {
      "epoch": 2.697090426174092,
      "grad_norm": 57.875030517578125,
      "learning_rate": 5.070911956038628e-06,
      "loss": 0.8774,
      "step": 22340
    },
    {
      "epoch": 2.698297718218037,
      "grad_norm": 15.151579856872559,
      "learning_rate": 5.050709119560386e-06,
      "loss": 0.6397,
      "step": 22350
    },
    {
      "epoch": 2.6995050102619826,
      "grad_norm": 5.054754734039307,
      "learning_rate": 5.030506283082145e-06,
      "loss": 1.4778,
      "step": 22360
    },
    {
      "epoch": 2.700712302305928,
      "grad_norm": 1.2760769128799438,
      "learning_rate": 5.010303446603903e-06,
      "loss": 0.9192,
      "step": 22370
    },
    {
      "epoch": 2.7019195943498735,
      "grad_norm": 13.200732231140137,
      "learning_rate": 4.990100610125662e-06,
      "loss": 0.6383,
      "step": 22380
    },
    {
      "epoch": 2.703126886393819,
      "grad_norm": 14.771918296813965,
      "learning_rate": 4.96989777364742e-06,
      "loss": 0.4579,
      "step": 22390
    },
    {
      "epoch": 2.7043341784377644,
      "grad_norm": 15.872856140136719,
      "learning_rate": 4.949694937169179e-06,
      "loss": 0.654,
      "step": 22400
    },
    {
      "epoch": 2.70554147048171,
      "grad_norm": 10.725876808166504,
      "learning_rate": 4.929492100690937e-06,
      "loss": 1.0454,
      "step": 22410
    },
    {
      "epoch": 2.706748762525655,
      "grad_norm": 13.3458890914917,
      "learning_rate": 4.909289264212696e-06,
      "loss": 0.8029,
      "step": 22420
    },
    {
      "epoch": 2.7079560545696006,
      "grad_norm": 12.241157531738281,
      "learning_rate": 4.889086427734454e-06,
      "loss": 0.7829,
      "step": 22430
    },
    {
      "epoch": 2.709163346613546,
      "grad_norm": 3.475360631942749,
      "learning_rate": 4.868883591256213e-06,
      "loss": 0.4523,
      "step": 22440
    },
    {
      "epoch": 2.7103706386574915,
      "grad_norm": 16.272232055664062,
      "learning_rate": 4.848680754777971e-06,
      "loss": 0.7703,
      "step": 22450
    },
    {
      "epoch": 2.711577930701437,
      "grad_norm": 20.661945343017578,
      "learning_rate": 4.8284779182997296e-06,
      "loss": 0.6478,
      "step": 22460
    },
    {
      "epoch": 2.7127852227453824,
      "grad_norm": 17.8594970703125,
      "learning_rate": 4.808275081821488e-06,
      "loss": 0.5175,
      "step": 22470
    },
    {
      "epoch": 2.713992514789328,
      "grad_norm": 17.136770248413086,
      "learning_rate": 4.7880722453432465e-06,
      "loss": 0.6361,
      "step": 22480
    },
    {
      "epoch": 2.715199806833273,
      "grad_norm": 26.285316467285156,
      "learning_rate": 4.767869408865005e-06,
      "loss": 1.1569,
      "step": 22490
    },
    {
      "epoch": 2.7164070988772187,
      "grad_norm": 23.89514923095703,
      "learning_rate": 4.7476665723867635e-06,
      "loss": 1.0319,
      "step": 22500
    },
    {
      "epoch": 2.717614390921164,
      "grad_norm": 20.616191864013672,
      "learning_rate": 4.727463735908522e-06,
      "loss": 1.1338,
      "step": 22510
    },
    {
      "epoch": 2.7188216829651095,
      "grad_norm": 22.41743278503418,
      "learning_rate": 4.70726089943028e-06,
      "loss": 0.8207,
      "step": 22520
    },
    {
      "epoch": 2.720028975009055,
      "grad_norm": 35.32134246826172,
      "learning_rate": 4.687058062952038e-06,
      "loss": 0.7702,
      "step": 22530
    },
    {
      "epoch": 2.7212362670530004,
      "grad_norm": 20.113290786743164,
      "learning_rate": 4.666855226473797e-06,
      "loss": 0.4736,
      "step": 22540
    },
    {
      "epoch": 2.722443559096946,
      "grad_norm": 22.93744468688965,
      "learning_rate": 4.646652389995556e-06,
      "loss": 0.9117,
      "step": 22550
    },
    {
      "epoch": 2.7236508511408912,
      "grad_norm": 25.353879928588867,
      "learning_rate": 4.626449553517314e-06,
      "loss": 0.7191,
      "step": 22560
    },
    {
      "epoch": 2.7248581431848367,
      "grad_norm": 20.134077072143555,
      "learning_rate": 4.606246717039073e-06,
      "loss": 0.5637,
      "step": 22570
    },
    {
      "epoch": 2.726065435228782,
      "grad_norm": 11.270496368408203,
      "learning_rate": 4.58604388056083e-06,
      "loss": 0.4035,
      "step": 22580
    },
    {
      "epoch": 2.7272727272727275,
      "grad_norm": 5.6114726066589355,
      "learning_rate": 4.56584104408259e-06,
      "loss": 0.4772,
      "step": 22590
    },
    {
      "epoch": 2.728480019316673,
      "grad_norm": 27.861053466796875,
      "learning_rate": 4.545638207604347e-06,
      "loss": 0.6486,
      "step": 22600
    },
    {
      "epoch": 2.7296873113606184,
      "grad_norm": 27.45098876953125,
      "learning_rate": 4.525435371126107e-06,
      "loss": 1.0234,
      "step": 22610
    },
    {
      "epoch": 2.730894603404564,
      "grad_norm": 5.040131092071533,
      "learning_rate": 4.505232534647865e-06,
      "loss": 0.7066,
      "step": 22620
    },
    {
      "epoch": 2.7321018954485092,
      "grad_norm": 11.255647659301758,
      "learning_rate": 4.485029698169623e-06,
      "loss": 0.6541,
      "step": 22630
    },
    {
      "epoch": 2.7333091874924547,
      "grad_norm": 29.218666076660156,
      "learning_rate": 4.464826861691382e-06,
      "loss": 0.9034,
      "step": 22640
    },
    {
      "epoch": 2.7345164795364,
      "grad_norm": 25.444873809814453,
      "learning_rate": 4.44462402521314e-06,
      "loss": 0.4275,
      "step": 22650
    },
    {
      "epoch": 2.7357237715803455,
      "grad_norm": 12.462470054626465,
      "learning_rate": 4.424421188734899e-06,
      "loss": 1.0304,
      "step": 22660
    },
    {
      "epoch": 2.736931063624291,
      "grad_norm": 39.20281982421875,
      "learning_rate": 4.404218352256657e-06,
      "loss": 0.7827,
      "step": 22670
    },
    {
      "epoch": 2.7381383556682364,
      "grad_norm": 16.545150756835938,
      "learning_rate": 4.384015515778415e-06,
      "loss": 0.9684,
      "step": 22680
    },
    {
      "epoch": 2.739345647712182,
      "grad_norm": 27.205568313598633,
      "learning_rate": 4.3638126793001745e-06,
      "loss": 0.9906,
      "step": 22690
    },
    {
      "epoch": 2.7405529397561272,
      "grad_norm": 12.680768013000488,
      "learning_rate": 4.343609842821932e-06,
      "loss": 1.28,
      "step": 22700
    },
    {
      "epoch": 2.7417602318000727,
      "grad_norm": 6.215914726257324,
      "learning_rate": 4.323407006343691e-06,
      "loss": 0.9242,
      "step": 22710
    },
    {
      "epoch": 2.742967523844018,
      "grad_norm": 25.616973876953125,
      "learning_rate": 4.303204169865449e-06,
      "loss": 0.7066,
      "step": 22720
    },
    {
      "epoch": 2.7441748158879635,
      "grad_norm": 12.596272468566895,
      "learning_rate": 4.2830013333872075e-06,
      "loss": 0.8823,
      "step": 22730
    },
    {
      "epoch": 2.745382107931909,
      "grad_norm": 54.96880340576172,
      "learning_rate": 4.262798496908967e-06,
      "loss": 0.702,
      "step": 22740
    },
    {
      "epoch": 2.7465893999758544,
      "grad_norm": 15.413534164428711,
      "learning_rate": 4.2425956604307245e-06,
      "loss": 0.9565,
      "step": 22750
    },
    {
      "epoch": 2.7477966920198,
      "grad_norm": 31.680713653564453,
      "learning_rate": 4.222392823952483e-06,
      "loss": 1.2125,
      "step": 22760
    },
    {
      "epoch": 2.7490039840637452,
      "grad_norm": 10.21345043182373,
      "learning_rate": 4.2021899874742414e-06,
      "loss": 0.7425,
      "step": 22770
    },
    {
      "epoch": 2.7502112761076907,
      "grad_norm": 21.1957950592041,
      "learning_rate": 4.181987150996e-06,
      "loss": 0.596,
      "step": 22780
    },
    {
      "epoch": 2.751418568151636,
      "grad_norm": 16.264373779296875,
      "learning_rate": 4.161784314517758e-06,
      "loss": 0.913,
      "step": 22790
    },
    {
      "epoch": 2.7526258601955815,
      "grad_norm": 4.253608703613281,
      "learning_rate": 4.141581478039517e-06,
      "loss": 0.3829,
      "step": 22800
    },
    {
      "epoch": 2.753833152239527,
      "grad_norm": 20.422269821166992,
      "learning_rate": 4.121378641561275e-06,
      "loss": 0.9628,
      "step": 22810
    },
    {
      "epoch": 2.7550404442834724,
      "grad_norm": 42.118228912353516,
      "learning_rate": 4.101175805083034e-06,
      "loss": 0.6965,
      "step": 22820
    },
    {
      "epoch": 2.756247736327418,
      "grad_norm": 8.46745491027832,
      "learning_rate": 4.080972968604792e-06,
      "loss": 0.6751,
      "step": 22830
    },
    {
      "epoch": 2.7574550283713632,
      "grad_norm": 11.796112060546875,
      "learning_rate": 4.060770132126551e-06,
      "loss": 0.8597,
      "step": 22840
    },
    {
      "epoch": 2.7586623204153087,
      "grad_norm": 6.251893520355225,
      "learning_rate": 4.040567295648309e-06,
      "loss": 1.0892,
      "step": 22850
    },
    {
      "epoch": 2.759869612459254,
      "grad_norm": 12.141739845275879,
      "learning_rate": 4.020364459170068e-06,
      "loss": 0.745,
      "step": 22860
    },
    {
      "epoch": 2.7610769045031995,
      "grad_norm": 7.207858085632324,
      "learning_rate": 4.000161622691826e-06,
      "loss": 0.6214,
      "step": 22870
    },
    {
      "epoch": 2.762284196547145,
      "grad_norm": 18.308454513549805,
      "learning_rate": 3.979958786213585e-06,
      "loss": 0.652,
      "step": 22880
    },
    {
      "epoch": 2.7634914885910904,
      "grad_norm": 0.6269408464431763,
      "learning_rate": 3.959755949735343e-06,
      "loss": 1.0198,
      "step": 22890
    },
    {
      "epoch": 2.764698780635036,
      "grad_norm": 18.16840934753418,
      "learning_rate": 3.939553113257102e-06,
      "loss": 0.4167,
      "step": 22900
    },
    {
      "epoch": 2.7659060726789813,
      "grad_norm": 27.222782135009766,
      "learning_rate": 3.919350276778859e-06,
      "loss": 0.8005,
      "step": 22910
    },
    {
      "epoch": 2.7671133647229267,
      "grad_norm": 2.03397274017334,
      "learning_rate": 3.8991474403006186e-06,
      "loss": 0.5999,
      "step": 22920
    },
    {
      "epoch": 2.768320656766872,
      "grad_norm": 3.166330337524414,
      "learning_rate": 3.878944603822377e-06,
      "loss": 0.6421,
      "step": 22930
    },
    {
      "epoch": 2.7695279488108175,
      "grad_norm": 33.515419006347656,
      "learning_rate": 3.8587417673441355e-06,
      "loss": 0.734,
      "step": 22940
    },
    {
      "epoch": 2.770735240854763,
      "grad_norm": 22.30040168762207,
      "learning_rate": 3.838538930865894e-06,
      "loss": 0.9398,
      "step": 22950
    },
    {
      "epoch": 2.7719425328987084,
      "grad_norm": 2.5903422832489014,
      "learning_rate": 3.818336094387652e-06,
      "loss": 0.558,
      "step": 22960
    },
    {
      "epoch": 2.773149824942654,
      "grad_norm": 11.485790252685547,
      "learning_rate": 3.798133257909411e-06,
      "loss": 0.5226,
      "step": 22970
    },
    {
      "epoch": 2.7743571169865993,
      "grad_norm": 44.14986801147461,
      "learning_rate": 3.777930421431169e-06,
      "loss": 0.8819,
      "step": 22980
    },
    {
      "epoch": 2.7755644090305447,
      "grad_norm": 10.0298433303833,
      "learning_rate": 3.7577275849529275e-06,
      "loss": 1.1996,
      "step": 22990
    },
    {
      "epoch": 2.77677170107449,
      "grad_norm": 15.662872314453125,
      "learning_rate": 3.7375247484746864e-06,
      "loss": 0.8833,
      "step": 23000
    },
    {
      "epoch": 2.7779789931184355,
      "grad_norm": 2.975010633468628,
      "learning_rate": 3.7173219119964444e-06,
      "loss": 1.0899,
      "step": 23010
    },
    {
      "epoch": 2.779186285162381,
      "grad_norm": 2.8923165798187256,
      "learning_rate": 3.697119075518203e-06,
      "loss": 0.6546,
      "step": 23020
    },
    {
      "epoch": 2.7803935772063264,
      "grad_norm": 13.618193626403809,
      "learning_rate": 3.676916239039961e-06,
      "loss": 0.4393,
      "step": 23030
    },
    {
      "epoch": 2.781600869250272,
      "grad_norm": 21.681699752807617,
      "learning_rate": 3.65671340256172e-06,
      "loss": 0.5826,
      "step": 23040
    },
    {
      "epoch": 2.7828081612942173,
      "grad_norm": 16.252573013305664,
      "learning_rate": 3.6365105660834787e-06,
      "loss": 0.6453,
      "step": 23050
    },
    {
      "epoch": 2.7840154533381627,
      "grad_norm": 21.905851364135742,
      "learning_rate": 3.6163077296052368e-06,
      "loss": 1.052,
      "step": 23060
    },
    {
      "epoch": 2.785222745382108,
      "grad_norm": 24.507797241210938,
      "learning_rate": 3.5961048931269953e-06,
      "loss": 0.9223,
      "step": 23070
    },
    {
      "epoch": 2.7864300374260536,
      "grad_norm": 1.465598225593567,
      "learning_rate": 3.5759020566487533e-06,
      "loss": 0.4018,
      "step": 23080
    },
    {
      "epoch": 2.787637329469999,
      "grad_norm": 11.160685539245605,
      "learning_rate": 3.555699220170512e-06,
      "loss": 0.5132,
      "step": 23090
    },
    {
      "epoch": 2.7888446215139444,
      "grad_norm": 24.53386688232422,
      "learning_rate": 3.5354963836922703e-06,
      "loss": 0.7379,
      "step": 23100
    },
    {
      "epoch": 2.79005191355789,
      "grad_norm": 22.422592163085938,
      "learning_rate": 3.515293547214029e-06,
      "loss": 0.6614,
      "step": 23110
    },
    {
      "epoch": 2.7912592056018353,
      "grad_norm": 21.590412139892578,
      "learning_rate": 3.4950907107357876e-06,
      "loss": 1.0693,
      "step": 23120
    },
    {
      "epoch": 2.7924664976457807,
      "grad_norm": 3.136813163757324,
      "learning_rate": 3.4748878742575457e-06,
      "loss": 0.5243,
      "step": 23130
    },
    {
      "epoch": 2.793673789689726,
      "grad_norm": 17.062559127807617,
      "learning_rate": 3.4546850377793046e-06,
      "loss": 0.876,
      "step": 23140
    },
    {
      "epoch": 2.7948810817336716,
      "grad_norm": 20.234739303588867,
      "learning_rate": 3.4344822013010626e-06,
      "loss": 1.0023,
      "step": 23150
    },
    {
      "epoch": 2.796088373777617,
      "grad_norm": 27.921627044677734,
      "learning_rate": 3.4142793648228215e-06,
      "loss": 0.7642,
      "step": 23160
    },
    {
      "epoch": 2.7972956658215624,
      "grad_norm": 2.9972870349884033,
      "learning_rate": 3.3940765283445796e-06,
      "loss": 1.1017,
      "step": 23170
    },
    {
      "epoch": 2.798502957865508,
      "grad_norm": 8.017427444458008,
      "learning_rate": 3.373873691866338e-06,
      "loss": 0.4162,
      "step": 23180
    },
    {
      "epoch": 2.7997102499094533,
      "grad_norm": 13.969335556030273,
      "learning_rate": 3.353670855388097e-06,
      "loss": 0.8823,
      "step": 23190
    },
    {
      "epoch": 2.8009175419533987,
      "grad_norm": 8.585166931152344,
      "learning_rate": 3.333468018909855e-06,
      "loss": 0.9161,
      "step": 23200
    },
    {
      "epoch": 2.802124833997344,
      "grad_norm": 31.29201316833496,
      "learning_rate": 3.3132651824316135e-06,
      "loss": 0.8761,
      "step": 23210
    },
    {
      "epoch": 2.8033321260412896,
      "grad_norm": 33.66498565673828,
      "learning_rate": 3.2930623459533715e-06,
      "loss": 0.6675,
      "step": 23220
    },
    {
      "epoch": 2.804539418085235,
      "grad_norm": 13.128753662109375,
      "learning_rate": 3.2728595094751304e-06,
      "loss": 1.0693,
      "step": 23230
    },
    {
      "epoch": 2.8057467101291804,
      "grad_norm": 2.004549741744995,
      "learning_rate": 3.2526566729968893e-06,
      "loss": 0.8379,
      "step": 23240
    },
    {
      "epoch": 2.806954002173126,
      "grad_norm": 36.93800735473633,
      "learning_rate": 3.2324538365186474e-06,
      "loss": 0.7774,
      "step": 23250
    },
    {
      "epoch": 2.8081612942170713,
      "grad_norm": 0.6580237746238708,
      "learning_rate": 3.212251000040406e-06,
      "loss": 0.9215,
      "step": 23260
    },
    {
      "epoch": 2.8093685862610167,
      "grad_norm": 26.313989639282227,
      "learning_rate": 3.192048163562164e-06,
      "loss": 0.9073,
      "step": 23270
    },
    {
      "epoch": 2.810575878304962,
      "grad_norm": 3.968627452850342,
      "learning_rate": 3.171845327083923e-06,
      "loss": 0.4221,
      "step": 23280
    },
    {
      "epoch": 2.8117831703489076,
      "grad_norm": 0.9853203296661377,
      "learning_rate": 3.151642490605681e-06,
      "loss": 0.7436,
      "step": 23290
    },
    {
      "epoch": 2.812990462392853,
      "grad_norm": 48.16170120239258,
      "learning_rate": 3.1314396541274398e-06,
      "loss": 0.763,
      "step": 23300
    },
    {
      "epoch": 2.8141977544367984,
      "grad_norm": 21.745222091674805,
      "learning_rate": 3.111236817649198e-06,
      "loss": 0.9709,
      "step": 23310
    },
    {
      "epoch": 2.815405046480744,
      "grad_norm": 39.600494384765625,
      "learning_rate": 3.0910339811709563e-06,
      "loss": 1.0824,
      "step": 23320
    },
    {
      "epoch": 2.8166123385246893,
      "grad_norm": 16.343358993530273,
      "learning_rate": 3.070831144692715e-06,
      "loss": 0.8031,
      "step": 23330
    },
    {
      "epoch": 2.8178196305686347,
      "grad_norm": 13.246718406677246,
      "learning_rate": 3.0506283082144737e-06,
      "loss": 0.5088,
      "step": 23340
    },
    {
      "epoch": 2.81902692261258,
      "grad_norm": 10.059104919433594,
      "learning_rate": 3.030425471736232e-06,
      "loss": 0.7274,
      "step": 23350
    },
    {
      "epoch": 2.8202342146565256,
      "grad_norm": 25.990707397460938,
      "learning_rate": 3.01022263525799e-06,
      "loss": 0.9511,
      "step": 23360
    },
    {
      "epoch": 2.821441506700471,
      "grad_norm": 26.931697845458984,
      "learning_rate": 2.9900197987797487e-06,
      "loss": 0.6364,
      "step": 23370
    },
    {
      "epoch": 2.8226487987444164,
      "grad_norm": 0.7795518040657043,
      "learning_rate": 2.969816962301507e-06,
      "loss": 0.5325,
      "step": 23380
    },
    {
      "epoch": 2.823856090788362,
      "grad_norm": 8.241957664489746,
      "learning_rate": 2.949614125823266e-06,
      "loss": 0.5118,
      "step": 23390
    },
    {
      "epoch": 2.8250633828323073,
      "grad_norm": 6.702354431152344,
      "learning_rate": 2.929411289345024e-06,
      "loss": 0.8339,
      "step": 23400
    },
    {
      "epoch": 2.8262706748762527,
      "grad_norm": 6.818068504333496,
      "learning_rate": 2.9092084528667826e-06,
      "loss": 0.9767,
      "step": 23410
    },
    {
      "epoch": 2.827477966920198,
      "grad_norm": 21.902935028076172,
      "learning_rate": 2.889005616388541e-06,
      "loss": 1.0451,
      "step": 23420
    },
    {
      "epoch": 2.8286852589641436,
      "grad_norm": 3.3882265090942383,
      "learning_rate": 2.8688027799102995e-06,
      "loss": 1.1567,
      "step": 23430
    },
    {
      "epoch": 2.829892551008089,
      "grad_norm": 20.919288635253906,
      "learning_rate": 2.848599943432058e-06,
      "loss": 0.6758,
      "step": 23440
    },
    {
      "epoch": 2.8310998430520344,
      "grad_norm": 49.04338455200195,
      "learning_rate": 2.8283971069538165e-06,
      "loss": 1.0175,
      "step": 23450
    },
    {
      "epoch": 2.83230713509598,
      "grad_norm": 2.381592035293579,
      "learning_rate": 2.808194270475575e-06,
      "loss": 0.8772,
      "step": 23460
    },
    {
      "epoch": 2.8335144271399253,
      "grad_norm": 20.163410186767578,
      "learning_rate": 2.7879914339973334e-06,
      "loss": 0.7478,
      "step": 23470
    },
    {
      "epoch": 2.8347217191838707,
      "grad_norm": 38.17195510864258,
      "learning_rate": 2.767788597519092e-06,
      "loss": 1.1389,
      "step": 23480
    },
    {
      "epoch": 2.835929011227816,
      "grad_norm": 3.0218889713287354,
      "learning_rate": 2.7475857610408504e-06,
      "loss": 0.7542,
      "step": 23490
    },
    {
      "epoch": 2.8371363032717616,
      "grad_norm": 13.981608390808105,
      "learning_rate": 2.7273829245626084e-06,
      "loss": 0.8801,
      "step": 23500
    },
    {
      "epoch": 2.838343595315707,
      "grad_norm": 23.690946578979492,
      "learning_rate": 2.707180088084367e-06,
      "loss": 0.6144,
      "step": 23510
    },
    {
      "epoch": 2.8395508873596524,
      "grad_norm": 31.39395523071289,
      "learning_rate": 2.686977251606126e-06,
      "loss": 0.698,
      "step": 23520
    },
    {
      "epoch": 2.840758179403598,
      "grad_norm": 4.945213317871094,
      "learning_rate": 2.6667744151278843e-06,
      "loss": 0.7814,
      "step": 23530
    },
    {
      "epoch": 2.8419654714475433,
      "grad_norm": 1.4600608348846436,
      "learning_rate": 2.6465715786496427e-06,
      "loss": 0.96,
      "step": 23540
    },
    {
      "epoch": 2.8431727634914887,
      "grad_norm": 1.9462136030197144,
      "learning_rate": 2.626368742171401e-06,
      "loss": 0.6431,
      "step": 23550
    },
    {
      "epoch": 2.844380055535434,
      "grad_norm": 20.473087310791016,
      "learning_rate": 2.6061659056931593e-06,
      "loss": 0.4897,
      "step": 23560
    },
    {
      "epoch": 2.8455873475793796,
      "grad_norm": 21.35914421081543,
      "learning_rate": 2.5859630692149177e-06,
      "loss": 0.6381,
      "step": 23570
    },
    {
      "epoch": 2.846794639623325,
      "grad_norm": 8.50961971282959,
      "learning_rate": 2.5657602327366766e-06,
      "loss": 0.9685,
      "step": 23580
    },
    {
      "epoch": 2.8480019316672704,
      "grad_norm": 20.052061080932617,
      "learning_rate": 2.5455573962584347e-06,
      "loss": 0.6682,
      "step": 23590
    },
    {
      "epoch": 2.849209223711216,
      "grad_norm": 27.84174346923828,
      "learning_rate": 2.525354559780193e-06,
      "loss": 1.1276,
      "step": 23600
    },
    {
      "epoch": 2.8504165157551613,
      "grad_norm": 20.635242462158203,
      "learning_rate": 2.5051517233019516e-06,
      "loss": 0.9109,
      "step": 23610
    },
    {
      "epoch": 2.8516238077991067,
      "grad_norm": 33.05842971801758,
      "learning_rate": 2.48494888682371e-06,
      "loss": 0.5883,
      "step": 23620
    },
    {
      "epoch": 2.852831099843052,
      "grad_norm": 11.741369247436523,
      "learning_rate": 2.4647460503454686e-06,
      "loss": 0.842,
      "step": 23630
    },
    {
      "epoch": 2.8540383918869976,
      "grad_norm": 14.876362800598145,
      "learning_rate": 2.444543213867227e-06,
      "loss": 0.9641,
      "step": 23640
    },
    {
      "epoch": 2.855245683930943,
      "grad_norm": 7.578466892242432,
      "learning_rate": 2.4243403773889855e-06,
      "loss": 1.4477,
      "step": 23650
    },
    {
      "epoch": 2.8564529759748885,
      "grad_norm": 20.323335647583008,
      "learning_rate": 2.404137540910744e-06,
      "loss": 0.4256,
      "step": 23660
    },
    {
      "epoch": 2.857660268018834,
      "grad_norm": 1.2333333492279053,
      "learning_rate": 2.3839347044325025e-06,
      "loss": 1.0358,
      "step": 23670
    },
    {
      "epoch": 2.8588675600627793,
      "grad_norm": 9.992804527282715,
      "learning_rate": 2.363731867954261e-06,
      "loss": 0.8056,
      "step": 23680
    },
    {
      "epoch": 2.8600748521067247,
      "grad_norm": 9.810280799865723,
      "learning_rate": 2.343529031476019e-06,
      "loss": 1.0339,
      "step": 23690
    },
    {
      "epoch": 2.86128214415067,
      "grad_norm": 28.12143325805664,
      "learning_rate": 2.323326194997778e-06,
      "loss": 1.3131,
      "step": 23700
    },
    {
      "epoch": 2.8624894361946156,
      "grad_norm": 38.31333923339844,
      "learning_rate": 2.3031233585195364e-06,
      "loss": 0.9485,
      "step": 23710
    },
    {
      "epoch": 2.863696728238561,
      "grad_norm": 29.09765625,
      "learning_rate": 2.282920522041295e-06,
      "loss": 1.0659,
      "step": 23720
    },
    {
      "epoch": 2.8649040202825065,
      "grad_norm": 28.854225158691406,
      "learning_rate": 2.2627176855630533e-06,
      "loss": 0.7775,
      "step": 23730
    },
    {
      "epoch": 2.866111312326452,
      "grad_norm": 14.76525592803955,
      "learning_rate": 2.2425148490848114e-06,
      "loss": 0.6289,
      "step": 23740
    },
    {
      "epoch": 2.8673186043703973,
      "grad_norm": 5.242019176483154,
      "learning_rate": 2.22231201260657e-06,
      "loss": 0.8181,
      "step": 23750
    },
    {
      "epoch": 2.8685258964143427,
      "grad_norm": 16.22235107421875,
      "learning_rate": 2.2021091761283283e-06,
      "loss": 0.9248,
      "step": 23760
    },
    {
      "epoch": 2.869733188458288,
      "grad_norm": 28.99667739868164,
      "learning_rate": 2.1819063396500872e-06,
      "loss": 0.4644,
      "step": 23770
    },
    {
      "epoch": 2.8709404805022336,
      "grad_norm": 30.718185424804688,
      "learning_rate": 2.1617035031718453e-06,
      "loss": 1.0037,
      "step": 23780
    },
    {
      "epoch": 2.872147772546179,
      "grad_norm": 11.010306358337402,
      "learning_rate": 2.1415006666936038e-06,
      "loss": 0.7039,
      "step": 23790
    },
    {
      "epoch": 2.8733550645901245,
      "grad_norm": 12.660778045654297,
      "learning_rate": 2.1212978302153622e-06,
      "loss": 0.5676,
      "step": 23800
    },
    {
      "epoch": 2.87456235663407,
      "grad_norm": 12.665000915527344,
      "learning_rate": 2.1010949937371207e-06,
      "loss": 0.8134,
      "step": 23810
    },
    {
      "epoch": 2.8757696486780153,
      "grad_norm": 7.3117499351501465,
      "learning_rate": 2.080892157258879e-06,
      "loss": 0.6334,
      "step": 23820
    },
    {
      "epoch": 2.8769769407219608,
      "grad_norm": 15.107014656066895,
      "learning_rate": 2.0606893207806377e-06,
      "loss": 0.9235,
      "step": 23830
    },
    {
      "epoch": 2.878184232765906,
      "grad_norm": 33.583805084228516,
      "learning_rate": 2.040486484302396e-06,
      "loss": 0.3769,
      "step": 23840
    },
    {
      "epoch": 2.8793915248098516,
      "grad_norm": 52.09463882446289,
      "learning_rate": 2.0202836478241546e-06,
      "loss": 0.9421,
      "step": 23850
    },
    {
      "epoch": 2.880598816853797,
      "grad_norm": 14.264322280883789,
      "learning_rate": 2.000080811345913e-06,
      "loss": 0.9542,
      "step": 23860
    },
    {
      "epoch": 2.8818061088977425,
      "grad_norm": 1.046788215637207,
      "learning_rate": 1.9798779748676716e-06,
      "loss": 0.8299,
      "step": 23870
    },
    {
      "epoch": 2.883013400941688,
      "grad_norm": 32.42605972290039,
      "learning_rate": 1.9596751383894296e-06,
      "loss": 0.7243,
      "step": 23880
    },
    {
      "epoch": 2.8842206929856333,
      "grad_norm": 20.745878219604492,
      "learning_rate": 1.9394723019111885e-06,
      "loss": 0.3755,
      "step": 23890
    },
    {
      "epoch": 2.8854279850295788,
      "grad_norm": 3.3034045696258545,
      "learning_rate": 1.919269465432947e-06,
      "loss": 0.5152,
      "step": 23900
    },
    {
      "epoch": 2.886635277073524,
      "grad_norm": 14.849128723144531,
      "learning_rate": 1.8990666289547055e-06,
      "loss": 0.8012,
      "step": 23910
    },
    {
      "epoch": 2.8878425691174696,
      "grad_norm": 48.19542694091797,
      "learning_rate": 1.8788637924764637e-06,
      "loss": 1.1743,
      "step": 23920
    },
    {
      "epoch": 2.889049861161415,
      "grad_norm": 21.161685943603516,
      "learning_rate": 1.8586609559982222e-06,
      "loss": 1.1177,
      "step": 23930
    },
    {
      "epoch": 2.8902571532053605,
      "grad_norm": 12.265286445617676,
      "learning_rate": 1.8384581195199805e-06,
      "loss": 0.5745,
      "step": 23940
    },
    {
      "epoch": 2.891464445249306,
      "grad_norm": 28.455659866333008,
      "learning_rate": 1.8182552830417394e-06,
      "loss": 0.7314,
      "step": 23950
    },
    {
      "epoch": 2.8926717372932513,
      "grad_norm": 19.606163024902344,
      "learning_rate": 1.7980524465634976e-06,
      "loss": 0.9118,
      "step": 23960
    },
    {
      "epoch": 2.8938790293371968,
      "grad_norm": 1.751171350479126,
      "learning_rate": 1.777849610085256e-06,
      "loss": 0.4256,
      "step": 23970
    },
    {
      "epoch": 2.895086321381142,
      "grad_norm": 41.09761428833008,
      "learning_rate": 1.7576467736070146e-06,
      "loss": 0.4949,
      "step": 23980
    },
    {
      "epoch": 2.8962936134250876,
      "grad_norm": 1.6189395189285278,
      "learning_rate": 1.7374439371287728e-06,
      "loss": 0.4477,
      "step": 23990
    },
    {
      "epoch": 2.897500905469033,
      "grad_norm": 1.2533290386199951,
      "learning_rate": 1.7172411006505313e-06,
      "loss": 0.591,
      "step": 24000
    },
    {
      "epoch": 2.8987081975129785,
      "grad_norm": 32.356964111328125,
      "learning_rate": 1.6970382641722898e-06,
      "loss": 0.8302,
      "step": 24010
    },
    {
      "epoch": 2.899915489556924,
      "grad_norm": 15.879366874694824,
      "learning_rate": 1.6768354276940485e-06,
      "loss": 0.9085,
      "step": 24020
    },
    {
      "epoch": 2.9011227816008693,
      "grad_norm": 9.08348560333252,
      "learning_rate": 1.6566325912158067e-06,
      "loss": 0.7531,
      "step": 24030
    },
    {
      "epoch": 2.9023300736448148,
      "grad_norm": 4.307328701019287,
      "learning_rate": 1.6364297547375652e-06,
      "loss": 0.8604,
      "step": 24040
    },
    {
      "epoch": 2.90353736568876,
      "grad_norm": 20.162485122680664,
      "learning_rate": 1.6162269182593237e-06,
      "loss": 0.7108,
      "step": 24050
    },
    {
      "epoch": 2.9047446577327056,
      "grad_norm": 15.785592079162598,
      "learning_rate": 1.596024081781082e-06,
      "loss": 0.7856,
      "step": 24060
    },
    {
      "epoch": 2.905951949776651,
      "grad_norm": 4.784730434417725,
      "learning_rate": 1.5758212453028404e-06,
      "loss": 0.8499,
      "step": 24070
    },
    {
      "epoch": 2.9071592418205965,
      "grad_norm": 11.784982681274414,
      "learning_rate": 1.555618408824599e-06,
      "loss": 0.9168,
      "step": 24080
    },
    {
      "epoch": 2.908366533864542,
      "grad_norm": 9.113356590270996,
      "learning_rate": 1.5354155723463576e-06,
      "loss": 1.0668,
      "step": 24090
    },
    {
      "epoch": 2.9095738259084873,
      "grad_norm": 17.0065975189209,
      "learning_rate": 1.515212735868116e-06,
      "loss": 0.8651,
      "step": 24100
    },
    {
      "epoch": 2.9107811179524328,
      "grad_norm": 3.3614354133605957,
      "learning_rate": 1.4950098993898743e-06,
      "loss": 0.6333,
      "step": 24110
    },
    {
      "epoch": 2.911988409996378,
      "grad_norm": 10.329329490661621,
      "learning_rate": 1.474807062911633e-06,
      "loss": 1.1333,
      "step": 24120
    },
    {
      "epoch": 2.9131957020403236,
      "grad_norm": 12.873364448547363,
      "learning_rate": 1.4546042264333913e-06,
      "loss": 0.849,
      "step": 24130
    },
    {
      "epoch": 2.914402994084269,
      "grad_norm": 9.731733322143555,
      "learning_rate": 1.4344013899551498e-06,
      "loss": 0.9708,
      "step": 24140
    },
    {
      "epoch": 2.9156102861282145,
      "grad_norm": 14.757841110229492,
      "learning_rate": 1.4141985534769082e-06,
      "loss": 0.8757,
      "step": 24150
    },
    {
      "epoch": 2.91681757817216,
      "grad_norm": 5.682179927825928,
      "learning_rate": 1.3939957169986667e-06,
      "loss": 0.4526,
      "step": 24160
    },
    {
      "epoch": 2.9180248702161053,
      "grad_norm": 18.788227081298828,
      "learning_rate": 1.3737928805204252e-06,
      "loss": 1.0879,
      "step": 24170
    },
    {
      "epoch": 2.919232162260051,
      "grad_norm": 24.06669807434082,
      "learning_rate": 1.3535900440421834e-06,
      "loss": 0.7252,
      "step": 24180
    },
    {
      "epoch": 2.920439454303996,
      "grad_norm": 27.022029876708984,
      "learning_rate": 1.3333872075639421e-06,
      "loss": 1.1794,
      "step": 24190
    },
    {
      "epoch": 2.9216467463479416,
      "grad_norm": 48.57954788208008,
      "learning_rate": 1.3131843710857004e-06,
      "loss": 0.7385,
      "step": 24200
    },
    {
      "epoch": 2.922854038391887,
      "grad_norm": 23.607837677001953,
      "learning_rate": 1.2929815346074589e-06,
      "loss": 0.8418,
      "step": 24210
    },
    {
      "epoch": 2.9240613304358325,
      "grad_norm": 20.550186157226562,
      "learning_rate": 1.2727786981292173e-06,
      "loss": 0.7148,
      "step": 24220
    },
    {
      "epoch": 2.925268622479778,
      "grad_norm": 28.820051193237305,
      "learning_rate": 1.2525758616509758e-06,
      "loss": 0.4264,
      "step": 24230
    },
    {
      "epoch": 2.9264759145237234,
      "grad_norm": 13.530130386352539,
      "learning_rate": 1.2323730251727343e-06,
      "loss": 0.6459,
      "step": 24240
    },
    {
      "epoch": 2.927683206567669,
      "grad_norm": 0.9471157789230347,
      "learning_rate": 1.2121701886944928e-06,
      "loss": 0.9581,
      "step": 24250
    },
    {
      "epoch": 2.928890498611614,
      "grad_norm": 11.320042610168457,
      "learning_rate": 1.1919673522162512e-06,
      "loss": 0.8247,
      "step": 24260
    },
    {
      "epoch": 2.9300977906555596,
      "grad_norm": 18.645793914794922,
      "learning_rate": 1.1717645157380095e-06,
      "loss": 0.5091,
      "step": 24270
    },
    {
      "epoch": 2.931305082699505,
      "grad_norm": 20.643190383911133,
      "learning_rate": 1.1515616792597682e-06,
      "loss": 1.0064,
      "step": 24280
    },
    {
      "epoch": 2.9325123747434505,
      "grad_norm": 29.072126388549805,
      "learning_rate": 1.1313588427815267e-06,
      "loss": 0.9801,
      "step": 24290
    },
    {
      "epoch": 2.933719666787396,
      "grad_norm": 9.969322204589844,
      "learning_rate": 1.111156006303285e-06,
      "loss": 0.7085,
      "step": 24300
    },
    {
      "epoch": 2.9349269588313414,
      "grad_norm": 41.35538864135742,
      "learning_rate": 1.0909531698250436e-06,
      "loss": 0.6847,
      "step": 24310
    },
    {
      "epoch": 2.936134250875287,
      "grad_norm": 18.740005493164062,
      "learning_rate": 1.0707503333468019e-06,
      "loss": 0.4901,
      "step": 24320
    },
    {
      "epoch": 2.937341542919232,
      "grad_norm": 3.393441915512085,
      "learning_rate": 1.0505474968685604e-06,
      "loss": 0.6946,
      "step": 24330
    },
    {
      "epoch": 2.9385488349631776,
      "grad_norm": 37.60076141357422,
      "learning_rate": 1.0303446603903188e-06,
      "loss": 0.9636,
      "step": 24340
    },
    {
      "epoch": 2.939756127007123,
      "grad_norm": 1.5639406442642212,
      "learning_rate": 1.0101418239120773e-06,
      "loss": 0.5411,
      "step": 24350
    },
    {
      "epoch": 2.9409634190510685,
      "grad_norm": 40.3783073425293,
      "learning_rate": 9.899389874338358e-07,
      "loss": 0.5571,
      "step": 24360
    },
    {
      "epoch": 2.942170711095014,
      "grad_norm": 15.966601371765137,
      "learning_rate": 9.697361509555943e-07,
      "loss": 0.5657,
      "step": 24370
    },
    {
      "epoch": 2.9433780031389594,
      "grad_norm": 28.036901473999023,
      "learning_rate": 9.495333144773527e-07,
      "loss": 0.597,
      "step": 24380
    },
    {
      "epoch": 2.944585295182905,
      "grad_norm": 45.4454231262207,
      "learning_rate": 9.293304779991111e-07,
      "loss": 0.7773,
      "step": 24390
    },
    {
      "epoch": 2.9457925872268502,
      "grad_norm": 2.66601300239563,
      "learning_rate": 9.091276415208697e-07,
      "loss": 0.7339,
      "step": 24400
    },
    {
      "epoch": 2.9469998792707957,
      "grad_norm": 25.57579803466797,
      "learning_rate": 8.88924805042628e-07,
      "loss": 0.8865,
      "step": 24410
    },
    {
      "epoch": 2.948207171314741,
      "grad_norm": 10.002870559692383,
      "learning_rate": 8.687219685643864e-07,
      "loss": 1.0785,
      "step": 24420
    },
    {
      "epoch": 2.9494144633586865,
      "grad_norm": 10.769933700561523,
      "learning_rate": 8.485191320861449e-07,
      "loss": 0.9873,
      "step": 24430
    },
    {
      "epoch": 2.950621755402632,
      "grad_norm": 14.110132217407227,
      "learning_rate": 8.283162956079034e-07,
      "loss": 0.5305,
      "step": 24440
    },
    {
      "epoch": 2.9518290474465774,
      "grad_norm": 10.063557624816895,
      "learning_rate": 8.081134591296618e-07,
      "loss": 0.8276,
      "step": 24450
    },
    {
      "epoch": 2.953036339490523,
      "grad_norm": 14.815621376037598,
      "learning_rate": 7.879106226514202e-07,
      "loss": 0.6987,
      "step": 24460
    },
    {
      "epoch": 2.9542436315344682,
      "grad_norm": 19.001821517944336,
      "learning_rate": 7.677077861731788e-07,
      "loss": 0.635,
      "step": 24470
    },
    {
      "epoch": 2.9554509235784137,
      "grad_norm": 10.607856750488281,
      "learning_rate": 7.475049496949372e-07,
      "loss": 0.5503,
      "step": 24480
    },
    {
      "epoch": 2.956658215622359,
      "grad_norm": 27.928333282470703,
      "learning_rate": 7.273021132166956e-07,
      "loss": 1.0071,
      "step": 24490
    },
    {
      "epoch": 2.9578655076663045,
      "grad_norm": 12.473018646240234,
      "learning_rate": 7.070992767384541e-07,
      "loss": 0.9008,
      "step": 24500
    },
    {
      "epoch": 2.95907279971025,
      "grad_norm": 1.9952353239059448,
      "learning_rate": 6.868964402602126e-07,
      "loss": 0.4408,
      "step": 24510
    },
    {
      "epoch": 2.9602800917541954,
      "grad_norm": 24.97207260131836,
      "learning_rate": 6.666936037819711e-07,
      "loss": 0.9502,
      "step": 24520
    },
    {
      "epoch": 2.961487383798141,
      "grad_norm": 9.32828140258789,
      "learning_rate": 6.464907673037294e-07,
      "loss": 0.5451,
      "step": 24530
    },
    {
      "epoch": 2.9626946758420862,
      "grad_norm": 5.814170837402344,
      "learning_rate": 6.262879308254879e-07,
      "loss": 0.9609,
      "step": 24540
    },
    {
      "epoch": 2.9639019678860317,
      "grad_norm": 17.481924057006836,
      "learning_rate": 6.060850943472464e-07,
      "loss": 0.5558,
      "step": 24550
    },
    {
      "epoch": 2.965109259929977,
      "grad_norm": 21.80217742919922,
      "learning_rate": 5.858822578690048e-07,
      "loss": 0.869,
      "step": 24560
    },
    {
      "epoch": 2.9663165519739225,
      "grad_norm": 12.477341651916504,
      "learning_rate": 5.656794213907633e-07,
      "loss": 0.7415,
      "step": 24570
    },
    {
      "epoch": 2.967523844017868,
      "grad_norm": 14.216306686401367,
      "learning_rate": 5.454765849125218e-07,
      "loss": 1.2044,
      "step": 24580
    },
    {
      "epoch": 2.9687311360618134,
      "grad_norm": 3.0837130546569824,
      "learning_rate": 5.252737484342802e-07,
      "loss": 0.3927,
      "step": 24590
    },
    {
      "epoch": 2.969938428105759,
      "grad_norm": 20.84140968322754,
      "learning_rate": 5.050709119560387e-07,
      "loss": 0.5982,
      "step": 24600
    },
    {
      "epoch": 2.9711457201497042,
      "grad_norm": 6.00036096572876,
      "learning_rate": 4.848680754777971e-07,
      "loss": 0.5652,
      "step": 24610
    },
    {
      "epoch": 2.9723530121936497,
      "grad_norm": 20.01803970336914,
      "learning_rate": 4.6466523899955555e-07,
      "loss": 1.0227,
      "step": 24620
    },
    {
      "epoch": 2.973560304237595,
      "grad_norm": 29.46135139465332,
      "learning_rate": 4.44462402521314e-07,
      "loss": 0.9088,
      "step": 24630
    },
    {
      "epoch": 2.9747675962815405,
      "grad_norm": 20.885726928710938,
      "learning_rate": 4.2425956604307245e-07,
      "loss": 0.6193,
      "step": 24640
    },
    {
      "epoch": 2.975974888325486,
      "grad_norm": 13.152467727661133,
      "learning_rate": 4.040567295648309e-07,
      "loss": 0.5242,
      "step": 24650
    },
    {
      "epoch": 2.9771821803694314,
      "grad_norm": 11.827566146850586,
      "learning_rate": 3.838538930865894e-07,
      "loss": 0.807,
      "step": 24660
    },
    {
      "epoch": 2.978389472413377,
      "grad_norm": 0.43027180433273315,
      "learning_rate": 3.636510566083478e-07,
      "loss": 1.1906,
      "step": 24670
    },
    {
      "epoch": 2.9795967644573222,
      "grad_norm": 18.257570266723633,
      "learning_rate": 3.434482201301063e-07,
      "loss": 0.6945,
      "step": 24680
    },
    {
      "epoch": 2.9808040565012677,
      "grad_norm": 1.7041082382202148,
      "learning_rate": 3.232453836518647e-07,
      "loss": 0.3921,
      "step": 24690
    },
    {
      "epoch": 2.982011348545213,
      "grad_norm": 10.323920249938965,
      "learning_rate": 3.030425471736232e-07,
      "loss": 0.8763,
      "step": 24700
    },
    {
      "epoch": 2.9832186405891585,
      "grad_norm": 17.336435317993164,
      "learning_rate": 2.8283971069538167e-07,
      "loss": 0.7154,
      "step": 24710
    },
    {
      "epoch": 2.984425932633104,
      "grad_norm": 3.589960813522339,
      "learning_rate": 2.626368742171401e-07,
      "loss": 0.7032,
      "step": 24720
    },
    {
      "epoch": 2.9856332246770494,
      "grad_norm": 39.42912292480469,
      "learning_rate": 2.4243403773889856e-07,
      "loss": 0.9092,
      "step": 24730
    },
    {
      "epoch": 2.986840516720995,
      "grad_norm": 24.309648513793945,
      "learning_rate": 2.22231201260657e-07,
      "loss": 1.2544,
      "step": 24740
    },
    {
      "epoch": 2.9880478087649402,
      "grad_norm": 5.731733322143555,
      "learning_rate": 2.0202836478241546e-07,
      "loss": 0.4304,
      "step": 24750
    },
    {
      "epoch": 2.9892551008088857,
      "grad_norm": 9.338282585144043,
      "learning_rate": 1.818255283041739e-07,
      "loss": 0.9187,
      "step": 24760
    },
    {
      "epoch": 2.990462392852831,
      "grad_norm": 34.498226165771484,
      "learning_rate": 1.6162269182593236e-07,
      "loss": 0.9183,
      "step": 24770
    },
    {
      "epoch": 2.9916696848967765,
      "grad_norm": 7.010932445526123,
      "learning_rate": 1.4141985534769083e-07,
      "loss": 0.7123,
      "step": 24780
    },
    {
      "epoch": 2.992876976940722,
      "grad_norm": 31.852649688720703,
      "learning_rate": 1.2121701886944928e-07,
      "loss": 1.0389,
      "step": 24790
    },
    {
      "epoch": 2.9940842689846674,
      "grad_norm": 9.63921070098877,
      "learning_rate": 1.0101418239120773e-07,
      "loss": 0.7995,
      "step": 24800
    },
    {
      "epoch": 2.995291561028613,
      "grad_norm": 21.90848731994629,
      "learning_rate": 8.081134591296618e-08,
      "loss": 0.5432,
      "step": 24810
    },
    {
      "epoch": 2.9964988530725583,
      "grad_norm": 17.434070587158203,
      "learning_rate": 6.060850943472464e-08,
      "loss": 1.0041,
      "step": 24820
    },
    {
      "epoch": 2.9977061451165037,
      "grad_norm": 14.60524845123291,
      "learning_rate": 4.040567295648309e-08,
      "loss": 1.051,
      "step": 24830
    },
    {
      "epoch": 2.998913437160449,
      "grad_norm": 30.907615661621094,
      "learning_rate": 2.0202836478241545e-08,
      "loss": 0.7545,
      "step": 24840
    }
  ],
  "logging_steps": 10,
  "max_steps": 24849,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3388987205708244.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}
